<div dir="ltr"><div dir="ltr">Hi Jacopo,<div><br></div><div>Thank you for your review comments.</div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Tue, 19 Jan 2021 at 16:59, Jacopo Mondi <<a href="mailto:jacopo@jmondi.org" target="_blank">jacopo@jmondi.org</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">Hi Naush,<br>
<br>
I know I have my tag here, and I'm possibly going to repeat some<br>
comments. I don't think any of them should block your series though<br>
<br>
On Tue, Jan 19, 2021 at 03:30:46PM +0000, Naushir Patuck wrote:<br>
> Add support for setting V4L2_CID_VBLANK appropriately when setting<br>
> V4L2_CID_EXPOSURE. This will allow adaptive framerates during<br>
> viewfinder use cases (e.g. when the exposure time goes above 33ms, we<br>
> can reduce the framerate to lower than 30fps).<br>
><br>
> The minimum and maximum frame durations are provided via libcamera<br>
> controls, and will prioritise exposure time limits over any AGC request.<br>
><br>
> V4L2_CID_VBLANK is controlled through the staggered writer, just like<br>
> the exposure and gain controls.<br>
><br>
> Signed-off-by: Naushir Patuck <<a href="mailto:naush@raspberrypi.com" target="_blank">naush@raspberrypi.com</a>><br>
> Reviewed-by: David Plowman <<a href="mailto:david.plowman@raspberrypi.com" target="_blank">david.plowman@raspberrypi.com</a>><br>
> Tested-by: David Plowman <<a href="mailto:david.plowman@raspberrypi.com" target="_blank">david.plowman@raspberrypi.com</a>><br>
> Reviewed-by: Jacopo Mondi <<a href="mailto:jacopo@jmondi.org" target="_blank">jacopo@jmondi.org</a>><br>
> ---<br>
> include/libcamera/ipa/raspberrypi.h | 1 +<br>
> src/ipa/raspberrypi/cam_helper.cpp | 35 +++++++++++-<br>
> src/ipa/raspberrypi/cam_helper.hpp | 15 ++++-<br>
> src/ipa/raspberrypi/cam_helper_imx219.cpp | 13 ++++-<br>
> src/ipa/raspberrypi/cam_helper_imx477.cpp | 11 +++-<br>
> src/ipa/raspberrypi/cam_helper_ov5647.cpp | 11 +++-<br>
> src/ipa/raspberrypi/raspberrypi.cpp | 56 ++++++++++++++++---<br>
> .../pipeline/raspberrypi/raspberrypi.cpp | 3 +-<br>
> 8 files changed, 130 insertions(+), 15 deletions(-)<br>
><br>
> diff --git a/include/libcamera/ipa/raspberrypi.h b/include/libcamera/ipa/raspberrypi.h<br>
> index 01fe5abce9a6..1de36039cee0 100644<br>
> --- a/include/libcamera/ipa/raspberrypi.h<br>
> +++ b/include/libcamera/ipa/raspberrypi.h<br>
> @@ -65,6 +65,7 @@ static const ControlInfoMap Controls = {<br>
> { &controls::Sharpness, ControlInfo(0.0f, 16.0f, 1.0f) },<br>
> { &controls::ColourCorrectionMatrix, ControlInfo(-16.0f, 16.0f) },<br>
> { &controls::ScalerCrop, ControlInfo(Rectangle{}, Rectangle(65535, 65535, 65535, 65535), Rectangle{}) },<br>
> + { &controls::FrameDurations, ControlInfo(1000, 1000000000) },<br>
<br>
The static initialization of the control limits worries me a bit.<br>
This should match the sensor limits. I can't ask you to go to fully<br>
implement this, but I fear the arbitrary initialization of the<br>
controls limits the RPi implementation does will be problematic for<br>
application that might what to know in which range they can safely<br>
operate.<br>
<br>
I presume your forthcoming applications will probably know those limits<br>
from a per-sensor configuration file ?<br></blockquote><div><br></div><div>You are correct here, the static initialization should match the current sensor mode limits. There currently exists no mechanism for this, so these values above are indeed somewhat arbitrary for now.</div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<br>
