[Webkit-unassigned] [Bug 232398] New: Some valid WebPs are failing to decode

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=232398

            Bug ID: 232398
           Summary: Some valid WebPs are failing to decode
           Product: WebKit
           Version: Safari 15
          Hardware: All
                OS: All
            Status: NEW
          Severity: Normal
          Priority: P2
         Component: Images
          Assignee: webkit-unassigned at lists.webkit.org
          Reporter: e at ericportis.com
                CC: sabouhallawa at apple.com

Created attachment 442623

  --> https://bugs.webkit.org/attachment.cgi?id=442623&action=review

Screenshot of nike.com looking different in Chrome/Safari, because of this bug

Some valid WebPs are failing to decode in both macOS Big Sur 11.6 and iOS 15.

As of this writing, this is currently affecting the home page on https://nike.com, which uses a single-color powder-blue image to create a background behind the "Member Benefits" section. Screenshot attached. In Chrome, this background appears blue, as intended. In Safari, it appears gray (the CSS background color behind the image).

So far I’ve only seen single-color palletized lossless WebPs fail.

For example, here’s a WebP that fails:

https://o.img.rodeo/w_356,h_356,b_gray/_.webp (FAILS)

If I make it any smaller in either dimension, it works:

https://o.img.rodeo/w_355,h_356,b_gray/_.webp (WORKS)
https://o.img.rodeo/w_100,h_100,b_gray/_.webp (WORKS)
https://o.img.rodeo/w_10,h_10,b_gray/_.webp (WORKS)

All larger sizes fail, too:

https://o.img.rodeo/w_1000,h_1000,b_gray/_.webp (FAILS)

All of the images that fail in mac/iOS are successfully decoded by `dwebp` and display fine in Chrome.

Interestingly, the boundary between success and failure does not just depend on the pixel count. Changing the color changes the boundary:

https://o.img.rodeo/w_355,h_356,b_red/_.webp (FAILS)

The issue seems similar to, but distinct from, the issue behind this report https://bugs.webkit.org/show_bug.cgi?id=219977.

This is definitely an underlying OS library/frameworks issue, and I have already filed FB9671920 about it. As such, I expect this WebKit issue to be WONTFIXed, but I'm filing in order to facilitate public discussion on the issue and possible workarounds.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20211027/c69525df/attachment.htm>