[FFmpeg-devel] [PATCH] AAC encoder: fix assertion error with prediction

Claudio Freire klaussfreire at gmail.com
Thu Dec 31 02:44:56 CET 2015


On Tue, Dec 29, 2015 at 1:12 PM, Claudio Freire <klaussfreire at gmail.com> wrote:
> On Tue, Dec 29, 2015 at 6:18 AM, Rostislav Pehlivanov
> <atomnuker at gmail.com> wrote:
>> Wouldn't it be simpler to just check if the maximum codebook was 0 after
>> calculating cost1 and skipping the rest of the code in the loop?
>>
>> On 29 December 2015 at 08:23, Claudio Freire <klaussfreire at gmail.com> wrote:
>>
>>> Fixes an assertion error reported in #2686 that happens when
>>> using prediction (either explicitly or implicitly by setting
>>> the AAC main profile), since prediction code would allow
>>> creating new zeroes or removing existing ones, without
>>> properly checking for SF delta violations.
>>>
>>> This patch forbids creating/removing zeroes, perhaps an
>>> overly conservative approach, but a safe one. More permissive
>>> and sophisticated approaches may be attempted in the future.
>
>
> It is possible, and in fact does happen and it helps considerably to
> hide some artifacts, to use prediction in zeroed bands.
>
> It basically uses the predicted coefficient with no correction.
>
> May well be better than using a zero.

If there are no further objections I may push it soon.

I've encoded and listened to lots of test samples without issue
(though I'm testing other patches too).


More information about the ffmpeg-devel mailing list