[libcamera-devel,v8,7/7] py: Add cam.py
diff mbox series

Message ID 20220506145414.99039-8-tomi.valkeinen@ideasonboard.com
State Accepted
Headers show
Series
  • Python bindings
Related show

Commit Message

Tomi Valkeinen May 6, 2022, 2:54 p.m. UTC
Add cam.py, which mimics the 'cam' tool. Four rendering backends are
added:

* null - Do nothing
* kms - Use KMS with dmabufs
* qt - SW render on a Qt window
* qtgl - OpenGL render on a Qt window

All the renderers handle only a few pixel formats, and especially the GL
renderer is just a prototype.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
---
 src/py/cam/cam.py        | 484 +++++++++++++++++++++++++++++++++++++++
 src/py/cam/cam_kms.py    | 183 +++++++++++++++
 src/py/cam/cam_null.py   |  47 ++++
 src/py/cam/cam_qt.py     | 354 ++++++++++++++++++++++++++++
 src/py/cam/cam_qtgl.py   | 385 +++++++++++++++++++++++++++++++
 src/py/cam/gl_helpers.py |  74 ++++++
 6 files changed, 1527 insertions(+)
 create mode 100755 src/py/cam/cam.py
 create mode 100644 src/py/cam/cam_kms.py
 create mode 100644 src/py/cam/cam_null.py
 create mode 100644 src/py/cam/cam_qt.py
 create mode 100644 src/py/cam/cam_qtgl.py
 create mode 100644 src/py/cam/gl_helpers.py

Comments

Laurent Pinchart May 6, 2022, 7:49 p.m. UTC | #1
Hi Tomi,

Thank you for the patch.

On Fri, May 06, 2022 at 05:54:14PM +0300, Tomi Valkeinen wrote:
> Add cam.py, which mimics the 'cam' tool. Four rendering backends are
> added:
> 
> * null - Do nothing
> * kms - Use KMS with dmabufs
> * qt - SW render on a Qt window
> * qtgl - OpenGL render on a Qt window
> 
> All the renderers handle only a few pixel formats, and especially the GL
> renderer is just a prototype.

I don't review this in much details as you mentioned it was a proof of
concept. We can improve things later, and high-level comments below
don't all need to (but can :-)) be addressed in a new version.

> Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
> ---
>  src/py/cam/cam.py        | 484 +++++++++++++++++++++++++++++++++++++++
>  src/py/cam/cam_kms.py    | 183 +++++++++++++++
>  src/py/cam/cam_null.py   |  47 ++++
>  src/py/cam/cam_qt.py     | 354 ++++++++++++++++++++++++++++
>  src/py/cam/cam_qtgl.py   | 385 +++++++++++++++++++++++++++++++
>  src/py/cam/gl_helpers.py |  74 ++++++
>  6 files changed, 1527 insertions(+)
>  create mode 100755 src/py/cam/cam.py
>  create mode 100644 src/py/cam/cam_kms.py
>  create mode 100644 src/py/cam/cam_null.py
>  create mode 100644 src/py/cam/cam_qt.py
>  create mode 100644 src/py/cam/cam_qtgl.py
>  create mode 100644 src/py/cam/gl_helpers.py
> 
> diff --git a/src/py/cam/cam.py b/src/py/cam/cam.py
> new file mode 100755
> index 00000000..5ea02850
> --- /dev/null
> +++ b/src/py/cam/cam.py
> @@ -0,0 +1,484 @@
> +#!/usr/bin/env python3
> +
> +# SPDX-License-Identifier: GPL-2.0-or-later
> +# Copyright (C) 2021, Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
> +
> +import argparse
> +import binascii
> +import libcamera as libcam
> +import os
> +import sys
> +
> +
> +class CustomCameraAction(argparse.Action):

This doesn't seem to be used.

> +    def __call__(self, parser, namespace, values, option_string=None):
> +        print(self.dest, values)
> +
> +        if 'camera' not in namespace or namespace.camera is None:
> +            setattr(namespace, 'camera', [])
> +
> +        previous = namespace.camera
> +        previous.append((self.dest, values))
> +        setattr(namespace, 'camera', previous)
> +
> +
> +class CustomAction(argparse.Action):
> +    def __init__(self, option_strings, dest, **kwargs):
> +        super().__init__(option_strings, dest, default={}, **kwargs)
> +
> +    def __call__(self, parser, namespace, values, option_string=None):
> +        if len(namespace.camera) == 0:
> +            print(f'Option {option_string} requires a --camera context')
> +            sys.exit(-1)
> +
> +        if self.type == bool:
> +            values = True
> +
> +        current = namespace.camera[-1]
> +
> +        data = getattr(namespace, self.dest)
> +
> +        if self.nargs == '+':
> +            if current not in data:
> +                data[current] = []
> +
> +            data[current] += values
> +        else:
> +            data[current] = values
> +
> +
> +def do_cmd_list(cm):
> +    print('Available cameras:')
> +
> +    for idx, c in enumerate(cm.cameras):
> +        print(f'{idx + 1}: {c.id}')
> +
> +
> +def do_cmd_list_props(ctx):
> +    camera = ctx['camera']
> +
> +    print('Properties for', ctx['id'])
> +
> +    for name, prop in camera.properties.items():
> +        print('\t{}: {}'.format(name, prop))
> +
> +
> +def do_cmd_list_controls(ctx):
> +    camera = ctx['camera']
> +
> +    print('Controls for', ctx['id'])
> +
> +    for name, prop in camera.controls.items():
> +        print('\t{}: {}'.format(name, prop))
> +
> +
> +def do_cmd_info(ctx):
> +    camera = ctx['camera']
> +
> +    print('Stream info for', ctx['id'])
> +
> +    roles = [libcam.StreamRole.Viewfinder]
> +
> +    camconfig = camera.generate_configuration(roles)
> +    if camconfig is None:
> +        raise Exception('Generating config failed')
> +
> +    for i, stream_config in enumerate(camconfig):
> +        print('\t{}: {}'.format(i, stream_config.to_string()))
> +
> +        formats = stream_config.formats
> +        for fmt in formats.pixel_formats:
> +            print('\t * Pixelformat:', fmt, formats.range(fmt))
> +
> +            for size in formats.sizes(fmt):
> +                print('\t  -', size)
> +
> +
> +def acquire(ctx):
> +    camera = ctx['camera']
> +
> +    camera.acquire()
> +
> +
> +def release(ctx):
> +    camera = ctx['camera']
> +
> +    camera.release()
> +
> +
> +def parse_streams(ctx):
> +    streams = []
> +
> +    for stream_desc in ctx['opt-stream']:
> +        stream_opts = {'role': libcam.StreamRole.Viewfinder}
> +
> +        for stream_opt in stream_desc.split(','):
> +            if stream_opt == 0:
> +                continue
> +
> +            arr = stream_opt.split('=')
> +            if len(arr) != 2:
> +                print('Bad stream option', stream_opt)
> +                sys.exit(-1)
> +
> +            key = arr[0]
> +            value = arr[1]
> +
> +            if key in ['width', 'height']:
> +                value = int(value)
> +            elif key == 'role':
> +                rolemap = {
> +                    'still': libcam.StreamRole.StillCapture,
> +                    'raw': libcam.StreamRole.Raw,
> +                    'video': libcam.StreamRole.VideoRecording,
> +                    'viewfinder': libcam.StreamRole.Viewfinder,
> +                }
> +
> +                role = rolemap.get(value.lower(), None)
> +
> +                if role is None:
> +                    print('Bad stream role', value)
> +                    sys.exit(-1)
> +
> +                value = role
> +            elif key == 'pixelformat':
> +                pass
> +            else:
> +                print('Bad stream option key', key)
> +                sys.exit(-1)
> +
> +            stream_opts[key] = value
> +
> +        streams.append(stream_opts)
> +
> +    return streams
> +
> +
> +def configure(ctx):
> +    camera = ctx['camera']
> +
> +    streams = parse_streams(ctx)
> +
> +    roles = [opts['role'] for opts in streams]
> +
> +    camconfig = camera.generate_configuration(roles)
> +    if camconfig is None:
> +        raise Exception('Generating config failed')
> +
> +    for idx, stream_opts in enumerate(streams):
> +        stream_config = camconfig.at(idx)
> +
> +        if 'width' in stream_opts and 'height' in stream_opts:
> +            stream_config.size = (stream_opts['width'], stream_opts['height'])
> +
> +        if 'pixelformat' in stream_opts:
> +            stream_config.pixel_format = stream_opts['pixelformat']
> +
> +    stat = camconfig.validate()
> +
> +    if stat == libcam.CameraConfiguration.Status.Invalid:
> +        print('Camera configuration invalid')
> +        exit(-1)
> +    elif stat == libcam.CameraConfiguration.Status.Adjusted:
> +        if ctx['opt-strict-formats']:
> +            print('Adjusting camera configuration disallowed by --strict-formats argument')
> +            exit(-1)
> +
> +        print('Camera configuration adjusted')
> +
> +    r = camera.configure(camconfig)
> +    if r != 0:
> +        raise Exception('Configure failed')
> +
> +    ctx['stream-names'] = {}
> +    ctx['streams'] = []
> +
> +    for idx, stream_config in enumerate(camconfig):
> +        stream = stream_config.stream
> +        ctx['streams'].append(stream)
> +        ctx['stream-names'][stream] = 'stream' + str(idx)
> +        print('{}-{}: stream config {}'.format(ctx['id'], ctx['stream-names'][stream], stream.configuration.to_string()))
> +
> +
> +def alloc_buffers(ctx):
> +    camera = ctx['camera']
> +
> +    allocator = libcam.FrameBufferAllocator(camera)
> +
> +    for idx, stream in enumerate(ctx['streams']):
> +        ret = allocator.allocate(stream)
> +        if ret < 0:
> +            print('Cannot allocate buffers')
> +            exit(-1)
> +
> +        allocated = len(allocator.buffers(stream))
> +
> +        print('{}-{}: Allocated {} buffers'.format(ctx['id'], ctx['stream-names'][stream], allocated))
> +
> +    ctx['allocator'] = allocator
> +
> +
> +def create_requests(ctx):
> +    camera = ctx['camera']
> +
> +    ctx['requests'] = []
> +
> +    # Identify the stream with the least number of buffers
> +    num_bufs = min([len(ctx['allocator'].buffers(stream)) for stream in ctx['streams']])
> +
> +    requests = []
> +
> +    for buf_num in range(num_bufs):
> +        request = camera.create_request(ctx['idx'])
> +
> +        if request is None:
> +            print('Can not create request')
> +            exit(-1)
> +
> +        for stream in ctx['streams']:
> +            buffers = ctx['allocator'].buffers(stream)
> +            buffer = buffers[buf_num]
> +
> +            ret = request.add_buffer(stream, buffer)
> +            if ret < 0:
> +                print('Can not set buffer for request')
> +                exit(-1)
> +
> +        requests.append(request)
> +
> +    ctx['requests'] = requests
> +
> +
> +def start(ctx):
> +    camera = ctx['camera']
> +
> +    camera.start()
> +
> +
> +def stop(ctx):
> +    camera = ctx['camera']
> +
> +    camera.stop()
> +
> +
> +def queue_requests(ctx):
> +    camera = ctx['camera']
> +
> +    for request in ctx['requests']:
> +        camera.queue_request(request)
> +        ctx['reqs-queued'] += 1
> +
> +    del ctx['requests']

It would be nice to group most functions operating on a context in a
class, with members of the context dict turned into class member. It
would make the next version easier to review.

> +
> +
> +def capture_init(contexts):
> +    for ctx in contexts:
> +        acquire(ctx)
> +
> +    for ctx in contexts:
> +        configure(ctx)
> +
> +    for ctx in contexts:
> +        alloc_buffers(ctx)
> +
> +    for ctx in contexts:
> +        create_requests(ctx)
> +
> +
> +def capture_start(contexts):
> +    for ctx in contexts:
> +        start(ctx)
> +
> +    for ctx in contexts:
> +        queue_requests(ctx)
> +
> +
> +# Called from renderer when there is a libcamera event
> +def event_handler(state):
> +    cm = state['cm']
> +    contexts = state['contexts']
> +
> +    os.read(cm.efd, 8)
> +
> +    reqs = cm.get_ready_requests()
> +
> +    for req in reqs:
> +        ctx = next(ctx for ctx in contexts if ctx['idx'] == req.cookie)
> +        request_handler(state, ctx, req)
> +
> +    running = any(ctx['reqs-completed'] < ctx['opt-capture'] for ctx in contexts)
> +    return running
> +
> +
> +def request_handler(state, ctx, req):
> +    if req.status != libcam.Request.Status.Complete:
> +        raise Exception('{}: Request failed: {}'.format(ctx['id'], req.status))
> +
> +    buffers = req.buffers
> +
> +    # Compute the frame rate. The timestamp is arbitrarily retrieved from
> +    # the first buffer, as all buffers should have matching timestamps.
> +    ts = buffers[next(iter(buffers))].metadata.timestamp
> +    last = ctx.get('last', 0)
> +    fps = 1000000000.0 / (ts - last) if (last != 0 and (ts - last) != 0) else 0
> +    ctx['last'] = ts
> +    ctx['fps'] = fps
> +
> +    for stream, fb in buffers.items():
> +        stream_name = ctx['stream-names'][stream]
> +
> +        crcs = []
> +        if ctx['opt-crc']:
> +            with fb.mmap() as mfb:
> +                plane_crcs = [binascii.crc32(p) for p in mfb.planes]
> +                crcs.append(plane_crcs)
> +
> +        meta = fb.metadata
> +
> +        print('{:.6f} ({:.2f} fps) {}-{}: seq {}, bytes {}, CRCs {}'
> +              .format(ts / 1000000000, fps,
> +                      ctx['id'], stream_name,
> +                      meta.sequence, meta.bytesused,
> +                      crcs))
> +
> +        if ctx['opt-metadata']:
> +            reqmeta = req.metadata
> +            for ctrl, val in reqmeta.items():
> +                print(f'\t{ctrl} = {val}')
> +
> +        if ctx['opt-save-frames']:
> +            with fb.mmap() as mfb:
> +                filename = 'frame-{}-{}-{}.data'.format(ctx['id'], stream_name, ctx['reqs-completed'])
> +                with open(filename, 'wb') as f:
> +                    for p in mfb.planes:
> +                        f.write(p)
> +
> +    state['renderer'].request_handler(ctx, req)
> +
> +    ctx['reqs-completed'] += 1
> +
> +
> +# Called from renderer when it has finished with a request
> +def request_prcessed(ctx, req):
> +    camera = ctx['camera']
> +
> +    if ctx['reqs-queued'] < ctx['opt-capture']:
> +        req.reuse()
> +        camera.queue_request(req)
> +        ctx['reqs-queued'] += 1
> +
> +
> +def capture_deinit(contexts):
> +    for ctx in contexts:
> +        stop(ctx)
> +
> +    for ctx in contexts:
> +        release(ctx)
> +
> +
> +def do_cmd_capture(state):
> +    capture_init(state['contexts'])
> +
> +    renderer = state['renderer']
> +
> +    renderer.setup()
> +
> +    capture_start(state['contexts'])
> +
> +    renderer.run()
> +
> +    capture_deinit(state['contexts'])
> +
> +
> +def main():
> +    parser = argparse.ArgumentParser()
> +    # global options
> +    parser.add_argument('-l', '--list', action='store_true', help='List all cameras')
> +    parser.add_argument('-c', '--camera', type=int, action='extend', nargs=1, default=[], help='Specify which camera to operate on, by index')
> +    parser.add_argument('-p', '--list-properties', action='store_true', help='List cameras properties')
> +    parser.add_argument('--list-controls', action='store_true', help='List cameras controls')
> +    parser.add_argument('-I', '--info', action='store_true', help='Display information about stream(s)')
> +    parser.add_argument('-R', '--renderer', default='null', help='Renderer (null, kms, qt, qtgl)')
> +
> +    # per camera options
> +    parser.add_argument('-C', '--capture', nargs='?', type=int, const=1000000, action=CustomAction, help='Capture until interrupted by user or until CAPTURE frames captured')
> +    parser.add_argument('--crc', nargs=0, type=bool, action=CustomAction, help='Print CRC32 for captured frames')
> +    parser.add_argument('--save-frames', nargs=0, type=bool, action=CustomAction, help='Save captured frames to files')
> +    parser.add_argument('--metadata', nargs=0, type=bool, action=CustomAction, help='Print the metadata for completed requests')
> +    parser.add_argument('--strict-formats', type=bool, nargs=0, action=CustomAction, help='Do not allow requested stream format(s) to be adjusted')
> +    parser.add_argument('-s', '--stream', nargs='+', action=CustomAction)
> +    args = parser.parse_args()
> +
> +    cm = libcam.CameraManager.singleton()
> +
> +    if args.list:
> +        do_cmd_list(cm)
> +
> +    contexts = []
> +
> +    for cam_idx in args.camera:
> +        camera = next((c for i, c in enumerate(cm.cameras) if i + 1 == cam_idx), None)
> +
> +        if camera is None:
> +            print('Unable to find camera', cam_idx)
> +            return -1
> +
> +        contexts.append({
> +                        'camera': camera,
> +                        'idx': cam_idx,
> +                        'id': 'cam' + str(cam_idx),
> +                        'reqs-queued': 0,
> +                        'reqs-completed': 0,
> +                        'opt-capture': args.capture.get(cam_idx, False),
> +                        'opt-crc': args.crc.get(cam_idx, False),
> +                        'opt-save-frames': args.save_frames.get(cam_idx, False),
> +                        'opt-metadata': args.metadata.get(cam_idx, False),
> +                        'opt-strict-formats': args.strict_formats.get(cam_idx, False),
> +                        'opt-stream': args.stream.get(cam_idx, ['role=viewfinder']),
> +                        })
> +
> +    for ctx in contexts:
> +        print('Using camera {} as {}'.format(ctx['camera'].id, ctx['id']))
> +
> +    for ctx in contexts:
> +        if args.list_properties:
> +            do_cmd_list_props(ctx)
> +        if args.list_controls:
> +            do_cmd_list_controls(ctx)
> +        if args.info:
> +            do_cmd_info(ctx)
> +
> +    if args.capture:
> +
> +        state = {
> +            'cm': cm,
> +            'contexts': contexts,
> +            'event_handler': event_handler,
> +            'request_prcessed': request_prcessed,
> +        }

Same here, a state class would be nice.

> +
> +        if args.renderer == 'null':
> +            import cam_null
> +            renderer = cam_null.NullRenderer(state)
> +        elif args.renderer == 'kms':
> +            import cam_kms
> +            renderer = cam_kms.KMSRenderer(state)
> +        elif args.renderer == 'qt':
> +            import cam_qt
> +            renderer = cam_qt.QtRenderer(state)
> +        elif args.renderer == 'qtgl':
> +            import cam_qtgl
> +            renderer = cam_qtgl.QtRenderer(state)
> +        else:
> +            print('Bad renderer', args.renderer)
> +            return -1
> +
> +        state['renderer'] = renderer
> +
> +        do_cmd_capture(state)
> +
> +    return 0
> +
> +
> +if __name__ == '__main__':
> +    sys.exit(main())
> diff --git a/src/py/cam/cam_kms.py b/src/py/cam/cam_kms.py
> new file mode 100644
> index 00000000..f4ee5a06
> --- /dev/null
> +++ b/src/py/cam/cam_kms.py
> @@ -0,0 +1,183 @@
> +# SPDX-License-Identifier: GPL-2.0-or-later
> +# Copyright (C) 2021, Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
> +
> +import pykms

Why am I not surprised ? :-D

> +import selectors
> +import sys
> +
> +FMT_MAP = {
> +    'RGB888': pykms.PixelFormat.RGB888,
> +    'YUYV': pykms.PixelFormat.YUYV,
> +    'ARGB8888': pykms.PixelFormat.ARGB8888,
> +    'XRGB8888': pykms.PixelFormat.XRGB8888,
> +}
> +
> +
> +class KMSRenderer:
> +    def __init__(self, state):
> +        self.state = state
> +
> +        self.cm = state['cm']
> +        self.contexts = state['contexts']
> +        self.running = False
> +
> +        card = pykms.Card()
> +
> +        res = pykms.ResourceManager(card)
> +        conn = res.reserve_connector()
> +        crtc = res.reserve_crtc(conn)
> +        mode = conn.get_default_mode()
> +        modeb = mode.to_blob(card)
> +
> +        req = pykms.AtomicReq(card)
> +        req.add_connector(conn, crtc)
> +        req.add_crtc(crtc, modeb)
> +        r = req.commit_sync(allow_modeset=True)
> +        assert(r == 0)
> +
> +        self.card = card
> +        self.resman = res
> +        self.crtc = crtc
> +        self.mode = mode
> +
> +        self.bufqueue = []
> +        self.current = None
> +        self.next = None
> +        self.cam_2_drm = {}
> +
> +    # KMS
> +
> +    def close(self):
> +        req = pykms.AtomicReq(self.card)
> +        for s in self.streams:
> +            req.add_plane(s['plane'], None, None, dst=(0, 0, 0, 0))
> +        req.commit()
> +
> +    def add_plane(self, req, stream, fb):
> +        s = next(s for s in self.streams if s['stream'] == stream)
> +        idx = s['idx']
> +        plane = s['plane']
> +
> +        if idx % 2 == 0:
> +            x = 0
> +        else:
> +            x = self.mode.hdisplay - fb.width
> +
> +        if idx // 2 == 0:
> +            y = 0
> +        else:
> +            y = self.mode.vdisplay - fb.height
> +
> +        req.add_plane(plane, fb, self.crtc, dst=(x, y, fb.width, fb.height))
> +
> +    def apply_request(self, drmreq):
> +
> +        buffers = drmreq['camreq'].buffers
> +
> +        for stream, fb in buffers.items():
> +            drmfb = self.cam_2_drm.get(fb, None)
> +
> +            req = pykms.AtomicReq(self.card)
> +            self.add_plane(req, stream, drmfb)
> +            req.commit()
> +
> +    def handle_page_flip(self, frame, time):
> +        old = self.current
> +        self.current = self.next
> +
> +        if len(self.bufqueue) > 0:
> +            self.next = self.bufqueue.pop(0)
> +        else:
> +            self.next = None
> +
> +        if self.next:
> +            drmreq = self.next
> +
> +            self.apply_request(drmreq)
> +
> +        if old:
> +            req = old['camreq']
> +            ctx = old['camctx']
> +            self.state['request_prcessed'](ctx, req)
> +
> +    def queue(self, drmreq):
> +        if not self.next:
> +            self.next = drmreq
> +            self.apply_request(drmreq)
> +        else:
> +            self.bufqueue.append(drmreq)
> +
> +    # libcamera
> +
> +    def setup(self):
> +        self.streams = []
> +
> +        idx = 0
> +        for ctx in self.contexts:
> +            for stream in ctx['streams']:
> +
> +                cfg = stream.configuration
> +                fmt = cfg.pixel_format
> +                fmt = FMT_MAP[fmt]
> +
> +                plane = self.resman.reserve_generic_plane(self.crtc, fmt)
> +                assert(plane is not None)
> +
> +                self.streams.append({
> +                    'idx': idx,
> +                    'stream': stream,
> +                    'plane': plane,
> +                    'fmt': fmt,
> +                    'size': cfg.size,
> +                })
> +
> +                for fb in ctx['allocator'].buffers(stream):
> +                    w, h = cfg.size
> +                    stride = cfg.stride
> +                    fd = fb.fd(0)
> +                    drmfb = pykms.DmabufFramebuffer(self.card, w, h, fmt,
> +                                                    [fd], [stride], [0])
> +                    self.cam_2_drm[fb] = drmfb
> +
> +                idx += 1
> +
> +    def readdrm(self, fileobj):
> +        for ev in self.card.read_events():
> +            if ev.type == pykms.DrmEventType.FLIP_COMPLETE:
> +                self.handle_page_flip(ev.seq, ev.time)
> +
> +    def readcam(self, fd):
> +        self.running = self.state['event_handler'](self.state)
> +
> +    def readkey(self, fileobj):
> +        sys.stdin.readline()
> +        self.running = False
> +
> +    def run(self):
> +        print('Capturing...')
> +
> +        self.running = True
> +
> +        sel = selectors.DefaultSelector()
> +        sel.register(self.card.fd, selectors.EVENT_READ, self.readdrm)
> +        sel.register(self.cm.efd, selectors.EVENT_READ, self.readcam)
> +        sel.register(sys.stdin, selectors.EVENT_READ, self.readkey)
> +
> +        print('Press enter to exit')
> +
> +        while self.running:
> +            events = sel.select()
> +            for key, mask in events:
> +                callback = key.data
> +                callback(key.fileobj)
> +
> +        print('Exiting...')
> +
> +    def request_handler(self, ctx, req):
> +
> +        drmreq = {
> +            'camctx': ctx,
> +            'camreq': req,
> +        }
> +
> +        self.queue(drmreq)
> diff --git a/src/py/cam/cam_null.py b/src/py/cam/cam_null.py
> new file mode 100644
> index 00000000..391397f6
> --- /dev/null
> +++ b/src/py/cam/cam_null.py
> @@ -0,0 +1,47 @@
> +# SPDX-License-Identifier: GPL-2.0-or-later
> +# Copyright (C) 2021, Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
> +
> +import selectors
> +import sys
> +
> +
> +class NullRenderer:
> +    def __init__(self, state):
> +        self.state = state
> +
> +        self.cm = state['cm']
> +        self.contexts = state['contexts']
> +
> +        self.running = False
> +
> +    def setup(self):
> +        pass
> +
> +    def run(self):
> +        print('Capturing...')
> +
> +        self.running = True
> +
> +        sel = selectors.DefaultSelector()
> +        sel.register(self.cm.efd, selectors.EVENT_READ, self.readcam)
> +        sel.register(sys.stdin, selectors.EVENT_READ, self.readkey)
> +
> +        print('Press enter to exit')
> +
> +        while self.running:
> +            events = sel.select()
> +            for key, mask in events:
> +                callback = key.data
> +                callback(key.fileobj)
> +
> +        print('Exiting...')
> +
> +    def readcam(self, fd):
> +        self.running = self.state['event_handler'](self.state)
> +
> +    def readkey(self, fileobj):
> +        sys.stdin.readline()
> +        self.running = False
> +
> +    def request_handler(self, ctx, req):
> +        self.state['request_prcessed'](ctx, req)
> diff --git a/src/py/cam/cam_qt.py b/src/py/cam/cam_qt.py
> new file mode 100644
> index 00000000..40044866
> --- /dev/null
> +++ b/src/py/cam/cam_qt.py
> @@ -0,0 +1,354 @@
> +# SPDX-License-Identifier: GPL-2.0-or-later
> +# Copyright (C) 2021, Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
> +#
> +# Debayering code from PiCamera documentation
> +
> +from io import BytesIO
> +from numpy.lib.stride_tricks import as_strided
> +from PIL import Image
> +from PIL.ImageQt import ImageQt
> +from PyQt5 import QtCore, QtGui, QtWidgets
> +import numpy as np
> +import sys
> +
> +
> +def rgb_to_pix(rgb):
> +    img = Image.frombuffer('RGB', (rgb.shape[1], rgb.shape[0]), rgb)
> +    qim = ImageQt(img).copy()
> +    pix = QtGui.QPixmap.fromImage(qim)
> +    return pix
> +
> +
> +def separate_components(data, r0, g0, g1, b0):
> +    # Now to split the data up into its red, green, and blue components. The
> +    # Bayer pattern of the OV5647 sensor is BGGR. In other words the first
> +    # row contains alternating green/blue elements, the second row contains
> +    # alternating red/green elements, and so on as illustrated below:
> +    #
> +    # GBGBGBGBGBGBGB
> +    # RGRGRGRGRGRGRG
> +    # GBGBGBGBGBGBGB
> +    # RGRGRGRGRGRGRG
> +    #
> +    # Please note that if you use vflip or hflip to change the orientation
> +    # of the capture, you must flip the Bayer pattern accordingly
> +
> +    rgb = np.zeros(data.shape + (3,), dtype=data.dtype)
> +    rgb[r0[1]::2, r0[0]::2, 0] = data[r0[1]::2, r0[0]::2]  # Red
> +    rgb[g0[1]::2, g0[0]::2, 1] = data[g0[1]::2, g0[0]::2]  # Green
> +    rgb[g1[1]::2, g1[0]::2, 1] = data[g1[1]::2, g1[0]::2]  # Green
> +    rgb[b0[1]::2, b0[0]::2, 2] = data[b0[1]::2, b0[0]::2]  # Blue
> +
> +    return rgb
> +
> +
> +def demosaic(rgb, r0, g0, g1, b0):
> +    # At this point we now have the raw Bayer data with the correct values
> +    # and colors but the data still requires de-mosaicing and
> +    # post-processing. If you wish to do this yourself, end the script here!
> +    #
> +    # Below we present a fairly naive de-mosaic method that simply
> +    # calculates the weighted average of a pixel based on the pixels
> +    # surrounding it. The weighting is provided b0[1] a b0[1]te representation of
> +    # the Bayer filter which we construct first:
> +
> +    bayer = np.zeros(rgb.shape, dtype=np.uint8)
> +    bayer[r0[1]::2, r0[0]::2, 0] = 1  # Red
> +    bayer[g0[1]::2, g0[0]::2, 1] = 1  # Green
> +    bayer[g1[1]::2, g1[0]::2, 1] = 1  # Green
> +    bayer[b0[1]::2, b0[0]::2, 2] = 1  # Blue
> +
> +    # Allocate an array to hold our output with the same shape as the input
> +    # data. After this we define the size of window that will be used to
> +    # calculate each weighted average (3x3). Then we pad out the rgb and
> +    # bayer arrays, adding blank pixels at their edges to compensate for the
> +    # size of the window when calculating averages for edge pixels.
> +
> +    output = np.empty(rgb.shape, dtype=rgb.dtype)
> +    window = (3, 3)
> +    borders = (window[0] - 1, window[1] - 1)
> +    border = (borders[0] // 2, borders[1] // 2)
> +
> +    # rgb_pad = np.zeros((
> +    #    rgb.shape[0] + borders[0],
> +    #    rgb.shape[1] + borders[1],
> +    #    rgb.shape[2]), dtype=rgb.dtype)
> +    # rgb_pad[
> +    #    border[0]:rgb_pad.shape[0] - border[0],
> +    #    border[1]:rgb_pad.shape[1] - border[1],
> +    #    :] = rgb
> +    # rgb = rgb_pad
> +    #
> +    # bayer_pad = np.zeros((
> +    #    bayer.shape[0] + borders[0],
> +    #    bayer.shape[1] + borders[1],
> +    #    bayer.shape[2]), dtype=bayer.dtype)
> +    # bayer_pad[
> +    #    border[0]:bayer_pad.shape[0] - border[0],
> +    #    border[1]:bayer_pad.shape[1] - border[1],
> +    #    :] = bayer
> +    # bayer = bayer_pad
> +
> +    # In numpy >=1.7.0 just use np.pad (version in Raspbian is 1.6.2 at the
> +    # time of writing...)
> +    #
> +    rgb = np.pad(rgb, [
> +        (border[0], border[0]),
> +        (border[1], border[1]),
> +        (0, 0),
> +    ], 'constant')
> +    bayer = np.pad(bayer, [
> +        (border[0], border[0]),
> +        (border[1], border[1]),
> +        (0, 0),
> +    ], 'constant')
> +
> +    # For each plane in the RGB data, we use a nifty numpy trick
> +    # (as_strided) to construct a view over the plane of 3x3 matrices. We do
> +    # the same for the bayer array, then use Einstein summation on each
> +    # (np.sum is simpler, but copies the data so it's slower), and divide
> +    # the results to get our weighted average:
> +
> +    for plane in range(3):
> +        p = rgb[..., plane]
> +        b = bayer[..., plane]
> +        pview = as_strided(p, shape=(
> +            p.shape[0] - borders[0],
> +            p.shape[1] - borders[1]) + window, strides=p.strides * 2)
> +        bview = as_strided(b, shape=(
> +            b.shape[0] - borders[0],
> +            b.shape[1] - borders[1]) + window, strides=b.strides * 2)
> +        psum = np.einsum('ijkl->ij', pview)
> +        bsum = np.einsum('ijkl->ij', bview)
> +        output[..., plane] = psum // bsum
> +
> +    return output
> +
> +
> +def to_rgb(fmt, size, data):
> +    w = size[0]
> +    h = size[1]
> +
> +    if fmt == 'YUYV':
> +        # YUV422
> +        yuyv = data.reshape((h, w // 2 * 4))
> +
> +        # YUV444
> +        yuv = np.empty((h, w, 3), dtype=np.uint8)
> +        yuv[:, :, 0] = yuyv[:, 0::2]                    # Y
> +        yuv[:, :, 1] = yuyv[:, 1::4].repeat(2, axis=1)  # U
> +        yuv[:, :, 2] = yuyv[:, 3::4].repeat(2, axis=1)  # V
> +
> +        m = np.array([
> +            [ 1.0, 1.0, 1.0],
> +            [-0.000007154783816076815, -0.3441331386566162, 1.7720025777816772],
> +            [ 1.4019975662231445, -0.7141380310058594 , 0.00001542569043522235]
> +        ])
> +
> +        rgb = np.dot(yuv, m)
> +        rgb[:, :, 0] -= 179.45477266423404
> +        rgb[:, :, 1] += 135.45870971679688
> +        rgb[:, :, 2] -= 226.8183044444304
> +        rgb = rgb.astype(np.uint8)
> +
> +    elif fmt == 'RGB888':
> +        rgb = data.reshape((h, w, 3))
> +        rgb[:, :, [0, 1, 2]] = rgb[:, :, [2, 1, 0]]
> +
> +    elif fmt == 'BGR888':
> +        rgb = data.reshape((h, w, 3))
> +
> +    elif fmt in ['ARGB8888', 'XRGB8888']:
> +        rgb = data.reshape((h, w, 4))
> +        rgb = np.flip(rgb, axis=2)
> +        # drop alpha component
> +        rgb = np.delete(rgb, np.s_[0::4], axis=2)
> +
> +    elif fmt.startswith('S'):
> +        bayer_pattern = fmt[1:5]
> +        bitspp = int(fmt[5:])
> +
> +        # TODO: shifting leaves the lowest bits 0
> +        if bitspp == 8:
> +            data = data.reshape((h, w))
> +            data = data.astype(np.uint16) << 8
> +        elif bitspp in [10, 12]:
> +            data = data.view(np.uint16)
> +            data = data.reshape((h, w))
> +            data = data << (16 - bitspp)
> +        else:
> +            raise Exception('Bad bitspp:' + str(bitspp))
> +
> +        idx = bayer_pattern.find('R')
> +        assert(idx != -1)
> +        r0 = (idx % 2, idx // 2)
> +
> +        idx = bayer_pattern.find('G')
> +        assert(idx != -1)
> +        g0 = (idx % 2, idx // 2)
> +
> +        idx = bayer_pattern.find('G', idx + 1)
> +        assert(idx != -1)
> +        g1 = (idx % 2, idx // 2)
> +
> +        idx = bayer_pattern.find('B')
> +        assert(idx != -1)
> +        b0 = (idx % 2, idx // 2)
> +
> +        rgb = separate_components(data, r0, g0, g1, b0)
> +        rgb = demosaic(rgb, r0, g0, g1, b0)
> +        rgb = (rgb >> 8).astype(np.uint8)
> +
> +    else:
> +        rgb = None
> +
> +    return rgb

I'd move those to a separate file.

> +
> +
> +class QtRenderer:
> +    def __init__(self, state):
> +        self.state = state
> +
> +        self.cm = state['cm']
> +        self.contexts = state['contexts']
> +
> +    def setup(self):
> +        self.app = QtWidgets.QApplication([])
> +
> +        windows = []
> +
> +        for ctx in self.contexts:
> +            camera = ctx['camera']
> +
> +            for stream in ctx['streams']:
> +                fmt = stream.configuration.pixel_format
> +                size = stream.configuration.size
> +
> +                window = MainWindow(ctx, stream)
> +                window.setAttribute(QtCore.Qt.WA_ShowWithoutActivating)
> +                window.show()
> +                windows.append(window)
> +
> +        self.windows = windows
> +
> +    def run(self):
> +        camnotif = QtCore.QSocketNotifier(self.cm.efd, QtCore.QSocketNotifier.Read)
> +        camnotif.activated.connect(lambda x: self.readcam())
> +
> +        keynotif = QtCore.QSocketNotifier(sys.stdin.fileno(), QtCore.QSocketNotifier.Read)
> +        keynotif.activated.connect(lambda x: self.readkey())
> +
> +        print('Capturing...')
> +
> +        self.app.exec()
> +
> +        print('Exiting...')
> +
> +    def readcam(self):
> +        running = self.state['event_handler'](self.state)
> +
> +        if not running:
> +            self.app.quit()
> +
> +    def readkey(self):
> +        sys.stdin.readline()
> +        self.app.quit()
> +
> +    def request_handler(self, ctx, req):
> +        buffers = req.buffers
> +
> +        for stream, fb in buffers.items():
> +            wnd = next(wnd for wnd in self.windows if wnd.stream == stream)
> +
> +            wnd.handle_request(stream, fb)
> +
> +        self.state['request_prcessed'](ctx, req)
> +
> +    def cleanup(self):
> +        for w in self.windows:
> +            w.close()
> +
> +
> +class MainWindow(QtWidgets.QWidget):
> +    def __init__(self, ctx, stream):
> +        super().__init__()
> +
> +        self.ctx = ctx
> +        self.stream = stream
> +
> +        self.label = QtWidgets.QLabel()
> +
> +        windowLayout = QtWidgets.QHBoxLayout()
> +        self.setLayout(windowLayout)
> +
> +        windowLayout.addWidget(self.label)
> +
> +        controlsLayout = QtWidgets.QVBoxLayout()
> +        windowLayout.addLayout(controlsLayout)
> +
> +        windowLayout.addStretch()
> +
> +        group = QtWidgets.QGroupBox('Info')
> +        groupLayout = QtWidgets.QVBoxLayout()
> +        group.setLayout(groupLayout)
> +        controlsLayout.addWidget(group)
> +
> +        lab = QtWidgets.QLabel(ctx['id'])
> +        groupLayout.addWidget(lab)
> +
> +        self.frameLabel = QtWidgets.QLabel()
> +        groupLayout.addWidget(self.frameLabel)
> +
> +        group = QtWidgets.QGroupBox('Properties')
> +        groupLayout = QtWidgets.QVBoxLayout()
> +        group.setLayout(groupLayout)
> +        controlsLayout.addWidget(group)
> +
> +        camera = ctx['camera']
> +
> +        for k, v in camera.properties.items():
> +            lab = QtWidgets.QLabel()
> +            lab.setText(k + ' = ' + str(v))
> +            groupLayout.addWidget(lab)
> +
> +        group = QtWidgets.QGroupBox('Controls')
> +        groupLayout = QtWidgets.QVBoxLayout()
> +        group.setLayout(groupLayout)
> +        controlsLayout.addWidget(group)
> +
> +        for k, (min, max, default) in camera.controls.items():
> +            lab = QtWidgets.QLabel()
> +            lab.setText('{} = {}/{}/{}'.format(k, min, max, default))
> +            groupLayout.addWidget(lab)
> +
> +        controlsLayout.addStretch()
> +
> +    def buf_to_qpixmap(self, stream, fb):
> +        with fb.mmap() as mfb:
> +            cfg = stream.configuration
> +            w, h = cfg.size
> +            pitch = cfg.stride
> +
> +            if cfg.pixel_format == 'MJPEG':
> +                img = Image.open(BytesIO(mfb.planes[0]))
> +                qim = ImageQt(img).copy()
> +                pix = QtGui.QPixmap.fromImage(qim)
> +            else:
> +                data = np.array(mfb.planes[0], dtype=np.uint8)
> +                rgb = to_rgb(cfg.pixel_format, cfg.size, data)
> +
> +                if rgb is None:
> +                    raise Exception('Format not supported: ' + cfg.pixel_format)
> +
> +                pix = rgb_to_pix(rgb)
> +
> +        return pix
> +
> +    def handle_request(self, stream, fb):
> +        ctx = self.ctx
> +
> +        pix = self.buf_to_qpixmap(stream, fb)
> +        self.label.setPixmap(pix)
> +
> +        self.frameLabel.setText('Queued: {}\nDone: {}\nFps: {:.2f}'
> +                       .format(ctx['reqs-queued'], ctx['reqs-completed'], ctx['fps']))
> diff --git a/src/py/cam/cam_qtgl.py b/src/py/cam/cam_qtgl.py
> new file mode 100644
> index 00000000..37b74d3f
> --- /dev/null
> +++ b/src/py/cam/cam_qtgl.py
> @@ -0,0 +1,385 @@
> +# SPDX-License-Identifier: GPL-2.0-or-later
> +# Copyright (C) 2021, Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
> +
> +from PyQt5 import QtCore, QtWidgets
> +from PyQt5.QtCore import Qt
> +
> +import math
> +import numpy as np
> +import os
> +import sys
> +
> +os.environ['PYOPENGL_PLATFORM'] = 'egl'
> +
> +import OpenGL
> +# OpenGL.FULL_LOGGING = True
> +
> +from OpenGL import GL as gl
> +from OpenGL.EGL.EXT.image_dma_buf_import import *
> +from OpenGL.EGL.KHR.image import *
> +from OpenGL.EGL.VERSION.EGL_1_0 import *
> +from OpenGL.EGL.VERSION.EGL_1_2 import *
> +from OpenGL.EGL.VERSION.EGL_1_3 import *
> +
> +from OpenGL.GLES2.OES.EGL_image import *
> +from OpenGL.GLES2.OES.EGL_image_external import *
> +from OpenGL.GLES2.VERSION.GLES2_2_0 import *
> +from OpenGL.GLES3.VERSION.GLES3_3_0 import *
> +
> +from OpenGL.GL import shaders
> +
> +from gl_helpers import *
> +
> +# libcamera format string -> DRM fourcc
> +FMT_MAP = {
> +    'RGB888': 'RG24',
> +    'XRGB8888': 'XR24',
> +    'ARGB8888': 'AR24',
> +    'YUYV': 'YUYV',
> +}
> +
> +
> +class EglState:
> +    def __init__(self):
> +        self.create_display()
> +        self.choose_config()
> +        self.create_context()
> +        self.check_extensions()
> +
> +    def create_display(self):
> +        xdpy = getEGLNativeDisplay()
> +        dpy = eglGetDisplay(xdpy)
> +        self.display = dpy

Could we use the Qt GL API to avoid managing the display manually ?