> };<br>
><br>
> } /* namespace RPi */<br>
> diff --git a/src/ipa/raspberrypi/cam_helper.cpp b/src/ipa/raspberrypi/cam_helper.cpp<br>
> index 6efa0d7fd030..b7b8faf09c69 100644<br>
> --- a/src/ipa/raspberrypi/cam_helper.cpp<br>
> +++ b/src/ipa/raspberrypi/cam_helper.cpp<br>
> @@ -34,8 +34,10 @@ CamHelper *CamHelper::Create(std::string const &cam_name)<br>
> return nullptr;<br>
> }<br>
><br>
> -CamHelper::CamHelper(MdParser *parser)<br>
> - : parser_(parser), initialized_(false)<br>
> +CamHelper::CamHelper(MdParser *parser, unsigned int maxFrameLength,<br>
> + unsigned int frameIntegrationDiff)<br>
> + : parser_(parser), initialized_(false), maxFrameLength_(maxFrameLength),<br>
> + frameIntegrationDiff_(frameIntegrationDiff)<br>
> {<br>
> }<br>
><br>
> @@ -56,6 +58,35 @@ double CamHelper::Exposure(uint32_t exposure_lines) const<br>
> return exposure_lines * mode_.line_length / 1000.0;<br>
> }<br>
><br>
> +uint32_t CamHelper::GetVBlanking(double &exposure, double minFrameDuration,<br>
> + double maxFrameDuration) const<br>
> +{<br>
> + uint32_t frameLengthMin, frameLengthMax, vblank;<br>
> + uint32_t exposureLines = ExposureLines(exposure);<br>
> +<br>
> + assert(initialized_);<br>
> +<br>
> + /*<br>
> + * Clamp frame length by the frame duration range and the maximum allowable<br>
> + * value in the sensor, given by maxFrameLength_.<br>
> + */<br>
> + frameLengthMin = std::clamp<uint32_t>(1e3 * minFrameDuration / mode_.line_length,<br>
> + mode_.height, maxFrameLength_);<br>
> + frameLengthMax = std::clamp<uint32_t>(1e3 * maxFrameDuration / mode_.line_length,<br>
> + mode_.height, maxFrameLength_);<br>
<br>
min and max frame durations are here adjusted to respect the sensor's<br>
configuration. Shouldn't this be reflected in the metadata ?<br></blockquote><div><br></div><div>Correct, they should - but currently don't. We discussed this in an earlier thread, but I have another series related to fps for plumbing in frame duration limits into our AGC - and with this change we return back the fully constrained durations in the metadata. There is a \todo in the handling of the FrameDurations control saying this is a pending change to be made.</div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<br>
> + /*<br>
> + * Limit the exposure to the maximum frame duration requested, and<br>
> + * re-calculate if it has been clipped.<br>
> + */<br>
> + exposureLines = std::min(frameLengthMax - frameIntegrationDiff_, exposureLines);<br>
<br>
Can't you here<br>
exposureLines = std::max(frameLengthMin, exposureLines + frameIntegrationDiff_);<br></blockquote><div><br></div><div>Is this correct? For rolling shutter sensors, it is typically the case that</div><div><br></div><div>frame length - exposure lines must be >= frame integration difference</div><div><br></div><div>The line in question is ensuring the maximum exposure cannot exceed the maximum frame length - frame integration difference. With the suggested change, we are instead saying that it might be possible to have exposureLine = frameLengthMin, so not accounting for
frameIntegrationDiff_. But also, if there is no clipping, we are increasing the requested exposure time by
frameIntegrationDiff_ lines. My head is spinning a bit with all of this, so please do correct me if what I have understood is rubbish :-)</div><div><br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<br>
