[v3,4/8] ipa: libipa: Add AgcMeanLuminance base class
diff mbox series

Message ID 20240424124917.1250837-5-dan.scally@ideasonboard.com
State New
Headers show
Series
  • Centralise Agc into libipa
Related show

Commit Message

Dan Scally April 24, 2024, 12:49 p.m. UTC
The Agc algorithms for the RkIsp1 and IPU3 IPAs do the same thing in
very large part; following the Rpi IPA's algorithm in spirit with a
few tunable values in that IPA being hardcoded in the libipa ones.
Add a new base class for AgcMeanLuminance which implements the same
algorithm and additionally parses yaml tuning files to inform an IPA
module's Agc algorithm about valid constraint and exposure modes and
their associated bounds.

Reviewed-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Signed-off-by: Daniel Scally <dan.scally@ideasonboard.com>
---
Changes in v3:

	- Finished renaming the class in all references to it (I hope...)
	- Updated documentation to try to be a bit clearer about how to use
	  the base class
	- Made functions const/private where possible
Changes in v2:

	- Renamed the class and files
	- Expanded the documentation
	- Added parseTuningData() so derived classes can call a single function
	  to cover all the parsing in ::init().

 src/ipa/libipa/agc_mean_luminance.cpp | 590 ++++++++++++++++++++++++++
 src/ipa/libipa/agc_mean_luminance.h   |  90 ++++
 src/ipa/libipa/meson.build            |   2 +
 3 files changed, 682 insertions(+)
 create mode 100644 src/ipa/libipa/agc_mean_luminance.cpp
 create mode 100644 src/ipa/libipa/agc_mean_luminance.h

Comments

Jacopo Mondi April 24, 2024, 5:28 p.m. UTC | #1
Hi Dan

On Wed, Apr 24, 2024 at 01:49:13PM +0100, Daniel Scally wrote:
> The Agc algorithms for the RkIsp1 and IPU3 IPAs do the same thing in
> very large part; following the Rpi IPA's algorithm in spirit with a
> few tunable values in that IPA being hardcoded in the libipa ones.
> Add a new base class for AgcMeanLuminance which implements the same
> algorithm and additionally parses yaml tuning files to inform an IPA
> module's Agc algorithm about valid constraint and exposure modes and
> their associated bounds.
>
> Reviewed-by: Stefan Klug <stefan.klug@ideasonboard.com>
> Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
> Signed-off-by: Daniel Scally <dan.scally@ideasonboard.com>
> ---
> Changes in v3:
>
> 	- Finished renaming the class in all references to it (I hope...)
> 	- Updated documentation to try to be a bit clearer about how to use
> 	  the base class
> 	- Made functions const/private where possible
> Changes in v2:
>
> 	- Renamed the class and files
> 	- Expanded the documentation
> 	- Added parseTuningData() so derived classes can call a single function
> 	  to cover all the parsing in ::init().
>
>  src/ipa/libipa/agc_mean_luminance.cpp | 590 ++++++++++++++++++++++++++
>  src/ipa/libipa/agc_mean_luminance.h   |  90 ++++
>  src/ipa/libipa/meson.build            |   2 +
>  3 files changed, 682 insertions(+)
>  create mode 100644 src/ipa/libipa/agc_mean_luminance.cpp
>  create mode 100644 src/ipa/libipa/agc_mean_luminance.h
>
> diff --git a/src/ipa/libipa/agc_mean_luminance.cpp b/src/ipa/libipa/agc_mean_luminance.cpp
> new file mode 100644
> index 00000000..7305db71
> --- /dev/null
> +++ b/src/ipa/libipa/agc_mean_luminance.cpp
> @@ -0,0 +1,590 @@
> +/* SPDX-License-Identifier: LGPL-2.1-or-later */
> +/*
> + * Copyright (C) 2024 Ideas on Board Oy
> + *
> + * agc_mean_luminance.cpp - Base class for mean luminance AGC algorithms
> + */
> +
> +#include "agc_mean_luminance.h"
> +
> +#include <cmath>
> +
> +#include <libcamera/base/log.h>
> +#include <libcamera/control_ids.h>
> +
> +#include "exposure_mode_helper.h"
> +
> +using namespace libcamera::controls;
> +
> +/**
> + * \file agc_mean_luminance.h
> + * \brief Base class implementing mean luminance AEGC
> + */
> +
> +namespace libcamera {
> +
> +using namespace std::literals::chrono_literals;
> +
> +LOG_DEFINE_CATEGORY(AgcMeanLuminance)
> +
> +namespace ipa {
> +
> +/*
> + * Number of frames for which to run the algorithm at full speed, before slowing
> + * down to prevent large and jarring changes in exposure from frame to frame.
> + */
> +static constexpr uint32_t kNumStartupFrames = 10;
> +
> +/*
> + * Default relative luminance target
> + *
> + * This value should be chosen so that when the camera points at a grey target,
> + * the resulting image brightness looks "right". Custom values can be passed
> + * as the relativeLuminanceTarget value in sensor tuning files.
> + */
> +static constexpr double kDefaultRelativeLuminanceTarget = 0.16;
> +
> +/**
> + * \struct AgcMeanLuminance::AgcConstraint
> + * \brief The boundaries and target for an AeConstraintMode constraint
> + *
> + * This structure describes an AeConstraintMode constraint for the purposes of
> + * this algorithm. These constraints are expressed as a pair of quantile
> + * boundaries for a histogram, along with a luminance target and a bounds-type.
> + * The algorithm uses the constraints by ensuring that the defined portion of a
> + * luminance histogram (I.E. lying between the two quantiles) is above or below
> + * the given luminance value.
> + */
> +
> +/**
> + * \enum AgcMeanLuminance::AgcConstraint::Bound
> + * \brief Specify whether the constraint defines a lower or upper bound
> + * \var AgcMeanLuminance::AgcConstraint::lower
> + * \brief The constraint defines a lower bound
> + * \var AgcMeanLuminance::AgcConstraint::upper
> + * \brief The constraint defines an upper bound
> + */
> +
> +/**
> + * \var AgcMeanLuminance::AgcConstraint::bound
> + * \brief The type of constraint bound
> + */
> +
> +/**
> + * \var AgcMeanLuminance::AgcConstraint::qLo
> + * \brief The lower quantile to use for the constraint
> + */
> +
> +/**
> + * \var AgcMeanLuminance::AgcConstraint::qHi
> + * \brief The upper quantile to use for the constraint
> + */
> +
> +/**
> + * \var AgcMeanLuminance::AgcConstraint::yTarget
> + * \brief The luminance target for the constraint
> + */
> +
> +/**
> + * \class AgcMeanLuminance
> + * \brief A mean-based auto-exposure algorithm
> + *
> + * This algorithm calculates a shutter time, analogue and digital gain such that
> + * the normalised mean luminance value of an image is driven towards a target,
> + * which itself is discovered from tuning data. The algorithm is a two-stage
> + * process.
> + *
> + * In the first stage, an initial gain value is derived by iteratively comparing
> + * the gain-adjusted mean luminance across an entire image against a target, and

s/an entire/the entire/

> + * selecting a value which pushes it as closely as possible towards the target.
> + *
> + * In the second stage we calculate the gain required to drive the average of a
> + * section of a histogram to a target value, where the target and the boundaries
> + * of the section of the histogram used in the calculation are taken from the
> + * values defined for the currently configured AeConstraintMode within the
> + * tuning data. This class provides a helper function to parse those tuning data
> + * to discover the constraints, and so requires a specific format for those
> + * data which is described in \ref parseTuningData(). The gain from the first
> + * stage is then clamped to the gain from this stage.

