Jon Sneyers
Jon Sneyers
E.g. the same ones `flif` takes for decoding, plus some to control the progressive rendering, etc.
... so others can independently implement FLIF. This can obviously only be completed once #85 is completed.
Repetitive patterns, in particular patterns that are the result of halftoning or color dithering (e.g. by starting from a GIF file or a high-resolution scan of a print), do not...
Because standard `./configure; make; make install` is nice, and it will make life much easier for distro package maintainers.
Major new version of the SSIMULACRA metric. Changes compared to the original version: - works in XYB color space - uses 2-norm and 4-norm (instead of 1-norm and max-norm-after-downscaling) -...
When converting the internal float sample values to uint ones (when JXL_TYPE_UINT8 or JXL_TYPE_UINT16 pixel formats are requested), currently this is done simply by picking the nearest integer value. This...
No implementation yet, just an API proposal. The aim is to make uint8 output more useful by allowing applications to let libjxl do the color conversion before conversion to uint8....
Adds a heuristic to use patches instead of VarDCT blocks for blocks that are better encoded with Modular. These heuristics still need some tweaking and improvement, but they're already somewhat...
Errors during reading jpeg bitstreams caused both a debug string to be printed and an error code enum to be set. However, these error codes are not used anywhere, and...
Same as https://github.com/libjxl/libjxl/pull/1149 but now without the change to ClampedGradient, which wasn't helping much anyway and it apparently wasn't safe when the numbers use the full int32_t range, like in...