To avoid the below clamp, which is partially useless as we're already<br>
sure exposureLines < (frameLengthMax - frameIntegrationDiff_)<br>
<br>
(and I'm not sure if frameIntegrationDiff_ has only to be considered<br>
when inspecting the upper limit or also when comparing to the bottom<br>
one)<br></blockquote><div><br></div><div>Ah, I *think* maybe I understand where you are going with this. In my calculations, I modify frame length to match the exposure requested (within allowable limits). Hence I only consider
frameIntegrationDiff_ on the upper limit, and explicitly subtract it for the vblank calculation below. I think your change is doing things the other way round, i.e. adjusting exposure to match the given frame length? If that is the case, I would prefer to keep it as is, so we get the exposure as close as possible to the request. If that is not the case, I'm sorry, but I have misunderstood :-)</div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<br>
it will also make sure the below recalculation of 'exposure' is in the limits<br>
<br>
<br>
<br>
> + exposure = Exposure(exposureLines);<br>
> +<br>
> + /* Limit the vblank to the range allowed by the frame length limits. */<br>
> + vblank = std::clamp(exposureLines + frameIntegrationDiff_,<br>
> + frameLengthMin, frameLengthMax) - mode_.height;<br>
<br>
Then you can simply<br>
return exposureLines - mode_.height;<br>
<br>
> + return vblank;<br>
> +}<br>
> +<br>
> void CamHelper::SetCameraMode(const CameraMode &mode)<br>
> {<br>
> mode_ = mode;<br>
> diff --git a/src/ipa/raspberrypi/cam_helper.hpp b/src/ipa/raspberrypi/cam_helper.hpp<br>
> index 044c28667f75..b1739a57e342 100644<br>
> --- a/src/ipa/raspberrypi/cam_helper.hpp<br>
> +++ b/src/ipa/raspberrypi/cam_helper.hpp<br>
> @@ -62,12 +62,15 @@ class CamHelper<br>
> {<br>
> public:<br>
> static CamHelper *Create(std::string const &cam_name);<br>
> - CamHelper(MdParser *parser);<br>
> + CamHelper(MdParser *parser, unsigned int maxFrameLength,<br>
> + unsigned int frameIntegrationDiff);<br>
> virtual ~CamHelper();<br>
> void SetCameraMode(const CameraMode &mode);<br>
> MdParser &Parser() const { return *parser_; }<br>
> uint32_t ExposureLines(double exposure_us) const;<br>
> double Exposure(uint32_t exposure_lines) const; // in us<br>
> + virtual uint32_t GetVBlanking(double &exposure_us, double minFrameDuration,<br>
<br>
The first parameter is called just 'exposure' in the function<br>
implementation. Mixing of snake_case and camelCase, but it's in many<br>
places already in the IPA, so...<br></blockquote><div><br></div><div>Indeed, this was also brought up in earlier comments. Our "controller" uses snake case throughout (mostly) and we do have a task of converting to camel case as some point to match our IPA (which is camel case throughout).</div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<br>
> + double maxFrameDuration) const;<br>
> virtual uint32_t GainCode(double gain) const = 0;<br>
> virtual double Gain(uint32_t gain_code) const = 0;<br>
> virtual void GetDelays(int &exposure_delay, int &gain_delay) const;<br>
> @@ -76,10 +79,20 @@ public:<br>
> virtual unsigned int HideFramesModeSwitch() const;<br>
> virtual unsigned int MistrustFramesStartup() const;<br>
> virtual unsigned int MistrustFramesModeSwitch() const;<br>
> +<br>
> protected:<br>
> MdParser *parser_;<br>
> CameraMode mode_;<br>
> +<br>
> +private:<br>
> bool initialized_;<br>
> + /* Largest possible frame length, in units of lines. */<br>
