That’s not an error. Let’s walk through it.
We only parse
which is, as per https://www.w3.org/Graphics/JPEG/itu-t81.pdf, page 11:
SOF3 lossless process frame marker, Huffman coding
And if we look in the
parseSOF(), we find that error message:
Let’s go back to the spec, page 36:
Tqi: Quantization table destination selector – Specifies one of four possible quantization table destinations
from which the quantization table to use for dequantization of DCT coefficients of component Ci is retrieved. If
the decoding process uses the dequantization procedure, this table shall have been installed in this destination
by the time the decoder is ready to decode the scan(s) containing component Ci. The destination shall not be respecified,
or its contents changed, until all scans containing Ci have been completed.
and the following
Table B.2 states that for
Tqi, within losslessly compressed image.
(and since we are in
SOF3, the image is losslessly compressed.) the valid
So unless i’m missing some fine-print that says
ignore the "Tq" value when in SOF3,
i’m inclined to say it is Yi who are at fault here.
Either they are just writing random garbage in that field, or using the wrong frame marker
(if that is indeed compressed with quantized components).
Maybe you should just mail them, and ask them? (do CC me please, if you do)