I'm not sure I would keep "The gain from..."

> + *
> + * The final gain is used to adjust the effective exposure value of the image,
> + * and that new exposure value is divided into shutter time, analogue gain and
> + * digital gain according to the selected AeExposureMode. This class uses the
> + * \ref ExposureModeHelper class to assist in that division, and expects the
> + * data needed to initialise that class to be present in tuning data in a
> + * format described in \ref parseTuningData().
> + *
> + * In order to be able to derive an AGC implementation from this class, an IPA

Or

In order to be able to use this algorithm and IPA module needs to be
able to do the following:

As we talk about inheritance and how to derive this class below

> + * needs to be able to do the following:
> + *
> + * 1. Provide a luminance estimation across an entire image.
> + * 2. Provide a luminance Histogram for the image to use in calculating
> + *    constraint compliance. The precision of the Histogram that is available
> + *    will determine the supportable precision of the constraints.
> + *
> + * Users should derive their Agc class from this class as well as their IPA's

s/User/IPA modules that want to use this class to implement their AEGC
algorithm/

s/should derive their Agc class from this class/should derive it/

s/as well as their IPA's Algorithm class//


> + * Algorithm class, and provide an overriding estimateLuminance() function for
> + * this class to use. They must call parseTuningData() in init(), and must also
> + * call setLimits() and resetFrameCounter() in configure(). They may then use
> + * calculateNewEv() in process(). If the limits passed to setLimits() change for
> + * any reason (for example, in response to a FrameDurationLimit control being
> + * passed in queueRequest()) then setLimits() must be called again with the new
> + * values.
> + */
> +
> +AgcMeanLuminance::AgcMeanLuminance()
> +	: frameCount_(0), filteredExposure_(0s), relativeLuminanceTarget_(0)
> +{
> +}
> +
> +/**
> + * \brief Parse the relative luminance target from the tuning data
> + * \param[in] tuningData The YamlObject holding the algorithm's tuning data
> + */

This and the below functions are private. Comments are always nice,
but I think you can de-doxygen them (it doens't hurt though)

> +void AgcMeanLuminance::parseRelativeLuminanceTarget(const YamlObject &tuningData)
> +{
> +	relativeLuminanceTarget_ =
> +		tuningData["relativeLuminanceTarget"].get<double>(kDefaultRelativeLuminanceTarget);
> +}
> +
> +/**
> + * \brief Parse an AeConstraintMode constraint from tuning data
> + * \param[in] modeDict the YamlObject holding the constraint data
> + * \param[in] id The constraint ID from AeConstraintModeEnum
> + */
> +void AgcMeanLuminance::parseConstraint(const YamlObject &modeDict, int32_t id)
> +{
> +	for (const auto &[boundName, content] : modeDict.asDict()) {
> +		if (boundName != "upper" && boundName != "lower") {
> +			LOG(AgcMeanLuminance, Warning)
> +				<< "Ignoring unknown constraint bound '" << boundName << "'";
> +			continue;
> +		}
> +
> +		unsigned int idx = static_cast<unsigned int>(boundName == "upper");
> +		AgcConstraint::Bound bound = static_cast<AgcConstraint::Bound>(idx);
> +		double qLo = content["qLo"].get<double>().value_or(0.98);
> +		double qHi = content["qHi"].get<double>().value_or(1.0);
> +		double yTarget =
> +			content["yTarget"].getList<double>().value_or(std::vector<double>{ 0.5 }).at(0);
> +
> +		AgcConstraint constraint = { bound, qLo, qHi, yTarget };
> +
> +		if (!constraintModes_.count(id))
> +			constraintModes_[id] = {};
> +
> +		if (idx)
> +			constraintModes_[id].push_back(constraint);
> +		else
> +			constraintModes_[id].insert(constraintModes_[id].begin(), constraint);
> +	}
> +}
> +
> +int AgcMeanLuminance::parseConstraintModes(const YamlObject &tuningData)
> +{
> +	std::vector<ControlValue> availableConstraintModes;
> +
> +	const YamlObject &yamlConstraintModes = tuningData[controls::AeConstraintMode.name()];
> +	if (yamlConstraintModes.isDictionary()) {
> +		for (const auto &[modeName, modeDict] : yamlConstraintModes.asDict()) {
> +			if (AeConstraintModeNameValueMap.find(modeName) ==
> +			    AeConstraintModeNameValueMap.end()) {
> +				LOG(AgcMeanLuminance, Warning)
> +					<< "Skipping unknown constraint mode '" << modeName << "'";
> +				continue;
> +			}
> +
> +			if (!modeDict.isDictionary()) {
> +				LOG(AgcMeanLuminance, Error)
> +					<< "Invalid constraint mode '" << modeName << "'";
> +				return -EINVAL;
> +			}
> +
> +			parseConstraint(modeDict,
> +					AeConstraintModeNameValueMap.at(modeName));
> +			availableConstraintModes.push_back(
> +				AeConstraintModeNameValueMap.at(modeName));
> +		}
> +	}
> +
> +	/*
> +	 * If the tuning data file contains no constraints then we use the
> +	 * default constraint that the IPU3/RkISP1 Agc algorithms were adhering
> +	 * to anyway before centralisation; this constraint forces the top 2% of
> +	 * the histogram to be at least 0.5.
> +	 */
> +	if (constraintModes_.empty()) {
> +		AgcConstraint constraint = {
> +			AgcConstraint::Bound::lower,
> +			0.98,
> +			1.0,
> +			0.5
> +		};
> +
> +		constraintModes_[controls::ConstraintNormal].insert(
> +			constraintModes_[controls::ConstraintNormal].begin(),
> +			constraint);
> +		availableConstraintModes.push_back(
> +			AeConstraintModeNameValueMap.at("ConstraintNormal"));
> +	}
> +
> +	controls_[&controls::AeConstraintMode] = ControlInfo(availableConstraintModes);
> +
> +	return 0;
> +}
> +
> +int AgcMeanLuminance::parseExposureModes(const YamlObject &tuningData)
> +{
> +	std::vector<ControlValue> availableExposureModes;
> +
> +	const YamlObject &yamlExposureModes = tuningData[controls::AeExposureMode.name()];
> +	if (yamlExposureModes.isDictionary()) {
> +		for (const auto &[modeName, modeValues] : yamlExposureModes.asDict()) {
> +			if (AeExposureModeNameValueMap.find(modeName) ==
> +			    AeExposureModeNameValueMap.end()) {
> +				LOG(AgcMeanLuminance, Warning)
> +					<< "Skipping unknown exposure mode '" << modeName << "'";
> +				continue;
> +			}
> +
> +			if (!modeValues.isDictionary()) {
> +				LOG(AgcMeanLuminance, Error)
> +					<< "Invalid exposure mode '" << modeName << "'";
> +				return -EINVAL;
> +			}
> +
> +			std::vector<uint32_t> shutters =
> +				modeValues["shutter"].getList<uint32_t>().value_or(std::vector<uint32_t>{});
> +			std::vector<double> gains =
> +				modeValues["gain"].getList<double>().value_or(std::vector<double>{});
> +
> +			if (shutters.size() != gains.size()) {
> +				LOG(AgcMeanLuminance, Error)
> +					<< "Shutter and gain array sizes unequal";
> +				return -EINVAL;
> +			}
> +
> +			if (shutters.empty()) {
> +				LOG(AgcMeanLuminance, Error)
> +					<< "Shutter and gain arrays are empty";
> +				return -EINVAL;
> +			}
> +
> +			std::vector<std::pair<utils::Duration, double>> stages;
> +			for (unsigned int i = 0; i < shutters.size(); i++) {
> +				stages.push_back({
> +					std::chrono::microseconds(shutters[i]),
> +					gains[i]
> +				});
> +			}
> +
> +			std::shared_ptr<ExposureModeHelper> helper =
> +				std::make_shared<ExposureModeHelper>(stages);

Using a shared_ptr<> avoids copying the newly created instance inside
the map but it's maybe an overkill.