> + unsigned int maxFrameLength_;<br>
> + /*<br>
> + * Smallest difference between the frame length and integration time,<br>
> + * in units of lines.<br>
> + */<br>
> + unsigned int frameIntegrationDiff_;<br>
<br>
nit: different declaration order than in the derived classes, but<br>
well, minor..<br></blockquote><div><br></div><div>Ack.</div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<br>
> };<br>
><br>
> // This is for registering camera helpers with the system, so that the<br>
> diff --git a/src/ipa/raspberrypi/cam_helper_imx219.cpp b/src/ipa/raspberrypi/cam_helper_imx219.cpp<br>
> index db8ab879deb7..8688ec091c44 100644<br>
> --- a/src/ipa/raspberrypi/cam_helper_imx219.cpp<br>
> +++ b/src/ipa/raspberrypi/cam_helper_imx219.cpp<br>
> @@ -49,13 +49,22 @@ public:<br>
> double Gain(uint32_t gain_code) const override;<br>
> unsigned int MistrustFramesModeSwitch() const override;<br>
> bool SensorEmbeddedDataPresent() const override;<br>
> +<br>
> +private:<br>
> + /*<br>
> + * Smallest difference between the frame length and integration time,<br>
> + * in units of lines.<br>
> + */<br>
> + static constexpr int frameIntegrationDiff = 4;<br>
> + /* Largest possible frame length, in units of lines. */<br>
> + static constexpr int maxFrameLength = 0xffff;<br>
> };<br>
><br>
> CamHelperImx219::CamHelperImx219()<br>
> #if ENABLE_EMBEDDED_DATA<br>
> - : CamHelper(new MdParserImx219())<br>
> + : CamHelper(new MdParserImx219(), maxFrameLength, frameIntegrationDiff)<br>
> #else<br>
> - : CamHelper(new MdParserRPi())<br>
> + : CamHelper(new MdParserRPi(), maxFrameLength, frameIntegrationDiff)<br>
> #endif<br>
> {<br>
> }<br>
> diff --git a/src/ipa/raspberrypi/cam_helper_imx477.cpp b/src/ipa/raspberrypi/cam_helper_imx477.cpp<br>
> index 0e896ac74f88..5396131003f1 100644<br>
> --- a/src/ipa/raspberrypi/cam_helper_imx477.cpp<br>
> +++ b/src/ipa/raspberrypi/cam_helper_imx477.cpp<br>
> @@ -38,10 +38,19 @@ public:<br>
> uint32_t GainCode(double gain) const override;<br>
> double Gain(uint32_t gain_code) const override;<br>
> bool SensorEmbeddedDataPresent() const override;<br>
> +<br>
> +private:<br>
> + /*<br>
> + * Smallest difference between the frame length and integration time,<br>
> + * in units of lines.<br>
> + */<br>
> + static constexpr int frameIntegrationDiff = 22;<br>
> + /* Largest possible frame length, in units of lines. */<br>
> + static constexpr int maxFrameLength = 0xffdc;<br>
> };<br>
><br>
> CamHelperImx477::CamHelperImx477()<br>
> - : CamHelper(new MdParserImx477())<br>
> + : CamHelper(new MdParserImx477(), maxFrameLength, frameIntegrationDiff)<br>
> {<br>
> }<br>
><br>
> diff --git a/src/ipa/raspberrypi/cam_helper_ov5647.cpp b/src/ipa/raspberrypi/cam_helper_ov5647.cpp<br>
> index 0b841cd175fa..2bd8a754f133 100644<br>
> --- a/src/ipa/raspberrypi/cam_helper_ov5647.cpp<br>
> +++ b/src/ipa/raspberrypi/cam_helper_ov5647.cpp<br>
> @@ -23,6 +23,15 @@ public:<br>
> unsigned int HideFramesModeSwitch() const override;<br>
> unsigned int MistrustFramesStartup() const override;<br>
> unsigned int MistrustFramesModeSwitch() const override;<br>
> +<br>
> +private:<br>
> + /*<br>
> + * Smallest difference between the frame length and integration time,<br>
> + * in units of lines.<br>
> + */<br>
> + static constexpr int frameIntegrationDiff = 4;<br>
> + /* Largest possible frame length, in units of lines. */<br>
> + static constexpr int maxFrameLength = 0xffff;<br>
> };<br>
><br>
> /*<br>
> @@ -31,7 +40,7 @@ public:<br>
> */<br>
><br>
> CamHelperOv5647::CamHelperOv5647()<br>
> - : CamHelper(new MdParserRPi())<br>
> + : CamHelper(new MdParserRPi(), maxFrameLength, frameIntegrationDiff)<br>
> {<br>
> }<br>
><br>
> diff --git a/src/ipa/raspberrypi/raspberrypi.cpp b/src/ipa/raspberrypi/raspberrypi.cpp<br>
> index d087b07e186b..ba9bc398ef87 100644<br>
> --- a/src/ipa/raspberrypi/raspberrypi.cpp<br>
> +++ b/src/ipa/raspberrypi/raspberrypi.cpp<br>
> @@ -58,6 +58,8 @@ namespace libcamera {<br>
> /* Configure the sensor with these values initially. */<br>
> constexpr double DefaultAnalogueGain = 1.0;<br>
> constexpr unsigned int DefaultExposureTime = 20000;<br>
> +constexpr double defaultMinFrameDuration = 1e6 / 30.0;<br>
> +constexpr double defaultMaxFrameDuration = 1e6 / 0.01;<br>
<br>
Quite arbitrary, isn't it ?<br></blockquote><div><br></div><div>Yes it is somewhat. I wanted to establish a default frameduration interval to start with in the absence of a request from the user. Now this could be variable framerate (as I have done here), or a fixed framerate. We have a preference to run at variable framerate by default, hence the values above. Note that these defaults are not necessarily what the sensor min and max are. For example, a sensor may be able to do 200fps, and we do not want to go to such extreme exposure times for a simple viewfinder as it would probably degrade image quality. But if an application really wants to do that, it is free to set the frame duration limits appropriately. In that respect, having the default values as a const in the IPA seemed to make more sense.</div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<br>
I would have expected this to be sent by the pipeline to the IPA by<br>
populating the FrameDuration ControlInfoMap at configure() time. The<br>
'sensorControls' provided at configure might very well contain the<br>
VBLANK control limits, which once combined with the sensor mode<br>
information should give you precise limits<br>
<br>
minFrameDuration = (mode.height + min(VBLANK)) * mode.line_lenght<br>
/ (mode.pixel_rate / 1e6)<br>
<br>
Or is this not required ?<br></blockquote><div><br></div><div>Correct, this is a calculation that is required, but not yet in place. My next patch series for fps does add this calculation in to control sensor limits. We discussed this previously, and decided not to introduce that series just yet to get this one through with basic functionality.</div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<br>
><br>
> LOG_DEFINE_CATEGORY(IPARPI)<br>
><br>
> @@ -150,6 +152,10 @@ private:<br>
><br>
> /* Distinguish the first camera start from others. */<br>
> bool firstStart_;<br>
> +<br>
> + /* Frame duration (1/fps) limits, given in microseconds. */<br>
> + double minFrameDuration_;<br>
> + double maxFrameDuration_;<br>
> };<br>
><br>
> int IPARPi::init(const IPASettings &settings)<br>
> @@ -332,7 +338,8 @@ void IPARPi::configure(const CameraSensorInfo &sensorInfo,<br>
> sensorMetadata = helper_->SensorEmbeddedDataPresent();<br>
><br>
> result->data.push_back(gainDelay);<br>
> - result->data.push_back(exposureDelay);<br>
> + result->data.push_back(exposureDelay); /* For EXPOSURE ctrl */<br>
> + result->data.push_back(exposureDelay); /* For VBLANK ctrl */<br>
> result->data.push_back(sensorMetadata);<br>
><br>
> result->operation |= RPi::IPA_CONFIG_STAGGERED_WRITE;<br>
> @@ -377,6 +384,9 @@ void IPARPi::configure(const CameraSensorInfo &sensorInfo,<br>
