Question on decoding process.

shinjite — Tue, 25 Aug 2009 15:25:17 GMT

So I have moved from the simple decoder to attempt to decode a whole video itself then just re-encode it. Reason being is to figure out how umc works. I'm using mpeg 2 so which uses yuv420... How much bytes do i advance per frame I've been reading this is the same as yv12 but backwords (or vice versa) - So does that mean i use same equation that h264 method uses (width * height * 3 / 2) which I would like to understand why because I have read some links that say yuv is 12 bytes per pixel... so wouldn't that be w*h*12...? I also read that 2 pixels are packed into this format? so wouldnt that make it w*h*6?

Next part of my problem is the simple decoder dies at frame 28 with status of UMC_ERR_INVALID_STREAM=-882, then that do loop exits... Well I made it so it doesnt exist and suprisingly the next frame after decodes!

This does this about 10 times again on 10 other frames - shortly after frame 428 out of probably a lot (10 second video) it crashs in ipp somewhere(after about 5 of these errors UMC::UMC_ERR_SYNC.

Any ideas?

Re: Question on decoding process.

Chao_Y_Intel — Mon, 31 Aug 2009 01:11:53 GMT

Hello,

For yu12 data, it is 12 bits per pixel (not 12 bytes). so the total size is width * height * 3/2.
check here for more info: http://www.fourcc.org/yuv.php

For the code die in the loop, could it be possible to attached your test bistream? so we can have a further check.

Thanks,
Chao

topic Question on decoding process. in Intel® Integrated Performance Primitives

Question on decoding process.

Re: Question on decoding process.