I don't remember if you went back and forth from passing stages to the
helper at construction or init time already, if that's the case, just
ignore me, but creating the helper instance in the map and init it with stages
just after would have saved usage of smart pointers.

> +
> +			exposureModeHelpers_[AeExposureModeNameValueMap.at(modeName)] = helper;
> +			availableExposureModes.push_back(AeExposureModeNameValueMap.at(modeName));
> +		}
> +	}
> +
> +	/*
> +	 * If we don't have any exposure modes in the tuning data we create an
> +	 * ExposureModeHelper using an empty vector of stages. This will result
> +	 * in the ExposureModeHelper simply driving the shutter as high as
> +	 * possible before touching gain.
> +	 */
> +	if (availableExposureModes.empty()) {
> +		int32_t exposureModeId = AeExposureModeNameValueMap.at("ExposureNormal");
> +		std::vector<std::pair<utils::Duration, double>> stages = { };
> +
> +		std::shared_ptr<ExposureModeHelper> helper =
> +			std::make_shared<ExposureModeHelper>(stages);
> +
> +		exposureModeHelpers_[exposureModeId] = helper;
> +		availableExposureModes.push_back(exposureModeId);
> +	}
> +
> +	controls_[&controls::AeExposureMode] = ControlInfo(availableExposureModes);
> +
> +	return 0;
> +}
> +
> +/**
> + * \brief Parse tuning data for AeConstraintMode and AeExposureMode controls
> + * \param[in] tuningData the YamlObject representing the tuning data
> + *
> + * This function parses tuning data to build the list of allowed values for the
> + * AeConstraintMode and AeExposureMode controls. Those tuning data must provide
> + * the data in a specific format; the Agc algorithm's tuning data should contain
> + * a dictionary called AeConstraintMode containing per-mode setting dictionaries
> + * with the key being a value from \ref controls::AeConstraintModeNameValueMap.
> + * Each mode dict may contain either a "lower" or "upper" key or both, for
> + * example:
> + *
> + * \code{.unparsed}
> + * algorithms:
> + *   - Agc:
> + *       AeConstraintMode:
> + *         ConstraintNormal:
> + *           lower:
> + *             qLo: 0.98
> + *             qHi: 1.0
> + *             yTarget: 0.5
> + *         ConstraintHighlight:
> + *           lower:
> + *             qLo: 0.98
> + *             qHi: 1.0
> + *             yTarget: 0.5
> + *           upper:
> + *             qLo: 0.98
> + *             qHi: 1.0
> + *             yTarget: 0.8
> + *
> + * \endcode
> + *
> + * For the AeExposureMode control the data should contain a dictionary called
> + * AeExposureMode containing per-mode setting dictionaries with the key being a
> + * value from \ref controls::AeExposureModeNameValueMap. Each mode dict should
> + * contain an array of shutter times with the key "shutter" and an array of gain
> + * values with the key "gain", in this format:
> + *
> + * \code{.unparsed}
> + * algorithms:
> + *   - Agc:
> + *       AeExposureMode:
> + *         ExposureNormal:
> + *           shutter: [ 100, 10000, 30000, 60000, 120000 ]
> + *           gain: [ 2.0, 4.0, 6.0, 8.0, 10.0 ]
> + *         ExposureShort:
> + *           shutter: [ 100, 10000, 30000, 60000, 120000 ]
> + *           gain: [ 2.0, 4.0, 6.0, 8.0, 10.0 ]
> + *
> + * \endcode
> + *
> + * \return 0 on success or a negative error code
> + */
> +int AgcMeanLuminance::parseTuningData(const YamlObject &tuningData)
> +{
> +	int ret;
> +
> +	parseRelativeLuminanceTarget(tuningData);
> +
> +	ret = parseConstraintModes(tuningData);
> +	if (ret)
> +		return ret;
> +
> +	ret = parseExposureModes(tuningData);

or just

        return parseExposureModes(tuningData);

