[theora] bitstream test suite

dos386 dos386 at gmail.com
Fri Jan 29 01:53:09 PST 2010


Loïc Le Loarer wrote (4 months ago):

> Does someone know where the original http://www.theora.org/test/test.zip is now ?

I've never seen this ZIP file, OTOH I had a deeper look into the "new"
Testsuite, and considering that 99% of the movies got encoded back in
2004, let me guess that the "new" Testsuite just recycles the content
of this ZIP ?

Philip Jägenstedt wrote:

> Any specific problem?

See other thread

> We use this test file by verifying that the first frame decoded
> matches a reference image and that it plays through to the end and
> stops at the full duration (10s in this case). It's difficult to do
> automated testing that playback is smooth and so on...

Appropriate content (see below) + human inspection

Silvia Pfeiffer wrote:

> this is an open community and you're very welcome to contribute
> further test material / provide improved material from existing test material.

I do have ideas for better movies and this has been on my ToDo list
for some time, just my pre-alpha video code doesn't progress that fast
as I'd like to ...

> That wiki page has just recently been created

While the movies seem marginally older ;-)

> to start a good collection and prepare for an automated test suite.

Good :-)

> BTW: there are further test files for transcoding at media.xiph.org.
> And at media.basilgohar.com (y4m files available at /preview)...

COOL (yesterday it was dead) ... BTW, I'd prefer a TAR or a bunch of
TAR's (per < 2 GiB) from 1'000'000's of single frames to download.

Gregory Maxwell wrote:

> The only 'players' that I'm aware of failing this is absolutely
> everything based on ffmpeg.

Why "players" ?

> Some players have had problems with still test material, so you can
> think of this as testing an additional case as well.

OK.

> There should be no important interaction with offset and motion, so I don't
> see any reason that particular test should move. There is another clip
> with both offset and motion, in any case.

Where ?

> Short videos are a pretty valid use case.  ::shurgs:: Long videos run
> into the concerns you raised about size.

5 seconds should fit, maybe time is needed to study the broken look ?



-- 
~~~ wow ~~~


More information about the theora mailing list