TurboJPEG Encode Uncorrect! But imencode is correct，but maybe encoder is not work

[NakanoSanku]

use opencv imencode is correct

    def __buffer2bytes(self):
        # Directly use the pixel buffer and reshape only once
        pixel_array = np.frombuffer(self.pixels, dtype=np.uint8).reshape((self.height.value, self.width.value, 4))
        flipped_rgb_pixel_array = pixel_array[::-1, :, [2, 1, 0]]
        _, data = cv2.imencode(self.encode, flipped_rgb_pixel_array)
        return data.tobytes()

maybe use this code

        String filename = tempfile();
        code = encoder->setDestination(filename);
        CV_Assert( code );

        code = encoder->write(image, params);
        encoder->throwOnEror();
        CV_Assert( code );

        FILE* f = fopen( filename.c_str(), "rb" );
        CV_Assert(f != 0);
        fseek( f, 0, SEEK_END );
        long pos = ftell(f);
        buf.resize((size_t)pos);
        fseek( f, 0, SEEK_SET );
        buf.resize(fread( &buf[0], 1, buf.size(), f ));
        fclose(f);
        remove(filename.c_str());

but use pyturbojpeg is uncorrect

    def __buffer2bytes(self):
        # Directly use the pixel buffer and reshape only once
        pixel_array = np.frombuffer(self.pixels, dtype=np.uint8).reshape(
            (self.height.value, self.width.value, 4))
        flipped_rgb_pixel_array = pixel_array[::-1, :, [2, 1, 0]]
        return self.__turboJpegEncoder.encode(
            flipped_rgb_pixel_array, quality=self.quality)

[lilohuang]

Hi @NakanoSanku ,

The cv2.imencode function calls cv2_convert.cpp to ensure that the non-contiguous input can be converted properly. However, PyTurboJPEG assumes that the input is already a contiguous buffer. You can try calling np.ascontiguousarray(flipped_rgb_pixel_array) before passing it to the PyTurboJPEG encoder. This should resolve the issue you encountered, and this will be natively supported in a future release.

e.g.,

return self.__turboJpegEncoder.encode(
            np.ascontiguousarray(flipped_rgb_pixel_array), quality=self.quality)

I'm going to close this issue. Please let me know if it doesn't work for you.

[NakanoSanku]

nice！thanks you ！