> +	if (ret)
> +		return ret;
> +
> +	return 0;
> +}
> +
> +/**
> + * \brief Set the ExposureModeHelper limits for this class
> + * \param[in] minShutter Minimum shutter time to allow
> + * \param[in] maxShutter Maximum shutter time to allow
> + * \param[in] minGain Minimum gain to allow
> + * \param[in] maxGain Maximum gain to allow
> + *
> + * This function calls \ref ExposureModeHelper::setLimits() for each
> + * ExposureModeHelper that has been created for this class.
> + */
> +void AgcMeanLuminance::setLimits(utils::Duration minShutter,
> +				 utils::Duration maxShutter,
> +				 double minGain, double maxGain)
> +{
> +	for (auto &[id, helper] : exposureModeHelpers_)
> +		helper->setLimits(minShutter, maxShutter, minGain, maxGain);
> +}
> +
> +/**
> + * \fn AgcMeanLuminance::constraintModes()
> + * \brief Get the constraint modes that have been parsed from tuning data
> + */
> +
> +/**
> + * \fn AgcMeanLuminance::exposureModeHelpers()
> + * \brief Get the ExposureModeHelpers that have been parsed from tuning data
> + */
> +
> +/**
> + * \fn AgcMeanLuminance::controls()
> + * \brief Get the controls that have been generated after parsing tuning data
> + */
> +
> +/**
> + * \fn AgcMeanLuminance::estimateLuminance(const double gain)
> + * \brief Estimate the luminance of an image, adjusted by a given gain
> + * \param[in] gain The gain with which to adjust the luminance estimate
> + *
> + * This function estimates the average relative luminance of the frame that
> + * would be output by the sensor if an additional \a gain was applied. It is a
> + * pure virtual function because estimation of luminance is a hardware-specific
> + * operation, which depends wholly on the format of the stats that are delivered
> + * to libcamera from the ISP. Derived classes must override this function with
> + * one that calculates the normalised mean luminance value across the entire
> + * image.
> + *
> + * \return The normalised relative luminance of the image
> + */
> +
> +/**
> + * \brief Estimate the initial gain needed to achieve a relative luminance
> + * target
> + * \return The calculated initial gain
> + */
> +double AgcMeanLuminance::estimateInitialGain() const
> +{
> +	double yTarget = relativeLuminanceTarget_;
> +	double yGain = 1.0;
> +
> +	/*
> +	* To account for non-linearity caused by saturation, the value needs to
> +	* be estimated in an iterative process, as multiplying by a gain will
> +	* not increase the relative luminance by the same factor if some image
> +	* regions are saturated.
> +	*/
> +	for (unsigned int i = 0; i < 8; i++) {
> +		double yValue = estimateLuminance(yGain);
> +		double extra_gain = std::min(10.0, yTarget / (yValue + .001));
> +
> +		yGain *= extra_gain;
> +		LOG(AgcMeanLuminance, Debug) << "Y value: " << yValue
> +				<< ", Y target: " << yTarget
> +				<< ", gives gain " << yGain;
> +
> +		if (utils::abs_diff(extra_gain, 1.0) < 0.01)
> +			break;
> +	}
> +
> +	return yGain;
> +}
> +
> +/**
> + * \brief Clamp gain within the bounds of a defined constraint
> + * \param[in] constraintModeIndex The index of the constraint to adhere to
> + * \param[in] hist A histogram over which to calculate inter-quantile means
> + * \param[in] gain The gain to clamp
> + *
> + * \return The gain clamped within the constraint bounds
> + */
> +double AgcMeanLuminance::constraintClampGain(uint32_t constraintModeIndex,
> +					     const Histogram &hist,
> +					     double gain)
> +{
> +	std::vector<AgcConstraint> &constraints = constraintModes_[constraintModeIndex];
> +	for (const AgcConstraint &constraint : constraints) {
> +		double newGain = constraint.yTarget * hist.bins() /
> +				 hist.interQuantileMean(constraint.qLo, constraint.qHi);
> +
> +		if (constraint.bound == AgcConstraint::Bound::lower &&
> +		    newGain > gain)
> +			gain = newGain;
> +
> +		if (constraint.bound == AgcConstraint::Bound::upper &&
> +		    newGain < gain)
> +			gain = newGain;
> +	}
> +
> +	return gain;
> +}
> +
> +/**
> + * \brief Apply a filter on the exposure value to limit the speed of changes
> + * \param[in] exposureValue The target exposure from the AGC algorithm
> + *
> + * The speed of the filter is adaptive, and will produce the target quicker
> + * during startup, or when the target exposure is within 20% of the most recent
> + * filter output.
> + *
> + * \return The filtered exposure
> + */
> +utils::Duration AgcMeanLuminance::filterExposure(utils::Duration exposureValue)
> +{
> +	double speed = 0.2;
> +
> +	/* Adapt instantly if we are in startup phase. */
> +	if (frameCount_ < kNumStartupFrames)
> +		speed = 1.0;
> +
> +	/*
> +	 * If we are close to the desired result, go faster to avoid making
> +	 * multiple micro-adjustments.
> +	 * \todo Make this customisable?
> +	 */
> +	if (filteredExposure_ < 1.2 * exposureValue &&
> +	    filteredExposure_ > 0.8 * exposureValue)
> +		speed = sqrt(speed);
> +
> +	filteredExposure_ = speed * exposureValue +
> +			    filteredExposure_ * (1.0 - speed);
> +
> +	return filteredExposure_;
> +}
> +
> +/**
> + * \brief Calculate the new exposure value

 and split it between shutter times and gains

> + * \param[in] constraintModeIndex The index of the current constraint mode
> + * \param[in] exposureModeIndex The index of the current exposure mode
> + * \param[in] yHist A Histogram from the ISP statistics to use in constraining
> + * the calculated gain
> + * \param[in] effectiveExposureValue The EV applied to the frame from which the
> + * statistics in use derive
> + *
> + * Calculate a new exposure value to try to obtain the target. The calculated
> + * exposure value is filtered to prevent rapid changes from frame to frame, and
> + * divided into shutter time, analogue and digital gain.
> + *
> + * \return Tuple of shutter time, analogue gain, and digital gain
> + */
> +std::tuple<utils::Duration, double, double>
> +AgcMeanLuminance::calculateNewEv(uint32_t constraintModeIndex,
> +				 uint32_t exposureModeIndex,
> +				 const Histogram &yHist,
> +				 utils::Duration effectiveExposureValue)
> +{
> +	/*
> +	 * The pipeline handler should validate that we have received an allowed
> +	 * value for AeExposureMode.
> +	 */
> +	std::shared_ptr<ExposureModeHelper> exposureModeHelper =
> +		exposureModeHelpers_.at(exposureModeIndex);
> +
> +	double gain = estimateInitialGain();
> +	gain = constraintClampGain(constraintModeIndex, yHist, gain);
> +
> +	/*
> +	 * We don't check whether we're already close to the target, because
> +	 * even if the effective exposure value is the same as the last frame's
> +	 * we could have switched to an exposure mode that would require a new
> +	 * pass through the splitExposure() function.
> +	 */
> +
> +	utils::Duration newExposureValue = effectiveExposureValue * gain;
> +	utils::Duration maxTotalExposure = exposureModeHelper->maxShutter()
> +					   * exposureModeHelper->maxGain();
> +	newExposureValue = std::min(newExposureValue, maxTotalExposure);
> +
> +	/*
> +	 * We filter the exposure value to make sure changes are not too jarring
> +	 * from frame to frame.
> +	 */
> +	newExposureValue = filterExposure(newExposureValue);
> +
> +	frameCount_++;
> +	return exposureModeHelper->splitExposure(newExposureValue);
> +}
> +
> +/**
> + * \fn AgcMeanLuminance::resetFrameCount()
> + * \brief Reset the frame counter
> + *
> + * This function resets the internal frame counter, which exists to help the
> + * algorithm decide whether it should respond instantly or not. The expectation
> + * is for derived classes to call this function before each camera start call,
> + * either in configure() or queueRequest() if the frame number is zero.

Isn't it enough to do it in configure() ?

> + */
> +
> +}; /* namespace ipa */
> +
> +}; /* namespace libcamera */
> diff --git a/src/ipa/libipa/agc_mean_luminance.h b/src/ipa/libipa/agc_mean_luminance.h
> new file mode 100644
> index 00000000..9f91e08f
> --- /dev/null
> +++ b/src/ipa/libipa/agc_mean_luminance.h
> @@ -0,0 +1,90 @@
> +/* SPDX-License-Identifier: LGPL-2.1-or-later */
> +/*
> + * Copyright (C) 2024 Ideas on Board Oy
> + *
> + agc_mean_luminance.h - Base class for mean luminance AGC algorithms
> + */
> +
> +#pragma once
> +

#include <map>
#include <memory> for shared_ptr


> +#include <tuple>
> +#include <vector>
> +
> +#include <libcamera/controls.h>
> +
> +#include "libcamera/internal/yaml_parser.h"
> +
> +#include "exposure_mode_helper.h"
> +#include "histogram.h"
> +
> +namespace libcamera {
> +
> +namespace ipa {
> +
> +class AgcMeanLuminance
> +{
> +public:
> +	AgcMeanLuminance();
> +	virtual ~AgcMeanLuminance() = default;
> +
> +	struct AgcConstraint {
> +		enum class Bound {
> +			lower = 0,
> +			upper = 1
> +		};
> +		Bound bound;
> +		double qLo;
> +		double qHi;
> +		double yTarget;
> +	};
> +
> +	int parseTuningData(const YamlObject &tuningData);
> +
> +	void setLimits(utils::Duration minShutter, utils::Duration maxShutter,
> +		       double minGain, double maxGain);
> +
> +	std::map<int32_t, std::vector<AgcConstraint>> constraintModes()
> +	{
> +		return constraintModes_;
> +	}
> +
> +	std::map<int32_t, std::shared_ptr<ExposureModeHelper>> exposureModeHelpers()
> +	{
> +		return exposureModeHelpers_;
> +	}
> +
> +	ControlInfoMap::Map controls()
> +	{
> +		return controls_;
> +	}
> +
> +	std::tuple<utils::Duration, double, double>
> +	calculateNewEv(uint32_t constraintModeIndex, uint32_t exposureModeIndex,
> +		       const Histogram &yHist, utils::Duration effectiveExposureValue);
> +	void resetFrameCount() { frameCount_ = 0; }