> +
> +    def choose_config(self):
> +        dpy = self.display
> +
> +        major, minor = EGLint(), EGLint()
> +
> +        b = eglInitialize(dpy, major, minor)
> +        assert(b)
> +
> +        print('EGL {} {}'.format(
> +              eglQueryString(dpy, EGL_VENDOR).decode(),
> +              eglQueryString(dpy, EGL_VERSION).decode()))
> +
> +        check_egl_extensions(dpy, ['EGL_EXT_image_dma_buf_import'])
> +
> +        b = eglBindAPI(EGL_OPENGL_ES_API)
> +        assert(b)
> +
> +        def print_config(dpy, cfg):
> +
> +            def _getconf(dpy, cfg, a):
> +                value = ctypes.c_long()
> +                eglGetConfigAttrib(dpy, cfg, a, value)
> +                return value.value
> +
> +            getconf = lambda a: _getconf(dpy, cfg, a)
> +
> +            print('EGL Config {}: color buf {}/{}/{}/{} = {}, depth {}, stencil {}, native visualid {}, native visualtype {}'.format(
> +                getconf(EGL_CONFIG_ID),
> +                getconf(EGL_ALPHA_SIZE),
> +                getconf(EGL_RED_SIZE),
> +                getconf(EGL_GREEN_SIZE),
> +                getconf(EGL_BLUE_SIZE),
> +                getconf(EGL_BUFFER_SIZE),
> +                getconf(EGL_DEPTH_SIZE),
> +                getconf(EGL_STENCIL_SIZE),
> +                getconf(EGL_NATIVE_VISUAL_ID),
> +                getconf(EGL_NATIVE_VISUAL_TYPE)))
> +
> +        if False:
> +            num_configs = ctypes.c_long()
> +            eglGetConfigs(dpy, None, 0, num_configs)
> +            print('{} configs'.format(num_configs.value))
> +
> +            configs = (EGLConfig * num_configs.value)()
> +            eglGetConfigs(dpy, configs, num_configs.value, num_configs)
> +            for config_id in configs:
> +                print_config(dpy, config_id)
> +
> +        config_attribs = [
> +            EGL_SURFACE_TYPE, EGL_WINDOW_BIT,
> +            EGL_RED_SIZE, 8,
> +            EGL_GREEN_SIZE, 8,
> +            EGL_BLUE_SIZE, 8,
> +            EGL_ALPHA_SIZE, 0,
> +            EGL_RENDERABLE_TYPE, EGL_OPENGL_ES2_BIT,
> +            EGL_NONE,
> +        ]
> +
> +        n = EGLint()
> +        configs = (EGLConfig * 1)()
> +        b = eglChooseConfig(dpy, config_attribs, configs, 1, n)
> +        assert(b and n.value == 1)
> +        config = configs[0]
> +
> +        print('Chosen Config:')
> +        print_config(dpy, config)
> +
> +        self.config = config
> +
> +    def create_context(self):
> +        dpy = self.display
> +
> +        context_attribs = [
> +            EGL_CONTEXT_CLIENT_VERSION, 2,
> +            EGL_NONE,
> +        ]
> +
> +        context = eglCreateContext(dpy, self.config, EGL_NO_CONTEXT, context_attribs)
> +        assert(context)
> +
> +        b = eglMakeCurrent(dpy, EGL_NO_SURFACE, EGL_NO_SURFACE, context)
> +        assert(b)
> +
> +        self.context = context
> +
> +    def check_extensions(self):
> +        check_gl_extensions(['GL_OES_EGL_image'])
> +
> +        assert(eglCreateImageKHR)
> +        assert(eglDestroyImageKHR)
> +        assert(glEGLImageTargetTexture2DOES)
> +
> +
> +class QtRenderer:
> +    def __init__(self, state):
> +        self.state = state
> +
> +    def setup(self):
> +        self.app = QtWidgets.QApplication([])
> +
> +        window = MainWindow(self.state)
> +        window.setAttribute(QtCore.Qt.WA_ShowWithoutActivating)
> +        window.show()
> +
> +        self.window = window
> +
> +    def run(self):
> +        camnotif = QtCore.QSocketNotifier(self.state['cm'].efd, QtCore.QSocketNotifier.Read)
> +        camnotif.activated.connect(lambda x: self.readcam())
> +
> +        keynotif = QtCore.QSocketNotifier(sys.stdin.fileno(), QtCore.QSocketNotifier.Read)
> +        keynotif.activated.connect(lambda x: self.readkey())
> +
> +        print('Capturing...')
> +
> +        self.app.exec()
> +
> +        print('Exiting...')
> +
> +    def readcam(self):
> +        running = self.state['event_handler'](self.state)
> +
> +        if not running:
> +            self.app.quit()
> +
> +    def readkey(self):
> +        sys.stdin.readline()
> +        self.app.quit()
> +
> +    def request_handler(self, ctx, req):
> +        self.window.handle_request(ctx, req)
> +
> +    def cleanup(self):
> +        self.window.close()
> +
> +
> +class MainWindow(QtWidgets.QWidget):
> +    def __init__(self, state):
> +        super().__init__()
> +
> +        self.setAttribute(Qt.WA_PaintOnScreen)
> +        self.setAttribute(Qt.WA_NativeWindow)
> +
> +        self.state = state
> +
> +        self.textures = {}
> +        self.reqqueue = {}
> +        self.current = {}
> +
> +        for ctx in self.state['contexts']:
> +
> +            self.reqqueue[ctx['idx']] = []
> +            self.current[ctx['idx']] = []
> +
> +            for stream in ctx['streams']:
> +                fmt = stream.configuration.pixel_format
> +                size = stream.configuration.size
> +
> +                if fmt not in FMT_MAP:
> +                    raise Exception('Unsupported pixel format: ' + str(fmt))
> +
> +                self.textures[stream] = None
> +
> +        num_tiles = len(self.textures)
> +        self.num_columns = math.ceil(math.sqrt(num_tiles))
> +        self.num_rows = math.ceil(num_tiles / self.num_columns)
> +
> +        self.egl = EglState()
> +
> +        self.surface = None
> +
> +    def paintEngine(self):
> +        return None
> +
> +    def create_surface(self):
> +        native_surface = c_void_p(self.winId().__int__())
> +        surface = eglCreateWindowSurface(self.egl.display, self.egl.config,
> +                                         native_surface, None)
> +
> +        b = eglMakeCurrent(self.egl.display, self.surface, self.surface, self.egl.context)
> +        assert(b)
> +
> +        self.surface = surface
> +
> +    def init_gl(self):
> +        self.create_surface()
> +
> +        vertShaderSrc = '''
> +            attribute vec2 aPosition;
> +            varying vec2 texcoord;
> +
> +            void main()
> +            {
> +                gl_Position = vec4(aPosition * 2.0 - 1.0, 0.0, 1.0);
> +                texcoord.x = aPosition.x;
> +                texcoord.y = 1.0 - aPosition.y;
> +            }
> +        '''
> +        fragShaderSrc = '''
> +            #extension GL_OES_EGL_image_external : enable
> +            precision mediump float;
> +            varying vec2 texcoord;
> +            uniform samplerExternalOES texture;
> +
> +            void main()
> +            {
> +                gl_FragColor = texture2D(texture, texcoord);
> +            }
> +        '''
> +
> +        program = shaders.compileProgram(
> +            shaders.compileShader(vertShaderSrc, GL_VERTEX_SHADER),
> +            shaders.compileShader(fragShaderSrc, GL_FRAGMENT_SHADER)
> +        )
> +
> +        glUseProgram(program)
> +
> +        glClearColor(0.5, 0.8, 0.7, 1.0)
> +
> +        vertPositions = [
> +            0.0, 0.0,
> +            1.0, 0.0,
> +            1.0, 1.0,
> +            0.0, 1.0
> +        ]
> +
> +        inputAttrib = glGetAttribLocation(program, 'aPosition')
> +        glVertexAttribPointer(inputAttrib, 2, GL_FLOAT, GL_FALSE, 0, vertPositions)
> +        glEnableVertexAttribArray(inputAttrib)
> +
> +    def create_texture(self, stream, fb):
> +        cfg = stream.configuration
> +        fmt = cfg.pixel_format
> +        fmt = str_to_fourcc(FMT_MAP[fmt])
> +        w, h = cfg.size
> +
> +        attribs = [
> +            EGL_WIDTH, w,
> +            EGL_HEIGHT, h,
> +            EGL_LINUX_DRM_FOURCC_EXT, fmt,
> +            EGL_DMA_BUF_PLANE0_FD_EXT, fb.fd(0),
> +            EGL_DMA_BUF_PLANE0_OFFSET_EXT, 0,
> +            EGL_DMA_BUF_PLANE0_PITCH_EXT, cfg.stride,
> +            EGL_NONE,
> +        ]
> +
> +        image = eglCreateImageKHR(self.egl.display,
> +                                  EGL_NO_CONTEXT,
> +                                  EGL_LINUX_DMA_BUF_EXT,
> +                                  None,
> +                                  attribs)
> +        assert(image)
> +
> +        textures = glGenTextures(1)
> +        glBindTexture(GL_TEXTURE_EXTERNAL_OES, textures)
> +        glTexParameteri(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_MAG_FILTER, GL_LINEAR)
> +        glTexParameteri(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_MIN_FILTER, GL_LINEAR)
> +        glTexParameteri(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE)
> +        glTexParameteri(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE)
> +        glEGLImageTargetTexture2DOES(GL_TEXTURE_EXTERNAL_OES, image)
> +
> +        return textures
> +
> +    def resizeEvent(self, event):
> +        size = event.size()
> +
> +        print('Resize', size)
> +
> +        super().resizeEvent(event)
> +
> +        if self.surface is None:
> +            return
> +
> +        glViewport(0, 0, size.width() // 2, size.height())
> +
> +    def paintEvent(self, event):
> +        if self.surface is None:
> +            self.init_gl()
> +
> +        for ctx_idx, queue in self.reqqueue.items():
> +            if len(queue) == 0:
> +                continue
> +
> +            ctx = next(ctx for ctx in self.state['contexts'] if ctx['idx'] == ctx_idx)
> +
> +            if self.current[ctx_idx]:
> +                old = self.current[ctx_idx]
> +                self.current[ctx_idx] = None
> +                self.state['request_prcessed'](ctx, old)
> +
> +            next_req = queue.pop(0)
> +            self.current[ctx_idx] = next_req
> +
> +            stream, fb = next(iter(next_req.buffers.items()))
> +
> +            self.textures[stream] = self.create_texture(stream, fb)
> +
> +        self.paint_gl()
> +
> +    def paint_gl(self):
> +        b = eglMakeCurrent(self.egl.display, self.surface, self.surface, self.egl.context)
> +        assert(b)
> +
> +        glClear(GL_COLOR_BUFFER_BIT)
> +
> +        size = self.size()
> +
> +        for idx, ctx in enumerate(self.state['contexts']):
> +            for stream in ctx['streams']:
> +                if self.textures[stream] is None:
> +                    continue
> +
> +                w = size.width() // self.num_columns
> +                h = size.height() // self.num_rows
> +
> +                x = idx % self.num_columns
> +                y = idx // self.num_columns
> +
> +                x *= w
> +                y *= h
> +
> +                glViewport(x, y, w, h)
> +
> +                glBindTexture(GL_TEXTURE_EXTERNAL_OES, self.textures[stream])
> +                glDrawArrays(GL_TRIANGLE_FAN, 0, 4)
> +
> +        b = eglSwapBuffers(self.egl.display, self.surface)
> +        assert(b)
> +
> +    def handle_request(self, ctx, req):
> +        self.reqqueue[ctx['idx']].append(req)
> +        self.update()
> diff --git a/src/py/cam/gl_helpers.py b/src/py/cam/gl_helpers.py
> new file mode 100644
> index 00000000..b377d735
> --- /dev/null
> +++ b/src/py/cam/gl_helpers.py
> @@ -0,0 +1,74 @@
> +# SPDX-License-Identifier: GPL-2.0-or-later
> +# Copyright (C) 2021, Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
> +
> +from OpenGL.EGL.VERSION.EGL_1_0 import EGLNativeDisplayType, eglGetProcAddress, eglQueryString, EGL_EXTENSIONS
> +
> +from OpenGL.raw.GLES2 import _types as _cs
> +from OpenGL.GLES2.VERSION.GLES2_2_0 import *
> +from OpenGL.GLES3.VERSION.GLES3_3_0 import *
> +from OpenGL import GL as gl
> +
> +from ctypes import c_int, c_char_p, c_void_p, cdll, POINTER, util, \
> +    pointer, CFUNCTYPE, c_bool
> +
> +
> +def getEGLNativeDisplay():
> +    _x11lib = cdll.LoadLibrary(util.find_library('X11'))
> +    XOpenDisplay = _x11lib.XOpenDisplay
> +    XOpenDisplay.argtypes = [c_char_p]
> +    XOpenDisplay.restype = POINTER(EGLNativeDisplayType)
> +
> +    xdpy = XOpenDisplay(None)

Did you mean

    return XOpenDisplay(None)

> +
> +
> +# Hack. PyOpenGL doesn't seem to manage to find glEGLImageTargetTexture2DOES.
> +def getglEGLImageTargetTexture2DOES():
> +    funcptr = eglGetProcAddress('glEGLImageTargetTexture2DOES')
> +    prototype = CFUNCTYPE(None, _cs.GLenum, _cs.GLeglImageOES)
> +    return prototype(funcptr)
> +
> +
> +glEGLImageTargetTexture2DOES = getglEGLImageTargetTexture2DOES()
> +
> +
> +def str_to_fourcc(str):
> +    assert(len(str) == 4)
> +    fourcc = 0
> +    for i, v in enumerate([ord(c) for c in str]):
> +        fourcc |= v << (i * 8)
> +    return fourcc

Is retrieving the numerical 4CC something that should be part of the
PixelFormat bindings ?

> +
> +
> +def get_gl_extensions():
> +    n = GLint()
> +    glGetIntegerv(GL_NUM_EXTENSIONS, n)
> +    gl_extensions = []
> +    for i in range(n.value):
> +        gl_extensions.append(gl.glGetStringi(GL_EXTENSIONS, i).decode())
> +    return gl_extensions
> +
> +
> +def check_gl_extensions(required_extensions):
> +    extensions = get_gl_extensions()
> +
> +    if False:
> +        print('GL EXTENSIONS: ', ' '.join(extensions))

Looks like you need some type of logging mechanism :-)

> +
> +    for ext in required_extensions:
> +        if ext not in extensions:
> +            raise Exception(ext + ' missing')
> +
> +
> +def get_egl_extensions(egl_display):
> +    return eglQueryString(egl_display, EGL_EXTENSIONS).decode().split(' ')
> +
> +
> +def check_egl_extensions(egl_display, required_extensions):
> +    extensions = get_egl_extensions(egl_display)
> +
> +    if False:
> +        print('EGL EXTENSIONS: ', ' '.join(extensions))
> +
> +    for ext in required_extensions:
> +        if ext not in extensions:
> +            raise Exception(ext + ' missing')
Tomi Valkeinen May 7, 2022, 7:57 a.m. UTC | #2
On 06/05/2022 22:49, Laurent Pinchart wrote:
> Hi Tomi,
> 
> Thank you for the patch.
> 
> On Fri, May 06, 2022 at 05:54:14PM +0300, Tomi Valkeinen wrote:
>> Add cam.py, which mimics the 'cam' tool. Four rendering backends are
>> added:
>>
>> * null - Do nothing
>> * kms - Use KMS with dmabufs
>> * qt - SW render on a Qt window
>> * qtgl - OpenGL render on a Qt window
>>
>> All the renderers handle only a few pixel formats, and especially the GL
>> renderer is just a prototype.
> 
> I don't review this in much details as you mentioned it was a proof of
> concept. We can improve things later, and high-level comments below
> don't all need to (but can :-)) be addressed in a new version.
> 
>> Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
>> ---
>>   src/py/cam/cam.py        | 484 +++++++++++++++++++++++++++++++++++++++
>>   src/py/cam/cam_kms.py    | 183 +++++++++++++++
>>   src/py/cam/cam_null.py   |  47 ++++
>>   src/py/cam/cam_qt.py     | 354 ++++++++++++++++++++++++++++
>>   src/py/cam/cam_qtgl.py   | 385 +++++++++++++++++++++++++++++++
>>   src/py/cam/gl_helpers.py |  74 ++++++
>>   6 files changed, 1527 insertions(+)
>>   create mode 100755 src/py/cam/cam.py
>>   create mode 100644 src/py/cam/cam_kms.py
>>   create mode 100644 src/py/cam/cam_null.py
>>   create mode 100644 src/py/cam/cam_qt.py
>>   create mode 100644 src/py/cam/cam_qtgl.py
>>   create mode 100644 src/py/cam/gl_helpers.py
>>
>> diff --git a/src/py/cam/cam.py b/src/py/cam/cam.py
>> new file mode 100755
>> index 00000000..5ea02850
>> --- /dev/null
>> +++ b/src/py/cam/cam.py
>> @@ -0,0 +1,484 @@
>> +#!/usr/bin/env python3
>> +
>> +# SPDX-License-Identifier: GPL-2.0-or-later
>> +# Copyright (C) 2021, Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
>> +
>> +import argparse
>> +import binascii
>> +import libcamera as libcam
>> +import os
>> +import sys
>> +
>> +
>> +class CustomCameraAction(argparse.Action):
> 
> This doesn't seem to be used.

Indeed.

>> +    def __call__(self, parser, namespace, values, option_string=None):
>> +        print(self.dest, values)
>> +
>> +        if 'camera' not in namespace or namespace.camera is None:
>> +            setattr(namespace, 'camera', [])
>> +
>> +        previous = namespace.camera
>> +        previous.append((self.dest, values))
>> +        setattr(namespace, 'camera', previous)
>> +
>> +
>> +class CustomAction(argparse.Action):
>> +    def __init__(self, option_strings, dest, **kwargs):
>> +        super().__init__(option_strings, dest, default={}, **kwargs)
>> +
>> +    def __call__(self, parser, namespace, values, option_string=None):
>> +        if len(namespace.camera) == 0:
>> +            print(f'Option {option_string} requires a --camera context')
>> +            sys.exit(-1)
>> +
>> +        if self.type == bool:
>> +            values = True
>> +
>> +        current = namespace.camera[-1]
>> +
>> +        data = getattr(namespace, self.dest)
>> +
>> +        if self.nargs == '+':
>> +            if current not in data:
>> +                data[current] = []
>> +
>> +            data[current] += values
>> +        else:
>> +            data[current] = values
>> +
>> +
>> +def do_cmd_list(cm):
>> +    print('Available cameras:')
>> +
>> +    for idx, c in enumerate(cm.cameras):
>> +        print(f'{idx + 1}: {c.id}')
>> +
>> +
>> +def do_cmd_list_props(ctx):
>> +    camera = ctx['camera']
>> +
>> +    print('Properties for', ctx['id'])
>> +
>> +    for name, prop in camera.properties.items():
>> +        print('\t{}: {}'.format(name, prop))
>> +
>> +
>> +def do_cmd_list_controls(ctx):
>> +    camera = ctx['camera']
>> +
>> +    print('Controls for', ctx['id'])
>> +
>> +    for name, prop in camera.controls.items():
>> +        print('\t{}: {}'.format(name, prop))
>> +
>> +
>> +def do_cmd_info(ctx):
>> +    camera = ctx['camera']
>> +
>> +    print('Stream info for', ctx['id'])
>> +
>> +    roles = [libcam.StreamRole.Viewfinder]
>> +
>> +    camconfig = camera.generate_configuration(roles)
>> +    if camconfig is None:
>> +        raise Exception('Generating config failed')
>> +
>> +    for i, stream_config in enumerate(camconfig):
>> +        print('\t{}: {}'.format(i, stream_config.to_string()))
>> +
>> +        formats = stream_config.formats
>> +        for fmt in formats.pixel_formats:
>> +            print('\t * Pixelformat:', fmt, formats.range(fmt))
>> +
>> +            for size in formats.sizes(fmt):
>> +                print('\t  -', size)
>> +
>> +
>> +def acquire(ctx):
>> +    camera = ctx['camera']
>> +
>> +    camera.acquire()
>> +
>> +
>> +def release(ctx):
>> +    camera = ctx['camera']
>> +
>> +    camera.release()
>> +
>> +
>> +def parse_streams(ctx):
>> +    streams = []
>> +
>> +    for stream_desc in ctx['opt-stream']:
>> +        stream_opts = {'role': libcam.StreamRole.Viewfinder}
>> +
>> +        for stream_opt in stream_desc.split(','):
>> +            if stream_opt == 0:
>> +                continue
>> +
>> +            arr = stream_opt.split('=')
>> +            if len(arr) != 2:
>> +                print('Bad stream option', stream_opt)
>> +                sys.exit(-1)
>> +
>> +            key = arr[0]
>> +            value = arr[1]
>> +
>> +            if key in ['width', 'height']:
>> +                value = int(value)
>> +            elif key == 'role':
>> +                rolemap = {
>> +                    'still': libcam.StreamRole.StillCapture,
>> +                    'raw': libcam.StreamRole.Raw,
>> +                    'video': libcam.StreamRole.VideoRecording,
>> +                    'viewfinder': libcam.StreamRole.Viewfinder,
>> +                }
>> +
>> +                role = rolemap.get(value.lower(), None)
>> +
>> +                if role is None:
>> +                    print('Bad stream role', value)
>> +                    sys.exit(-1)
>> +
>> +                value = role
>> +            elif key == 'pixelformat':
>> +                pass
>> +            else:
>> +                print('Bad stream option key', key)
>> +                sys.exit(-1)
>> +
>> +            stream_opts[key] = value
>> +
>> +        streams.append(stream_opts)
>> +
>> +    return streams
>> +
>> +
>> +def configure(ctx):
>> +    camera = ctx['camera']
>> +
>> +    streams = parse_streams(ctx)
>> +
>> +    roles = [opts['role'] for opts in streams]
>> +
>> +    camconfig = camera.generate_configuration(roles)
>> +    if camconfig is None:
>> +        raise Exception('Generating config failed')
>> +
>> +    for idx, stream_opts in enumerate(streams):
>> +        stream_config = camconfig.at(idx)
>> +
>> +        if 'width' in stream_opts and 'height' in stream_opts:
>> +            stream_config.size = (stream_opts['width'], stream_opts['height'])
>> +
>> +        if 'pixelformat' in stream_opts:
>> +            stream_config.pixel_format = stream_opts['pixelformat']
>> +
>> +    stat = camconfig.validate()
>> +
>> +    if stat == libcam.CameraConfiguration.Status.Invalid:
>> +        print('Camera configuration invalid')
>> +        exit(-1)
>> +    elif stat == libcam.CameraConfiguration.Status.Adjusted:
>> +        if ctx['opt-strict-formats']:
>> +            print('Adjusting camera configuration disallowed by --strict-formats argument')
>> +            exit(-1)
>> +
>> +        print('Camera configuration adjusted')
>> +
>> +    r = camera.configure(camconfig)
>> +    if r != 0:
>> +        raise Exception('Configure failed')
>> +
>> +    ctx['stream-names'] = {}
>> +    ctx['streams'] = []
>> +
>> +    for idx, stream_config in enumerate(camconfig):
>> +        stream = stream_config.stream
>> +        ctx['streams'].append(stream)
>> +        ctx['stream-names'][stream] = 'stream' + str(idx)
>> +        print('{}-{}: stream config {}'.format(ctx['id'], ctx['stream-names'][stream], stream.configuration.to_string()))
>> +
>> +
>> +def alloc_buffers(ctx):
>> +    camera = ctx['camera']
>> +
>> +    allocator = libcam.FrameBufferAllocator(camera)
>> +
>> +    for idx, stream in enumerate(ctx['streams']):
>> +        ret = allocator.allocate(stream)
>> +        if ret < 0:
>> +            print('Cannot allocate buffers')
>> +            exit(-1)
>> +
>> +        allocated = len(allocator.buffers(stream))
>> +
>> +        print('{}-{}: Allocated {} buffers'.format(ctx['id'], ctx['stream-names'][stream], allocated))
>> +
>> +    ctx['allocator'] = allocator
>> +
>> +
>> +def create_requests(ctx):
>> +    camera = ctx['camera']
>> +
>> +    ctx['requests'] = []
>> +
>> +    # Identify the stream with the least number of buffers
>> +    num_bufs = min([len(ctx['allocator'].buffers(stream)) for stream in ctx['streams']])
>> +
>> +    requests = []
>> +
>> +    for buf_num in range(num_bufs):
>> +        request = camera.create_request(ctx['idx'])
>> +
>> +        if request is None:
>> +            print('Can not create request')
>> +            exit(-1)
>> +
>> +        for stream in ctx['streams']:
>> +            buffers = ctx['allocator'].buffers(stream)
>> +            buffer = buffers[buf_num]
>> +
>> +            ret = request.add_buffer(stream, buffer)
>> +            if ret < 0:
>> +                print('Can not set buffer for request')
>> +                exit(-1)
>> +
>> +        requests.append(request)
>> +
>> +    ctx['requests'] = requests
>> +
>> +
>> +def start(ctx):
>> +    camera = ctx['camera']
>> +
>> +    camera.start()
>> +
>> +
>> +def stop(ctx):
>> +    camera = ctx['camera']
>> +
>> +    camera.stop()
>> +
>> +
>> +def queue_requests(ctx):
>> +    camera = ctx['camera']
>> +
>> +    for request in ctx['requests']:
>> +        camera.queue_request(request)
>> +        ctx['reqs-queued'] += 1
>> +
>> +    del ctx['requests']
> 
> It would be nice to group most functions operating on a context in a
> class, with members of the context dict turned into class member. It
> would make the next version easier to review.

