Add texture compression interface and ETC1 encoder.

Add texture compression interface and ETC1 encoder.

This CL introduces a CC texture compression API and adds an ETC1
encoder that will be used for compressing tile textures.

BUG=434699
Committed: https://crrev.com/7957c6df0e1f3bd7d023d8a60e0d62d457065e48
Cr-Commit-Position: refs/heads/master@{#321788}

[christiank, 2015-03-19 10:26:59]

christiank@opera.com changed reviewers:
+ reveman@chromium.org

[christiank, 2015-03-19 10:27:00]

Hi David,

This CL continues the work auygun started in issue: 820713006.

This CL contains the same API that evolved in 820713006 but instead of using
auygun's DXT/ATC encoders I have supplied the ETC1 encoder I have been working
on.

Thanks!

Regards,
Christian

[reveman, 2015-03-20 15:15:55]

Looks great! Do we need the texture_compress/ directory? If cc/resources/ is
getting too large, then maybe we should instead figure out how to split that
directory into 2 top-level directories instead..

https://codereview.chromium.org/1015373003/diff/1/cc/resources/texture_compre...
File cc/resources/texture_compress/texture_compressor_perftest.cc (right):

https://codereview.chromium.org/1015373003/diff/1/cc/resources/texture_compre...
cc/resources/texture_compress/texture_compressor_perftest.cc:65: if
(!compressor_)
nit: can you remove this check?

[christiank, 2015-03-23 07:36:28]

On 2015/03/20 15:15:55, reveman wrote:
> Looks great! Do we need the texture_compress/ directory? If cc/resources/ is
> getting too large, then maybe we should instead figure out how to split that
> directory into 2 top-level directories instead..

Thanks! No, we can move the files anywhere. I am not sure what's the best place
though. Should I move them to cc/resources/?

[christiank, 2015-03-23 07:36:56]

https://codereview.chromium.org/1015373003/diff/1/cc/resources/texture_compre...
File cc/resources/texture_compress/texture_compressor_perftest.cc (right):

https://codereview.chromium.org/1015373003/diff/1/cc/resources/texture_compre...
cc/resources/texture_compress/texture_compressor_perftest.cc:65: if
(!compressor_)
On 2015/03/20 15:15:55, reveman wrote:
> nit: can you remove this check?

Sure, now removed.

[reveman, 2015-03-23 14:27:23]

Let's move the files to cc/resources/ for now and we can figure out later if
that directory needs to be split.

lgtm after moving the files

https://codereview.chromium.org/1015373003/diff/20001/cc/resources/texture_co...
File cc/resources/texture_compress/texture_compressor_etc1.h (right):

https://codereview.chromium.org/1015373003/diff/20001/cc/resources/texture_co...
cc/resources/texture_compress/texture_compressor_etc1.h:8: #include <stdint.h>
nit: unnecessary

[christiank, 2015-03-23 14:54:56]

The CQ bit was checked by christiank@opera.com

[christiank, 2015-03-23 14:54:56]

The patchset sent to the CQ was uploaded after l-g-t-m from reveman@chromium.org
Link to the patchset: https://codereview.chromium.org/1015373003/#ps60001
(title: "Move files from cc/resources/texture_compress to cc/resources/")

[commit-bot: I haz the power, 2015-03-23 14:55:08]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1015373003/60001

[commit-bot: I haz the power, 2015-03-23 16:20:01]

Committed patchset #4 (id:60001)

[commit-bot: I haz the power, 2015-03-23 16:20:22]

Patchset 4 (id:??) landed as
https://crrev.com/7957c6df0e1f3bd7d023d8a60e0d62d457065e48
Cr-Commit-Position: refs/heads/master@{#321788}

[radu.velea, 2015-04-10 13:37:52]

radu.velea@intel.com changed reviewers:
+ radu.velea@intel.com

[radu.velea, 2015-04-10 13:37:53]

Shouldn't there be a break here:
cc/resources/texture_compressor_etc1.cc:414? If first 2 channels have a big diff
and the
last doesn't the flag will be false but colors 555.

https://codereview.chromium.org/1015373003/diff/60001/cc/resources/texture_co...
File cc/resources/texture_compressor_etc1.cc (right):

https://codereview.chromium.org/1015373003/diff/60001/cc/resources/texture_co...
cc/resources/texture_compressor_etc1.cc:414: sub_block_avg[j] =
MakeColor444(avg_color_1);
Shouldn't there be a break here? If first 2 channels have a big diff and the
last doesn't the flag will be false but colors 555.