On 7/12/22 16:34, Nicolas Dufresne wrote:
> Le mardi 12 juillet 2022 à 17:53 +0530, Viswanath Boma a écrit :
>> From: Stanimir Varbanov <stanimir.varbanov@xxxxxxxxxx>
>>
>> V4l2 encoder compliance expecting default values of colormetry for the control.
>
> nit: colormetry -> colorimetry
>
>>
>> Change-Id: I1db0d4940b54e033d646ce39d60dc488afba8d58
>> Signed-off-by: Viswanath Boma <quic_vboma@xxxxxxxxxxx>
>> ---
>> drivers/media/platform/qcom/venus/venc_ctrls.c | 4 +++-
>> 1 file changed, 3 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/media/platform/qcom/venus/venc_ctrls.c b/drivers/media/platform/qcom/venus/venc_ctrls.c
>> index ea5805e71c143..37ba7d97f99b2 100644
>> --- a/drivers/media/platform/qcom/venus/venc_ctrls.c
>> +++ b/drivers/media/platform/qcom/venus/venc_ctrls.c
>> @@ -352,6 +352,8 @@ static const struct v4l2_ctrl_ops venc_ctrl_ops = {
>> int venc_ctrl_init(struct venus_inst *inst)
>> {
>> int ret;
>> + struct v4l2_ctrl_hdr10_mastering_display p_hdr10_mastering = { {34000, 13250, 7500 },
>> + { 16000, 34500, 3000 }, 15635, 16450, 10000000, 500 };
>
> What is the origin of these values ? Should this be done in the control
> framework instead ?
We needed some safe default values to satisfy v4l2 control. I took the
values from here [1].
I'm not sure do we need them in v4l2 control core (most probably it is a
good to have it), Hans ?
>
>>
>> ret = v4l2_ctrl_handler_init(&inst->ctrl_handler, 58);
>> if (ret)
>> @@ -580,7 +582,7 @@ int venc_ctrl_init(struct venus_inst *inst)
>>
>> v4l2_ctrl_new_std_compound(&inst->ctrl_handler, &venc_ctrl_ops,
>> V4L2_CID_COLORIMETRY_HDR10_MASTERING_DISPLAY,
>> - v4l2_ctrl_ptr_create(NULL));
>> + v4l2_ctrl_ptr_create((void *)&p_hdr10_mastering));
>>
>> v4l2_ctrl_new_std(&inst->ctrl_handler, &venc_ctrl_ops,
>> V4L2_CID_MPEG_VIDEO_INTRA_REFRESH_PERIOD, 0,
>
--
regards,
Stan
[1] https://www.elecard.com/page/article_hdr