Yes, definitely. The app has grown from my initial experimental scripts, 
as perhaps can be seen... =)

>> +
>> +
>> +def capture_init(contexts):
>> +    for ctx in contexts:
>> +        acquire(ctx)
>> +
>> +    for ctx in contexts:
>> +        configure(ctx)
>> +
>> +    for ctx in contexts:
>> +        alloc_buffers(ctx)
>> +
>> +    for ctx in contexts:
>> +        create_requests(ctx)
>> +
>> +
>> +def capture_start(contexts):
>> +    for ctx in contexts:
>> +        start(ctx)
>> +
>> +    for ctx in contexts:
>> +        queue_requests(ctx)
>> +
>> +
>> +# Called from renderer when there is a libcamera event
>> +def event_handler(state):
>> +    cm = state['cm']
>> +    contexts = state['contexts']
>> +
>> +    os.read(cm.efd, 8)
>> +
>> +    reqs = cm.get_ready_requests()
>> +
>> +    for req in reqs:
>> +        ctx = next(ctx for ctx in contexts if ctx['idx'] == req.cookie)
>> +        request_handler(state, ctx, req)
>> +
>> +    running = any(ctx['reqs-completed'] < ctx['opt-capture'] for ctx in contexts)
>> +    return running
>> +
>> +
>> +def request_handler(state, ctx, req):
>> +    if req.status != libcam.Request.Status.Complete:
>> +        raise Exception('{}: Request failed: {}'.format(ctx['id'], req.status))
>> +
>> +    buffers = req.buffers
>> +
>> +    # Compute the frame rate. The timestamp is arbitrarily retrieved from
>> +    # the first buffer, as all buffers should have matching timestamps.
>> +    ts = buffers[next(iter(buffers))].metadata.timestamp
>> +    last = ctx.get('last', 0)
>> +    fps = 1000000000.0 / (ts - last) if (last != 0 and (ts - last) != 0) else 0
>> +    ctx['last'] = ts
>> +    ctx['fps'] = fps
>> +
>> +    for stream, fb in buffers.items():
>> +        stream_name = ctx['stream-names'][stream]
>> +
>> +        crcs = []
>> +        if ctx['opt-crc']:
>> +            with fb.mmap() as mfb:
>> +                plane_crcs = [binascii.crc32(p) for p in mfb.planes]
>> +                crcs.append(plane_crcs)
>> +
>> +        meta = fb.metadata
>> +
>> +        print('{:.6f} ({:.2f} fps) {}-{}: seq {}, bytes {}, CRCs {}'
>> +              .format(ts / 1000000000, fps,
>> +                      ctx['id'], stream_name,
>> +                      meta.sequence, meta.bytesused,
>> +                      crcs))
>> +
>> +        if ctx['opt-metadata']:
>> +            reqmeta = req.metadata
>> +            for ctrl, val in reqmeta.items():
>> +                print(f'\t{ctrl} = {val}')
>> +
>> +        if ctx['opt-save-frames']:
>> +            with fb.mmap() as mfb:
>> +                filename = 'frame-{}-{}-{}.data'.format(ctx['id'], stream_name, ctx['reqs-completed'])
>> +                with open(filename, 'wb') as f:
>> +                    for p in mfb.planes:
>> +                        f.write(p)
>> +
>> +    state['renderer'].request_handler(ctx, req)
>> +
>> +    ctx['reqs-completed'] += 1
>> +
>> +
>> +# Called from renderer when it has finished with a request
>> +def request_prcessed(ctx, req):
>> +    camera = ctx['camera']
>> +
>> +    if ctx['reqs-queued'] < ctx['opt-capture']:
>> +        req.reuse()
>> +        camera.queue_request(req)
>> +        ctx['reqs-queued'] += 1
>> +
>> +
>> +def capture_deinit(contexts):
>> +    for ctx in contexts:
>> +        stop(ctx)
>> +
>> +    for ctx in contexts:
>> +        release(ctx)
>> +
>> +
>> +def do_cmd_capture(state):
>> +    capture_init(state['contexts'])
>> +
>> +    renderer = state['renderer']
>> +
>> +    renderer.setup()
>> +
>> +    capture_start(state['contexts'])
>> +
>> +    renderer.run()
>> +
>> +    capture_deinit(state['contexts'])
>> +
>> +
>> +def main():
>> +    parser = argparse.ArgumentParser()
>> +    # global options
>> +    parser.add_argument('-l', '--list', action='store_true', help='List all cameras')
>> +    parser.add_argument('-c', '--camera', type=int, action='extend', nargs=1, default=[], help='Specify which camera to operate on, by index')
>> +    parser.add_argument('-p', '--list-properties', action='store_true', help='List cameras properties')
>> +    parser.add_argument('--list-controls', action='store_true', help='List cameras controls')
>> +    parser.add_argument('-I', '--info', action='store_true', help='Display information about stream(s)')
>> +    parser.add_argument('-R', '--renderer', default='null', help='Renderer (null, kms, qt, qtgl)')
>> +
>> +    # per camera options
>> +    parser.add_argument('-C', '--capture', nargs='?', type=int, const=1000000, action=CustomAction, help='Capture until interrupted by user or until CAPTURE frames captured')
>> +    parser.add_argument('--crc', nargs=0, type=bool, action=CustomAction, help='Print CRC32 for captured frames')
>> +    parser.add_argument('--save-frames', nargs=0, type=bool, action=CustomAction, help='Save captured frames to files')
>> +    parser.add_argument('--metadata', nargs=0, type=bool, action=CustomAction, help='Print the metadata for completed requests')
>> +    parser.add_argument('--strict-formats', type=bool, nargs=0, action=CustomAction, help='Do not allow requested stream format(s) to be adjusted')
>> +    parser.add_argument('-s', '--stream', nargs='+', action=CustomAction)
>> +    args = parser.parse_args()
>> +
>> +    cm = libcam.CameraManager.singleton()
>> +
>> +    if args.list:
>> +        do_cmd_list(cm)
>> +
>> +    contexts = []
>> +
>> +    for cam_idx in args.camera:
>> +        camera = next((c for i, c in enumerate(cm.cameras) if i + 1 == cam_idx), None)
>> +
>> +        if camera is None:
>> +            print('Unable to find camera', cam_idx)
>> +            return -1
>> +
>> +        contexts.append({
>> +                        'camera': camera,
>> +                        'idx': cam_idx,
>> +                        'id': 'cam' + str(cam_idx),
>> +                        'reqs-queued': 0,
>> +                        'reqs-completed': 0,
>> +                        'opt-capture': args.capture.get(cam_idx, False),
>> +                        'opt-crc': args.crc.get(cam_idx, False),
>> +                        'opt-save-frames': args.save_frames.get(cam_idx, False),
>> +                        'opt-metadata': args.metadata.get(cam_idx, False),
>> +                        'opt-strict-formats': args.strict_formats.get(cam_idx, False),
>> +                        'opt-stream': args.stream.get(cam_idx, ['role=viewfinder']),
>> +                        })
>> +
>> +    for ctx in contexts:
>> +        print('Using camera {} as {}'.format(ctx['camera'].id, ctx['id']))
>> +
>> +    for ctx in contexts:
>> +        if args.list_properties:
>> +            do_cmd_list_props(ctx)
>> +        if args.list_controls:
>> +            do_cmd_list_controls(ctx)
>> +        if args.info:
>> +            do_cmd_info(ctx)
>> +
>> +    if args.capture:
>> +
>> +        state = {
>> +            'cm': cm,
>> +            'contexts': contexts,
>> +            'event_handler': event_handler,
>> +            'request_prcessed': request_prcessed,
>> +        }
> 
> Same here, a state class would be nice.
> 
>> +
>> +        if args.renderer == 'null':
>> +            import cam_null
>> +            renderer = cam_null.NullRenderer(state)
>> +        elif args.renderer == 'kms':
>> +            import cam_kms
>> +            renderer = cam_kms.KMSRenderer(state)
>> +        elif args.renderer == 'qt':
>> +            import cam_qt
>> +            renderer = cam_qt.QtRenderer(state)
>> +        elif args.renderer == 'qtgl':
>> +            import cam_qtgl
>> +            renderer = cam_qtgl.QtRenderer(state)
>> +        else:
>> +            print('Bad renderer', args.renderer)
>> +            return -1
>> +
>> +        state['renderer'] = renderer
>> +
>> +        do_cmd_capture(state)
>> +
>> +    return 0
>> +
>> +
>> +if __name__ == '__main__':
>> +    sys.exit(main())
>> diff --git a/src/py/cam/cam_kms.py b/src/py/cam/cam_kms.py
>> new file mode 100644
>> index 00000000..f4ee5a06
>> --- /dev/null
>> +++ b/src/py/cam/cam_kms.py
>> @@ -0,0 +1,183 @@
>> +# SPDX-License-Identifier: GPL-2.0-or-later
>> +# Copyright (C) 2021, Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
>> +
>> +import pykms
> 
> Why am I not surprised ? :-D
> 
>> +import selectors
>> +import sys
>> +
>> +FMT_MAP = {
>> +    'RGB888': pykms.PixelFormat.RGB888,
>> +    'YUYV': pykms.PixelFormat.YUYV,
>> +    'ARGB8888': pykms.PixelFormat.ARGB8888,
>> +    'XRGB8888': pykms.PixelFormat.XRGB8888,
>> +}
>> +
>> +
>> +class KMSRenderer:
>> +    def __init__(self, state):
>> +        self.state = state
>> +
>> +        self.cm = state['cm']
>> +        self.contexts = state['contexts']
>> +        self.running = False
>> +
>> +        card = pykms.Card()
>> +
>> +        res = pykms.ResourceManager(card)
>> +        conn = res.reserve_connector()
>> +        crtc = res.reserve_crtc(conn)
>> +        mode = conn.get_default_mode()
>> +        modeb = mode.to_blob(card)
>> +
>> +        req = pykms.AtomicReq(card)
>> +        req.add_connector(conn, crtc)
>> +        req.add_crtc(crtc, modeb)
>> +        r = req.commit_sync(allow_modeset=True)
>> +        assert(r == 0)
>> +
>> +        self.card = card
>> +        self.resman = res
>> +        self.crtc = crtc
>> +        self.mode = mode
>> +
>> +        self.bufqueue = []
>> +        self.current = None
>> +        self.next = None
>> +        self.cam_2_drm = {}
>> +
>> +    # KMS
>> +
>> +    def close(self):
>> +        req = pykms.AtomicReq(self.card)
>> +        for s in self.streams:
>> +            req.add_plane(s['plane'], None, None, dst=(0, 0, 0, 0))
>> +        req.commit()
>> +
>> +    def add_plane(self, req, stream, fb):
>> +        s = next(s for s in self.streams if s['stream'] == stream)
>> +        idx = s['idx']
>> +        plane = s['plane']
>> +
>> +        if idx % 2 == 0:
>> +            x = 0
>> +        else:
>> +            x = self.mode.hdisplay - fb.width
>> +
>> +        if idx // 2 == 0:
>> +            y = 0
>> +        else:
>> +            y = self.mode.vdisplay - fb.height
>> +
>> +        req.add_plane(plane, fb, self.crtc, dst=(x, y, fb.width, fb.height))
>> +
>> +    def apply_request(self, drmreq):
>> +
>> +        buffers = drmreq['camreq'].buffers
>> +
>> +        for stream, fb in buffers.items():
>> +            drmfb = self.cam_2_drm.get(fb, None)
>> +
>> +            req = pykms.AtomicReq(self.card)
>> +            self.add_plane(req, stream, drmfb)
>> +            req.commit()
>> +
>> +    def handle_page_flip(self, frame, time):
>> +        old = self.current
>> +        self.current = self.next
>> +
>> +        if len(self.bufqueue) > 0:
>> +            self.next = self.bufqueue.pop(0)
>> +        else:
>> +            self.next = None
>> +
>> +        if self.next:
>> +            drmreq = self.next
>> +
>> +            self.apply_request(drmreq)
>> +
>> +        if old:
>> +            req = old['camreq']
>> +            ctx = old['camctx']
>> +            self.state['request_prcessed'](ctx, req)
>> +
>> +    def queue(self, drmreq):
>> +        if not self.next:
>> +            self.next = drmreq
>> +            self.apply_request(drmreq)
>> +        else:
>> +            self.bufqueue.append(drmreq)
>> +
>> +    # libcamera
>> +
>> +    def setup(self):
>> +        self.streams = []
>> +
>> +        idx = 0
>> +        for ctx in self.contexts:
>> +            for stream in ctx['streams']:
>> +
>> +                cfg = stream.configuration
>> +                fmt = cfg.pixel_format
>> +                fmt = FMT_MAP[fmt]
>> +
>> +                plane = self.resman.reserve_generic_plane(self.crtc, fmt)
>> +                assert(plane is not None)
>> +
>> +                self.streams.append({
>> +                    'idx': idx,
>> +                    'stream': stream,
>> +                    'plane': plane,
>> +                    'fmt': fmt,
>> +                    'size': cfg.size,
>> +                })
>> +
>> +                for fb in ctx['allocator'].buffers(stream):
>> +                    w, h = cfg.size
>> +                    stride = cfg.stride
>> +                    fd = fb.fd(0)
>> +                    drmfb = pykms.DmabufFramebuffer(self.card, w, h, fmt,
>> +                                                    [fd], [stride], [0])
>> +                    self.cam_2_drm[fb] = drmfb
>> +
>> +                idx += 1
>> +
>> +    def readdrm(self, fileobj):
>> +        for ev in self.card.read_events():
>> +            if ev.type == pykms.DrmEventType.FLIP_COMPLETE:
>> +                self.handle_page_flip(ev.seq, ev.time)
>> +
>> +    def readcam(self, fd):
>> +        self.running = self.state['event_handler'](self.state)
>> +
>> +    def readkey(self, fileobj):
>> +        sys.stdin.readline()
>> +        self.running = False
>> +
>> +    def run(self):
>> +        print('Capturing...')
>> +
>> +        self.running = True
>> +
>> +        sel = selectors.DefaultSelector()
>> +        sel.register(self.card.fd, selectors.EVENT_READ, self.readdrm)
>> +        sel.register(self.cm.efd, selectors.EVENT_READ, self.readcam)
>> +        sel.register(sys.stdin, selectors.EVENT_READ, self.readkey)
>> +
>> +        print('Press enter to exit')
>> +
>> +        while self.running:
>> +            events = sel.select()
>> +            for key, mask in events:
>> +                callback = key.data
>> +                callback(key.fileobj)
>> +
>> +        print('Exiting...')
>> +
>> +    def request_handler(self, ctx, req):
>> +
>> +        drmreq = {
>> +            'camctx': ctx,
>> +            'camreq': req,
>> +        }
>> +
>> +        self.queue(drmreq)
>> diff --git a/src/py/cam/cam_null.py b/src/py/cam/cam_null.py
>> new file mode 100644
>> index 00000000..391397f6
>> --- /dev/null
>> +++ b/src/py/cam/cam_null.py
>> @@ -0,0 +1,47 @@
>> +# SPDX-License-Identifier: GPL-2.0-or-later
>> +# Copyright (C) 2021, Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
>> +
>> +import selectors
>> +import sys
>> +
>> +
>> +class NullRenderer:
>> +    def __init__(self, state):
>> +        self.state = state
>> +
>> +        self.cm = state['cm']
>> +        self.contexts = state['contexts']
>> +
>> +        self.running = False
>> +
>> +    def setup(self):
>> +        pass
>> +
>> +    def run(self):
>> +        print('Capturing...')
>> +
>> +        self.running = True
>> +
>> +        sel = selectors.DefaultSelector()
>> +        sel.register(self.cm.efd, selectors.EVENT_READ, self.readcam)
>> +        sel.register(sys.stdin, selectors.EVENT_READ, self.readkey)
>> +
>> +        print('Press enter to exit')
>> +
>> +        while self.running:
>> +            events = sel.select()
>> +            for key, mask in events:
>> +                callback = key.data
>> +                callback(key.fileobj)
>> +
>> +        print('Exiting...')
>> +
>> +    def readcam(self, fd):
>> +        self.running = self.state['event_handler'](self.state)
>> +
>> +    def readkey(self, fileobj):
>> +        sys.stdin.readline()
>> +        self.running = False
>> +
>> +    def request_handler(self, ctx, req):
>> +        self.state['request_prcessed'](ctx, req)
>> diff --git a/src/py/cam/cam_qt.py b/src/py/cam/cam_qt.py
>> new file mode 100644
>> index 00000000..40044866
>> --- /dev/null
>> +++ b/src/py/cam/cam_qt.py
>> @@ -0,0 +1,354 @@
>> +# SPDX-License-Identifier: GPL-2.0-or-later
>> +# Copyright (C) 2021, Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
>> +#
>> +# Debayering code from PiCamera documentation
>> +
>> +from io import BytesIO
>> +from numpy.lib.stride_tricks import as_strided
>> +from PIL import Image
>> +from PIL.ImageQt import ImageQt
>> +from PyQt5 import QtCore, QtGui, QtWidgets
>> +import numpy as np
>> +import sys
>> +
>> +
>> +def rgb_to_pix(rgb):
>> +    img = Image.frombuffer('RGB', (rgb.shape[1], rgb.shape[0]), rgb)
>> +    qim = ImageQt(img).copy()
>> +    pix = QtGui.QPixmap.fromImage(qim)
>> +    return pix
>> +
>> +
>> +def separate_components(data, r0, g0, g1, b0):
>> +    # Now to split the data up into its red, green, and blue components. The
>> +    # Bayer pattern of the OV5647 sensor is BGGR. In other words the first
>> +    # row contains alternating green/blue elements, the second row contains
>> +    # alternating red/green elements, and so on as illustrated below:
>> +    #
>> +    # GBGBGBGBGBGBGB
>> +    # RGRGRGRGRGRGRG
>> +    # GBGBGBGBGBGBGB
>> +    # RGRGRGRGRGRGRG
>> +    #
>> +    # Please note that if you use vflip or hflip to change the orientation
>> +    # of the capture, you must flip the Bayer pattern accordingly
>> +
>> +    rgb = np.zeros(data.shape + (3,), dtype=data.dtype)
>> +    rgb[r0[1]::2, r0[0]::2, 0] = data[r0[1]::2, r0[0]::2]  # Red
>> +    rgb[g0[1]::2, g0[0]::2, 1] = data[g0[1]::2, g0[0]::2]  # Green
>> +    rgb[g1[1]::2, g1[0]::2, 1] = data[g1[1]::2, g1[0]::2]  # Green
>> +    rgb[b0[1]::2, b0[0]::2, 2] = data[b0[1]::2, b0[0]::2]  # Blue
>> +
>> +    return rgb
>> +
>> +
>> +def demosaic(rgb, r0, g0, g1, b0):
>> +    # At this point we now have the raw Bayer data with the correct values
>> +    # and colors but the data still requires de-mosaicing and
>> +    # post-processing. If you wish to do this yourself, end the script here!
>> +    #
>> +    # Below we present a fairly naive de-mosaic method that simply
>> +    # calculates the weighted average of a pixel based on the pixels
>> +    # surrounding it. The weighting is provided b0[1] a b0[1]te representation of
>> +    # the Bayer filter which we construct first:
>> +
>> +    bayer = np.zeros(rgb.shape, dtype=np.uint8)
>> +    bayer[r0[1]::2, r0[0]::2, 0] = 1  # Red
>> +    bayer[g0[1]::2, g0[0]::2, 1] = 1  # Green
>> +    bayer[g1[1]::2, g1[0]::2, 1] = 1  # Green
>> +    bayer[b0[1]::2, b0[0]::2, 2] = 1  # Blue
>> +
>> +    # Allocate an array to hold our output with the same shape as the input
>> +    # data. After this we define the size of window that will be used to
>> +    # calculate each weighted average (3x3). Then we pad out the rgb and
>> +    # bayer arrays, adding blank pixels at their edges to compensate for the
>> +    # size of the window when calculating averages for edge pixels.
>> +
>> +    output = np.empty(rgb.shape, dtype=rgb.dtype)
>> +    window = (3, 3)
>> +    borders = (window[0] - 1, window[1] - 1)
>> +    border = (borders[0] // 2, borders[1] // 2)
>> +
>> +    # rgb_pad = np.zeros((
>> +    #    rgb.shape[0] + borders[0],
>> +    #    rgb.shape[1] + borders[1],
>> +    #    rgb.shape[2]), dtype=rgb.dtype)
>> +    # rgb_pad[
>> +    #    border[0]:rgb_pad.shape[0] - border[0],
>> +    #    border[1]:rgb_pad.shape[1] - border[1],
>> +    #    :] = rgb
>> +    # rgb = rgb_pad
>> +    #
>> +    # bayer_pad = np.zeros((
>> +    #    bayer.shape[0] + borders[0],
>> +    #    bayer.shape[1] + borders[1],
>> +    #    bayer.shape[2]), dtype=bayer.dtype)
>> +    # bayer_pad[
>> +    #    border[0]:bayer_pad.shape[0] - border[0],
>> +    #    border[1]:bayer_pad.shape[1] - border[1],
>> +    #    :] = bayer
>> +    # bayer = bayer_pad
>> +
>> +    # In numpy >=1.7.0 just use np.pad (version in Raspbian is 1.6.2 at the
>> +    # time of writing...)
>> +    #
>> +    rgb = np.pad(rgb, [
>> +        (border[0], border[0]),
>> +        (border[1], border[1]),
>> +        (0, 0),
>> +    ], 'constant')
>> +    bayer = np.pad(bayer, [
>> +        (border[0], border[0]),
>> +        (border[1], border[1]),
>> +        (0, 0),
>> +    ], 'constant')
>> +
>> +    # For each plane in the RGB data, we use a nifty numpy trick
>> +    # (as_strided) to construct a view over the plane of 3x3 matrices. We do
>> +    # the same for the bayer array, then use Einstein summation on each
>> +    # (np.sum is simpler, but copies the data so it's slower), and divide
>> +    # the results to get our weighted average:
>> +
>> +    for plane in range(3):
>> +        p = rgb[..., plane]
>> +        b = bayer[..., plane]
>> +        pview = as_strided(p, shape=(
>> +            p.shape[0] - borders[0],
>> +            p.shape[1] - borders[1]) + window, strides=p.strides * 2)
>> +        bview = as_strided(b, shape=(
>> +            b.shape[0] - borders[0],
>> +            b.shape[1] - borders[1]) + window, strides=b.strides * 2)
>> +        psum = np.einsum('ijkl->ij', pview)
>> +        bsum = np.einsum('ijkl->ij', bview)
>> +        output[..., plane] = psum // bsum
>> +
>> +    return output
>> +
>> +
>> +def to_rgb(fmt, size, data):
>> +    w = size[0]
>> +    h = size[1]
>> +
>> +    if fmt == 'YUYV':
>> +        # YUV422
>> +        yuyv = data.reshape((h, w // 2 * 4))
>> +
>> +        # YUV444
>> +        yuv = np.empty((h, w, 3), dtype=np.uint8)
>> +        yuv[:, :, 0] = yuyv[:, 0::2]                    # Y
>> +        yuv[:, :, 1] = yuyv[:, 1::4].repeat(2, axis=1)  # U
>> +        yuv[:, :, 2] = yuyv[:, 3::4].repeat(2, axis=1)  # V
>> +
>> +        m = np.array([
>> +            [ 1.0, 1.0, 1.0],
>> +            [-0.000007154783816076815, -0.3441331386566162, 1.7720025777816772],
>> +            [ 1.4019975662231445, -0.7141380310058594 , 0.00001542569043522235]
>> +        ])
>> +
>> +        rgb = np.dot(yuv, m)
>> +        rgb[:, :, 0] -= 179.45477266423404
>> +        rgb[:, :, 1] += 135.45870971679688
>> +        rgb[:, :, 2] -= 226.8183044444304
>> +        rgb = rgb.astype(np.uint8)
>> +
>> +    elif fmt == 'RGB888':
>> +        rgb = data.reshape((h, w, 3))
>> +        rgb[:, :, [0, 1, 2]] = rgb[:, :, [2, 1, 0]]
>> +
>> +    elif fmt == 'BGR888':
>> +        rgb = data.reshape((h, w, 3))
>> +
>> +    elif fmt in ['ARGB8888', 'XRGB8888']:
>> +        rgb = data.reshape((h, w, 4))
>> +        rgb = np.flip(rgb, axis=2)
>> +        # drop alpha component
>> +        rgb = np.delete(rgb, np.s_[0::4], axis=2)
>> +
>> +    elif fmt.startswith('S'):
>> +        bayer_pattern = fmt[1:5]
>> +        bitspp = int(fmt[5:])
>> +
>> +        # TODO: shifting leaves the lowest bits 0
>> +        if bitspp == 8:
>> +            data = data.reshape((h, w))
>> +            data = data.astype(np.uint16) << 8
>> +        elif bitspp in [10, 12]:
>> +            data = data.view(np.uint16)
>> +            data = data.reshape((h, w))
>> +            data = data << (16 - bitspp)
>> +        else:
>> +            raise Exception('Bad bitspp:' + str(bitspp))
>> +
>> +        idx = bayer_pattern.find('R')
>> +        assert(idx != -1)
>> +        r0 = (idx % 2, idx // 2)
>> +
>> +        idx = bayer_pattern.find('G')
>> +        assert(idx != -1)
>> +        g0 = (idx % 2, idx // 2)
>> +
>> +        idx = bayer_pattern.find('G', idx + 1)
>> +        assert(idx != -1)
>> +        g1 = (idx % 2, idx // 2)
>> +
>> +        idx = bayer_pattern.find('B')
>> +        assert(idx != -1)
>> +        b0 = (idx % 2, idx // 2)
>> +
>> +        rgb = separate_components(data, r0, g0, g1, b0)
>> +        rgb = demosaic(rgb, r0, g0, g1, b0)
>> +        rgb = (rgb >> 8).astype(np.uint8)
>> +
>> +    else:
>> +        rgb = None
>> +
>> +    return rgb
> 
> I'd move those to a separate file.
> 
>> +
>> +
>> +class QtRenderer:
>> +    def __init__(self, state):
>> +        self.state = state
>> +
>> +        self.cm = state['cm']
>> +        self.contexts = state['contexts']
>> +
>> +    def setup(self):
>> +        self.app = QtWidgets.QApplication([])
>> +
>> +        windows = []
>> +
>> +        for ctx in self.contexts:
>> +            camera = ctx['camera']
>> +
>> +            for stream in ctx['streams']:
>> +                fmt = stream.configuration.pixel_format
>> +                size = stream.configuration.size
>> +
>> +                window = MainWindow(ctx, stream)
>> +                window.setAttribute(QtCore.Qt.WA_ShowWithoutActivating)
>> +                window.show()
>> +                windows.append(window)
>> +
>> +        self.windows = windows
>> +
>> +    def run(self):
>> +        camnotif = QtCore.QSocketNotifier(self.cm.efd, QtCore.QSocketNotifier.Read)
>> +        camnotif.activated.connect(lambda x: self.readcam())
>> +
>> +        keynotif = QtCore.QSocketNotifier(sys.stdin.fileno(), QtCore.QSocketNotifier.Read)
>> +        keynotif.activated.connect(lambda x: self.readkey())
>> +
>> +        print('Capturing...')
>> +
>> +        self.app.exec()
>> +
>> +        print('Exiting...')
>> +
>> +    def readcam(self):
>> +        running = self.state['event_handler'](self.state)
>> +
>> +        if not running:
>> +            self.app.quit()
>> +
>> +    def readkey(self):
>> +        sys.stdin.readline()
>> +        self.app.quit()
>> +
>> +    def request_handler(self, ctx, req):
>> +        buffers = req.buffers
>> +
>> +        for stream, fb in buffers.items():
>> +            wnd = next(wnd for wnd in self.windows if wnd.stream == stream)
>> +
>> +            wnd.handle_request(stream, fb)
>> +
>> +        self.state['request_prcessed'](ctx, req)
>> +
>> +    def cleanup(self):
>> +        for w in self.windows:
>> +            w.close()
>> +
>> +
>> +class MainWindow(QtWidgets.QWidget):
>> +    def __init__(self, ctx, stream):
>> +        super().__init__()
>> +
>> +        self.ctx = ctx
>> +        self.stream = stream
>> +
>> +        self.label = QtWidgets.QLabel()
>> +
>> +        windowLayout = QtWidgets.QHBoxLayout()
>> +        self.setLayout(windowLayout)
>> +
>> +        windowLayout.addWidget(self.label)
>> +
>> +        controlsLayout = QtWidgets.QVBoxLayout()
>> +        windowLayout.addLayout(controlsLayout)
>> +
>> +        windowLayout.addStretch()
>> +
>> +        group = QtWidgets.QGroupBox('Info')
>> +        groupLayout = QtWidgets.QVBoxLayout()
>> +        group.setLayout(groupLayout)
>> +        controlsLayout.addWidget(group)
>> +
>> +        lab = QtWidgets.QLabel(ctx['id'])
>> +        groupLayout.addWidget(lab)
>> +
>> +        self.frameLabel = QtWidgets.QLabel()
>> +        groupLayout.addWidget(self.frameLabel)
>> +
>> +        group = QtWidgets.QGroupBox('Properties')
>> +        groupLayout = QtWidgets.QVBoxLayout()
>> +        group.setLayout(groupLayout)
>> +        controlsLayout.addWidget(group)
>> +
>> +        camera = ctx['camera']
>> +
>> +        for k, v in camera.properties.items():
>> +            lab = QtWidgets.QLabel()
>> +            lab.setText(k + ' = ' + str(v))
>> +            groupLayout.addWidget(lab)
>> +
>> +        group = QtWidgets.QGroupBox('Controls')
>> +        groupLayout = QtWidgets.QVBoxLayout()
>> +        group.setLayout(groupLayout)
>> +        controlsLayout.addWidget(group)
>> +
>> +        for k, (min, max, default) in camera.controls.items():
>> +            lab = QtWidgets.QLabel()
>> +            lab.setText('{} = {}/{}/{}'.format(k, min, max, default))
>> +            groupLayout.addWidget(lab)
>> +
>> +        controlsLayout.addStretch()
>> +
>> +    def buf_to_qpixmap(self, stream, fb):
>> +        with fb.mmap() as mfb:
>> +            cfg = stream.configuration
>> +            w, h = cfg.size
>> +            pitch = cfg.stride
>> +
>> +            if cfg.pixel_format == 'MJPEG':
>> +                img = Image.open(BytesIO(mfb.planes[0]))
>> +                qim = ImageQt(img).copy()
>> +                pix = QtGui.QPixmap.fromImage(qim)
>> +            else:
>> +                data = np.array(mfb.planes[0], dtype=np.uint8)
>> +                rgb = to_rgb(cfg.pixel_format, cfg.size, data)
>> +
>> +                if rgb is None:
>> +                    raise Exception('Format not supported: ' + cfg.pixel_format)
>> +
>> +                pix = rgb_to_pix(rgb)
>> +
>> +        return pix
>> +
>> +    def handle_request(self, stream, fb):
>> +        ctx = self.ctx
>> +
>> +        pix = self.buf_to_qpixmap(stream, fb)
>> +        self.label.setPixmap(pix)
>> +
>> +        self.frameLabel.setText('Queued: {}\nDone: {}\nFps: {:.2f}'
>> +                       .format(ctx['reqs-queued'], ctx['reqs-completed'], ctx['fps']))
>> diff --git a/src/py/cam/cam_qtgl.py b/src/py/cam/cam_qtgl.py
>> new file mode 100644
>> index 00000000..37b74d3f
>> --- /dev/null
>> +++ b/src/py/cam/cam_qtgl.py
>> @@ -0,0 +1,385 @@
>> +# SPDX-License-Identifier: GPL-2.0-or-later
>> +# Copyright (C) 2021, Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
>> +
>> +from PyQt5 import QtCore, QtWidgets
>> +from PyQt5.QtCore import Qt
>> +
>> +import math
>> +import numpy as np
>> +import os
>> +import sys
>> +
>> +os.environ['PYOPENGL_PLATFORM'] = 'egl'
>> +
>> +import OpenGL
>> +# OpenGL.FULL_LOGGING = True
>> +
>> +from OpenGL import GL as gl
>> +from OpenGL.EGL.EXT.image_dma_buf_import import *
>> +from OpenGL.EGL.KHR.image import *
>> +from OpenGL.EGL.VERSION.EGL_1_0 import *
>> +from OpenGL.EGL.VERSION.EGL_1_2 import *
>> +from OpenGL.EGL.VERSION.EGL_1_3 import *
>> +
>> +from OpenGL.GLES2.OES.EGL_image import *
>> +from OpenGL.GLES2.OES.EGL_image_external import *
>> +from OpenGL.GLES2.VERSION.GLES2_2_0 import *
>> +from OpenGL.GLES3.VERSION.GLES3_3_0 import *
>> +
>> +from OpenGL.GL import shaders
>> +
>> +from gl_helpers import *
>> +
>> +# libcamera format string -> DRM fourcc
>> +FMT_MAP = {
>> +    'RGB888': 'RG24',
>> +    'XRGB8888': 'XR24',
>> +    'ARGB8888': 'AR24',
>> +    'YUYV': 'YUYV',
>> +}
>> +
>> +
>> +class EglState:
>> +    def __init__(self):
>> +        self.create_display()
>> +        self.choose_config()
>> +        self.create_context()
>> +        self.check_extensions()
>> +
>> +    def create_display(self):
>> +        xdpy = getEGLNativeDisplay()
>> +        dpy = eglGetDisplay(xdpy)
>> +        self.display = dpy
> 
> Could we use the Qt GL API to avoid managing the display manually ?

