You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The image-rendering property supports values such as smooth and high-quality but these are not specific enough. Bilinear is preferable in some cases (example), but not others, and we get bugs about this in chrome (crbug.com/562162). To change our defaults to be cubic in more cases, we need a way for users to request other algorithms.
Could we support the following new values for image-rendering?
nearest-neighbor
bilinear
trilinear (using mipmaps)
cubic-catmull-rom
The first three of these are ubiquitous and are typically available in hardware. An alternative to cubic-catmull-rom is to just specify cubic and leave the details up to implementations (see this page).
The text was updated successfully, but these errors were encountered:
I also agree that bicubic catmull-rom is the more generally applicable of the various cubic options.
Lastly, an opt-in to specify resampling in linear-light rather than gamma-encoded space is desirable. The demo page you linked for bicubic casually mentions this was done but does not show the result of not doing it.
Currently the values are intentionally vague because implementations didn't want to require specific algorithms, both because it would potentially require more implementation work they didn't think was very necessary, and because they wanted the freedom to upgrade and downgrade their scaling algorithms as needed to accommodate the page and device load.
I'm happy to add more specific algorithm requirements (possibly behind specifically-named keywords), but I'd want some impl support beyond Chrome. ^_^
Uh oh!
There was an error while loading. Please reload this page.
The
image-rendering
property supports values such assmooth
andhigh-quality
but these are not specific enough. Bilinear is preferable in some cases (example), but not others, and we get bugs about this in chrome (crbug.com/562162). To change our defaults to be cubic in more cases, we need a way for users to request other algorithms.Could we support the following new values for
image-rendering
?nearest-neighbor
bilinear
trilinear
(using mipmaps)cubic-catmull-rom
The first three of these are ubiquitous and are typically available in hardware. An alternative to
cubic-catmull-rom
is to just specifycubic
and leave the details up to implementations (see this page).The text was updated successfully, but these errors were encountered: