mpeg vs avi captures

[jerets]

ive been always wondering why so many people go with mpeg for captures when ive only been able to make sense of the use of avi. i was wondering what some of the benefits were to mpeg captures.

also, w/ avi captures ive always ran into interlacing issues (which ive corrected w/ virtualdubs blend filter..which does a decent job of covering it up) - but does mpeg have interlacing issues?

ive always thought avi made more sense if your going to watch your captures on the pc.. b/c i've always saw it from the point of view, im able to capture more quality and pack it in less file space than i would be able to in mpeg.

Always wanted to see the other side of it.
thanks in advance for any helpful responses
jerets

[jerets]

any1 know any benefits?

[sterno]

Do you mean to capture video, or to store and watch captured video? For storing and watching, people use MPEG because if you burn to VCD, SVCD, or DVD you can play it in a set-top player on a television with all the convenience of a device that was designed to do nothing but play movies. An AVI file will generally require a computer for playback. Personally, I have a 17" computer screen with cheap speakers and a 27" TV with a good Dolby receiver and decent speakers, so I prefer to watch them on my TV.

Many people actually capture in AVI with high-quality codecs like HuffyUV or MJPEG (or DV-AVI with a firewire card) and then convert to MPEG. These codecs generally produce files that can easily be edited without loss of quality, but which use enormous amounts of disc space. Since you mention higher quality and smaller size, you're probably talking about AVIs with DivX (or some variant of it). AVI is just a generic way of putting video and audio together, both can be compressed with a variety of different codecs with different characteristics (or uncompressed).

Interlacing is unrelated to the format used. However, SVCD and DVD allow interlaced video, so if you are only going to watch on an interlaced display like a TV you don't actually need to deinterlace.

[pepegot1]

Direct capture to MPEG is not very good, unless you have some expensive equipment. It's advantage, of course, is small file size, but editing MPEG with Home editors like VS 6.0, Pinnacle Studio, etc produces poor quality video. MPEG is difficult to edit in. I edited a DV file in Ulead's VS 6.0, converted and burned a VCD that was pure garbage. I directly captured in MPEG and that was horrible. Now, I capture, with Scenalyzer to DV2 (can use Virtual Dub on this , if needed), use VS 6.02 the edit/effects/titles/etc., save in original format, which uses the Main Concept codec 2.04, I installed. Then, I convert to mpeg using either TMPGenc or the Main Concept Encoder. The results are very good video & sound. Main Concept(MC) is much faster than TMPGenc, both have their advantages. Both result in good video, with MC being a tad more consistent than TMPGenc. Probably, because the whole job is done by MC. The only problem is big files in DV2 format, and a 2 gig limit for DV2 vs DV1(4 gig limit). DV is somewhat smaller than raw AVI

[hwoodwar]

I think the main reason people are attracted to MPEG captures it that the encoding is done in real time. There is no waiting XXX hours to re-encode after capture.

I have one of the many ATI cards that support capture to MPEG-1 and now MPEG-2 format. I made a number of test comparisons when I got the card, and again when the software was upgraded to "standard" MPEG-2 support. In my test comparisons when I captured to avi (using the HuffyUV codec) and then re-encoded to mpeg using TMPGenc, I got a much higher quality picture, afterwards to save to VCD, CVD or SVCD. Since those tests I have never gone back to capturing strait to mpeg format.

[jerets]

yea.. i was referring more to the end result (encoding). a lot of helpful responses! thanks guys

sterno: yea.. we just got some bose speakers.. so i can see what you mean. and they do sound nicer then pc speakers

Thanks guys for your helpful responses!
i appreciate the help
thanks again !
jerets

[Bradskey]

Capturing MPEG with various versions of the included ATI MMC is crap, yes, but that doesn't mean capturing to MPEG is always crap. ATI software actually makes the signal from their own products look worse than it is. Its really pathetic that you have to get a third-party TV viewer to get the most out of an ATI TV Wonder or an ATI All-in-Wonder. A good PVR application and WDM card combo can produce results from analog sources with only imperceptable quality loss compared to uncompressed AVI (on sufficiently powerful systems). This is without expensive equipment.

And frankly I don't know what in the world you are talking about when you start saying your AVI caps take up less space than MPEG, unless you're using some lossy codec that's bound to be far worse than MPEG, or DivX, which is MPEG4. A good Huffy cap is going to be hundreds of megabytes per minute video. The equivalent MPEG will be anywhere from 40-60 megabytes per minute video. This is one of the benefits of MPEG -- smaller file sizes. Especially if you have a Win9x system, many caps can be done without file splits or having to have 60+ GB hard drives.

As for interlacing issues, its not an issue. The source video is interlaced, and MPEG2 supports interlace video, so I cap to MPEG2 and leave it interlaced because the file will be be played in a DVD player and displayed on an interlaced television set. If its some clip you just want to save on PC your MPEG encoder should be able to de-interlace it for you.