I had a lot of trouble getting something working with Python + GLES + 
EGL combination. And I think this only works on some platforms/drivers 
(if I recall right, Ubuntu + nvidia driver doesn't work, Ubuntu + 
nouveau works). Qt GL was one I tried.

I think the main issues were about the extensions to use a dmabuf 
directly, without copying. Which is the main reason for this whole exercise.

If someone gets this to work with Qt GL, that's great and I support it. 
But I personally am not going to spend more time on this, as this works 
(quite ok), and I have already gotten my share of gray hair on this...

Also, KMS + GLES + EGL would be nice to have too, and for that Qt GL is 
not an option.

>> +
>> +    def choose_config(self):
>> +        dpy = self.display
>> +
>> +        major, minor = EGLint(), EGLint()
>> +
>> +        b = eglInitialize(dpy, major, minor)
>> +        assert(b)
>> +
>> +        print('EGL {} {}'.format(
>> +              eglQueryString(dpy, EGL_VENDOR).decode(),
>> +              eglQueryString(dpy, EGL_VERSION).decode()))
>> +
>> +        check_egl_extensions(dpy, ['EGL_EXT_image_dma_buf_import'])
>> +
>> +        b = eglBindAPI(EGL_OPENGL_ES_API)
>> +        assert(b)
>> +
>> +        def print_config(dpy, cfg):
>> +
>> +            def _getconf(dpy, cfg, a):
>> +                value = ctypes.c_long()
>> +                eglGetConfigAttrib(dpy, cfg, a, value)
>> +                return value.value
>> +
>> +            getconf = lambda a: _getconf(dpy, cfg, a)
>> +
>> +            print('EGL Config {}: color buf {}/{}/{}/{} = {}, depth {}, stencil {}, native visualid {}, native visualtype {}'.format(
>> +                getconf(EGL_CONFIG_ID),
>> +                getconf(EGL_ALPHA_SIZE),
>> +                getconf(EGL_RED_SIZE),
>> +                getconf(EGL_GREEN_SIZE),
>> +                getconf(EGL_BLUE_SIZE),
>> +                getconf(EGL_BUFFER_SIZE),
>> +                getconf(EGL_DEPTH_SIZE),
>> +                getconf(EGL_STENCIL_SIZE),
>> +                getconf(EGL_NATIVE_VISUAL_ID),
>> +                getconf(EGL_NATIVE_VISUAL_TYPE)))
>> +
>> +        if False:
>> +            num_configs = ctypes.c_long()
>> +            eglGetConfigs(dpy, None, 0, num_configs)
>> +            print('{} configs'.format(num_configs.value))
>> +
>> +            configs = (EGLConfig * num_configs.value)()
>> +            eglGetConfigs(dpy, configs, num_configs.value, num_configs)
>> +            for config_id in configs:
>> +                print_config(dpy, config_id)
>> +
>> +        config_attribs = [
>> +            EGL_SURFACE_TYPE, EGL_WINDOW_BIT,
>> +            EGL_RED_SIZE, 8,
>> +            EGL_GREEN_SIZE, 8,
>> +            EGL_BLUE_SIZE, 8,
>> +            EGL_ALPHA_SIZE, 0,
>> +            EGL_RENDERABLE_TYPE, EGL_OPENGL_ES2_BIT,
>> +            EGL_NONE,
>> +        ]
>> +
>> +        n = EGLint()
>> +        configs = (EGLConfig * 1)()
>> +        b = eglChooseConfig(dpy, config_attribs, configs, 1, n)
>> +        assert(b and n.value == 1)
>> +        config = configs[0]
>> +
>> +        print('Chosen Config:')
>> +        print_config(dpy, config)
>> +
>> +        self.config = config
>> +
>> +    def create_context(self):
>> +        dpy = self.display
>> +
>> +        context_attribs = [
>> +            EGL_CONTEXT_CLIENT_VERSION, 2,
>> +            EGL_NONE,
>> +        ]
>> +
>> +        context = eglCreateContext(dpy, self.config, EGL_NO_CONTEXT, context_attribs)
>> +        assert(context)
>> +
>> +        b = eglMakeCurrent(dpy, EGL_NO_SURFACE, EGL_NO_SURFACE, context)
>> +        assert(b)
>> +
>> +        self.context = context
>> +
>> +    def check_extensions(self):
>> +        check_gl_extensions(['GL_OES_EGL_image'])
>> +
>> +        assert(eglCreateImageKHR)
>> +        assert(eglDestroyImageKHR)
>> +        assert(glEGLImageTargetTexture2DOES)
>> +
>> +
>> +class QtRenderer:
>> +    def __init__(self, state):
>> +        self.state = state
>> +
>> +    def setup(self):
>> +        self.app = QtWidgets.QApplication([])
>> +
>> +        window = MainWindow(self.state)
>> +        window.setAttribute(QtCore.Qt.WA_ShowWithoutActivating)
>> +        window.show()
>> +
>> +        self.window = window
>> +
>> +    def run(self):
>> +        camnotif = QtCore.QSocketNotifier(self.state['cm'].efd, QtCore.QSocketNotifier.Read)
>> +        camnotif.activated.connect(lambda x: self.readcam())
>> +
>> +        keynotif = QtCore.QSocketNotifier(sys.stdin.fileno(), QtCore.QSocketNotifier.Read)
>> +        keynotif.activated.connect(lambda x: self.readkey())
>> +
>> +        print('Capturing...')
>> +
>> +        self.app.exec()
>> +
>> +        print('Exiting...')
>> +
>> +    def readcam(self):
>> +        running = self.state['event_handler'](self.state)
>> +
>> +        if not running:
>> +            self.app.quit()
>> +
>> +    def readkey(self):
>> +        sys.stdin.readline()
>> +        self.app.quit()
>> +
>> +    def request_handler(self, ctx, req):
>> +        self.window.handle_request(ctx, req)
>> +
>> +    def cleanup(self):
>> +        self.window.close()
>> +
>> +
>> +class MainWindow(QtWidgets.QWidget):
>> +    def __init__(self, state):
>> +        super().__init__()
>> +
>> +        self.setAttribute(Qt.WA_PaintOnScreen)
>> +        self.setAttribute(Qt.WA_NativeWindow)
>> +
>> +        self.state = state
>> +
>> +        self.textures = {}
>> +        self.reqqueue = {}
>> +        self.current = {}
>> +
>> +        for ctx in self.state['contexts']:
>> +
>> +            self.reqqueue[ctx['idx']] = []
>> +            self.current[ctx['idx']] = []
>> +
>> +            for stream in ctx['streams']:
>> +                fmt = stream.configuration.pixel_format
>> +                size = stream.configuration.size
>> +
>> +                if fmt not in FMT_MAP:
>> +                    raise Exception('Unsupported pixel format: ' + str(fmt))
>> +
>> +                self.textures[stream] = None
>> +
>> +        num_tiles = len(self.textures)
>> +        self.num_columns = math.ceil(math.sqrt(num_tiles))
>> +        self.num_rows = math.ceil(num_tiles / self.num_columns)
>> +
>> +        self.egl = EglState()
>> +
>> +        self.surface = None
>> +
>> +    def paintEngine(self):
>> +        return None
>> +
>> +    def create_surface(self):
>> +        native_surface = c_void_p(self.winId().__int__())
>> +        surface = eglCreateWindowSurface(self.egl.display, self.egl.config,
>> +                                         native_surface, None)
>> +
>> +        b = eglMakeCurrent(self.egl.display, self.surface, self.surface, self.egl.context)
>> +        assert(b)
>> +
>> +        self.surface = surface
>> +
>> +    def init_gl(self):
>> +        self.create_surface()
>> +
>> +        vertShaderSrc = '''
>> +            attribute vec2 aPosition;
>> +            varying vec2 texcoord;
>> +
>> +            void main()
>> +            {
>> +                gl_Position = vec4(aPosition * 2.0 - 1.0, 0.0, 1.0);
>> +                texcoord.x = aPosition.x;
>> +                texcoord.y = 1.0 - aPosition.y;
>> +            }
>> +        '''
>> +        fragShaderSrc = '''
>> +            #extension GL_OES_EGL_image_external : enable
>> +            precision mediump float;
>> +            varying vec2 texcoord;
>> +            uniform samplerExternalOES texture;
>> +
>> +            void main()
>> +            {
>> +                gl_FragColor = texture2D(texture, texcoord);
>> +            }
>> +        '''
>> +
>> +        program = shaders.compileProgram(
>> +            shaders.compileShader(vertShaderSrc, GL_VERTEX_SHADER),
>> +            shaders.compileShader(fragShaderSrc, GL_FRAGMENT_SHADER)
>> +        )
>> +
>> +        glUseProgram(program)
>> +
>> +        glClearColor(0.5, 0.8, 0.7, 1.0)
>> +
>> +        vertPositions = [
>> +            0.0, 0.0,
>> +            1.0, 0.0,
>> +            1.0, 1.0,
>> +            0.0, 1.0
>> +        ]
>> +
>> +        inputAttrib = glGetAttribLocation(program, 'aPosition')
>> +        glVertexAttribPointer(inputAttrib, 2, GL_FLOAT, GL_FALSE, 0, vertPositions)
>> +        glEnableVertexAttribArray(inputAttrib)
>> +
>> +    def create_texture(self, stream, fb):
>> +        cfg = stream.configuration
>> +        fmt = cfg.pixel_format
>> +        fmt = str_to_fourcc(FMT_MAP[fmt])
>> +        w, h = cfg.size
>> +
>> +        attribs = [
>> +            EGL_WIDTH, w,
>> +            EGL_HEIGHT, h,
>> +            EGL_LINUX_DRM_FOURCC_EXT, fmt,
>> +            EGL_DMA_BUF_PLANE0_FD_EXT, fb.fd(0),
>> +            EGL_DMA_BUF_PLANE0_OFFSET_EXT, 0,
>> +            EGL_DMA_BUF_PLANE0_PITCH_EXT, cfg.stride,
>> +            EGL_NONE,
>> +        ]
>> +
>> +        image = eglCreateImageKHR(self.egl.display,
>> +                                  EGL_NO_CONTEXT,
>> +                                  EGL_LINUX_DMA_BUF_EXT,
>> +                                  None,
>> +                                  attribs)
>> +        assert(image)
>> +
>> +        textures = glGenTextures(1)
>> +        glBindTexture(GL_TEXTURE_EXTERNAL_OES, textures)
>> +        glTexParameteri(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_MAG_FILTER, GL_LINEAR)
>> +        glTexParameteri(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_MIN_FILTER, GL_LINEAR)
>> +        glTexParameteri(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE)
>> +        glTexParameteri(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE)
>> +        glEGLImageTargetTexture2DOES(GL_TEXTURE_EXTERNAL_OES, image)
>> +
>> +        return textures
>> +
>> +    def resizeEvent(self, event):
>> +        size = event.size()
>> +
>> +        print('Resize', size)
>> +
>> +        super().resizeEvent(event)
>> +
>> +        if self.surface is None:
>> +            return
>> +
>> +        glViewport(0, 0, size.width() // 2, size.height())
>> +
>> +    def paintEvent(self, event):
>> +        if self.surface is None:
>> +            self.init_gl()
>> +
>> +        for ctx_idx, queue in self.reqqueue.items():
>> +            if len(queue) == 0:
>> +                continue
>> +
>> +            ctx = next(ctx for ctx in self.state['contexts'] if ctx['idx'] == ctx_idx)
>> +
>> +            if self.current[ctx_idx]:
>> +                old = self.current[ctx_idx]
>> +                self.current[ctx_idx] = None
>> +                self.state['request_prcessed'](ctx, old)
>> +
>> +            next_req = queue.pop(0)
>> +            self.current[ctx_idx] = next_req
>> +
>> +            stream, fb = next(iter(next_req.buffers.items()))
>> +
>> +            self.textures[stream] = self.create_texture(stream, fb)
>> +
>> +        self.paint_gl()
>> +
>> +    def paint_gl(self):
>> +        b = eglMakeCurrent(self.egl.display, self.surface, self.surface, self.egl.context)
>> +        assert(b)
>> +
>> +        glClear(GL_COLOR_BUFFER_BIT)
>> +
>> +        size = self.size()
>> +
>> +        for idx, ctx in enumerate(self.state['contexts']):
>> +            for stream in ctx['streams']:
>> +                if self.textures[stream] is None:
>> +                    continue
>> +
>> +                w = size.width() // self.num_columns
>> +                h = size.height() // self.num_rows
>> +
>> +                x = idx % self.num_columns
>> +                y = idx // self.num_columns
>> +
>> +                x *= w
>> +                y *= h
>> +
>> +                glViewport(x, y, w, h)
>> +
>> +                glBindTexture(GL_TEXTURE_EXTERNAL_OES, self.textures[stream])
>> +                glDrawArrays(GL_TRIANGLE_FAN, 0, 4)
>> +
>> +        b = eglSwapBuffers(self.egl.display, self.surface)
>> +        assert(b)
>> +
>> +    def handle_request(self, ctx, req):
>> +        self.reqqueue[ctx['idx']].append(req)
>> +        self.update()
>> diff --git a/src/py/cam/gl_helpers.py b/src/py/cam/gl_helpers.py
>> new file mode 100644
>> index 00000000..b377d735
>> --- /dev/null
>> +++ b/src/py/cam/gl_helpers.py
>> @@ -0,0 +1,74 @@
>> +# SPDX-License-Identifier: GPL-2.0-or-later
>> +# Copyright (C) 2021, Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
>> +
>> +from OpenGL.EGL.VERSION.EGL_1_0 import EGLNativeDisplayType, eglGetProcAddress, eglQueryString, EGL_EXTENSIONS
>> +
>> +from OpenGL.raw.GLES2 import _types as _cs
>> +from OpenGL.GLES2.VERSION.GLES2_2_0 import *
>> +from OpenGL.GLES3.VERSION.GLES3_3_0 import *
>> +from OpenGL import GL as gl
>> +
>> +from ctypes import c_int, c_char_p, c_void_p, cdll, POINTER, util, \
>> +    pointer, CFUNCTYPE, c_bool
>> +
>> +
>> +def getEGLNativeDisplay():
>> +    _x11lib = cdll.LoadLibrary(util.find_library('X11'))
>> +    XOpenDisplay = _x11lib.XOpenDisplay
>> +    XOpenDisplay.argtypes = [c_char_p]
>> +    XOpenDisplay.restype = POINTER(EGLNativeDisplayType)
>> +
>> +    xdpy = XOpenDisplay(None)
> 
> Did you mean
> 
>      return XOpenDisplay(None)

Yes... It's funny it still works =)

>> +
>> +
>> +# Hack. PyOpenGL doesn't seem to manage to find glEGLImageTargetTexture2DOES.
>> +def getglEGLImageTargetTexture2DOES():
>> +    funcptr = eglGetProcAddress('glEGLImageTargetTexture2DOES')
>> +    prototype = CFUNCTYPE(None, _cs.GLenum, _cs.GLeglImageOES)
>> +    return prototype(funcptr)
>> +
>> +
>> +glEGLImageTargetTexture2DOES = getglEGLImageTargetTexture2DOES()
>> +
>> +
>> +def str_to_fourcc(str):
>> +    assert(len(str) == 4)
>> +    fourcc = 0
>> +    for i, v in enumerate([ord(c) for c in str]):
>> +        fourcc |= v << (i * 8)
>> +    return fourcc
> 
> Is retrieving the numerical 4CC something that should be part of the
> PixelFormat bindings ?

If it's part of libcamera, yes. But I think we should keep the bindings 
to reflect libcamera as far as possible. Utility features can be built 
to a pure python library on top.

