diff --git a/include/libcamera/internal/v4l2_videodevice.h b/include/libcamera/internal/v4l2_videodevice.h
index dc259523599c..de4745982e94 100644
--- a/include/libcamera/internal/v4l2_videodevice.h
+++ b/include/libcamera/internal/v4l2_videodevice.h
@@ -187,7 +187,7 @@ public:
 
 	int getFormat(V4L2DeviceFormat *format);
 	int setFormat(V4L2DeviceFormat *format);
-	std::map<V4L2PixelFormat, std::vector<SizeRange>> formats(uint32_t code = 0);
+	ImageFormats<V4L2PixelFormat> formats(uint32_t code = 0);
 
 	int setSelection(unsigned int target, Rectangle *rect);
 
diff --git a/src/libcamera/pipeline/raspberrypi/raspberrypi.cpp b/src/libcamera/pipeline/raspberrypi/raspberrypi.cpp
index e16a9c7f10d3..64364afb3f7e 100644
--- a/src/libcamera/pipeline/raspberrypi/raspberrypi.cpp
+++ b/src/libcamera/pipeline/raspberrypi/raspberrypi.cpp
@@ -37,8 +37,6 @@ namespace libcamera {
 
 LOG_DEFINE_CATEGORY(RPI)
 
-using V4L2PixFmtMap = std::map<V4L2PixelFormat, std::vector<SizeRange>>;
-
 namespace {
 
 bool isRaw(PixelFormat &pixFmt)
@@ -67,7 +65,7 @@ double scoreFormat(double desired, double actual)
 	return score;
 }
 
-V4L2DeviceFormat findBestMode(V4L2PixFmtMap &formatsMap, const Size &req)
+V4L2DeviceFormat findBestMode(ImageFormats<V4L2PixelFormat> &formatsMap, const Size &req)
 {
 	double bestScore = 9e9, score;
 	V4L2DeviceFormat bestMode = {};
@@ -79,12 +77,11 @@ V4L2DeviceFormat findBestMode(V4L2PixFmtMap &formatsMap, const Size &req)
 #define PENALTY_UNPACKED	 500.0
 
 	/* Calculate the closest/best mode from the user requested size. */
-	for (const auto &iter : formatsMap) {
-		V4L2PixelFormat v4l2Format = iter.first;
+	for (const auto &v4l2Format : formatsMap.formats()) {
 		PixelFormat pixelFormat = v4l2Format.toPixelFormat();
 		const PixelFormatInfo &info = PixelFormatInfo::info(pixelFormat);
 
-		for (const SizeRange &sz : iter.second) {
+		for (const SizeRange &sz : formatsMap.sizes(v4l2Format)) {
 			double modeWidth = sz.contains(req) ? req.width : sz.max.width;
 			double modeHeight = sz.contains(req) ? req.height : sz.max.height;
 			double reqAr = static_cast<double>(req.width) / req.height;
@@ -427,7 +424,7 @@ CameraConfiguration::Status RPiCameraConfiguration::validate()
 			 * Calculate the best sensor mode we can use based on
 			 * the user request.
 			 */
-			V4L2PixFmtMap fmts = data_->unicam_[Unicam::Image].dev()->formats();
+			ImageFormats<V4L2PixelFormat> fmts = data_->unicam_[Unicam::Image].dev()->formats();
 			V4L2DeviceFormat sensorFormat = findBestMode(fmts, cfg.size);
 			PixelFormat sensorPixFormat = sensorFormat.fourcc.toPixelFormat();
 			if (cfg.size != sensorFormat.size ||
@@ -481,7 +478,7 @@ CameraConfiguration::Status RPiCameraConfiguration::validate()
 		 *
 		 */
 		PixelFormat &cfgPixFmt = config_.at(outSize[i].first).pixelFormat;
-		V4L2PixFmtMap fmts;
+		ImageFormats<V4L2PixelFormat> fmts;
 
