[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Public WebGL] Promote ASTC extension to community approved now ?



Great. https://github.com/KhronosGroup/WebGL/pull/2204 uploaded.


On Mon, Dec 12, 2016 at 2:27 PM, Jeff Gilbert <jgilbert@mozilla.com> wrote:
No objections here.
I actually thought it went CA already.

On Mon, Dec 12, 2016 at 8:51 AM, Kenneth Russell <kbr@google.com> wrote:
> Apologies for the delay but David's excellent test has now been verified and
> merged in https://github.com/KhronosGroup/WebGL/pull/2172 .
>
> It would be great to promote this extension to community approved now. Any
> objections or other comments?
>
> -Ken
>
>
>
> On Tue, Nov 22, 2016 at 4:52 PM, Kenneth Russell <kbr@google.com> wrote:
>>
>> Thank you David for this excellent contribution!
>>
>> We'll verify it and merge as soon as possible -- by next Monday the
>> latest, due to the US holidays. If there are certain formats that aren't
>> working well on many GPU types, we should probably file bugs against the
>> various GPU vendors and temporarily disable testing just those formats, so
>> the rest of the tests pass 100% (thereby acting as regression tests).
>>
>> -Ken
>>
>>
>>
>> On Tue, Nov 22, 2016 at 8:48 AM, David Peicho <dadouvic2@gmail.com> wrote:
>>>
>>> We have worked on the conformance tests, and the pull request is now
>>> available (https://github.com/KhronosGroup/WebGL/pull/2172).
>>>
>>> When the flag --enable-webgl-draft-extensions is activated, the extension
>>> is accessible and the texture uploading works with LDR/HDR encoded ASTC.
>>>
>>> However, on an Intel Core i7-6700HQ and with a 16x16 red/green test
>>> image, only few compression formats were working well
>>> (COMPRESSED_RGBA_ASTC_4x4_KHR, COMPRESSED_RGBA_ASTC_5x4_KHRs and
>>> COMPRESSED_RGBA_ASTC_5x5_KHR), others giving strange decoding result.
>>>
>>> David,
>>> Sketchfab
>>>
>>> On Fri, Nov 11, 2016 at 12:19 AM, Kenneth Russell <kbr@google.com> wrote:
>>>>
>>>> Sorry, clarification: the tests are under
>>>> sdk/tests/conformance/extensions/ in the Github repository.
>>>>
>>>> -Ken
>>>>
>>>>
>>>> On Thu, Nov 10, 2016 at 2:52 PM, Kenneth Russell <kbr@google.com> wrote:
>>>>>
>>>>> Could someone please provide a conformance test for this extension? We
>>>>> don't have an ASTC pipeline conveniently in house.
>>>>>
>>>>> The existing compressed texture extension tests in
>>>>> https://github.com/KhronosGroup/WebGL under conformance/extensions/ should
>>>>> be a good starting point. Chrome already has this extension implemented
>>>>> behind the --enable-webgl-draft-extensions flag. We're basically waiting for
>>>>> a test to be able to validate the implementation and give a thumbs up.
>>>>>
>>>>> Thanks,
>>>>>
>>>>> -Ken
>>>>>
>>>>>
>>>>> On Thu, Nov 10, 2016 at 2:30 PM, Christophe Riccio
>>>>> <christophe.riccio@unity3d.com> wrote:
>>>>>>
>>>>>> There quite a few Android
>>>>>> http://opengles.gpuinfo.org/gles_listreports.php?extension=GL_KHR_texture_compression_astc_ldr.
>>>>>> and iOS (with PowerVR 6XT chips) mobile devices with ASTC support.
>>>>>>
>>>>>> But yes, I don't believe there is any web browser exposing ASTC. I
>>>>>> think that would be a great first step.
>>>>>>
>>>>>> On Thu, Nov 10, 2016 at 11:29 AM, Florian Bösch <pyalot@gmail.com>
>>>>>> wrote:
>>>>>>>
>>>>>>> I did not record any device/webgl implementation exposing the ASTC
>>>>>>> extension (expected for draft extensions). The Samsung S7 lists astc support
>>>>>>> (though it seems to be spread across 3 extensions and one that isn't in the
>>>>>>> ES registry). Desktop linux driver for nvidia does not expose astc. iOS does
>>>>>>> not have astc.
>>>>>>>
>>>>>>> Since draft extensions are not exposed by default, I can't quantify
>>>>>>> what the current implementation status across platforms and UA vendors is.
>>>>>>> Could the UA vendors (apple, google, microsoft, mozilla) respectively
>>>>>>> comment on:
>>>>>>>
>>>>>>> What their implementation status is (and across which platforms)
>>>>>>> How well supported ASTC currently is on devices they see (by usage)
>>>>>>> If the spec is final
>>>>>>
>>>>>>
>>>>>
>>>>
>>>
>>
>