The initial signs were wrong for the deltas of f(x), the ellipse
equation. This seemed to be fine for larger circles/ellipses but
broke things at a small scale, this was previously fixed with a
horrible "error = error / 4" hack. With this change, all ellipses
look a little better :^)
This also fixed a signed integer overflow Andreas found with UBSAN,
which happened for circles with a 1px radius.
With this, common v4 profiles, such as embedded into jpgs by iPhones
(when configured to write jpegs) or Pixel phones, are identical to
the input when reexported :^)
Restart markers are supposed to be applied every restart interval.
However, in these loops macroblocks are counted taking the luma sampling
factor into consideration. Meaning that we need to correct this factor
when testing if we should reset the DC.
Also, the decoder was discarding the first byte of every scan with a set
restart interval, as `0 % n == 0` is always true.
I'm not sure why this isn't caught on other people's setups or CI, but
when building on NixOS it fails with:
error: comparison of integer expressions of different signedness:
‘size_t’ {aka ‘long unsigned int’} and ‘int’ [-Werror=sign-compare]
With this, simple v4 matrix profiles using parametric curves, such as
Compact-ICC-Profiles/profiles/sRGB-v4.icc, can be completely serialized
and the serialized file can be read again by `icc` :^)
The patch also contains modifications on several classes, functions or
files that are related to the `JPGLoader`.
Renaming include:
- JPGLoader{.h, .cpp}
- JPGImageDecoderPlugin
- JPGLoadingContext
- JPG_DEBUG
- decode_jpg
- FuzzJPGLoader.cpp
- Few string literals or texts
I can see small speedup of about 6-7% with this for some gradient
painting, though Tim has been able to see an ~30% speedup in some
resizing cases so this seems like a worthwhile change.
Co-authored-by: Tim Ledbetter <timledbetter@gmail.com>
For now, this can write the profile header.
`icc` refuses to dump its contents since the required copyrightTag,
profileDescriptionTag, and required per-device-class tags are missing,
but it looks ok in a hex editor.
header_size is already set to 0 if context.is_included_in_ico is true
and to bmp_header_size else, so we can remove some redundant checks
for context.is_included_in_ico.
No behavior change.
The method was only used in `load_from_file(StringView path)` and
replacing it with `load_from_file(NonnullOwnPtr<Core::File>)` is very
straightforward.
I had missed this in 21cc0c0cb2 because this tag is missing in
"Table 32 — Tag list" in the v2 ICC spec O_o. But it's in
"6.4.26 namedColorTag".
Also add a comment pointing to a page saying that all these tags
are very deprecated and not even recommended for new v2 profiles.
(That's how I noticed that namedColorTag was missing.)