Other inline functions are

	void resetFrameCount()
        {
                frameCount_ = 0;
        }

All minors
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>

Thanks
  j


> +
> +private:
> +	virtual double estimateLuminance(const double gain) const = 0;
> +
> +	void parseRelativeLuminanceTarget(const YamlObject &tuningData);
> +	void parseConstraint(const YamlObject &modeDict, int32_t id);
> +	int parseConstraintModes(const YamlObject &tuningData);
> +	int parseExposureModes(const YamlObject &tuningData);
> +	double estimateInitialGain() const;
> +	double constraintClampGain(uint32_t constraintModeIndex,
> +				   const Histogram &hist,
> +				   double gain);
> +	utils::Duration filterExposure(utils::Duration exposureValue);
> +
> +	uint64_t frameCount_;
> +	utils::Duration filteredExposure_;
> +	double relativeLuminanceTarget_;
> +
> +	std::map<int32_t, std::vector<AgcConstraint>> constraintModes_;
> +	std::map<int32_t, std::shared_ptr<ExposureModeHelper>> exposureModeHelpers_;
> +	ControlInfoMap::Map controls_;
> +};
> +
> +}; /* namespace ipa */
> +
> +}; /* namespace libcamera */
> diff --git a/src/ipa/libipa/meson.build b/src/ipa/libipa/meson.build
> index 37fbd177..7ce885da 100644
> --- a/src/ipa/libipa/meson.build
> +++ b/src/ipa/libipa/meson.build
> @@ -1,6 +1,7 @@
>  # SPDX-License-Identifier: CC0-1.0
>
>  libipa_headers = files([
> +    'agc_mean_luminance.h',
>      'algorithm.h',
>      'camera_sensor_helper.h',
>      'exposure_mode_helper.h',
> @@ -10,6 +11,7 @@ libipa_headers = files([
>  ])
>
>  libipa_sources = files([
> +    'agc_mean_luminance.cpp',
>      'algorithm.cpp',
>      'camera_sensor_helper.cpp',
>      'exposure_mode_helper.cpp',
> --
> 2.34.1
>