 		if (i == maxIndex)
 			fmts = data_->isp_[Isp::Output0].dev()->formats();
@@ -518,7 +515,7 @@ CameraConfiguration *PipelineHandlerRPi::generateConfiguration(Camera *camera,
 	RPiCameraData *data = cameraData(camera);
 	CameraConfiguration *config = new RPiCameraConfiguration(data);
 	V4L2DeviceFormat sensorFormat;
-	V4L2PixFmtMap fmts;
+	ImageFormats<V4L2PixelFormat> fmts;
 
 	if (roles.empty())
 		return config;
@@ -605,7 +602,7 @@ int PipelineHandlerRPi::configure(Camera *camera, CameraConfiguration *config)
 	}
 
 	/* First calculate the best sensor mode we can use based on the user request. */
-	V4L2PixFmtMap fmts = data->unicam_[Unicam::Image].dev()->formats();
+	ImageFormats<V4L2PixelFormat> fmts = data->unicam_[Unicam::Image].dev()->formats();
 	V4L2DeviceFormat sensorFormat = findBestMode(fmts, rawStream ? sensorSize : maxSize);
 
 	/*
diff --git a/src/libcamera/pipeline/simple/simple.cpp b/src/libcamera/pipeline/simple/simple.cpp
index 1ec8d0f7de03..f3e03ba60196 100644
--- a/src/libcamera/pipeline/simple/simple.cpp
+++ b/src/libcamera/pipeline/simple/simple.cpp
@@ -275,8 +275,7 @@ int SimpleCameraData::init()
 			return ret;
 		}
 
-		std::map<V4L2PixelFormat, std::vector<SizeRange>> videoFormats =
-			video_->formats(format.mbus_code);
+		ImageFormats<V4L2PixelFormat> videoFormats = video_->formats(format.mbus_code);
 
 		LOG(SimplePipeline, Debug)
 			<< "Adding configuration for " << format.size.toString()
diff --git a/src/libcamera/pipeline/uvcvideo/uvcvideo.cpp b/src/libcamera/pipeline/uvcvideo/uvcvideo.cpp
index a074909499f1..dbd835f5c5ef 100644
--- a/src/libcamera/pipeline/uvcvideo/uvcvideo.cpp
+++ b/src/libcamera/pipeline/uvcvideo/uvcvideo.cpp
@@ -159,8 +159,7 @@ CameraConfiguration *PipelineHandlerUVC::generateConfiguration(Camera *camera,
 	if (roles.empty())
 		return config;
 
-	std::map<V4L2PixelFormat, std::vector<SizeRange>> v4l2Formats =
-		data->video_->formats();
+	ImageFormats<V4L2PixelFormat> v4l2Formats = data->video_->formats();
 	std::map<PixelFormat, std::vector<SizeRange>> deviceFormats;
 	std::transform(v4l2Formats.begin(), v4l2Formats.end(),
 		       std::inserter(deviceFormats, deviceFormats.begin()),
diff --git a/src/libcamera/v4l2_videodevice.cpp b/src/libcamera/v4l2_videodevice.cpp
index 3614b2ed1cbc..ea952444e0ad 100644
--- a/src/libcamera/v4l2_videodevice.cpp
+++ b/src/libcamera/v4l2_videodevice.cpp
@@ -925,9 +925,9 @@ int V4L2VideoDevice::setFormatSingleplane(V4L2DeviceFormat *format)
  *
  * \return A list of the supported video device formats
  */
-std::map<V4L2PixelFormat, std::vector<SizeRange>> V4L2VideoDevice::formats(uint32_t code)
+ImageFormats<V4L2PixelFormat> V4L2VideoDevice::formats(uint32_t code)
 {
-	std::map<V4L2PixelFormat, std::vector<SizeRange>> formats;
+	ImageFormats<V4L2PixelFormat> formats;
 
 	for (V4L2PixelFormat pixelFormat : enumPixelformats(code)) {
 		std::vector<SizeRange> sizes = enumSizes(pixelFormat);