> controller_.Initialise();<br>
> controllerInit_ = true;<br>
><br>
> + minFrameDuration_ = defaultMinFrameDuration;<br>
> + maxFrameDuration_ = defaultMaxFrameDuration;<br>
> +<br>
> /* Supply initial values for gain and exposure. */<br>
> ControlList ctrls(sensorCtrls_);<br>
> AgcStatus agcStatus;<br>
> @@ -524,7 +534,8 @@ bool IPARPi::validateSensorControls()<br>
> {<br>
> static const uint32_t ctrls[] = {<br>
> V4L2_CID_ANALOGUE_GAIN,<br>
> - V4L2_CID_EXPOSURE<br>
> + V4L2_CID_EXPOSURE,<br>
> + V4L2_CID_VBLANK,<br>
> };<br>
><br>
> for (auto c : ctrls) {<br>
> @@ -551,7 +562,7 @@ bool IPARPi::validateIspControls()<br>
> V4L2_CID_USER_BCM2835_ISP_DENOISE,<br>
> V4L2_CID_USER_BCM2835_ISP_SHARPEN,<br>
> V4L2_CID_USER_BCM2835_ISP_DPC,<br>
> - V4L2_CID_USER_BCM2835_ISP_LENS_SHADING<br>
> + V4L2_CID_USER_BCM2835_ISP_LENS_SHADING,<br>
<br>
Unrelated it seems<br></blockquote><div><br></div><div>Ack. Thank Kieran for pointing it out :-)</div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<br>
> };<br>
><br>
> for (auto c : ctrls) {<br>
> @@ -804,6 +815,25 @@ void IPARPi::queueRequest(const ControlList &controls)<br>
> break;<br>
> }<br>
><br>
> + case controls::FRAME_DURATIONS: {<br>
> + auto frameDurations = ctrl.second.get<Span<const int64_t>>();<br>
> +<br>
> + /* This will be applied once AGC recalculations occur. */<br>
> + minFrameDuration_ = frameDurations[0] ? frameDurations[0] : defaultMinFrameDuration;<br>
> + maxFrameDuration_ = frameDurations[1] ? frameDurations[1] : defaultMaxFrameDuration;<br>
> + maxFrameDuration_ = std::max(maxFrameDuration_, minFrameDuration_);<br>
<br>
Why this last line ?<br>
</blockquote><div><br></div><div>That last line was a recent addition. Laurent correctly pointed out that bad things would happen if
maxFrameDuration_ <
minFrameDuration_.</div><div></div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">Also, it really seems seems arbitrary defaults clips the exposure to<br>
pretty abitrary values. If you don't want to do the calculation of the<br>
defaults, what about skipping the clipping completely ? Or is this<br>
your intetion ? "Pick default values large enough that no clipping<br>
occours" ?<br></blockquote><div><br></div><div>The idea was that setting a 0 for either min or max would revert to the ipa defaults. With default values for either/both min and max values, we still need to ensure
maxFrameDuration_ >= minFrameDuration_. Perhaps the disconnect here is the notion of using IPA provided default values?</div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<br>
> +<br>
> + /*<br>
> + * \todo The values returned in the metadata below must be<br>
> + * correctly clipped by what the sensor mode suppots and<br>
> + * what the AGC exposure mode or manual shutter speed limits<br>
> + */<br>
<br>
Why this metadata can't be set in reportMetadata() which is called<br>
after processStat() which calls GetVBlanking() and updates the limits<br>
?<br></blockquote><div><br></div><div>reportMetadata() is used for per-frame dynamic metadata. FrameDurationLimits metadata is only set once when a request comes through. This is to inform the app if any clipping has occurred over the user requested values. I know we had quite a long discussion about what exactly the return metadata value should be, and I think we agreed on this? The next patch series for fps does call GetVBlanking() from here to correctly clip the return values.</div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<br>
