[FFmpeg-devel] [PATCH] dpx: fix some stupid typos

[Holger Hummel | Celluloid VFX]

Georg Lippitsch wrote:
> Am 28.08.2012, 16:05 Uhr, schrieb Paul B Mahol <onemda at gmail.com>:
>
>>>          case 16:
>>>              if (planar) {
>>> -                int source_bpc = target_packet_size / elements;
>>> +                int source_bpc = source_packet_size / elements;
>>>                  int target_bpc = target_packet_size / elements;
>>>                  for (x = 0; x < avctx->height; x++) {
>>>                      uint8_t *dst[AV_NUM_DATA_POINTERS];
>>>                      for (i=0; i<elements; i++)
>>>                          dst[i] = ptr[i];
>>>                      for (y = 0; y < avctx->width; y++) {
>>> -                        for (i=0; i<3; i++) {
>>> +                        for (i=0; i<elements; i++) {
>>
>> Shouldn't this crash if there is alpha channel and gbrp is used? gbrp
>> only have 3 components.
>
> Indeed. I also found out that there are some more flaws with the 12 
> Bit decoder (but 10 Bit seem to work well).
> I'll submit some more patches in the next days.
i'm not a developer/programmer. just following this list a bit out of 
interest in what's going on (e.g. dpx support).
just a note from a "vfx professional's" point of view: dpx with alpha 
will/should be a very rare case. nonetheless they can be created with 
the various software packages that support it.
in case that's of any help for you guys i can offer to send dpx examples 
with/out alpha (written by TheFoundry's Nuke) - either little or big 
endian, 8/10/12 or 16bit - or all of those variations.
just let me know if i should upload them somewhere (and in case of 'yes' 
also where to upload to).

cheers,
Holger

-- 
Holger Hummel  -  holger at celluloid-vfx.com

Celluloid Visual Effects, Paul-Lincke-Ufer 39/40, 10999 Berlin
phone +49 (0)30 / 54 735 220  -  mail at celluloid-vfx.com