Missing linearization
- Dominant language
- Rust
- Stars
- 11
- Forks
- 0
- PR merge metrics
- No merged PRs in 30d
Description
I'm currently going through the Rust ecosystem of crates performing image resampling to make a list of which ones actually qualify to get this right.
`u8`/channel color data will always have some gamma for obvious reasons.
I.e. the results of the downsampling using this crate are just always wrong as data must be linearized for the filtering math to produced meaningful results.
See http://www.ericbrasseur.org/gamma.html for a detailed explanation. Also see https://imagemagick.org/Usage/resize/#resize_colorspace for some less extreme but still pretty obvious examples (Earth City Lights image).
This is the test image from first website above:

The expected result:

And what the code in this crate produces instead:

Also see [this issue](https://github.com/Traverse-Research/ispc-downsampler/issues/25) for a discussion that may help you with a fix. :)
Contributor guide
No contributing guide indexed for this repository
Assessment
This issue has not been assessed yet.