<br>
Sorry for the many questions. The only thing that worries me is the<br>
ControlInfo intialization, but that's not something you should address<br>
at the moment. Other than that with the metadata and default value<br>
assignement clarified, you can retain my tag.<br></blockquote><div><br></div><div>Not a problem. Good to ensure things are right by everyone. Please let me know your thoughts on my reply, there may be more to discuss before submitting this.</div><div><br></div><div>Regards,</div><div>Naush</div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<br>
Thanks<br>
j<br>
<br>
> + libcameraMetadata_.set(controls::FrameDurations,<br>
> + { static_cast<int64_t>(minFrameDuration_),<br>
> + static_cast<int64_t>(maxFrameDuration_) });<br>
> + break;<br>
> + }<br>
> +<br>
> default:<br>
> LOG(IPARPI, Warning)<br>
> << "Ctrl " << controls::<a href="http://controls.at" rel="noreferrer" target="_blank">controls.at</a>(ctrl.first)->name()<br>
> @@ -962,15 +992,27 @@ void IPARPi::applyAWB(const struct AwbStatus *awbStatus, ControlList &ctrls)<br>
> void IPARPi::applyAGC(const struct AgcStatus *agcStatus, ControlList &ctrls)<br>
> {<br>
> int32_t gainCode = helper_->GainCode(agcStatus->analogue_gain);<br>
> - int32_t exposureLines = helper_->ExposureLines(agcStatus->shutter_time);<br>
><br>
> - LOG(IPARPI, Debug) << "Applying AGC Exposure: " << agcStatus->shutter_time<br>
> - << " (Shutter lines: " << exposureLines << ") Gain: "<br>
> + /* GetVBlanking might clip exposure time to the fps limits. */<br>
> + double exposure = agcStatus->shutter_time;<br>
> + int32_t vblanking = helper_->GetVBlanking(exposure, minFrameDuration_,<br>
> + maxFrameDuration_);<br>
> + int32_t exposureLines = helper_->ExposureLines(exposure);<br>
> +<br>
> + LOG(IPARPI, Debug) << "Applying AGC Exposure: " << exposure<br>
> + << " (Shutter lines: " << exposureLines << ", AGC requested "<br>
> + << agcStatus->shutter_time << ") Gain: "<br>
> << agcStatus->analogue_gain << " (Gain Code: "<br>
> << gainCode << ")";<br>
><br>
> - ctrls.set(V4L2_CID_ANALOGUE_GAIN, gainCode);<br>
> + /*<br>
> + * Due to the behavior of V4L2, the current value of VBLANK could clip the<br>
> + * exposure time without us knowing. The next time though this function should<br>
> + * clip exposure correctly.<br>
> + */<br>
> + ctrls.set(V4L2_CID_VBLANK, vblanking);<br>
> ctrls.set(V4L2_CID_EXPOSURE, exposureLines);<br>
> + ctrls.set(V4L2_CID_ANALOGUE_GAIN, gainCode);<br>
> }<br>
><br>
> void IPARPi::applyDG(const struct AgcStatus *dgStatus, ControlList &ctrls)<br>
> diff --git a/src/libcamera/pipeline/raspberrypi/raspberrypi.cpp b/src/libcamera/pipeline/raspberrypi/raspberrypi.cpp<br>
> index 7a5f5881b9b3..252cab64023e 100644<br>
> --- a/src/libcamera/pipeline/raspberrypi/raspberrypi.cpp<br>
> +++ b/src/libcamera/pipeline/raspberrypi/raspberrypi.cpp<br>
> @@ -1233,7 +1233,8 @@ int RPiCameraData::configureIPA(const CameraConfiguration *config)<br>
> if (!staggeredCtrl_) {<br>
> staggeredCtrl_.init(unicam_[Unicam::Image].dev(),<br>
> { { V4L2_CID_ANALOGUE_GAIN, result.data[resultIdx++] },<br>
> - { V4L2_CID_EXPOSURE, result.data[resultIdx++] } });<br>
> + { V4L2_CID_EXPOSURE, result.data[resultIdx++] },<br>
> + { V4L2_CID_VBLANK, result.data[resultIdx++] } });<br>
> sensorMetadata_ = result.data[resultIdx++];<br>
> }<br>
> }<br>
> --<br>
> 2.25.1<br>
><br>
</blockquote></div></div>