h264 - don't warn about deblocking type 1 without multiple slices

If you enable multithreaded decoding for h.264, you get the
warning
"Cannot parallelize deblocking type 1, decoding such
frames in  
sequential order",
even if the video has only one slice and can't be
parallelized anyway.

This patch disables that bit of code for the first slice in
a frame.  
I tested it on
single-slice (that it can't parallelize) and multi-slice
(that it  
can), but not on
type 1 + multi-slice since I can't find any. Is there any on
mplayerhq?


_______________________________________________
ffmpeg-devel mailing list
ffmpeg-develmplayerhq.hu
http://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-devel

Alexander Strange wrote:
> If you enable multithreaded decoding for h.264, you get
the warning
> "Cannot parallelize deblocking type 1, decoding
such frames in 
> sequential order",
> even if the video has only one slice and can't be
parallelized anyway.
> 
> This patch disables that bit of code for the first
slice in a frame. I 
> tested it on
> single-slice (that it can't parallelize) and
multi-slice (that it can), 
> but not on
> type 1 + multi-slice since I can't find any. Is there
any on mplayerhq?

Dunno about mphq, but apples movietrailers are 1 +
multi_spice, and
for those files the warning is still printed with your patch
(as
expected I guess)

Having said that, I'm a little in doubt if the line should
be around
at all (in the current form) since there are various other
types of
content that cannot be parallelized (which it dont
"warn" abount).

Perhaps it should be extended to warn if a user requests
multithreading but the content is single-sliced.

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-develmplayerhq.hu
http://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-devel

Hello,
On Sat, Oct 27, 2007 at 12:13:07PM +0200, Andreas Öman
wrote:
[...]
> Having said that, I'm a little in doubt if the line
should be around
> at all (in the current form) since there are various
other types of
> content that cannot be parallelized (which it dont
"warn" abount).

I think the critical difference is that using -fast or
skiploopfilter
allows parallel decoding in this case, so it might be worth
informing
the user. There is nothing the user can do for single-sliced
content.

Greetings,
Reimar Döffinger
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-develmplayerhq.hu
http://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-devel

Hi

On Fri, Oct 26, 2007 at 05:40:01PM -0400, Alexander Strange
wrote:
> If you enable multithreaded decoding for h.264, you get
the warning
> "Cannot parallelize deblocking type 1, decoding
such frames in sequential 
> order",
> even if the video has only one slice and can't be
parallelized anyway.
>
> This patch disables that bit of code for the first
slice in a frame. I 
> tested it on
> single-slice (that it can't parallelize) and
multi-slice (that it can), but 
> not on
> type 1 + multi-slice since I can't find any. Is there
any on mplayerhq?

this patch is wrong, it will disable the CODEC_FLAG2_FAST
behavior of
deblocking_filter == 1 slices

[...]
-- 
Michael     GnuPG fingerprint:
9FF2128B147EF6730BADF133611EC787040B0FAB

Let us carefully observe those good qualities wherein our
enemies excel us
and endeavor to excel them, by avoiding what is faulty, and
imitating what
is excellent in them. -- Plutarch

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-develmplayerhq.hu
http://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-devel

On Oct 27, 2007, at 3:18 PM, Michael Niedermayer wrote:

> Hi
>
> On Fri, Oct 26, 2007 at 05:40:01PM -0400, Alexander
Strange wrote:
>> If you enable multithreaded decoding for h.264, you
get the warning
>> "Cannot parallelize deblocking type 1,
decoding such frames in  
>> sequential
>> order",
>> even if the video has only one slice and can't be
parallelized  
>> anyway.
>>
>> This patch disables that bit of code for the first
slice in a  
>> frame. I
>> tested it on
>> single-slice (that it can't parallelize) and
multi-slice (that it  
>> can), but
>> not on
>> type 1 + multi-slice since I can't find any. Is
there any on  
>> mplayerhq?
>
> this patch is wrong, it will disable the
CODEC_FLAG2_FAST behavior of
> deblocking_filter == 1 slices
>
> [...]

Looking at the h264 spec, deblocking_filter 1 is the same as
2 on  
single-slice,
so I think it's safe to make that change on everything.

I found a sample at http://www.apple.com/quicktime/guide/hd/nasaspacesh
uttle.html 
.





_______________________________________________
ffmpeg-devel mailing list
ffmpeg-develmplayerhq.hu
http://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-devel

Hi

On Sun, Oct 28, 2007 at 01:18:57PM -0400, Alexander Strange
wrote:
>
> On Oct 27, 2007, at 3:18 PM, Michael Niedermayer
wrote:
>
>> Hi
>>
>> On Fri, Oct 26, 2007 at 05:40:01PM -0400, Alexander
Strange wrote:
>>> If you enable multithreaded decoding for h.264,
you get the warning
>>> "Cannot parallelize deblocking type 1,
decoding such frames in sequential
>>> order",
>>> even if the video has only one slice and can't
be parallelized anyway.
>>>
>>> This patch disables that bit of code for the
first slice in a frame. I
>>> tested it on
>>> single-slice (that it can't parallelize) and
multi-slice (that it can), 
>>> but
>>> not on
>>> type 1 + multi-slice since I can't find any. Is
there any on mplayerhq?
>>
>> this patch is wrong, it will disable the
CODEC_FLAG2_FAST behavior of
>> deblocking_filter == 1 slices
>>
>> [...]
>
> Looking at the h264 spec, deblocking_filter 1 is the
same as 2 on 
> single-slice,
> so I think it's safe to make that change on
everything.

no, this is still wrong
now the code would change to deblocking_filter = 2
(=incorrect decoding)
even if no multithreading (=no speed gain) is possible

[...]
-- 
Michael     GnuPG fingerprint:
9FF2128B147EF6730BADF133611EC787040B0FAB

I count him braver who overcomes his desires than him who
conquers his
enemies for the hardest victory is over self. -- Aristotle

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-develmplayerhq.hu
http://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-devel