Intel® Integrated Performance Primitives
Community support and discussions relating to developing high-performance vision, signal, security, and storage applications.

Video transcoding

ssingh179
Beginner
94 Views
Hi,

If I write an application for transcoding video (let's say, H.263 -> H.264 ), I can, of course, do a full decode to YUV, and then do a full encode, YUV -> H.264 bitstream.

However, if I wish transcode by only parsing/partially decoding the H.263 bitstream, and then "processing" this partially decoded stream to create a legitimate H.264 bitstream, does IPP provide any in-built support for such operations?

Essentially, the idea is to transcode as efficiently as possible, using IPP.

Thanks,
S
0 Kudos
2 Replies
shyaki
Beginner
94 Views
Quoting - ssingh179
Hi,

If I write an application for transcoding video (let's say, H.263 -> H.264 ), I can, of course, do a full decode to YUV, and then do a full encode, YUV -> H.264 bitstream.

However, if I wish transcode by only parsing/partially decoding the H.263 bitstream, and then "processing" this partially decoded stream to create a legitimate H.264 bitstream, does IPP provide any in-built support for such operations?

Essentially, the idea is to transcode as efficiently as possible, using IPP.

Thanks,
S

either H.264 or H.263 is an IBP codec which in general involves motion estimation/compesation to remove the temp redundency, and DCT-like transform to remove the spatical redundency. I don't think it is possible to transcode withoutuncompress the source and re-compress it to the new format. At least, 8x8DCT is used in h.263 while 4x4 integer transform is used in h.264. They are fundamentally different.

h.264->mpeg2 may be possible with only syntax translation, but I am not sure.
shyaki
Beginner
94 Views
Quoting - shyaki

either H.264 or H.263 is an IBP codec which in general involves motion estimation/compesation to remove the temp redundency, and DCT-like transform to remove the spatical redundency. I don't think it is possible to transcode withoutuncompress the source and re-compress it to the new format. At least, 8x8DCT is used in h.263 while 4x4 integer transform is used in h.264. They are fundamentally different.

h.264->mpeg2 may be possible with only syntax translation, but I am not sure.

I mean h.263->mpeg2.
Reply