Issue 298 in webp: cwebp 0.5.0 compatibility regression

[vladapilny@gmail.com via Monorail]

Status: New
Owner: ----
Labels: Type-Defect Priority-Medium

New issue 298 by vladapi...@gmail.com: cwebp 0.5.0 compatibility regression
https://bugs.chromium.org/p/webp/issues/detail?id=298

What steps will reproduce the problem?
1. get source image c4_source.png
2. libwebp-0.4.4\cwebp.exe c4_source.png -o c4_0.4.4.webp
3. libwebp-0.5.0\cwebp.exe c4_source.png -o c4_0.5.0.webp
4. compare resulting images in multiple Chrome/Chromium/Android browsers

What is the expected output? What do you see instead?
0.4.4 displays always OK, 0.5.0 is in multiple places broken, see attached screenshots.

What version of the product are you using? On what operating system?
0.5.0 BROKEN in:
Chromium 50.0.2661.102 (Gentoo linux)
Xamarin Android Player - Nexus 7 Marshmallow (emulator run on Win 7)

Graphical glitches appears only on some converted images.
Observed as always WORKING (at least with these samples):
Chrome 51.0.2704.63 (Gentoo linux)
Chromium 53.0.2744.0 (Win 7)
Xamarin Android Player - Nexus 4 KitKat (Win 7)

Attachments:
c4_source.png 174 KB
c4_0.4.4.webp 16.9 KB
c4_0.5.0.webp 15.8 KB
c4_0.4.4.log 820 bytes
c4_0.5.0.log 873 bytes
screenshot_c4_0.4.4_Chromium50_linux.png 86.3 KB
screenshot_c4_0.5.0_Chromium50_linux.png 95.6 KB
screenshot_c4_0.4.4_Xamarin-Nexus7-Marshmallow.png 140 KB
screenshot_c4_0.5.0_Xamarin-Nexus7-Marshmallow.png 152 KB

--
You received this message because:
1. The project was configured to send all issue notifications to this address

You may adjust your notification preferences at:
https://bugs.chromium.org/hosting/settings

[urvang@google.com via Monorail]

Comment #1 on issue 298 by urv...@google.com: cwebp 0.5.0 compatibility regression
https://bugs.chromium.org/p/webp/issues/detail?id=298#c1

I just viewed your c4_0.5.0.webp on Chrome 50 on Ubuntu, and it looks fine to me. (screenshot attached).

Are you sure there isn't something wrong with the display you are testing on? (e.g. It may be ignoring alpha in the image?)

I'll try on a Nexus 7 next week too, to be sure.

Attachments:
screenshot_cwebp_0.5.0_chrome_50_ubuntu.png 583 KB

[vladapilny@gmail.com via Monorail]

Comment #2 on issue 298 by vladapi...@gmail.com: cwebp 0.5.0 compatibility regression
https://bugs.chromium.org/p/webp/issues/detail?id=298#c2

But it happens on 2 different PC. Chrome in linux is OK, but Chromium not.

[jzern@google.com via Monorail]

Comment #3 on issue 298 by jz...@google.com: cwebp 0.5.0 compatibility regression
https://bugs.chromium.org/p/webp/issues/detail?id=298#c3

Sounds like issue #291 [1]. Can you try a cwebp built from the tip of the tree?
The color-cache change that was re-enabled in 0.5.0 was disabled again in preparation for 0.5.1 [2].

[1] https://bugs.chromium.org/p/webp/issues/detail?id=291
[2] https://chromium-review.googlesource.com/#/c/333270/
97934e2 Revert "Re-enable encoding of alpha plane with color cache for next release."

[vladapilny@gmail.com via Monorail]

Comment #4 on issue 298 by vladapi...@gmail.com: cwebp 0.5.0 compatibility regression
https://bugs.chromium.org/p/webp/issues/detail?id=298#c4

Bingo! Thank you, this was the problem. Please release official 0.5.1, there is everything OK. .-)

I'm sorry for duplication. The strange behavior was, 0.5.0 was decoded OK on old Android 4.x, but not on latest Marshmallow.

[urvang@google.com via Monorail]

Updates:
Mergedinto: 291
Status: Duplicate

Comment #5 on issue 298 by urv...@google.com: cwebp 0.5.0 compatibility regression
https://bugs.chromium.org/p/webp/issues/detail?id=298#c5

Thanks for confirming!
I'm marking it as a duplicate then. Please await 0.5.1 release.