I will say that capturing MPEG directly to S/VCD format with the intent of immediately burning to S/VCD yields inferior results. Not always terrible, but not the best possible. MPEG-2 cap should be done at high bitrates and resolutions (at least 5-8 Mbps) and then frameserved into a good MPEG encoder for good quality final output. I also must admit that MPEG is not easy to edit. For AVI you can easily do frame-accurate edits and cuts using VirtualDub. MPEG cutting is usually done with less-than-accurate tools like the included TMPGEnc merge and cut tool. Such tools are usually only able to edit at key frames or on 1 second boundaries. For enough $$$ you can get pretty good tools like MPEG2VCR or M2-Edit to perform frame-accurate edits and cuts on MPEGs similar to the way you edit AVIs with VirtualDub.

[macquest]

hi grabbers

AVI or MPG ??? well that sure depends on what you want.. in the end.
ill tell you about my experiences.
Ever since i bought one of the cheopo DVD VCD SVCD players, i want to have a playable CD. Also.. i like to exchange some if them with friends. So the standard is VCD or SVCD.

First, about the MPG way. After a LOT of testing, i found WinDVR to be the only and best Software for realtime encoding. Not that i realy like it, theres a lot of trouble... another story. But, it grabs VCD in realtime, in a pretty good quality. Id say it is about 95 percent of a TMPGenc encoded MPG1. Anything else... like PowerVCR, etc etc... is garbage.
The disadvantages... well you cant manipulate the material. Theres no sense in re-encoding, so you have to live with the quality. I usualy cut the material in M2-Edit, which is fantastic. But thats all. You cant render a logo away, you cant change size... etc. (btw i bought the Terratec Cinergy 400 that comes with WinDVR, but i grab from a cheopo Medion card, because the Terratec produces a kind of slimy picture)

Thats why i checked the "AVI way". Here i have tried first the well known grabbers, like VirtualDUB, iuVCR, Video Capturix, FlyDS; but they all had issues with stability. talking about the big freeze in the midth of a nice movie. (well, i dropped virtual dub because of the horror tuner interface)
Then i gave ShowShifter a try, and it runs and runs. ShowShifter comes with the PICvideo Codec, and thats also my fav one.
The "AVI to MPG" way needs a lot of time, but now i have near to perfect results. I load the grabbed PICvideo AVI-s into VirtualDUB, render the TV Station logo away, i can enlarge if i want, etc etc... then i frameserve it to TMPGenc... the best MPG encoder out there. burn with Nero, of course.

So, for now, if i want to grab standard stuff, or i want to grab a whole night, i prefer direct to MPG1 grabbing with WinDVR. If i want the best i can get, i grab with Showshifter and encode with TMPGenc.

[Gazorgan]

When I capture, I generally use Huffyuv and am 'virtually lossless'. This equates to 10-20 GB/hour depending on the resolution I'm capturing to.

Pros:
No artifacts
I can encode to whatever format I want
No need to frameserve, I can open it in Vdub or Tmpgenc and go

Cons:
Takes up a LOT of space
Stiff Disk Subsystem requirements
Need W2K or XP with NTFS, 98 just doesn't handle this as well, and the 4 GB file splits are jsut a workaround
Need to re-encode it, but then you would need to anyway to apply filters that are nearly always needed in captures

The one 'pro' in capturing to mpg is that you don't need to re-encode it. But then, you are getting a single pass encode of an analog source, so the quality is mediocre at best. You can capture to VCD on a fast machine with pretty good results, similiar to what a VCR looks like. Capturing to SVCD never looks as good as capping to huffyuv and doing a 2-pass encode.

I guess the choice depends on the final product. If you jsut want to watch tv eps, then capture to MPG. If you want to archive a tape, then avi and re-encode.

[Nelson37]

AVI cap, lossless or no codec, gives best quality for HUGE filesize, dropped frames are a significant problem at high resolutions. Editing is easy.

You can lower the resolution or use a lossy codec at the cost of some quality, and most likely eliminate the framedrops.

Frame dropping becomes more significant when IVTC is used. IVTC is very important for reducing filesize AND increasing quality.

Capture to DIVX has drawbacks but for PC only display, hi-quality modes are an option.

MPEG-1 has no support for interlacing.

MPEG-2 has many advantages for capturing. Smaller filesize allows for higher resolution with NO frame drops. Software often gives significant control over quality options. Multiple videos can be captured before filling available space.

REAL-TIME capture to VCD specs yields poor quality for most, SVCD can be tweaked to barely acceptable.

The best of both worlds? MPEG-2, 12Mbps, CBR, I-frame only, quality and CPU usage set to highest, Motion Search vectors quadrupled, capture 720x480 with overscan clipped off for 672x448, Process just like a DVD Vob file, run thru VDUB and edit just like an AVI, use IVTC and Multipass encoding. SVCD and CVD made this way I cannot tell from same film captured as AVI and processed with same parameters.

Many people use smoothing filters on AVI and other files, particularly when creating VCD. Careful use can give very good output, but something that occurred to me is this - What is possibly the most sophisticated smoothing filter available? MPEG. Designed specifically to through out UNSEEN detail to reduce filisize. Movies from Digital Cable, or DVD for that matter, I actually MPEG compress the second and third time. Sure there mathematically is a loss in quality but I can't SEE it. Processed AVI file looks worse unless smoothing or higher bitrate is applied. Even at the higher bitrates, difference is marginal, and for the same bitrate on the MPEG capture I can use a higher resolution, so comparing same size finished file its a tossup.