[FFmpeg-cvslog] r23647 - in trunk: Changelog libavcodec/Makefile libavcodec/aacdec.c libavcodec/aacsbr.c libavcodec/avcodec.h libavcodec/mpeg4audio.c libavcodec/ps.c libavcodec/ps.h libavcodec/ps_tablegen.c libavc...

[Alex Converse]

On Tue, Jun 22, 2010 at 4:06 AM, Diego Biurrun <diego at biurrun.de> wrote:
> On Sat, Jun 19, 2010 at 04:14:51PM +0200, alexc wrote:
>>
>> Log:
>> Add HE-AAC v2 support to the AAC decoder.
>
> Excellent :)
>
> Do you have speed comparisons to libfaad and/or CoreAAC?
>

on my system linux/amd64 it is faster than faad. That may be from
having a faster LC/SBR core.

>> --- trunk/Changelog ? Sat Jun 19 11:56:05 2010 ? ? ? ?(r23646)
>> +++ trunk/Changelog ? Sat Jun 19 16:14:51 2010 ? ? ? ?(r23647)
>> @@ -11,6 +11,7 @@ version <next>:
>> ?- CODEC_CAP_EXPERIMENTAL added
>> ?- Demuxer for On2's IVF format
>> ?- Pictor/PC Paint decoder
>> +- HE-AAC v2 decoder
>
> Could you mention the other names this format has? ?I think it's also
> called PS and I'm probably not the only person that confuses them all
> the time.
>

HE-AACv2 is HE-AACv1 mono + PS

PS is a generic (but deprecated?) tool that can run on top of any
codec using SBR as well as MPEG-4 SSC.

>> --- trunk/libavcodec/Makefile Sat Jun 19 11:56:05 2010 ? ? ? ?(r23646)
>> +++ trunk/libavcodec/Makefile Sat Jun 19 16:14:51 2010 ? ? ? ?(r23647)
>> @@ -42,7 +42,7 @@ OBJS-$(CONFIG_VAAPI) ? ? ? ? ? ? ? ? ? +
>>
>> -OBJS-$(CONFIG_AAC_DECODER) ? ? ? ? ? ? += aacdec.o aactab.o aacsbr.o
>> +OBJS-$(CONFIG_AAC_DECODER) ? ? ? ? ? ? += aacdec.o aactab.o aacsbr.o ps.o
>
> Is this general? ?Otherwise I think it should be aacps.c.
>

See above but already moved for the time being.

>> --- /dev/null 00:00:00 1970 ? (empty, because file is newly added)
>> +++ trunk/libavcodec/ps.c ? ? Sat Jun 19 16:14:51 2010 ? ? ? ?(r23647)
>> @@ -0,0 +1,1124 @@
>> +static int iid_data(AVCodecContext *avctx, GetBitContext *gb, PSContext *ps, int e, int dt)
>
> Here and below: extra good karma for breaking these long lines where
> easily possible.
>
>> + ? ? ? ? ? ?re_op += filter[j+1] * (in[i+j+1][0] + in[12-j-1+i][0]);
>> + ? ? ? ? ? ?im_op += filter[j+1] * (in[i+j+1][1] + in[12-j-1+i][1]);
>
> IMO this mass of vars and operators is unreadable without spaces.
>

The "i"s are gone.

>> + ? ?if(is34) {
>
> if (
>
> same in other places
>

"if(" globally fixed
"for(" globally fixed

--Alex