Patch
diff mbox series

diff --git a/src/ipa/libipa/agc_mean_luminance.cpp b/src/ipa/libipa/agc_mean_luminance.cpp
new file mode 100644
index 00000000..7305db71
--- /dev/null
+++ b/src/ipa/libipa/agc_mean_luminance.cpp
@@ -0,0 +1,590 @@ 
+/* SPDX-License-Identifier: LGPL-2.1-or-later */
+/*
+ * Copyright (C) 2024 Ideas on Board Oy
+ *
+ * agc_mean_luminance.cpp - Base class for mean luminance AGC algorithms
+ */
+
+#include "agc_mean_luminance.h"
+
+#include <cmath>
+
+#include <libcamera/base/log.h>
+#include <libcamera/control_ids.h>
+
+#include "exposure_mode_helper.h"
+
+using namespace libcamera::controls;
+
+/**
+ * \file agc_mean_luminance.h
+ * \brief Base class implementing mean luminance AEGC
+ */
+
+namespace libcamera {
+
+using namespace std::literals::chrono_literals;
+
+LOG_DEFINE_CATEGORY(AgcMeanLuminance)
+
+namespace ipa {
+
+/*
+ * Number of frames for which to run the algorithm at full speed, before slowing
+ * down to prevent large and jarring changes in exposure from frame to frame.
+ */
+static constexpr uint32_t kNumStartupFrames = 10;
+
+/*
+ * Default relative luminance target
+ *
+ * This value should be chosen so that when the camera points at a grey target,
+ * the resulting image brightness looks "right". Custom values can be passed
+ * as the relativeLuminanceTarget value in sensor tuning files.
+ */
+static constexpr double kDefaultRelativeLuminanceTarget = 0.16;
+
+/**
+ * \struct AgcMeanLuminance::AgcConstraint
+ * \brief The boundaries and target for an AeConstraintMode constraint
+ *
+ * This structure describes an AeConstraintMode constraint for the purposes of
+ * this algorithm. These constraints are expressed as a pair of quantile
+ * boundaries for a histogram, along with a luminance target and a bounds-type.
+ * The algorithm uses the constraints by ensuring that the defined portion of a
+ * luminance histogram (I.E. lying between the two quantiles) is above or below
+ * the given luminance value.
+ */
+
+/**
+ * \enum AgcMeanLuminance::AgcConstraint::Bound
+ * \brief Specify whether the constraint defines a lower or upper bound
+ * \var AgcMeanLuminance::AgcConstraint::lower
+ * \brief The constraint defines a lower bound
+ * \var AgcMeanLuminance::AgcConstraint::upper
+ * \brief The constraint defines an upper bound
+ */
+
+/**
+ * \var AgcMeanLuminance::AgcConstraint::bound
+ * \brief The type of constraint bound
+ */
+
+/**
+ * \var AgcMeanLuminance::AgcConstraint::qLo
+ * \brief The lower quantile to use for the constraint
+ */
+
+/**
+ * \var AgcMeanLuminance::AgcConstraint::qHi
+ * \brief The upper quantile to use for the constraint
+ */
+
+/**
+ * \var AgcMeanLuminance::AgcConstraint::yTarget
+ * \brief The luminance target for the constraint
+ */
+
+/**
+ * \class AgcMeanLuminance
+ * \brief A mean-based auto-exposure algorithm
+ *
+ * This algorithm calculates a shutter time, analogue and digital gain such that
+ * the normalised mean luminance value of an image is driven towards a target,
+ * which itself is discovered from tuning data. The algorithm is a two-stage
+ * process.
+ *
+ * In the first stage, an initial gain value is derived by iteratively comparing
+ * the gain-adjusted mean luminance across an entire image against a target, and
+ * selecting a value which pushes it as closely as possible towards the target.
+ *
+ * In the second stage we calculate the gain required to drive the average of a
+ * section of a histogram to a target value, where the target and the boundaries
+ * of the section of the histogram used in the calculation are taken from the
+ * values defined for the currently configured AeConstraintMode within the
+ * tuning data. This class provides a helper function to parse those tuning data
+ * to discover the constraints, and so requires a specific format for those
+ * data which is described in \ref parseTuningData(). The gain from the first
+ * stage is then clamped to the gain from this stage.
+ *
+ * The final gain is used to adjust the effective exposure value of the image,
+ * and that new exposure value is divided into shutter time, analogue gain and
+ * digital gain according to the selected AeExposureMode. This class uses the
+ * \ref ExposureModeHelper class to assist in that division, and expects the
+ * data needed to initialise that class to be present in tuning data in a
+ * format described in \ref parseTuningData().
+ *
+ * In order to be able to derive an AGC implementation from this class, an IPA
+ * needs to be able to do the following:
+ *
+ * 1. Provide a luminance estimation across an entire image.
+ * 2. Provide a luminance Histogram for the image to use in calculating
+ *    constraint compliance. The precision of the Histogram that is available
+ *    will determine the supportable precision of the constraints.
+ *
+ * Users should derive their Agc class from this class as well as their IPA's
+ * Algorithm class, and provide an overriding estimateLuminance() function for
+ * this class to use. They must call parseTuningData() in init(), and must also
+ * call setLimits() and resetFrameCounter() in configure(). They may then use
+ * calculateNewEv() in process(). If the limits passed to setLimits() change for
+ * any reason (for example, in response to a FrameDurationLimit control being
+ * passed in queueRequest()) then setLimits() must be called again with the new
+ * values.
+ */
+
+AgcMeanLuminance::AgcMeanLuminance()
+	: frameCount_(0), filteredExposure_(0s), relativeLuminanceTarget_(0)
+{
+}
+
+/**
+ * \brief Parse the relative luminance target from the tuning data
+ * \param[in] tuningData The YamlObject holding the algorithm's tuning data
+ */
+void AgcMeanLuminance::parseRelativeLuminanceTarget(const YamlObject &tuningData)
+{
+	relativeLuminanceTarget_ =
+		tuningData["relativeLuminanceTarget"].get<double>(kDefaultRelativeLuminanceTarget);
+}
+
+/**
+ * \brief Parse an AeConstraintMode constraint from tuning data
+ * \param[in] modeDict the YamlObject holding the constraint data
+ * \param[in] id The constraint ID from AeConstraintModeEnum
+ */
+void AgcMeanLuminance::parseConstraint(const YamlObject &modeDict, int32_t id)
+{
+	for (const auto &[boundName, content] : modeDict.asDict()) {
+		if (boundName != "upper" && boundName != "lower") {
+			LOG(AgcMeanLuminance, Warning)
+				<< "Ignoring unknown constraint bound '" << boundName << "'";
+			continue;
+		}
+
+		unsigned int idx = static_cast<unsigned int>(boundName == "upper");
+		AgcConstraint::Bound bound = static_cast<AgcConstraint::Bound>(idx);
+		double qLo = content["qLo"].get<double>().value_or(0.98);
+		double qHi = content["qHi"].get<double>().value_or(1.0);
+		double yTarget =
+			content["yTarget"].getList<double>().value_or(std::vector<double>{ 0.5 }).at(0);
+
+		AgcConstraint constraint = { bound, qLo, qHi, yTarget };
+
+		if (!constraintModes_.count(id))
+			constraintModes_[id] = {};
+
+		if (idx)
+			constraintModes_[id].push_back(constraint);
+		else
+			constraintModes_[id].insert(constraintModes_[id].begin(), constraint);
+	}
+}
+
+int AgcMeanLuminance::parseConstraintModes(const YamlObject &tuningData)
+{
+	std::vector<ControlValue> availableConstraintModes;
+
+	const YamlObject &yamlConstraintModes = tuningData[controls::AeConstraintMode.name()];
+	if (yamlConstraintModes.isDictionary()) {
+		for (const auto &[modeName, modeDict] : yamlConstraintModes.asDict()) {
+			if (AeConstraintModeNameValueMap.find(modeName) ==
+			    AeConstraintModeNameValueMap.end()) {
+				LOG(AgcMeanLuminance, Warning)
+					<< "Skipping unknown constraint mode '" << modeName << "'";
+				continue;
+			}
+
+			if (!modeDict.isDictionary()) {
+				LOG(AgcMeanLuminance, Error)
+					<< "Invalid constraint mode '" << modeName << "'";
+				return -EINVAL;
+			}
+
+			parseConstraint(modeDict,
+					AeConstraintModeNameValueMap.at(modeName));
