Sensor.alloc_extra_fb returned error when using JPEG and VGA snapshot

Hi. I try to create a clone of the snapshot image and process it instead so don’t modify the original and save it to display. It works fine if I set the sensor with the RGB565 and QQVGA settings, but when I bumped to JPEG and VGA, it returned the error “OSError: Image is not mutable!” like in this photo. Do you have any idea how to fix this issue? Thanks in advance.

Hi, it’s because we have no support for JPEG/Bayer decoding. I have adding support for this stuff coming soon this year. Once it lands you’ll be able to do more stuff with JPEG/bayer images than just displaying them.

Hi. Thanks a lot for clearing that out. it sounds great and I’m looking forwards to it.

I am very interested in being able to load jpeg to use for tf.classify
It would be good to have an eta for this feature.

keep up the good work

It’s right around the corner. Larry has already finished Jpeg decoding support. Hardware decoding will not come that quick… but, software will be in the next release.

Once this feature is added all methods that use draw_image internally in the firmware as a front end will support jpegs.

That’s great news, we have a use case for testing on the H7 + and we need to be able to re review images stored onto the SD card as a form of validation.
Im looking forwards to the next release.

Has the support for the same been added? I am trying to do a similar thing for .jpg where I load an image from the storage and try to make a copy of the same. But it throws the same error at me.

We don’t have support form drawing on JPEG images.

That requires decompressing the image and then compressing per call…

Use the to_rgb565() method to make it an RGB565 image.