video: Patch SDMA & capture sample to arbitrate height and width #96986
Conversation
zephyrbot
added
platform: NXP
zephyrbot
requested review from
avolmat-st,
dbaluta,
decsny,
dleach02,
JarmouniA,
josuah,
kartben,
loicpoulain,
manuargue,
mmahadevan108,
nashif and
ngphibang
zacck
force-pushed
the
pr-patch-sdma-for-mcxn236
branch
from
85af2ed to
91be28e
Compare
zacck
force-pushed
the
pr-patch-sdma-for-mcxn236
branch
from
91be28e to
bafe2fb
Compare
Thank you, this is done. |
Ensure Image sensors using video_sdma only use the valid RGB565 format. Signed-off-by: Zacck Osiemo <[email protected]>
This board uses sdma as its video driver and that requires this size frame to be set for the sample to work Signed-off-by: Zacck Osiemo <[email protected]>
zacck
force-pushed
the
pr-patch-sdma-for-mcxn236
branch
from
bafe2fb to
aa93037
Compare
|
| /* Verify that format is RGB565 */ | ||
| if ((fmt->pixelformat != fmts[0].pixelformat) || | ||
| (fmt->width != fmts[0].width_min) || | ||
| (fmt->height != fmts[0].height_min)) { | ||
| return -ENOTSUP; | ||
| ret = video_set_format(config->sensor_dev, fmt); | ||
|
|
||
| if (ret < 0) { | ||
| LOG_ERR("Sensor device does not support RGB565"); | ||
| return ret; | ||
| } |
There was a problem hiding this comment.
The fix for this is ultimately in the capture sample IMHO, but for the sake of a workaround that stll makes sense.
Another compromise in the meantime is to configure the default format in _init() and make this call always return the default supported format without asking the source?
Thanks for the help!
There was a problem hiding this comment.
If you wanted to make sure that fmt = RGB565 320x240, you would also need to use fmts[0] instead of fmt: as what you propose seems like doing this?
if (/* format is not what is expected */) {
/* use the incorrect fmt */
}So that would need a tmp = { /* correct fmt */ } if wanting to go that path (but +1 for simply moving that in init()).
The video capture sample does not work when using a combination of the FRDM-MCXN236 and the OV5640 camera and this is due to the sensor and the underlying interface having different defaults. This patch sets the interface resolution to that selected by the sample which in turn is set in the CONFIG.