Closed GoogleCodeExporter closed 9 years ago
I couldn't reproduce here (MacBook pro), the compression just completed ok:
cwebp PIA13894.png -m 5 -lossless
File: PIA13894.png
Dimension: 2610 x 4104
Output: 6823212 bytes
Lossless-ARGB compressed size: 6823212 bytes
* Lossless features used: PREDICTION CROSS-COLOR-TRANSFORM
* Precision Bits: histogram=7 transform=3 cache=0
cwebp PIA13894.png -m 6 -lossless
File: PIA13894.png
Dimension: 2610 x 4104
Output: 6823212 bytes
Lossless-ARGB compressed size: 6823212 bytes
* Lossless features used: PREDICTION CROSS-COLOR-TRANSFORM
* Precision Bits: histogram=7 transform=3 cache=0
Could it be it's "just" a memory problem?
Original comment by pascal.m...@gmail.com
on 25 Sep 2012 at 8:44
I'd advise to try options in reverse order first,
cwebp.exe -lossless -m 5 PIA13894.png -o PIA13894.wp5
then try the same on Linux, then on Windows. My machine is Windows XP
Pro SP3, with 3Gb RAM. Intel Core i7 CPU.
With -m 0...3 compressed size is 6682232.
Alex
Original comment by gral...@gmail.com
on 25 Sep 2012 at 8:53
Hi Alex,
I tried to run this on my Linux box (Ubuntu 12.04) with order of options you
specified, and works fine for me too:
cwebp -lossless -m 6 PIA13894.back.png -o PIA13894.back.png.6.webp
Saving file 'PIA13894.back.png.6.webp'
File: PIA13894.back.png
Dimension: 2610 x 4104
Output: 6823212 bytes
Lossless-ARGB compressed size: 6823212 bytes
* Lossless features used: PREDICTION CROSS-COLOR-TRANSFORM
* Precision Bits: histogram=7 transform=3 cache=0
Hopefully, someone with a windows machine will try it next.
Original comment by urv...@google.com
on 25 Sep 2012 at 9:39
I can reproduce this. I'll see about putting a fix together.
Original comment by jz...@google.com
on 3 Oct 2012 at 6:37
Original comment by jz...@google.com
on 3 Oct 2012 at 6:38
This is a problem in all 32-bit builds. There were a couple cases of integer
overflow, the core issue is fixed by:
https://gerrit.chromium.org/gerrit/#/c/34535/
Original comment by jz...@google.com
on 3 Oct 2012 at 7:32
This commit:
80237c4 GetHistoBits: fix integer overflow
has been merged (v0.2.0-27-g7fee5d1).
Two additional fixes came out of this:
a6ae04d VP8LAllocateHistogramSet: fix overflow in size calculation
8a99723 EncodeImageInternal: fix uninitialized free
Original comment by jz...@google.com
on 3 Oct 2012 at 10:38
Original issue reported on code.google.com by
pascal.m...@gmail.com
on 25 Sep 2012 at 8:36Attachments: