Improve resize kernel docs and expose missing (bi)linear kernel

[einarakerlind]

Question about an existing feature

What are you trying to achieve?

I noticed that i get identical results for all kernels except nearest when upscaling an image. Is this expected behaviour?
Version: 0.33.3

When you searched for similar issues, what did you find that might be related?

—

Please provide a minimal, standalone code sample, without other dependencies, that demonstrates this question

sharp("path-to-image/example.png").resize({ width: 600, height: 600, background: { alpha: 0, r: 0, g: 0, b: 0 }, kernel: 'lanczos2' });

Please provide sample image(s) that help explain this question

Original image

Scaled using cubic kernel

Scaled using lanczos2 kernel

[lovell]

Yes, the underlying logic that maps downsampling kernels to upsampling interpolators within libvips can be seen at:

https://github.com/libvips/libvips/blob/c2ede6c110ffcdb47606e8366b8f6a5ff7515b76/libvips/resample/resize.c#L113-L132

However this mapping was missing from the sharp documentation, which commit 6257994 addresses (as well as exposing the missing bilinear kernel). Thanks for reporting.

[einarakerlind]

Thanks for clarifying!