[x264-devel] Re: Patch for Scenarist HD compatibility

[Loren Merritt]

On Mon, 8 Jan 2007, Ian Caulfield wrote:
> On 1/8/07, Loren Merritt <lorenm at u.washington.edu> wrote:
>> 
>> Not a bug. i_count is the number of bits passed in i_bits, which is a
>> uint32_t, thus i_count<=32. i_left>0 because whenever it is 0 we increment
>> the bitstream pointer and refill it. So the shift can't exceed 31.
>
> Sorry, I should have explained further - I had the problem when bs_write_ue
> was being called with a value having 23 significant bits - bs_write was then
> being called with a bit count of 45. I assumed that the intended behaviour
> was to zero-extend the value when the count exceeded 31. I could modify the
> patch to make bs_write_ue make two calls to bs_write if you'd prefer.

Yes, 2 calls of <=32 bits is better. What's the largest number you need 
to encode with bs_write_ue?

>>> - move the version info SEI header after the SPS and PPS headers
>>> (required for Scenarist to identify the stream as an H264 ES)
>> 
>> Bugs should be fixed where the bug is. This is a bug in Scenarist, so out
>> of principle I won't work around it in x264.
>
> Could it be a requirement of the HD DVD spec that H264 streams start with
> AUD, SPS, PPS? I don't have access to that spec...

Conceivably, but I won't believe it until someone has a copy of the spec.

--Loren Merritt

-- 
This is the x264-devel mailing-list
To unsubscribe, go to: http://developers.videolan.org/lists.html