+			availableConstraintModes.push_back(
+				AeConstraintModeNameValueMap.at(modeName));
+		}
+	}
+
+	/*
+	 * If the tuning data file contains no constraints then we use the
+	 * default constraint that the IPU3/RkISP1 Agc algorithms were adhering
+	 * to anyway before centralisation; this constraint forces the top 2% of
+	 * the histogram to be at least 0.5.
+	 */
+	if (constraintModes_.empty()) {
+		AgcConstraint constraint = {
+			AgcConstraint::Bound::lower,
+			0.98,
+			1.0,
+			0.5
+		};
+
+		constraintModes_[controls::ConstraintNormal].insert(
+			constraintModes_[controls::ConstraintNormal].begin(),
+			constraint);
+		availableConstraintModes.push_back(
+			AeConstraintModeNameValueMap.at("ConstraintNormal"));
+	}
+
+	controls_[&controls::AeConstraintMode] = ControlInfo(availableConstraintModes);
+
+	return 0;
+}
+
+int AgcMeanLuminance::parseExposureModes(const YamlObject &tuningData)
+{
+	std::vector<ControlValue> availableExposureModes;
+
+	const YamlObject &yamlExposureModes = tuningData[controls::AeExposureMode.name()];
+	if (yamlExposureModes.isDictionary()) {
+		for (const auto &[modeName, modeValues] : yamlExposureModes.asDict()) {
+			if (AeExposureModeNameValueMap.find(modeName) ==
+			    AeExposureModeNameValueMap.end()) {
+				LOG(AgcMeanLuminance, Warning)
+					<< "Skipping unknown exposure mode '" << modeName << "'";
+				continue;
+			}
+
+			if (!modeValues.isDictionary()) {
+				LOG(AgcMeanLuminance, Error)
+					<< "Invalid exposure mode '" << modeName << "'";
+				return -EINVAL;
+			}
+
+			std::vector<uint32_t> shutters =
+				modeValues["shutter"].getList<uint32_t>().value_or(std::vector<uint32_t>{});
+			std::vector<double> gains =
+				modeValues["gain"].getList<double>().value_or(std::vector<double>{});
+
+			if (shutters.size() != gains.size()) {
+				LOG(AgcMeanLuminance, Error)
+					<< "Shutter and gain array sizes unequal";
+				return -EINVAL;
+			}
+
+			if (shutters.empty()) {
+				LOG(AgcMeanLuminance, Error)
+					<< "Shutter and gain arrays are empty";
+				return -EINVAL;
+			}
+
+			std::vector<std::pair<utils::Duration, double>> stages;
+			for (unsigned int i = 0; i < shutters.size(); i++) {
+				stages.push_back({
+					std::chrono::microseconds(shutters[i]),
+					gains[i]
+				});
+			}
+
+			std::shared_ptr<ExposureModeHelper> helper =
+				std::make_shared<ExposureModeHelper>(stages);
+
+			exposureModeHelpers_[AeExposureModeNameValueMap.at(modeName)] = helper;
+			availableExposureModes.push_back(AeExposureModeNameValueMap.at(modeName));
+		}
+	}
+
+	/*
+	 * If we don't have any exposure modes in the tuning data we create an
+	 * ExposureModeHelper using an empty vector of stages. This will result
+	 * in the ExposureModeHelper simply driving the shutter as high as
+	 * possible before touching gain.
+	 */
+	if (availableExposureModes.empty()) {
+		int32_t exposureModeId = AeExposureModeNameValueMap.at("ExposureNormal");
+		std::vector<std::pair<utils::Duration, double>> stages = { };
+
+		std::shared_ptr<ExposureModeHelper> helper =
+			std::make_shared<ExposureModeHelper>(stages);
+
+		exposureModeHelpers_[exposureModeId] = helper;
+		availableExposureModes.push_back(exposureModeId);
+	}
+
+	controls_[&controls::AeExposureMode] = ControlInfo(availableExposureModes);
+
+	return 0;
+}
+
+/**
+ * \brief Parse tuning data for AeConstraintMode and AeExposureMode controls
+ * \param[in] tuningData the YamlObject representing the tuning data
+ *
+ * This function parses tuning data to build the list of allowed values for the
+ * AeConstraintMode and AeExposureMode controls. Those tuning data must provide
+ * the data in a specific format; the Agc algorithm's tuning data should contain
+ * a dictionary called AeConstraintMode containing per-mode setting dictionaries
+ * with the key being a value from \ref controls::AeConstraintModeNameValueMap.
+ * Each mode dict may contain either a "lower" or "upper" key or both, for
+ * example:
+ *
+ * \code{.unparsed}
+ * algorithms:
+ *   - Agc:
+ *       AeConstraintMode:
+ *         ConstraintNormal:
+ *           lower:
+ *             qLo: 0.98
+ *             qHi: 1.0
+ *             yTarget: 0.5
+ *         ConstraintHighlight:
+ *           lower:
+ *             qLo: 0.98
+ *             qHi: 1.0
+ *             yTarget: 0.5
+ *           upper:
+ *             qLo: 0.98
+ *             qHi: 1.0
+ *             yTarget: 0.8
+ *
+ * \endcode
+ *
+ * For the AeExposureMode control the data should contain a dictionary called
+ * AeExposureMode containing per-mode setting dictionaries with the key being a
+ * value from \ref controls::AeExposureModeNameValueMap. Each mode dict should
+ * contain an array of shutter times with the key "shutter" and an array of gain
+ * values with the key "gain", in this format:
+ *
+ * \code{.unparsed}
+ * algorithms:
+ *   - Agc:
+ *       AeExposureMode:
+ *         ExposureNormal:
+ *           shutter: [ 100, 10000, 30000, 60000, 120000 ]
+ *           gain: [ 2.0, 4.0, 6.0, 8.0, 10.0 ]
+ *         ExposureShort:
+ *           shutter: [ 100, 10000, 30000, 60000, 120000 ]
+ *           gain: [ 2.0, 4.0, 6.0, 8.0, 10.0 ]
+ *
+ * \endcode
+ *
+ * \return 0 on success or a negative error code
+ */
+int AgcMeanLuminance::parseTuningData(const YamlObject &tuningData)
+{
+	int ret;
+
+	parseRelativeLuminanceTarget(tuningData);
+
+	ret = parseConstraintModes(tuningData);
+	if (ret)
+		return ret;
+
+	ret = parseExposureModes(tuningData);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+/**
+ * \brief Set the ExposureModeHelper limits for this class
+ * \param[in] minShutter Minimum shutter time to allow
+ * \param[in] maxShutter Maximum shutter time to allow
+ * \param[in] minGain Minimum gain to allow
+ * \param[in] maxGain Maximum gain to allow
+ *
+ * This function calls \ref ExposureModeHelper::setLimits() for each
+ * ExposureModeHelper that has been created for this class.
+ */
+void AgcMeanLuminance::setLimits(utils::Duration minShutter,
+				 utils::Duration maxShutter,
+				 double minGain, double maxGain)
+{
+	for (auto &[id, helper] : exposureModeHelpers_)
+		helper->setLimits(minShutter, maxShutter, minGain, maxGain);
+}
+
+/**
+ * \fn AgcMeanLuminance::constraintModes()
+ * \brief Get the constraint modes that have been parsed from tuning data
+ */
+
+/**
+ * \fn AgcMeanLuminance::exposureModeHelpers()
+ * \brief Get the ExposureModeHelpers that have been parsed from tuning data
+ */
+
+/**
+ * \fn AgcMeanLuminance::controls()
+ * \brief Get the controls that have been generated after parsing tuning data
+ */
+
+/**
+ * \fn AgcMeanLuminance::estimateLuminance(const double gain)
+ * \brief Estimate the luminance of an image, adjusted by a given gain
+ * \param[in] gain The gain with which to adjust the luminance estimate
+ *
+ * This function estimates the average relative luminance of the frame that
+ * would be output by the sensor if an additional \a gain was applied. It is a
+ * pure virtual function because estimation of luminance is a hardware-specific
+ * operation, which depends wholly on the format of the stats that are delivered
+ * to libcamera from the ISP. Derived classes must override this function with
+ * one that calculates the normalised mean luminance value across the entire
+ * image.
+ *
+ * \return The normalised relative luminance of the image
+ */
+
+/**
+ * \brief Estimate the initial gain needed to achieve a relative luminance
+ * target
+ * \return The calculated initial gain
+ */
+double AgcMeanLuminance::estimateInitialGain() const
+{
+	double yTarget = relativeLuminanceTarget_;
+	double yGain = 1.0;
+
+	/*
+	* To account for non-linearity caused by saturation, the value needs to
+	* be estimated in an iterative process, as multiplying by a gain will
+	* not increase the relative luminance by the same factor if some image
+	* regions are saturated.
