Klaus Post
Klaus Post
@LibRaw - "RawDecoder->uncorrectedRawValues" should satisfy the latter. A dither option should satisfy the former. But please open an issue, so I don't forget when this is closed.
Any decoder can insert known bad pixels into `RawImage->mBadPixelPositions`, but only RW2 has zero is bad. The "bad" pixels are inserted [while the image is being decoded](https://github.com/klauspost/rawspeed/blob/develop/RawSpeed/Rw2Decoder.cpp#L170). That way we...
Implemented.
oh. #8 is already here ;)
Yes - however, it is only "TiffIFDBE" and "TiffEntryBE". These classes actually swap endianness. I am not referring to the function names in RawDecoder, which are correct, and will remain...
Whether or not a TiffEntry is "TiffEntry" or "TiffEntryBE" _doesn't_ actually say if the data is LE or BE. On a BE system it will be the other way around....
Yes, after some back&forth, I agree that would be both simpler and more reliable, and not something that should have any big performance impact.
I will see if there is a non-messy way to add this. Seems like we will be forced to add this as a second update to the metadata.
@harshavardhana That is not how it works. For PutObject and Multipart we can set it before we have received the data since it is always provided. If it mismatches the...
(also copy to same object to update metadata we don't even read the data).