  Tomi
Laurent Pinchart May 7, 2022, 11:56 a.m. UTC | #3
Hi Tomi,

On Sat, May 07, 2022 at 10:57:37AM +0300, Tomi Valkeinen wrote:
> On 06/05/2022 22:49, Laurent Pinchart wrote:
> > On Fri, May 06, 2022 at 05:54:14PM +0300, Tomi Valkeinen wrote:
> >> Add cam.py, which mimics the 'cam' tool. Four rendering backends are
> >> added:
> >>
> >> * null - Do nothing
> >> * kms - Use KMS with dmabufs
> >> * qt - SW render on a Qt window
> >> * qtgl - OpenGL render on a Qt window
> >>
> >> All the renderers handle only a few pixel formats, and especially the GL
> >> renderer is just a prototype.
> > 
> > I don't review this in much details as you mentioned it was a proof of
> > concept. We can improve things later, and high-level comments below
> > don't all need to (but can :-)) be addressed in a new version.
> > 
> >> Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
> >> ---
> >>   src/py/cam/cam.py        | 484 +++++++++++++++++++++++++++++++++++++++
> >>   src/py/cam/cam_kms.py    | 183 +++++++++++++++
> >>   src/py/cam/cam_null.py   |  47 ++++
> >>   src/py/cam/cam_qt.py     | 354 ++++++++++++++++++++++++++++
> >>   src/py/cam/cam_qtgl.py   | 385 +++++++++++++++++++++++++++++++
> >>   src/py/cam/gl_helpers.py |  74 ++++++
> >>   6 files changed, 1527 insertions(+)
> >>   create mode 100755 src/py/cam/cam.py
> >>   create mode 100644 src/py/cam/cam_kms.py
> >>   create mode 100644 src/py/cam/cam_null.py
> >>   create mode 100644 src/py/cam/cam_qt.py
> >>   create mode 100644 src/py/cam/cam_qtgl.py
> >>   create mode 100644 src/py/cam/gl_helpers.py

[snip]

> >> diff --git a/src/py/cam/cam_qtgl.py b/src/py/cam/cam_qtgl.py
> >> new file mode 100644
> >> index 00000000..37b74d3f
> >> --- /dev/null
> >> +++ b/src/py/cam/cam_qtgl.py
> >> @@ -0,0 +1,385 @@
> >> +# SPDX-License-Identifier: GPL-2.0-or-later
> >> +# Copyright (C) 2021, Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
> >> +
> >> +from PyQt5 import QtCore, QtWidgets
> >> +from PyQt5.QtCore import Qt
> >> +
> >> +import math
> >> +import numpy as np
> >> +import os
> >> +import sys
> >> +
> >> +os.environ['PYOPENGL_PLATFORM'] = 'egl'
> >> +
> >> +import OpenGL
> >> +# OpenGL.FULL_LOGGING = True
> >> +
> >> +from OpenGL import GL as gl
> >> +from OpenGL.EGL.EXT.image_dma_buf_import import *
> >> +from OpenGL.EGL.KHR.image import *
> >> +from OpenGL.EGL.VERSION.EGL_1_0 import *
> >> +from OpenGL.EGL.VERSION.EGL_1_2 import *
> >> +from OpenGL.EGL.VERSION.EGL_1_3 import *
> >> +
> >> +from OpenGL.GLES2.OES.EGL_image import *
> >> +from OpenGL.GLES2.OES.EGL_image_external import *
> >> +from OpenGL.GLES2.VERSION.GLES2_2_0 import *
> >> +from OpenGL.GLES3.VERSION.GLES3_3_0 import *
> >> +
> >> +from OpenGL.GL import shaders
> >> +
> >> +from gl_helpers import *
> >> +
> >> +# libcamera format string -> DRM fourcc
> >> +FMT_MAP = {
> >> +    'RGB888': 'RG24',
> >> +    'XRGB8888': 'XR24',
> >> +    'ARGB8888': 'AR24',
> >> +    'YUYV': 'YUYV',
> >> +}
> >> +
> >> +
> >> +class EglState:
> >> +    def __init__(self):
> >> +        self.create_display()
> >> +        self.choose_config()
> >> +        self.create_context()
> >> +        self.check_extensions()
> >> +
> >> +    def create_display(self):
> >> +        xdpy = getEGLNativeDisplay()
> >> +        dpy = eglGetDisplay(xdpy)
> >> +        self.display = dpy
> > 
> > Could we use the Qt GL API to avoid managing the display manually ?
> 
> I had a lot of trouble getting something working with Python + GLES + 
> EGL combination. And I think this only works on some platforms/drivers 
> (if I recall right, Ubuntu + nvidia driver doesn't work, Ubuntu + 
> nouveau works). Qt GL was one I tried.
> 
> I think the main issues were about the extensions to use a dmabuf 
> directly, without copying. Which is the main reason for this whole exercise.
> 
> If someone gets this to work with Qt GL, that's great and I support it. 
> But I personally am not going to spend more time on this, as this works 
> (quite ok), and I have already gotten my share of gray hair on this...
> 
> Also, KMS + GLES + EGL would be nice to have too, and for that Qt GL is 
> not an option.

Agreed, let's leave it as-is. It would be nice to split the GL part out
in a separate class later. Candidates for improvements would be
integration with other windowing systems (I'm thinking of Wayland, but
also pure KMS + GLES). Qt should make that transparent, but for plain
KMS we'll need something else.

> >> +
> >> +    def choose_config(self):
> >> +        dpy = self.display
> >> +
> >> +        major, minor = EGLint(), EGLint()
> >> +
> >> +        b = eglInitialize(dpy, major, minor)
> >> +        assert(b)
> >> +
> >> +        print('EGL {} {}'.format(
> >> +              eglQueryString(dpy, EGL_VENDOR).decode(),
> >> +              eglQueryString(dpy, EGL_VERSION).decode()))
> >> +
> >> +        check_egl_extensions(dpy, ['EGL_EXT_image_dma_buf_import'])
> >> +
> >> +        b = eglBindAPI(EGL_OPENGL_ES_API)
> >> +        assert(b)
> >> +
> >> +        def print_config(dpy, cfg):
> >> +
> >> +            def _getconf(dpy, cfg, a):
> >> +                value = ctypes.c_long()
> >> +                eglGetConfigAttrib(dpy, cfg, a, value)
> >> +                return value.value
> >> +
> >> +            getconf = lambda a: _getconf(dpy, cfg, a)
> >> +
> >> +            print('EGL Config {}: color buf {}/{}/{}/{} = {}, depth {}, stencil {}, native visualid {}, native visualtype {}'.format(
> >> +                getconf(EGL_CONFIG_ID),
> >> +                getconf(EGL_ALPHA_SIZE),
> >> +                getconf(EGL_RED_SIZE),
> >> +                getconf(EGL_GREEN_SIZE),
> >> +                getconf(EGL_BLUE_SIZE),
> >> +                getconf(EGL_BUFFER_SIZE),
> >> +                getconf(EGL_DEPTH_SIZE),
> >> +                getconf(EGL_STENCIL_SIZE),
> >> +                getconf(EGL_NATIVE_VISUAL_ID),
> >> +                getconf(EGL_NATIVE_VISUAL_TYPE)))
> >> +
> >> +        if False:
> >> +            num_configs = ctypes.c_long()
> >> +            eglGetConfigs(dpy, None, 0, num_configs)
> >> +            print('{} configs'.format(num_configs.value))
> >> +
> >> +            configs = (EGLConfig * num_configs.value)()
> >> +            eglGetConfigs(dpy, configs, num_configs.value, num_configs)
> >> +            for config_id in configs:
> >> +                print_config(dpy, config_id)
> >> +
> >> +        config_attribs = [
> >> +            EGL_SURFACE_TYPE, EGL_WINDOW_BIT,
> >> +            EGL_RED_SIZE, 8,
> >> +            EGL_GREEN_SIZE, 8,
> >> +            EGL_BLUE_SIZE, 8,
> >> +            EGL_ALPHA_SIZE, 0,
> >> +            EGL_RENDERABLE_TYPE, EGL_OPENGL_ES2_BIT,
> >> +            EGL_NONE,
> >> +        ]
> >> +
> >> +        n = EGLint()
> >> +        configs = (EGLConfig * 1)()
> >> +        b = eglChooseConfig(dpy, config_attribs, configs, 1, n)
> >> +        assert(b and n.value == 1)
> >> +        config = configs[0]
> >> +
> >> +        print('Chosen Config:')
> >> +        print_config(dpy, config)
> >> +
> >> +        self.config = config
> >> +
> >> +    def create_context(self):
> >> +        dpy = self.display
> >> +
> >> +        context_attribs = [
> >> +            EGL_CONTEXT_CLIENT_VERSION, 2,
> >> +            EGL_NONE,
> >> +        ]
> >> +
> >> +        context = eglCreateContext(dpy, self.config, EGL_NO_CONTEXT, context_attribs)
> >> +        assert(context)
> >> +
> >> +        b = eglMakeCurrent(dpy, EGL_NO_SURFACE, EGL_NO_SURFACE, context)
> >> +        assert(b)
> >> +
> >> +        self.context = context
> >> +
> >> +    def check_extensions(self):
> >> +        check_gl_extensions(['GL_OES_EGL_image'])
> >> +
> >> +        assert(eglCreateImageKHR)
> >> +        assert(eglDestroyImageKHR)
> >> +        assert(glEGLImageTargetTexture2DOES)
> >> +
> >> +
> >> +class QtRenderer:
> >> +    def __init__(self, state):
> >> +        self.state = state
> >> +
> >> +    def setup(self):
> >> +        self.app = QtWidgets.QApplication([])
> >> +
> >> +        window = MainWindow(self.state)
> >> +        window.setAttribute(QtCore.Qt.WA_ShowWithoutActivating)
> >> +        window.show()
> >> +
> >> +        self.window = window
> >> +
> >> +    def run(self):
> >> +        camnotif = QtCore.QSocketNotifier(self.state['cm'].efd, QtCore.QSocketNotifier.Read)
> >> +        camnotif.activated.connect(lambda x: self.readcam())
> >> +
> >> +        keynotif = QtCore.QSocketNotifier(sys.stdin.fileno(), QtCore.QSocketNotifier.Read)
> >> +        keynotif.activated.connect(lambda x: self.readkey())
> >> +
> >> +        print('Capturing...')
> >> +
> >> +        self.app.exec()
> >> +
> >> +        print('Exiting...')
> >> +
> >> +    def readcam(self):
> >> +        running = self.state['event_handler'](self.state)
> >> +
> >> +        if not running:
> >> +            self.app.quit()
> >> +
> >> +    def readkey(self):
> >> +        sys.stdin.readline()
> >> +        self.app.quit()
> >> +
> >> +    def request_handler(self, ctx, req):
> >> +        self.window.handle_request(ctx, req)
> >> +
> >> +    def cleanup(self):
> >> +        self.window.close()
> >> +
> >> +
> >> +class MainWindow(QtWidgets.QWidget):
> >> +    def __init__(self, state):
> >> +        super().__init__()
> >> +
> >> +        self.setAttribute(Qt.WA_PaintOnScreen)
> >> +        self.setAttribute(Qt.WA_NativeWindow)
> >> +
> >> +        self.state = state
> >> +
> >> +        self.textures = {}
> >> +        self.reqqueue = {}
> >> +        self.current = {}
> >> +
> >> +        for ctx in self.state['contexts']:
> >> +
> >> +            self.reqqueue[ctx['idx']] = []
> >> +            self.current[ctx['idx']] = []
> >> +
> >> +            for stream in ctx['streams']:
> >> +                fmt = stream.configuration.pixel_format
> >> +                size = stream.configuration.size
> >> +
> >> +                if fmt not in FMT_MAP:
> >> +                    raise Exception('Unsupported pixel format: ' + str(fmt))
> >> +
> >> +                self.textures[stream] = None
> >> +
> >> +        num_tiles = len(self.textures)
> >> +        self.num_columns = math.ceil(math.sqrt(num_tiles))
> >> +        self.num_rows = math.ceil(num_tiles / self.num_columns)
> >> +
> >> +        self.egl = EglState()
> >> +
> >> +        self.surface = None
> >> +
> >> +    def paintEngine(self):
> >> +        return None
> >> +
> >> +    def create_surface(self):
> >> +        native_surface = c_void_p(self.winId().__int__())
> >> +        surface = eglCreateWindowSurface(self.egl.display, self.egl.config,
> >> +                                         native_surface, None)
> >> +
> >> +        b = eglMakeCurrent(self.egl.display, self.surface, self.surface, self.egl.context)
> >> +        assert(b)
> >> +
> >> +        self.surface = surface
> >> +
> >> +    def init_gl(self):
> >> +        self.create_surface()
> >> +
> >> +        vertShaderSrc = '''
> >> +            attribute vec2 aPosition;
> >> +            varying vec2 texcoord;
> >> +
> >> +            void main()
> >> +            {
> >> +                gl_Position = vec4(aPosition * 2.0 - 1.0, 0.0, 1.0);
> >> +                texcoord.x = aPosition.x;
> >> +                texcoord.y = 1.0 - aPosition.y;
> >> +            }
> >> +        '''
> >> +        fragShaderSrc = '''
> >> +            #extension GL_OES_EGL_image_external : enable
> >> +            precision mediump float;
> >> +            varying vec2 texcoord;
> >> +            uniform samplerExternalOES texture;
> >> +
> >> +            void main()
> >> +            {
> >> +                gl_FragColor = texture2D(texture, texcoord);
> >> +            }
> >> +        '''
> >> +
> >> +        program = shaders.compileProgram(
> >> +            shaders.compileShader(vertShaderSrc, GL_VERTEX_SHADER),
> >> +            shaders.compileShader(fragShaderSrc, GL_FRAGMENT_SHADER)
> >> +        )
> >> +
> >> +        glUseProgram(program)
> >> +
> >> +        glClearColor(0.5, 0.8, 0.7, 1.0)
> >> +
> >> +        vertPositions = [
> >> +            0.0, 0.0,
> >> +            1.0, 0.0,
> >> +            1.0, 1.0,
> >> +            0.0, 1.0
> >> +        ]
> >> +
> >> +        inputAttrib = glGetAttribLocation(program, 'aPosition')
> >> +        glVertexAttribPointer(inputAttrib, 2, GL_FLOAT, GL_FALSE, 0, vertPositions)
> >> +        glEnableVertexAttribArray(inputAttrib)
> >> +
> >> +    def create_texture(self, stream, fb):
> >> +        cfg = stream.configuration
> >> +        fmt = cfg.pixel_format
> >> +        fmt = str_to_fourcc(FMT_MAP[fmt])
> >> +        w, h = cfg.size
> >> +
> >> +        attribs = [
> >> +            EGL_WIDTH, w,
> >> +            EGL_HEIGHT, h,
> >> +            EGL_LINUX_DRM_FOURCC_EXT, fmt,
> >> +            EGL_DMA_BUF_PLANE0_FD_EXT, fb.fd(0),
> >> +            EGL_DMA_BUF_PLANE0_OFFSET_EXT, 0,
> >> +            EGL_DMA_BUF_PLANE0_PITCH_EXT, cfg.stride,
> >> +            EGL_NONE,
> >> +        ]
> >> +
> >> +        image = eglCreateImageKHR(self.egl.display,
> >> +                                  EGL_NO_CONTEXT,
> >> +                                  EGL_LINUX_DMA_BUF_EXT,
> >> +                                  None,
> >> +                                  attribs)
> >> +        assert(image)
> >> +
> >> +        textures = glGenTextures(1)
> >> +        glBindTexture(GL_TEXTURE_EXTERNAL_OES, textures)
> >> +        glTexParameteri(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_MAG_FILTER, GL_LINEAR)
> >> +        glTexParameteri(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_MIN_FILTER, GL_LINEAR)
> >> +        glTexParameteri(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE)
> >> +        glTexParameteri(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE)
> >> +        glEGLImageTargetTexture2DOES(GL_TEXTURE_EXTERNAL_OES, image)
> >> +
> >> +        return textures
> >> +
> >> +    def resizeEvent(self, event):
> >> +        size = event.size()
> >> +
> >> +        print('Resize', size)
> >> +
> >> +        super().resizeEvent(event)
> >> +
> >> +        if self.surface is None:
> >> +            return
> >> +
> >> +        glViewport(0, 0, size.width() // 2, size.height())
> >> +
> >> +    def paintEvent(self, event):
> >> +        if self.surface is None:
> >> +            self.init_gl()
> >> +
> >> +        for ctx_idx, queue in self.reqqueue.items():
> >> +            if len(queue) == 0:
> >> +                continue
> >> +
> >> +            ctx = next(ctx for ctx in self.state['contexts'] if ctx['idx'] == ctx_idx)
> >> +
> >> +            if self.current[ctx_idx]:
> >> +                old = self.current[ctx_idx]
> >> +                self.current[ctx_idx] = None
> >> +                self.state['request_prcessed'](ctx, old)
> >> +
> >> +            next_req = queue.pop(0)
> >> +            self.current[ctx_idx] = next_req
> >> +
> >> +            stream, fb = next(iter(next_req.buffers.items()))
> >> +
> >> +            self.textures[stream] = self.create_texture(stream, fb)
> >> +
> >> +        self.paint_gl()
> >> +
> >> +    def paint_gl(self):
> >> +        b = eglMakeCurrent(self.egl.display, self.surface, self.surface, self.egl.context)
> >> +        assert(b)
> >> +
> >> +        glClear(GL_COLOR_BUFFER_BIT)
> >> +
> >> +        size = self.size()
> >> +
> >> +        for idx, ctx in enumerate(self.state['contexts']):
> >> +            for stream in ctx['streams']:
> >> +                if self.textures[stream] is None:
> >> +                    continue
> >> +
> >> +                w = size.width() // self.num_columns
> >> +                h = size.height() // self.num_rows
> >> +
> >> +                x = idx % self.num_columns
> >> +                y = idx // self.num_columns
> >> +
> >> +                x *= w
> >> +                y *= h
> >> +
> >> +                glViewport(x, y, w, h)
> >> +
> >> +                glBindTexture(GL_TEXTURE_EXTERNAL_OES, self.textures[stream])
> >> +                glDrawArrays(GL_TRIANGLE_FAN, 0, 4)
> >> +
> >> +        b = eglSwapBuffers(self.egl.display, self.surface)
> >> +        assert(b)
> >> +
> >> +    def handle_request(self, ctx, req):
> >> +        self.reqqueue[ctx['idx']].append(req)
> >> +        self.update()
> >> diff --git a/src/py/cam/gl_helpers.py b/src/py/cam/gl_helpers.py
> >> new file mode 100644
> >> index 00000000..b377d735
> >> --- /dev/null
> >> +++ b/src/py/cam/gl_helpers.py
> >> @@ -0,0 +1,74 @@
> >> +# SPDX-License-Identifier: GPL-2.0-or-later
> >> +# Copyright (C) 2021, Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
> >> +
> >> +from OpenGL.EGL.VERSION.EGL_1_0 import EGLNativeDisplayType, eglGetProcAddress, eglQueryString, EGL_EXTENSIONS
> >> +
> >> +from OpenGL.raw.GLES2 import _types as _cs
> >> +from OpenGL.GLES2.VERSION.GLES2_2_0 import *
> >> +from OpenGL.GLES3.VERSION.GLES3_3_0 import *
> >> +from OpenGL import GL as gl
> >> +
> >> +from ctypes import c_int, c_char_p, c_void_p, cdll, POINTER, util, \
> >> +    pointer, CFUNCTYPE, c_bool
> >> +
> >> +
> >> +def getEGLNativeDisplay():
> >> +    _x11lib = cdll.LoadLibrary(util.find_library('X11'))
> >> +    XOpenDisplay = _x11lib.XOpenDisplay
> >> +    XOpenDisplay.argtypes = [c_char_p]
> >> +    XOpenDisplay.restype = POINTER(EGLNativeDisplayType)
> >> +
> >> +    xdpy = XOpenDisplay(None)
> > 
> > Did you mean
> > 
> >      return XOpenDisplay(None)
> 
> Yes... It's funny it still works =)
> 
> >> +
> >> +
> >> +# Hack. PyOpenGL doesn't seem to manage to find glEGLImageTargetTexture2DOES.
> >> +def getglEGLImageTargetTexture2DOES():
> >> +    funcptr = eglGetProcAddress('glEGLImageTargetTexture2DOES')
> >> +    prototype = CFUNCTYPE(None, _cs.GLenum, _cs.GLeglImageOES)
> >> +    return prototype(funcptr)
> >> +
> >> +
> >> +glEGLImageTargetTexture2DOES = getglEGLImageTargetTexture2DOES()
> >> +
> >> +
> >> +def str_to_fourcc(str):
> >> +    assert(len(str) == 4)
> >> +    fourcc = 0
> >> +    for i, v in enumerate([ord(c) for c in str]):
> >> +        fourcc |= v << (i * 8)
> >> +    return fourcc
> > 
> > Is retrieving the numerical 4CC something that should be part of the
> > PixelFormat bindings ?
> 
> If it's part of libcamera, yes. But I think we should keep the bindings 
> to reflect libcamera as far as possible. Utility features can be built 
> to a pure python library on top.

It is part of libcamera :-) See PixelFormat::fourcc() and
PixelFormat::modifier().
Tomi Valkeinen May 9, 2022, 9:13 a.m. UTC | #4
On 07/05/2022 14:56, Laurent Pinchart wrote:

>>>> +def str_to_fourcc(str):
>>>> +    assert(len(str) == 4)
>>>> +    fourcc = 0
>>>> +    for i, v in enumerate([ord(c) for c in str]):
>>>> +        fourcc |= v << (i * 8)
>>>> +    return fourcc
>>>
>>> Is retrieving the numerical 4CC something that should be part of the
>>> PixelFormat bindings ?
>>
>> If it's part of libcamera, yes. But I think we should keep the bindings
>> to reflect libcamera as far as possible. Utility features can be built
>> to a pure python library on top.
> 
> It is part of libcamera :-) See PixelFormat::fourcc() and
> PixelFormat::modifier().

I'll leave the above as it is for now, but I have added a todo item to 
implement PixelFormat. At the moment the python bindings only use a 
fourcc string as the pixel format, and that's obviously not enough (but 
was quick to implement...).