+	*/
+	for (unsigned int i = 0; i < 8; i++) {
+		double yValue = estimateLuminance(yGain);
+		double extra_gain = std::min(10.0, yTarget / (yValue + .001));
+
+		yGain *= extra_gain;
+		LOG(AgcMeanLuminance, Debug) << "Y value: " << yValue
+				<< ", Y target: " << yTarget
+				<< ", gives gain " << yGain;
+
+		if (utils::abs_diff(extra_gain, 1.0) < 0.01)
+			break;
+	}
+
+	return yGain;
+}
+
+/**
+ * \brief Clamp gain within the bounds of a defined constraint
+ * \param[in] constraintModeIndex The index of the constraint to adhere to
+ * \param[in] hist A histogram over which to calculate inter-quantile means
+ * \param[in] gain The gain to clamp
+ *
+ * \return The gain clamped within the constraint bounds
+ */
+double AgcMeanLuminance::constraintClampGain(uint32_t constraintModeIndex,
+					     const Histogram &hist,
+					     double gain)
+{
+	std::vector<AgcConstraint> &constraints = constraintModes_[constraintModeIndex];
+	for (const AgcConstraint &constraint : constraints) {
+		double newGain = constraint.yTarget * hist.bins() /
+				 hist.interQuantileMean(constraint.qLo, constraint.qHi);
+
+		if (constraint.bound == AgcConstraint::Bound::lower &&
+		    newGain > gain)
+			gain = newGain;
+
+		if (constraint.bound == AgcConstraint::Bound::upper &&
+		    newGain < gain)
+			gain = newGain;
+	}
+
+	return gain;
+}
+
+/**
+ * \brief Apply a filter on the exposure value to limit the speed of changes
+ * \param[in] exposureValue The target exposure from the AGC algorithm
+ *
+ * The speed of the filter is adaptive, and will produce the target quicker
+ * during startup, or when the target exposure is within 20% of the most recent
+ * filter output.
+ *
+ * \return The filtered exposure
+ */
+utils::Duration AgcMeanLuminance::filterExposure(utils::Duration exposureValue)
+{
+	double speed = 0.2;
+
+	/* Adapt instantly if we are in startup phase. */
+	if (frameCount_ < kNumStartupFrames)
+		speed = 1.0;
+
+	/*
+	 * If we are close to the desired result, go faster to avoid making
+	 * multiple micro-adjustments.
+	 * \todo Make this customisable?
+	 */
+	if (filteredExposure_ < 1.2 * exposureValue &&
+	    filteredExposure_ > 0.8 * exposureValue)
+		speed = sqrt(speed);
+
+	filteredExposure_ = speed * exposureValue +
+			    filteredExposure_ * (1.0 - speed);
+
+	return filteredExposure_;
+}
+
+/**
+ * \brief Calculate the new exposure value
+ * \param[in] constraintModeIndex The index of the current constraint mode
+ * \param[in] exposureModeIndex The index of the current exposure mode
+ * \param[in] yHist A Histogram from the ISP statistics to use in constraining
+ * the calculated gain
+ * \param[in] effectiveExposureValue The EV applied to the frame from which the
+ * statistics in use derive
+ *
+ * Calculate a new exposure value to try to obtain the target. The calculated
+ * exposure value is filtered to prevent rapid changes from frame to frame, and
+ * divided into shutter time, analogue and digital gain.
+ *
+ * \return Tuple of shutter time, analogue gain, and digital gain
+ */
+std::tuple<utils::Duration, double, double>
+AgcMeanLuminance::calculateNewEv(uint32_t constraintModeIndex,
+				 uint32_t exposureModeIndex,
+				 const Histogram &yHist,
+				 utils::Duration effectiveExposureValue)
+{
+	/*
+	 * The pipeline handler should validate that we have received an allowed
+	 * value for AeExposureMode.
+	 */
+	std::shared_ptr<ExposureModeHelper> exposureModeHelper =
+		exposureModeHelpers_.at(exposureModeIndex);
+
+	double gain = estimateInitialGain();
+	gain = constraintClampGain(constraintModeIndex, yHist, gain);
+
+	/*
+	 * We don't check whether we're already close to the target, because
+	 * even if the effective exposure value is the same as the last frame's
+	 * we could have switched to an exposure mode that would require a new
+	 * pass through the splitExposure() function.
+	 */
+
+	utils::Duration newExposureValue = effectiveExposureValue * gain;
+	utils::Duration maxTotalExposure = exposureModeHelper->maxShutter()
+					   * exposureModeHelper->maxGain();
+	newExposureValue = std::min(newExposureValue, maxTotalExposure);
+
+	/*
+	 * We filter the exposure value to make sure changes are not too jarring
+	 * from frame to frame.
+	 */
+	newExposureValue = filterExposure(newExposureValue);
+
+	frameCount_++;
+	return exposureModeHelper->splitExposure(newExposureValue);
+}
+
+/**
+ * \fn AgcMeanLuminance::resetFrameCount()
+ * \brief Reset the frame counter
+ *
+ * This function resets the internal frame counter, which exists to help the
+ * algorithm decide whether it should respond instantly or not. The expectation
+ * is for derived classes to call this function before each camera start call,
+ * either in configure() or queueRequest() if the frame number is zero.
+ */
+
+}; /* namespace ipa */
+
+}; /* namespace libcamera */
diff --git a/src/ipa/libipa/agc_mean_luminance.h b/src/ipa/libipa/agc_mean_luminance.h
new file mode 100644
index 00000000..9f91e08f
--- /dev/null
+++ b/src/ipa/libipa/agc_mean_luminance.h
@@ -0,0 +1,90 @@ 
+/* SPDX-License-Identifier: LGPL-2.1-or-later */
+/*
+ * Copyright (C) 2024 Ideas on Board Oy
+ *
+ agc_mean_luminance.h - Base class for mean luminance AGC algorithms
+ */
+
+#pragma once
+
+#include <tuple>
+#include <vector>
+
+#include <libcamera/controls.h>
+
+#include "libcamera/internal/yaml_parser.h"
+
+#include "exposure_mode_helper.h"
+#include "histogram.h"
+
+namespace libcamera {
+
+namespace ipa {
+
+class AgcMeanLuminance
+{
+public:
+	AgcMeanLuminance();
+	virtual ~AgcMeanLuminance() = default;
+
+	struct AgcConstraint {
+		enum class Bound {
+			lower = 0,
+			upper = 1
+		};
+		Bound bound;
+		double qLo;
+		double qHi;
+		double yTarget;
+	};
+
+	int parseTuningData(const YamlObject &tuningData);
+
+	void setLimits(utils::Duration minShutter, utils::Duration maxShutter,
+		       double minGain, double maxGain);
+
+	std::map<int32_t, std::vector<AgcConstraint>> constraintModes()
+	{
+		return constraintModes_;
+	}
+
+	std::map<int32_t, std::shared_ptr<ExposureModeHelper>> exposureModeHelpers()
+	{
+		return exposureModeHelpers_;
+	}
+
+	ControlInfoMap::Map controls()
+	{
+		return controls_;
+	}
+
+	std::tuple<utils::Duration, double, double>
+	calculateNewEv(uint32_t constraintModeIndex, uint32_t exposureModeIndex,
+		       const Histogram &yHist, utils::Duration effectiveExposureValue);
+	void resetFrameCount() { frameCount_ = 0; }
+
+private:
+	virtual double estimateLuminance(const double gain) const = 0;
+
+	void parseRelativeLuminanceTarget(const YamlObject &tuningData);
+	void parseConstraint(const YamlObject &modeDict, int32_t id);
+	int parseConstraintModes(const YamlObject &tuningData);
+	int parseExposureModes(const YamlObject &tuningData);
+	double estimateInitialGain() const;
+	double constraintClampGain(uint32_t constraintModeIndex,
+				   const Histogram &hist,
+				   double gain);
+	utils::Duration filterExposure(utils::Duration exposureValue);
+
+	uint64_t frameCount_;
+	utils::Duration filteredExposure_;
+	double relativeLuminanceTarget_;
+
+	std::map<int32_t, std::vector<AgcConstraint>> constraintModes_;
+	std::map<int32_t, std::shared_ptr<ExposureModeHelper>> exposureModeHelpers_;
+	ControlInfoMap::Map controls_;
+};
+
+}; /* namespace ipa */
+
+}; /* namespace libcamera */
diff --git a/src/ipa/libipa/meson.build b/src/ipa/libipa/meson.build
index 37fbd177..7ce885da 100644
--- a/src/ipa/libipa/meson.build
+++ b/src/ipa/libipa/meson.build
@@ -1,6 +1,7 @@ 
 # SPDX-License-Identifier: CC0-1.0
 
 libipa_headers = files([
+    'agc_mean_luminance.h',
     'algorithm.h',
     'camera_sensor_helper.h',
     'exposure_mode_helper.h',
@@ -10,6 +11,7 @@  libipa_headers = files([
 ])
 
 libipa_sources = files([
+    'agc_mean_luminance.cpp',
     'algorithm.cpp',
     'camera_sensor_helper.cpp',
     'exposure_mode_helper.cpp',