i-frame only capturing like the pro's!

[((( atom )))]

hi all,

i just had a little idea, wich might be usefull for all people dealing with capturing different materials.

it is most common to use mpeg1 for the initial capture and do the final processing afterwords.

broadcast people use something called "i-frame only mpeg2" when they have to perform similar tasks, since a: the quality is unbeatable, and b: you can edit the stuff perfectly of course. i guess the same rules apply for mpeg4 and it might be of interest for some people.

for 1-frame only xvid simply set max and min i-frame intervals to 1 and voila: i-frame only encoding.

i'll just tried it with a dvd-source and it works fine. it should have 2 major advantages over mpeg1:
- you can use all the goodies, the codec has, already when capturing.
- that should lead to less transcoding artifacts, since you stay natively in one format.

i myself can not even make big use of it, since i don't usually capture any thing, but i thought it might be something for others..

my regards...

[mf]

Err... encoding speed? Framedrop?

[((( atom )))]

Err... encoding speed? Framedrop?

dunno xactly, since i don't capture. my test-encode did ~30fps. that was mpeg2-source via avisynth with a rescaler involved (640x256 res.) on a 2ghz athlon.

if one has a proper processor, i am quite sure he will be able to capture in a decent resolution.

just try for yourself..

[jggimi]

I used to capture I-mode only in MPEG2 at high bitrates. Images were similar to MJPEG/AVI, but consumed far more resources, leading to frequent framedrops for me. So I've stuck with AVI capture.

[((( atom )))]

there are some people who capture directly into xvid from a tv-source for example using i- and p-frames, wich i suppose to already consume quite some more cpu that i-frame only.

what do you mean by "avi-capture"? with xvid i capture avi.

[jggimi]

AVI capture: I mean capture in an AVI container with HuffYUV or MJPEG video codecs, and PCM audio. I don't directly capture into an MPEG-4 codec for the same reasons I don't bother with MPEG-2. Since I post-process, I don't mind using a lossless (or nearly lossless) codec for capture.

[Swan]

My experience with i-frame only on-the-fly capturing is with Mpeg-2. I can say that you need a hell of a high bitrate to get good results.
With Mpeg-2, and i-frame only, you need to go to at least 10.000 kbps video bitrate, preferably higher, and if you go beyond that, the risk of losing frames increases. I get better results when capturing with I, B and P frames and having the app close the GOPS, so editing is still possible. But all'n all, capturing in Mpeg-2 isn't my choice for video that is to be edited.

[((( atom )))]

well, that mpeg2 is a totally different story ain't no big news. that is why i came up with the idea of doing that in mpeg4. no need for such crazy bitrates and still extremely high quality - isn't that what we all are after?

[dar1us]

I have been capping i-frame only since I had a TVWonder from ATI - this was in excess of about 3-4 years ago. This is under MPEG-4 codecs (XVID for me). The quality is pretty good, just set XVID to minimal everything, min/max i-frames both set to 1.

The quality isn't as good as Huff results, but are a lot smaller. 20 gigs manages about 4 hours under UYUY (VIVO card ATM).

For me, it is either XVID at i-frames 1/1, HuffYUV or MJPEG (but I am not satisfied with results compared to XVID at I-Frames 1/1). HUFFYUV if I have the space, that is about 30 gigs for 2 hours, I have an 80 gig disk for windows/capturing (don't knock it, 0 dropped frames+nice transfer bus (ATA133/Raid) to my working disk (100 gigs).


harrison