  Tomi

Patch
diff mbox series

diff --git a/src/py/cam/cam.py b/src/py/cam/cam.py
new file mode 100755
index 00000000..5ea02850
--- /dev/null
+++ b/src/py/cam/cam.py
@@ -0,0 +1,484 @@ 
+#!/usr/bin/env python3
+
+# SPDX-License-Identifier: GPL-2.0-or-later
+# Copyright (C) 2021, Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
+
+import argparse
+import binascii
+import libcamera as libcam
+import os
+import sys
+
+
+class CustomCameraAction(argparse.Action):
+    def __call__(self, parser, namespace, values, option_string=None):
+        print(self.dest, values)
+
+        if 'camera' not in namespace or namespace.camera is None:
+            setattr(namespace, 'camera', [])
+
+        previous = namespace.camera
+        previous.append((self.dest, values))
+        setattr(namespace, 'camera', previous)
+
+
+class CustomAction(argparse.Action):
+    def __init__(self, option_strings, dest, **kwargs):
+        super().__init__(option_strings, dest, default={}, **kwargs)
+
+    def __call__(self, parser, namespace, values, option_string=None):
+        if len(namespace.camera) == 0:
+            print(f'Option {option_string} requires a --camera context')
+            sys.exit(-1)
+
+        if self.type == bool:
+            values = True
+
+        current = namespace.camera[-1]
+
+        data = getattr(namespace, self.dest)
+
+        if self.nargs == '+':
+            if current not in data:
+                data[current] = []
+
+            data[current] += values
+        else:
+            data[current] = values
+
+
+def do_cmd_list(cm):
+    print('Available cameras:')
+
+    for idx, c in enumerate(cm.cameras):
+        print(f'{idx + 1}: {c.id}')
+
+
+def do_cmd_list_props(ctx):
+    camera = ctx['camera']
+
+    print('Properties for', ctx['id'])
+
+    for name, prop in camera.properties.items():
+        print('\t{}: {}'.format(name, prop))
+
+
+def do_cmd_list_controls(ctx):
+    camera = ctx['camera']
+
+    print('Controls for', ctx['id'])
+
+    for name, prop in camera.controls.items():
+        print('\t{}: {}'.format(name, prop))
+
+
+def do_cmd_info(ctx):
+    camera = ctx['camera']
+
+    print('Stream info for', ctx['id'])
+
+    roles = [libcam.StreamRole.Viewfinder]
+
+    camconfig = camera.generate_configuration(roles)
+    if camconfig is None:
+        raise Exception('Generating config failed')
+
+    for i, stream_config in enumerate(camconfig):
+        print('\t{}: {}'.format(i, stream_config.to_string()))
+
+        formats = stream_config.formats
+        for fmt in formats.pixel_formats:
+            print('\t * Pixelformat:', fmt, formats.range(fmt))
+
+            for size in formats.sizes(fmt):
+                print('\t  -', size)
+
+
+def acquire(ctx):
+    camera = ctx['camera']
+
+    camera.acquire()
+
+
+def release(ctx):
+    camera = ctx['camera']
+
+    camera.release()
+
+
+def parse_streams(ctx):
+    streams = []
+
+    for stream_desc in ctx['opt-stream']:
+        stream_opts = {'role': libcam.StreamRole.Viewfinder}
+
+        for stream_opt in stream_desc.split(','):
+            if stream_opt == 0:
+                continue
+
+            arr = stream_opt.split('=')
+            if len(arr) != 2:
+                print('Bad stream option', stream_opt)
+                sys.exit(-1)
+
+            key = arr[0]
+            value = arr[1]
+
+            if key in ['width', 'height']:
+                value = int(value)
+            elif key == 'role':
+                rolemap = {
+                    'still': libcam.StreamRole.StillCapture,
+                    'raw': libcam.StreamRole.Raw,
+                    'video': libcam.StreamRole.VideoRecording,
+                    'viewfinder': libcam.StreamRole.Viewfinder,
+                }
+
+                role = rolemap.get(value.lower(), None)
+
+                if role is None:
+                    print('Bad stream role', value)
+                    sys.exit(-1)
+
+                value = role
+            elif key == 'pixelformat':
+                pass
+            else:
+                print('Bad stream option key', key)
+                sys.exit(-1)
+
+            stream_opts[key] = value
+
+        streams.append(stream_opts)
+
+    return streams
+
+
+def configure(ctx):
+    camera = ctx['camera']
+
+    streams = parse_streams(ctx)
+
+    roles = [opts['role'] for opts in streams]
+
+    camconfig = camera.generate_configuration(roles)
+    if camconfig is None:
+        raise Exception('Generating config failed')
+
+    for idx, stream_opts in enumerate(streams):
+        stream_config = camconfig.at(idx)
+
+        if 'width' in stream_opts and 'height' in stream_opts:
+            stream_config.size = (stream_opts['width'], stream_opts['height'])
+
+        if 'pixelformat' in stream_opts:
+            stream_config.pixel_format = stream_opts['pixelformat']
+
+    stat = camconfig.validate()
+
+    if stat == libcam.CameraConfiguration.Status.Invalid:
+        print('Camera configuration invalid')
+        exit(-1)
+    elif stat == libcam.CameraConfiguration.Status.Adjusted:
+        if ctx['opt-strict-formats']:
+            print('Adjusting camera configuration disallowed by --strict-formats argument')
+            exit(-1)
+
+        print('Camera configuration adjusted')
+
+    r = camera.configure(camconfig)
+    if r != 0:
+        raise Exception('Configure failed')
+
+    ctx['stream-names'] = {}
+    ctx['streams'] = []
+
+    for idx, stream_config in enumerate(camconfig):
+        stream = stream_config.stream
+        ctx['streams'].append(stream)
+        ctx['stream-names'][stream] = 'stream' + str(idx)
+        print('{}-{}: stream config {}'.format(ctx['id'], ctx['stream-names'][stream], stream.configuration.to_string()))
+
+
+def alloc_buffers(ctx):
+    camera = ctx['camera']
+
+    allocator = libcam.FrameBufferAllocator(camera)
+
+    for idx, stream in enumerate(ctx['streams']):
+        ret = allocator.allocate(stream)
+        if ret < 0:
+            print('Cannot allocate buffers')
+            exit(-1)
+
+        allocated = len(allocator.buffers(stream))
+
+        print('{}-{}: Allocated {} buffers'.format(ctx['id'], ctx['stream-names'][stream], allocated))
+
+    ctx['allocator'] = allocator
+
+
+def create_requests(ctx):
+    camera = ctx['camera']
+
+    ctx['requests'] = []
+
+    # Identify the stream with the least number of buffers
+    num_bufs = min([len(ctx['allocator'].buffers(stream)) for stream in ctx['streams']])
+
+    requests = []
+
+    for buf_num in range(num_bufs):
+        request = camera.create_request(ctx['idx'])
+
+        if request is None:
+            print('Can not create request')
+            exit(-1)
+
+        for stream in ctx['streams']:
+            buffers = ctx['allocator'].buffers(stream)
+            buffer = buffers[buf_num]
+
+            ret = request.add_buffer(stream, buffer)
+            if ret < 0:
+                print('Can not set buffer for request')
+                exit(-1)
+
+        requests.append(request)
+
+    ctx['requests'] = requests
+
+
+def start(ctx):
+    camera = ctx['camera']
+
+    camera.start()
+
+
+def stop(ctx):
+    camera = ctx['camera']
+
+    camera.stop()
+
+
+def queue_requests(ctx):
+    camera = ctx['camera']
+
+    for request in ctx['requests']:
+        camera.queue_request(request)
+        ctx['reqs-queued'] += 1
+
+    del ctx['requests']
+
+
+def capture_init(contexts):
+    for ctx in contexts:
+        acquire(ctx)
+
+    for ctx in contexts:
+        configure(ctx)
+
+    for ctx in contexts:
+        alloc_buffers(ctx)
+
+    for ctx in contexts:
+        create_requests(ctx)
+
+
+def capture_start(contexts):
+    for ctx in contexts:
+        start(ctx)
+
+    for ctx in contexts:
+        queue_requests(ctx)
+
+
+# Called from renderer when there is a libcamera event
+def event_handler(state):
+    cm = state['cm']
+    contexts = state['contexts']
+
+    os.read(cm.efd, 8)
+
+    reqs = cm.get_ready_requests()
+
+    for req in reqs:
+        ctx = next(ctx for ctx in contexts if ctx['idx'] == req.cookie)
+        request_handler(state, ctx, req)
+
+    running = any(ctx['reqs-completed'] < ctx['opt-capture'] for ctx in contexts)
+    return running
+
+
+def request_handler(state, ctx, req):
+    if req.status != libcam.Request.Status.Complete:
+        raise Exception('{}: Request failed: {}'.format(ctx['id'], req.status))
+
+    buffers = req.buffers
+
+    # Compute the frame rate. The timestamp is arbitrarily retrieved from
+    # the first buffer, as all buffers should have matching timestamps.
+    ts = buffers[next(iter(buffers))].metadata.timestamp
+    last = ctx.get('last', 0)
+    fps = 1000000000.0 / (ts - last) if (last != 0 and (ts - last) != 0) else 0
+    ctx['last'] = ts
+    ctx['fps'] = fps
+
+    for stream, fb in buffers.items():
+        stream_name = ctx['stream-names'][stream]
+
+        crcs = []
+        if ctx['opt-crc']:
+            with fb.mmap() as mfb:
+                plane_crcs = [binascii.crc32(p) for p in mfb.planes]
+                crcs.append(plane_crcs)
+
+        meta = fb.metadata
+
+        print('{:.6f} ({:.2f} fps) {}-{}: seq {}, bytes {}, CRCs {}'
+              .format(ts / 1000000000, fps,
+                      ctx['id'], stream_name,
+                      meta.sequence, meta.bytesused,
+                      crcs))
+
+        if ctx['opt-metadata']:
+            reqmeta = req.metadata
+            for ctrl, val in reqmeta.items():
+                print(f'\t{ctrl} = {val}')
+
+        if ctx['opt-save-frames']:
+            with fb.mmap() as mfb:
+                filename = 'frame-{}-{}-{}.data'.format(ctx['id'], stream_name, ctx['reqs-completed'])
+                with open(filename, 'wb') as f:
+                    for p in mfb.planes:
+                        f.write(p)
+
+    state['renderer'].request_handler(ctx, req)
+
+    ctx['reqs-completed'] += 1
+
+
+# Called from renderer when it has finished with a request
+def request_prcessed(ctx, req):
+    camera = ctx['camera']
+
+    if ctx['reqs-queued'] < ctx['opt-capture']:
+        req.reuse()
+        camera.queue_request(req)
+        ctx['reqs-queued'] += 1
+
+
+def capture_deinit(contexts):
+    for ctx in contexts:
+        stop(ctx)
+
+    for ctx in contexts:
+        release(ctx)
+
+
+def do_cmd_capture(state):
+    capture_init(state['contexts'])
+
+    renderer = state['renderer']
+
+    renderer.setup()
+
+    capture_start(state['contexts'])
+
+    renderer.run()
+
+    capture_deinit(state['contexts'])
+
+
+def main():
+    parser = argparse.ArgumentParser()
+    # global options
+    parser.add_argument('-l', '--list', action='store_true', help='List all cameras')
+    parser.add_argument('-c', '--camera', type=int, action='extend', nargs=1, default=[], help='Specify which camera to operate on, by index')
+    parser.add_argument('-p', '--list-properties', action='store_true', help='List cameras properties')
+    parser.add_argument('--list-controls', action='store_true', help='List cameras controls')
+    parser.add_argument('-I', '--info', action='store_true', help='Display information about stream(s)')
+    parser.add_argument('-R', '--renderer', default='null', help='Renderer (null, kms, qt, qtgl)')
+
+    # per camera options
+    parser.add_argument('-C', '--capture', nargs='?', type=int, const=1000000, action=CustomAction, help='Capture until interrupted by user or until CAPTURE frames captured')
+    parser.add_argument('--crc', nargs=0, type=bool, action=CustomAction, help='Print CRC32 for captured frames')
+    parser.add_argument('--save-frames', nargs=0, type=bool, action=CustomAction, help='Save captured frames to files')
+    parser.add_argument('--metadata', nargs=0, type=bool, action=CustomAction, help='Print the metadata for completed requests')
+    parser.add_argument('--strict-formats', type=bool, nargs=0, action=CustomAction, help='Do not allow requested stream format(s) to be adjusted')
+    parser.add_argument('-s', '--stream', nargs='+', action=CustomAction)
+    args = parser.parse_args()
+
+    cm = libcam.CameraManager.singleton()
+
+    if args.list:
+        do_cmd_list(cm)
+
+    contexts = []
+
+    for cam_idx in args.camera:
+        camera = next((c for i, c in enumerate(cm.cameras) if i + 1 == cam_idx), None)
+
+        if camera is None:
+            print('Unable to find camera', cam_idx)
+            return -1
+
+        contexts.append({
+                        'camera': camera,
+                        'idx': cam_idx,
+                        'id': 'cam' + str(cam_idx),
+                        'reqs-queued': 0,
+                        'reqs-completed': 0,
+                        'opt-capture': args.capture.get(cam_idx, False),
+                        'opt-crc': args.crc.get(cam_idx, False),
+                        'opt-save-frames': args.save_frames.get(cam_idx, False),
+                        'opt-metadata': args.metadata.get(cam_idx, False),
+                        'opt-strict-formats': args.strict_formats.get(cam_idx, False),
+                        'opt-stream': args.stream.get(cam_idx, ['role=viewfinder']),
+                        })
+
+    for ctx in contexts:
+        print('Using camera {} as {}'.format(ctx['camera'].id, ctx['id']))
+
+    for ctx in contexts:
+        if args.list_properties:
+            do_cmd_list_props(ctx)
+        if args.list_controls:
+            do_cmd_list_controls(ctx)
+        if args.info:
+            do_cmd_info(ctx)
+
+    if args.capture:
+
+        state = {
+            'cm': cm,
+            'contexts': contexts,
+            'event_handler': event_handler,
+            'request_prcessed': request_prcessed,
+        }
+
+        if args.renderer == 'null':
+            import cam_null
+            renderer = cam_null.NullRenderer(state)
+        elif args.renderer == 'kms':
+            import cam_kms
+            renderer = cam_kms.KMSRenderer(state)
+        elif args.renderer == 'qt':
+            import cam_qt
+            renderer = cam_qt.QtRenderer(state)
+        elif args.renderer == 'qtgl':
+            import cam_qtgl
+            renderer = cam_qtgl.QtRenderer(state)
+        else:
+            print('Bad renderer', args.renderer)
+            return -1
+
+        state['renderer'] = renderer
+
+        do_cmd_capture(state)
+
+    return 0
+
+
+if __name__ == '__main__':
+    sys.exit(main())
diff --git a/src/py/cam/cam_kms.py b/src/py/cam/cam_kms.py
new file mode 100644
index 00000000..f4ee5a06
--- /dev/null
+++ b/src/py/cam/cam_kms.py
@@ -0,0 +1,183 @@ 
+# SPDX-License-Identifier: GPL-2.0-or-later
+# Copyright (C) 2021, Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
+
+import pykms
+import selectors
+import sys
+
+FMT_MAP = {
+    'RGB888': pykms.PixelFormat.RGB888,
+    'YUYV': pykms.PixelFormat.YUYV,
+    'ARGB8888': pykms.PixelFormat.ARGB8888,
+    'XRGB8888': pykms.PixelFormat.XRGB8888,
+}
+
+
+class KMSRenderer:
+    def __init__(self, state):
+        self.state = state
+
+        self.cm = state['cm']
+        self.contexts = state['contexts']
+        self.running = False
+
+        card = pykms.Card()
+
+        res = pykms.ResourceManager(card)
+        conn = res.reserve_connector()
+        crtc = res.reserve_crtc(conn)
+        mode = conn.get_default_mode()
+        modeb = mode.to_blob(card)
+
+        req = pykms.AtomicReq(card)
+        req.add_connector(conn, crtc)
+        req.add_crtc(crtc, modeb)
+        r = req.commit_sync(allow_modeset=True)
+        assert(r == 0)
+
+        self.card = card
+        self.resman = res
+        self.crtc = crtc
+        self.mode = mode
+
+        self.bufqueue = []
+        self.current = None
+        self.next = None
+        self.cam_2_drm = {}
+
+    # KMS
+
+    def close(self):
+        req = pykms.AtomicReq(self.card)
+        for s in self.streams:
+            req.add_plane(s['plane'], None, None, dst=(0, 0, 0, 0))
+        req.commit()
+
+    def add_plane(self, req, stream, fb):
+        s = next(s for s in self.streams if s['stream'] == stream)
+        idx = s['idx']
+        plane = s['plane']
+
+        if idx % 2 == 0:
+            x = 0
+        else:
+            x = self.mode.hdisplay - fb.width
+
+        if idx // 2 == 0:
+            y = 0
+        else:
+            y = self.mode.vdisplay - fb.height
+
+        req.add_plane(plane, fb, self.crtc, dst=(x, y, fb.width, fb.height))
+
+    def apply_request(self, drmreq):
+
+        buffers = drmreq['camreq'].buffers
+
+        for stream, fb in buffers.items():
+            drmfb = self.cam_2_drm.get(fb, None)
+
+            req = pykms.AtomicReq(self.card)
+            self.add_plane(req, stream, drmfb)
+            req.commit()
+
+    def handle_page_flip(self, frame, time):
+        old = self.current
+        self.current = self.next
+
+        if len(self.bufqueue) > 0:
+            self.next = self.bufqueue.pop(0)
+        else:
+            self.next = None
+
+        if self.next:
+            drmreq = self.next
+
+            self.apply_request(drmreq)
+
+        if old:
+            req = old['camreq']
+            ctx = old['camctx']
+            self.state['request_prcessed'](ctx, req)
+
+    def queue(self, drmreq):
+        if not self.next:
+            self.next = drmreq
+            self.apply_request(drmreq)
+        else:
+            self.bufqueue.append(drmreq)
+
+    # libcamera
+
+    def setup(self):
+        self.streams = []
+
+        idx = 0
+        for ctx in self.contexts:
+            for stream in ctx['streams']:
+
+                cfg = stream.configuration
+                fmt = cfg.pixel_format
+                fmt = FMT_MAP[fmt]
+
+                plane = self.resman.reserve_generic_plane(self.crtc, fmt)
+                assert(plane is not None)
+
+                self.streams.append({
+                    'idx': idx,
+                    'stream': stream,
+                    'plane': plane,
+                    'fmt': fmt,
+                    'size': cfg.size,
+                })
+
+                for fb in ctx['allocator'].buffers(stream):
+                    w, h = cfg.size
+                    stride = cfg.stride
+                    fd = fb.fd(0)
+                    drmfb = pykms.DmabufFramebuffer(self.card, w, h, fmt,
+                                                    [fd], [stride], [0])
+                    self.cam_2_drm[fb] = drmfb
+
+                idx += 1
+
+    def readdrm(self, fileobj):
+        for ev in self.card.read_events():
+            if ev.type == pykms.DrmEventType.FLIP_COMPLETE:
+                self.handle_page_flip(ev.seq, ev.time)
+
+    def readcam(self, fd):
+        self.running = self.state['event_handler'](self.state)
+
+    def readkey(self, fileobj):
+        sys.stdin.readline()
+        self.running = False
+
+    def run(self):
+        print('Capturing...')
+
+        self.running = True
+
+        sel = selectors.DefaultSelector()
+        sel.register(self.card.fd, selectors.EVENT_READ, self.readdrm)
+        sel.register(self.cm.efd, selectors.EVENT_READ, self.readcam)
+        sel.register(sys.stdin, selectors.EVENT_READ, self.readkey)
+
+        print('Press enter to exit')
+
+        while self.running:
+            events = sel.select()
+            for key, mask in events:
+                callback = key.data
+                callback(key.fileobj)
+
+        print('Exiting...')
+
+    def request_handler(self, ctx, req):
+
+        drmreq = {
+            'camctx': ctx,
+            'camreq': req,
+        }
+
+        self.queue(drmreq)
diff --git a/src/py/cam/cam_null.py b/src/py/cam/cam_null.py
new file mode 100644
index 00000000..391397f6
--- /dev/null
+++ b/src/py/cam/cam_null.py
@@ -0,0 +1,47 @@ 
+# SPDX-License-Identifier: GPL-2.0-or-later
+# Copyright (C) 2021, Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
+
+import selectors
+import sys
+
+
+class NullRenderer:
+    def __init__(self, state):
+        self.state = state
+
+        self.cm = state['cm']
+        self.contexts = state['contexts']
+
+        self.running = False
+
+    def setup(self):
+        pass
+
+    def run(self):
+        print('Capturing...')
+
+        self.running = True
+
+        sel = selectors.DefaultSelector()
+        sel.register(self.cm.efd, selectors.EVENT_READ, self.readcam)
+        sel.register(sys.stdin, selectors.EVENT_READ, self.readkey)
+
+        print('Press enter to exit')
+
+        while self.running:
+            events = sel.select()
+            for key, mask in events:
+                callback = key.data
+                callback(key.fileobj)
+
+        print('Exiting...')
+
+    def readcam(self, fd):
+        self.running = self.state['event_handler'](self.state)
+
+    def readkey(self, fileobj):
+        sys.stdin.readline()
+        self.running = False
+
+    def request_handler(self, ctx, req):
+        self.state['request_prcessed'](ctx, req)
diff --git a/src/py/cam/cam_qt.py b/src/py/cam/cam_qt.py
new file mode 100644
index 00000000..40044866
--- /dev/null
+++ b/src/py/cam/cam_qt.py
@@ -0,0 +1,354 @@ 
+# SPDX-License-Identifier: GPL-2.0-or-later
+# Copyright (C) 2021, Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
+#
+# Debayering code from PiCamera documentation
+
+from io import BytesIO
+from numpy.lib.stride_tricks import as_strided
+from PIL import Image
+from PIL.ImageQt import ImageQt
+from PyQt5 import QtCore, QtGui, QtWidgets
+import numpy as np
+import sys
+
+
+def rgb_to_pix(rgb):
+    img = Image.frombuffer('RGB', (rgb.shape[1], rgb.shape[0]), rgb)
+    qim = ImageQt(img).copy()
+    pix = QtGui.QPixmap.fromImage(qim)
+    return pix
+
+
+def separate_components(data, r0, g0, g1, b0):
+    # Now to split the data up into its red, green, and blue components. The
+    # Bayer pattern of the OV5647 sensor is BGGR. In other words the first
+    # row contains alternating green/blue elements, the second row contains
+    # alternating red/green elements, and so on as illustrated below:
+    #
+    # GBGBGBGBGBGBGB
+    # RGRGRGRGRGRGRG
+    # GBGBGBGBGBGBGB
+    # RGRGRGRGRGRGRG
+    #
+    # Please note that if you use vflip or hflip to change the orientation
+    # of the capture, you must flip the Bayer pattern accordingly
+
+    rgb = np.zeros(data.shape + (3,), dtype=data.dtype)
+    rgb[r0[1]::2, r0[0]::2, 0] = data[r0[1]::2, r0[0]::2]  # Red
+    rgb[g0[1]::2, g0[0]::2, 1] = data[g0[1]::2, g0[0]::2]  # Green
+    rgb[g1[1]::2, g1[0]::2, 1] = data[g1[1]::2, g1[0]::2]  # Green
+    rgb[b0[1]::2, b0[0]::2, 2] = data[b0[1]::2, b0[0]::2]  # Blue
+
+    return rgb
+
+
+def demosaic(rgb, r0, g0, g1, b0):
+    # At this point we now have the raw Bayer data with the correct values
+    # and colors but the data still requires de-mosaicing and
+    # post-processing. If you wish to do this yourself, end the script here!
+    #
+    # Below we present a fairly naive de-mosaic method that simply
+    # calculates the weighted average of a pixel based on the pixels
+    # surrounding it. The weighting is provided b0[1] a b0[1]te representation of
+    # the Bayer filter which we construct first:
+
+    bayer = np.zeros(rgb.shape, dtype=np.uint8)
+    bayer[r0[1]::2, r0[0]::2, 0] = 1  # Red
+    bayer[g0[1]::2, g0[0]::2, 1] = 1  # Green
+    bayer[g1[1]::2, g1[0]::2, 1] = 1  # Green
+    bayer[b0[1]::2, b0[0]::2, 2] = 1  # Blue
+
+    # Allocate an array to hold our output with the same shape as the input
+    # data. After this we define the size of window that will be used to
+    # calculate each weighted average (3x3). Then we pad out the rgb and
+    # bayer arrays, adding blank pixels at their edges to compensate for the
+    # size of the window when calculating averages for edge pixels.
+
+    output = np.empty(rgb.shape, dtype=rgb.dtype)
+    window = (3, 3)
+    borders = (window[0] - 1, window[1] - 1)
+    border = (borders[0] // 2, borders[1] // 2)
+
+    # rgb_pad = np.zeros((
+    #    rgb.shape[0] + borders[0],
+    #    rgb.shape[1] + borders[1],
+    #    rgb.shape[2]), dtype=rgb.dtype)
+    # rgb_pad[
+    #    border[0]:rgb_pad.shape[0] - border[0],
+    #    border[1]:rgb_pad.shape[1] - border[1],
+    #    :] = rgb
+    # rgb = rgb_pad
+    #
+    # bayer_pad = np.zeros((
+    #    bayer.shape[0] + borders[0],
+    #    bayer.shape[1] + borders[1],
+    #    bayer.shape[2]), dtype=bayer.dtype)
+    # bayer_pad[
+    #    border[0]:bayer_pad.shape[0] - border[0],
+    #    border[1]:bayer_pad.shape[1] - border[1],
+    #    :] = bayer
+    # bayer = bayer_pad
+
+    # In numpy >=1.7.0 just use np.pad (version in Raspbian is 1.6.2 at the
+    # time of writing...)
+    #
+    rgb = np.pad(rgb, [
+        (border[0], border[0]),
+        (border[1], border[1]),
+        (0, 0),
+    ], 'constant')
+    bayer = np.pad(bayer, [
+        (border[0], border[0]),
+        (border[1], border[1]),
+        (0, 0),
+    ], 'constant')
+
+    # For each plane in the RGB data, we use a nifty numpy trick
+    # (as_strided) to construct a view over the plane of 3x3 matrices. We do
+    # the same for the bayer array, then use Einstein summation on each
+    # (np.sum is simpler, but copies the data so it's slower), and divide
+    # the results to get our weighted average:
+
+    for plane in range(3):
+        p = rgb[..., plane]
+        b = bayer[..., plane]
+        pview = as_strided(p, shape=(
+            p.shape[0] - borders[0],
+            p.shape[1] - borders[1]) + window, strides=p.strides * 2)
+        bview = as_strided(b, shape=(
+            b.shape[0] - borders[0],
+            b.shape[1] - borders[1]) + window, strides=b.strides * 2)
+        psum = np.einsum('ijkl->ij', pview)
+        bsum = np.einsum('ijkl->ij', bview)
+        output[..., plane] = psum // bsum
+
+    return output
+
+
+def to_rgb(fmt, size, data):
+    w = size[0]
+    h = size[1]
+
+    if fmt == 'YUYV':
+        # YUV422
+        yuyv = data.reshape((h, w // 2 * 4))
+
+        # YUV444
+        yuv = np.empty((h, w, 3), dtype=np.uint8)
+        yuv[:, :, 0] = yuyv[:, 0::2]                    # Y
+        yuv[:, :, 1] = yuyv[:, 1::4].repeat(2, axis=1)  # U
+        yuv[:, :, 2] = yuyv[:, 3::4].repeat(2, axis=1)  # V
+
+        m = np.array([
+            [ 1.0, 1.0, 1.0],
+            [-0.000007154783816076815, -0.3441331386566162, 1.7720025777816772],
+            [ 1.4019975662231445, -0.7141380310058594 , 0.00001542569043522235]
+        ])
+
+        rgb = np.dot(yuv, m)
+        rgb[:, :, 0] -= 179.45477266423404
+        rgb[:, :, 1] += 135.45870971679688
+        rgb[:, :, 2] -= 226.8183044444304
+        rgb = rgb.astype(np.uint8)
+
+    elif fmt == 'RGB888':
+        rgb = data.reshape((h, w, 3))
+        rgb[:, :, [0, 1, 2]] = rgb[:, :, [2, 1, 0]]
+
+    elif fmt == 'BGR888':
+        rgb = data.reshape((h, w, 3))
+
+    elif fmt in ['ARGB8888', 'XRGB8888']:
+        rgb = data.reshape((h, w, 4))
+        rgb = np.flip(rgb, axis=2)
+        # drop alpha component
+        rgb = np.delete(rgb, np.s_[0::4], axis=2)
+
+    elif fmt.startswith('S'):
+        bayer_pattern = fmt[1:5]
+        bitspp = int(fmt[5:])
+
+        # TODO: shifting leaves the lowest bits 0
+        if bitspp == 8:
+            data = data.reshape((h, w))
+            data = data.astype(np.uint16) << 8
+        elif bitspp in [10, 12]:
+            data = data.view(np.uint16)
+            data = data.reshape((h, w))
+            data = data << (16 - bitspp)
+        else:
+            raise Exception('Bad bitspp:' + str(bitspp))
+
+        idx = bayer_pattern.find('R')
+        assert(idx != -1)
+        r0 = (idx % 2, idx // 2)
+
+        idx = bayer_pattern.find('G')
+        assert(idx != -1)
+        g0 = (idx % 2, idx // 2)
+
+        idx = bayer_pattern.find('G', idx + 1)
+        assert(idx != -1)
+        g1 = (idx % 2, idx // 2)
+
+        idx = bayer_pattern.find('B')
+        assert(idx != -1)
+        b0 = (idx % 2, idx // 2)
+
+        rgb = separate_components(data, r0, g0, g1, b0)
+        rgb = demosaic(rgb, r0, g0, g1, b0)
+        rgb = (rgb >> 8).astype(np.uint8)
+
+    else:
+        rgb = None
+
+    return rgb
+
+
+class QtRenderer:
+    def __init__(self, state):
+        self.state = state
+
+        self.cm = state['cm']
+        self.contexts = state['contexts']
+
+    def setup(self):
+        self.app = QtWidgets.QApplication([])
+
+        windows = []
+
+        for ctx in self.contexts:
+            camera = ctx['camera']
+
+            for stream in ctx['streams']:
+                fmt = stream.configuration.pixel_format
+                size = stream.configuration.size
+
+                window = MainWindow(ctx, stream)
+                window.setAttribute(QtCore.Qt.WA_ShowWithoutActivating)
+                window.show()
+                windows.append(window)
+
+        self.windows = windows
+
+    def run(self):
+        camnotif = QtCore.QSocketNotifier(self.cm.efd, QtCore.QSocketNotifier.Read)
+        camnotif.activated.connect(lambda x: self.readcam())
+
+        keynotif = QtCore.QSocketNotifier(sys.stdin.fileno(), QtCore.QSocketNotifier.Read)
+        keynotif.activated.connect(lambda x: self.readkey())
+
+        print('Capturing...')
+
+        self.app.exec()
+
+        print('Exiting...')
+
+    def readcam(self):
+        running = self.state['event_handler'](self.state)
+
+        if not running:
+            self.app.quit()
+
+    def readkey(self):
+        sys.stdin.readline()
+        self.app.quit()
+
+    def request_handler(self, ctx, req):
+        buffers = req.buffers
+
+        for stream, fb in buffers.items():
+            wnd = next(wnd for wnd in self.windows if wnd.stream == stream)
+
+            wnd.handle_request(stream, fb)
+
+        self.state['request_prcessed'](ctx, req)
+
+    def cleanup(self):
+        for w in self.windows:
+            w.close()
+
+
+class MainWindow(QtWidgets.QWidget):
+    def __init__(self, ctx, stream):
+        super().__init__()
+
+        self.ctx = ctx
+        self.stream = stream
+
+        self.label = QtWidgets.QLabel()
+
+        windowLayout = QtWidgets.QHBoxLayout()
+        self.setLayout(windowLayout)
+
+        windowLayout.addWidget(self.label)
+
+        controlsLayout = QtWidgets.QVBoxLayout()
+        windowLayout.addLayout(controlsLayout)
+
+        windowLayout.addStretch()
+
+        group = QtWidgets.QGroupBox('Info')
+        groupLayout = QtWidgets.QVBoxLayout()
+        group.setLayout(groupLayout)
+        controlsLayout.addWidget(group)
+
+        lab = QtWidgets.QLabel(ctx['id'])
+        groupLayout.addWidget(lab)
+
+        self.frameLabel = QtWidgets.QLabel()
+        groupLayout.addWidget(self.frameLabel)
+
+        group = QtWidgets.QGroupBox('Properties')
+        groupLayout = QtWidgets.QVBoxLayout()
+        group.setLayout(groupLayout)
+        controlsLayout.addWidget(group)
+
+        camera = ctx['camera']
+
+        for k, v in camera.properties.items():
+            lab = QtWidgets.QLabel()
+            lab.setText(k + ' = ' + str(v))
+            groupLayout.addWidget(lab)
+
+        group = QtWidgets.QGroupBox('Controls')
+        groupLayout = QtWidgets.QVBoxLayout()
+        group.setLayout(groupLayout)
+        controlsLayout.addWidget(group)
+
+        for k, (min, max, default) in camera.controls.items():
+            lab = QtWidgets.QLabel()
+            lab.setText('{} = {}/{}/{}'.format(k, min, max, default))
+            groupLayout.addWidget(lab)
+
+        controlsLayout.addStretch()
+
+    def buf_to_qpixmap(self, stream, fb):
+        with fb.mmap() as mfb:
+            cfg = stream.configuration
+            w, h = cfg.size
+            pitch = cfg.stride
+
+            if cfg.pixel_format == 'MJPEG':
+                img = Image.open(BytesIO(mfb.planes[0]))
+                qim = ImageQt(img).copy()
+                pix = QtGui.QPixmap.fromImage(qim)
+            else:
+                data = np.array(mfb.planes[0], dtype=np.uint8)
+                rgb = to_rgb(cfg.pixel_format, cfg.size, data)
+
+                if rgb is None:
+                    raise Exception('Format not supported: ' + cfg.pixel_format)
+
+                pix = rgb_to_pix(rgb)
+
+        return pix
+
+    def handle_request(self, stream, fb):
+        ctx = self.ctx
+
+        pix = self.buf_to_qpixmap(stream, fb)
+        self.label.setPixmap(pix)
+
+        self.frameLabel.setText('Queued: {}\nDone: {}\nFps: {:.2f}'
+                       .format(ctx['reqs-queued'], ctx['reqs-completed'], ctx['fps']))
diff --git a/src/py/cam/cam_qtgl.py b/src/py/cam/cam_qtgl.py
new file mode 100644
index 00000000..37b74d3f
--- /dev/null
+++ b/src/py/cam/cam_qtgl.py
@@ -0,0 +1,385 @@ 
+# SPDX-License-Identifier: GPL-2.0-or-later
+# Copyright (C) 2021, Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
+
+from PyQt5 import QtCore, QtWidgets
+from PyQt5.QtCore import Qt
+
+import math
+import numpy as np
+import os
+import sys
+
+os.environ['PYOPENGL_PLATFORM'] = 'egl'
+
+import OpenGL
+# OpenGL.FULL_LOGGING = True
+
+from OpenGL import GL as gl
+from OpenGL.EGL.EXT.image_dma_buf_import import *
+from OpenGL.EGL.KHR.image import *
+from OpenGL.EGL.VERSION.EGL_1_0 import *
+from OpenGL.EGL.VERSION.EGL_1_2 import *
+from OpenGL.EGL.VERSION.EGL_1_3 import *
+
+from OpenGL.GLES2.OES.EGL_image import *
+from OpenGL.GLES2.OES.EGL_image_external import *
+from OpenGL.GLES2.VERSION.GLES2_2_0 import *
+from OpenGL.GLES3.VERSION.GLES3_3_0 import *
+
+from OpenGL.GL import shaders
+
+from gl_helpers import *
+
+# libcamera format string -> DRM fourcc
+FMT_MAP = {
+    'RGB888': 'RG24',
+    'XRGB8888': 'XR24',
+    'ARGB8888': 'AR24',
+    'YUYV': 'YUYV',
+}
+
+
+class EglState:
+    def __init__(self):
+        self.create_display()
+        self.choose_config()
+        self.create_context()
+        self.check_extensions()
+
+    def create_display(self):
+        xdpy = getEGLNativeDisplay()
+        dpy = eglGetDisplay(xdpy)
+        self.display = dpy
+
+    def choose_config(self):
+        dpy = self.display
+
+        major, minor = EGLint(), EGLint()
+
+        b = eglInitialize(dpy, major, minor)
+        assert(b)
+
+        print('EGL {} {}'.format(
+              eglQueryString(dpy, EGL_VENDOR).decode(),
+              eglQueryString(dpy, EGL_VERSION).decode()))
+
+        check_egl_extensions(dpy, ['EGL_EXT_image_dma_buf_import'])
+
+        b = eglBindAPI(EGL_OPENGL_ES_API)
+        assert(b)
+
+        def print_config(dpy, cfg):
+
+            def _getconf(dpy, cfg, a):
+                value = ctypes.c_long()
+                eglGetConfigAttrib(dpy, cfg, a, value)
+                return value.value
+
+            getconf = lambda a: _getconf(dpy, cfg, a)
+
+            print('EGL Config {}: color buf {}/{}/{}/{} = {}, depth {}, stencil {}, native visualid {}, native visualtype {}'.format(
+                getconf(EGL_CONFIG_ID),
+                getconf(EGL_ALPHA_SIZE),
+                getconf(EGL_RED_SIZE),
+                getconf(EGL_GREEN_SIZE),
+                getconf(EGL_BLUE_SIZE),
+                getconf(EGL_BUFFER_SIZE),
+                getconf(EGL_DEPTH_SIZE),
+                getconf(EGL_STENCIL_SIZE),
+                getconf(EGL_NATIVE_VISUAL_ID),
+                getconf(EGL_NATIVE_VISUAL_TYPE)))
+
+        if False:
+            num_configs = ctypes.c_long()
+            eglGetConfigs(dpy, None, 0, num_configs)
+            print('{} configs'.format(num_configs.value))
+
+            configs = (EGLConfig * num_configs.value)()
+            eglGetConfigs(dpy, configs, num_configs.value, num_configs)
+            for config_id in configs:
+                print_config(dpy, config_id)
+
+        config_attribs = [
+            EGL_SURFACE_TYPE, EGL_WINDOW_BIT,
+            EGL_RED_SIZE, 8,
+            EGL_GREEN_SIZE, 8,
+            EGL_BLUE_SIZE, 8,
+            EGL_ALPHA_SIZE, 0,
+            EGL_RENDERABLE_TYPE, EGL_OPENGL_ES2_BIT,
+            EGL_NONE,
+        ]
+
+        n = EGLint()
+        configs = (EGLConfig * 1)()
+        b = eglChooseConfig(dpy, config_attribs, configs, 1, n)
+        assert(b and n.value == 1)
+        config = configs[0]
+
+        print('Chosen Config:')
+        print_config(dpy, config)
+
+        self.config = config
+
+    def create_context(self):
+        dpy = self.display
+
+        context_attribs = [
+            EGL_CONTEXT_CLIENT_VERSION, 2,
+            EGL_NONE,
+        ]
+
+        context = eglCreateContext(dpy, self.config, EGL_NO_CONTEXT, context_attribs)
+        assert(context)
+
+        b = eglMakeCurrent(dpy, EGL_NO_SURFACE, EGL_NO_SURFACE, context)
+        assert(b)
+
+        self.context = context
+
+    def check_extensions(self):
+        check_gl_extensions(['GL_OES_EGL_image'])
+
+        assert(eglCreateImageKHR)
+        assert(eglDestroyImageKHR)
+        assert(glEGLImageTargetTexture2DOES)
+
+
+class QtRenderer:
+    def __init__(self, state):
+        self.state = state
+
+    def setup(self):
+        self.app = QtWidgets.QApplication([])
+
+        window = MainWindow(self.state)
+        window.setAttribute(QtCore.Qt.WA_ShowWithoutActivating)
+        window.show()
+
+        self.window = window
+
+    def run(self):
+        camnotif = QtCore.QSocketNotifier(self.state['cm'].efd, QtCore.QSocketNotifier.Read)
+        camnotif.activated.connect(lambda x: self.readcam())
+
+        keynotif = QtCore.QSocketNotifier(sys.stdin.fileno(), QtCore.QSocketNotifier.Read)
+        keynotif.activated.connect(lambda x: self.readkey())
+
+        print('Capturing...')
+
+        self.app.exec()
+
+        print('Exiting...')
+
+    def readcam(self):
+        running = self.state['event_handler'](self.state)
+
+        if not running:
+            self.app.quit()
+
+    def readkey(self):
+        sys.stdin.readline()
+        self.app.quit()
+
+    def request_handler(self, ctx, req):
+        self.window.handle_request(ctx, req)
+
+    def cleanup(self):
+        self.window.close()
+
+
+class MainWindow(QtWidgets.QWidget):
+    def __init__(self, state):
+        super().__init__()
+
+        self.setAttribute(Qt.WA_PaintOnScreen)
+        self.setAttribute(Qt.WA_NativeWindow)
+
+        self.state = state
+
+        self.textures = {}
+        self.reqqueue = {}
+        self.current = {}
+
+        for ctx in self.state['contexts']:
+
+            self.reqqueue[ctx['idx']] = []
+            self.current[ctx['idx']] = []
+
+            for stream in ctx['streams']:
+                fmt = stream.configuration.pixel_format
+                size = stream.configuration.size
+
+                if fmt not in FMT_MAP:
+                    raise Exception('Unsupported pixel format: ' + str(fmt))
+
+                self.textures[stream] = None
+
+        num_tiles = len(self.textures)
+        self.num_columns = math.ceil(math.sqrt(num_tiles))
+        self.num_rows = math.ceil(num_tiles / self.num_columns)
+
+        self.egl = EglState()
+
+        self.surface = None
+
+    def paintEngine(self):
+        return None
+
+    def create_surface(self):
+        native_surface = c_void_p(self.winId().__int__())
+        surface = eglCreateWindowSurface(self.egl.display, self.egl.config,
+                                         native_surface, None)
+
+        b = eglMakeCurrent(self.egl.display, self.surface, self.surface, self.egl.context)
+        assert(b)
+
+        self.surface = surface
+
+    def init_gl(self):
+        self.create_surface()
+
+        vertShaderSrc = '''
+            attribute vec2 aPosition;
+            varying vec2 texcoord;
+
+            void main()
+            {
+                gl_Position = vec4(aPosition * 2.0 - 1.0, 0.0, 1.0);
+                texcoord.x = aPosition.x;
+                texcoord.y = 1.0 - aPosition.y;
+            }
+        '''
+        fragShaderSrc = '''
+            #extension GL_OES_EGL_image_external : enable
+            precision mediump float;
+            varying vec2 texcoord;
+            uniform samplerExternalOES texture;
+
+            void main()
+            {
+                gl_FragColor = texture2D(texture, texcoord);
+            }
+        '''
+
+        program = shaders.compileProgram(
+            shaders.compileShader(vertShaderSrc, GL_VERTEX_SHADER),
+            shaders.compileShader(fragShaderSrc, GL_FRAGMENT_SHADER)
+        )
+
+        glUseProgram(program)
+
+        glClearColor(0.5, 0.8, 0.7, 1.0)
+
+        vertPositions = [
+            0.0, 0.0,
+            1.0, 0.0,
+            1.0, 1.0,
+            0.0, 1.0
+        ]
+
+        inputAttrib = glGetAttribLocation(program, 'aPosition')
+        glVertexAttribPointer(inputAttrib, 2, GL_FLOAT, GL_FALSE, 0, vertPositions)
+        glEnableVertexAttribArray(inputAttrib)
+
+    def create_texture(self, stream, fb):
+        cfg = stream.configuration
+        fmt = cfg.pixel_format
+        fmt = str_to_fourcc(FMT_MAP[fmt])
+        w, h = cfg.size
+
+        attribs = [
+            EGL_WIDTH, w,
+            EGL_HEIGHT, h,
+            EGL_LINUX_DRM_FOURCC_EXT, fmt,
+            EGL_DMA_BUF_PLANE0_FD_EXT, fb.fd(0),
+            EGL_DMA_BUF_PLANE0_OFFSET_EXT, 0,
+            EGL_DMA_BUF_PLANE0_PITCH_EXT, cfg.stride,
+            EGL_NONE,
+        ]
+
+        image = eglCreateImageKHR(self.egl.display,
+                                  EGL_NO_CONTEXT,
+                                  EGL_LINUX_DMA_BUF_EXT,
+                                  None,
+                                  attribs)
+        assert(image)
+
+        textures = glGenTextures(1)
+        glBindTexture(GL_TEXTURE_EXTERNAL_OES, textures)
+        glTexParameteri(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_MAG_FILTER, GL_LINEAR)
+        glTexParameteri(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_MIN_FILTER, GL_LINEAR)
+        glTexParameteri(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE)
+        glTexParameteri(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE)
+        glEGLImageTargetTexture2DOES(GL_TEXTURE_EXTERNAL_OES, image)
+
+        return textures
+
+    def resizeEvent(self, event):
+        size = event.size()
+
+        print('Resize', size)
+
+        super().resizeEvent(event)
+
+        if self.surface is None:
+            return
+
+        glViewport(0, 0, size.width() // 2, size.height())
+
+    def paintEvent(self, event):
+        if self.surface is None:
+            self.init_gl()
+
+        for ctx_idx, queue in self.reqqueue.items():
+            if len(queue) == 0:
+                continue
+
+            ctx = next(ctx for ctx in self.state['contexts'] if ctx['idx'] == ctx_idx)
+
+            if self.current[ctx_idx]:
+                old = self.current[ctx_idx]
+                self.current[ctx_idx] = None
+                self.state['request_prcessed'](ctx, old)
+
+            next_req = queue.pop(0)
+            self.current[ctx_idx] = next_req
+
+            stream, fb = next(iter(next_req.buffers.items()))
+
+            self.textures[stream] = self.create_texture(stream, fb)
+
+        self.paint_gl()
+
+    def paint_gl(self):
+        b = eglMakeCurrent(self.egl.display, self.surface, self.surface, self.egl.context)
+        assert(b)
+
+        glClear(GL_COLOR_BUFFER_BIT)
+
+        size = self.size()
+
+        for idx, ctx in enumerate(self.state['contexts']):
+            for stream in ctx['streams']:
+                if self.textures[stream] is None:
+                    continue
+
+                w = size.width() // self.num_columns
+                h = size.height() // self.num_rows
+
+                x = idx % self.num_columns
+                y = idx // self.num_columns
+
+                x *= w
+                y *= h
+
+                glViewport(x, y, w, h)
+
+                glBindTexture(GL_TEXTURE_EXTERNAL_OES, self.textures[stream])
+                glDrawArrays(GL_TRIANGLE_FAN, 0, 4)
+
+        b = eglSwapBuffers(self.egl.display, self.surface)
+        assert(b)
+
+    def handle_request(self, ctx, req):
+        self.reqqueue[ctx['idx']].append(req)
+        self.update()
diff --git a/src/py/cam/gl_helpers.py b/src/py/cam/gl_helpers.py
new file mode 100644
index 00000000..b377d735
--- /dev/null
+++ b/src/py/cam/gl_helpers.py
@@ -0,0 +1,74 @@ 
+# SPDX-License-Identifier: GPL-2.0-or-later
+# Copyright (C) 2021, Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
+
+from OpenGL.EGL.VERSION.EGL_1_0 import EGLNativeDisplayType, eglGetProcAddress, eglQueryString, EGL_EXTENSIONS
+
+from OpenGL.raw.GLES2 import _types as _cs
+from OpenGL.GLES2.VERSION.GLES2_2_0 import *
+from OpenGL.GLES3.VERSION.GLES3_3_0 import *
+from OpenGL import GL as gl
+
+from ctypes import c_int, c_char_p, c_void_p, cdll, POINTER, util, \
+    pointer, CFUNCTYPE, c_bool
+
+
+def getEGLNativeDisplay():
+    _x11lib = cdll.LoadLibrary(util.find_library('X11'))
+    XOpenDisplay = _x11lib.XOpenDisplay
+    XOpenDisplay.argtypes = [c_char_p]
+    XOpenDisplay.restype = POINTER(EGLNativeDisplayType)
+
+    xdpy = XOpenDisplay(None)
+
+
+# Hack. PyOpenGL doesn't seem to manage to find glEGLImageTargetTexture2DOES.
+def getglEGLImageTargetTexture2DOES():
+    funcptr = eglGetProcAddress('glEGLImageTargetTexture2DOES')
+    prototype = CFUNCTYPE(None, _cs.GLenum, _cs.GLeglImageOES)
+    return prototype(funcptr)
+
+
+glEGLImageTargetTexture2DOES = getglEGLImageTargetTexture2DOES()
+
+
+def str_to_fourcc(str):
+    assert(len(str) == 4)
+    fourcc = 0
+    for i, v in enumerate([ord(c) for c in str]):
+        fourcc |= v << (i * 8)
+    return fourcc
+
+
+def get_gl_extensions():
+    n = GLint()
+    glGetIntegerv(GL_NUM_EXTENSIONS, n)
+    gl_extensions = []
+    for i in range(n.value):
+        gl_extensions.append(gl.glGetStringi(GL_EXTENSIONS, i).decode())
+    return gl_extensions
+
+
+def check_gl_extensions(required_extensions):
+    extensions = get_gl_extensions()
+
+    if False:
+        print('GL EXTENSIONS: ', ' '.join(extensions))
+
+    for ext in required_extensions:
+        if ext not in extensions:
+            raise Exception(ext + ' missing')
+
+
+def get_egl_extensions(egl_display):
+    return eglQueryString(egl_display, EGL_EXTENSIONS).decode().split(' ')
+
+
+def check_egl_extensions(egl_display, required_extensions):
+    extensions = get_egl_extensions(egl_display)
+
+    if False:
+        print('EGL EXTENSIONS: ', ' '.join(extensions))
+
+    for ext in required_extensions:
+        if ext not in extensions:
+            raise Exception(ext + ' missing')