AVI to MPEG or real time MPEG capturing, What's better ????

[rpvc]

Hi there,

it might sound like a stupid question but I have always captured in AVI with a DC30+ and then compressed to MPEG I or II with TMPEGenc getting a satisfactory job done despite the amount of time needed to make an AVI into MPEG.

However, in the last few months I have seen new video cards that allow you to capture directly in MPEG I or II. Do their real time capturing provide the same quality as with AVI to MPEG ????

If so I am going to throw the DC30+ out of the window. I calculated that I would save about 70% of my time.

PLEASE I NEED AN ANSWER

[Jeomite]

I'm pretty sure capturing to AVI and then converting MPEG-1 is better quality, yet more time consuming.

While direct MPEG-1 capturing can result in many non-irreversible problems such as Audio Desyncs...

That's just my primary thoughts on the subject.

[Freddie100]

I have the Hauppauge TV Win PVR, Marvel G400 and Dazzle II.
I am very impressed with the Dazzle II. I get DVD quality when I use my own templates and can burn directly from capture to cd using Nero. I use to do AVI to Mpeg but the quality wasn't as good as I am getting now. Plus I eliminated a lot of processing time.

[Stinky]

I find capturing to AVI will give you better quality and more options for editing

capturing to mpeg can be more convenient but making changes/edits are harder

I go with VirtualDub / ATI captures when I want great quality and the ablitly to do a lot of edits

I go with MPEG-1 captures when I want to watch it right away and just have to edit out a few things (like commercials)

[SHS]

Yup not only eliminated a lot of processing time but also save a lot disk space in the long run.
been you need 22.5gig avg for each hour for AVI.

[kinneera]

It's practically guaranteed that you're going to sacrifice some quality if you opt for real-time capture. There's a reason most MPEG encoders don't operate in realtime...

If you intend to be making VideoCDs, and need the ability to edit your video first, stick with AVI captures. Editing VideoCD compliant MPEGs, or and MPEG for that matter, can be extremely tricky. I know from many frustrating hours of unsuccessful experience...

You don't need 22GB for an hour of AVI if you use a good codec like Huffyuv or MJPEG. I can fit nearly an 90 min into less than 10GB using Huffyuv.

You also want to make sure your system can handle realtime MPEG capture. MPEG2 can be particularly demanding.

[SHS]

kinneera you so full B.S. "sacrifice some quality" I think not, And yes MPEG2 capture card dose to do realtime operate, MPEG2 can be eazy edit just like AVI.
Huffyuv Junk, MJPEG not free, RGB24 is better codec.
If he gose Hardware base MPEG Capture it is far faster no if or but.
"90 min into less than 10GB using Huffyuv" Tell me win a hose can fly I try use it.
Gee that funny I push 704x480 MPEG2 12.0MBit/sec FullD1 VBR "WinTV-PVR" and only eat up 2to3% CPU usage on old 200MMX and harddisk only gose blink, blink where I do AVI 352x240 the CPU usage MAX OUT and harddisk is runing like bad out of hell on P3 600 WoW what a major diff in real-time capture let see you try do that, but on the other hand SoftBase MPEG capture is a diff story like WinDVR, PowerVCR, WinVCR is not very good option.

[kinneera]

MPEG encoding is a process that involves filtering out massive amounts of data while trying to maintain decent quality, by describing most frames relative to the previous or last frame. If you force this to happen in realtime, you will absolutely always, and I mean always, suffer some quality degradation. Whether it is noticeable, or within your standards of quality is entirely subjective.

MPEGs are a total pain to edit. Look anywhere in just about any of these forums, but especially the editing forum, and you will see the evidence loud and clear. Audio sync problems abound, inconsistencies in the actual format of an MPEG proliferate, and most of the tools are expensive. Editing in AVI, using VirtualDub, or in TMPGEnc by setting the Source Range is far easier.

Huffyuv is a very good quality codec, maybe not the best, but certainly far better than any MPEG, because it quite simply ISN'T DROPPING DATA. Every frame is fully described, albeit in a compressed format, whereas 80% of frames in an MPEG file will be totally useless without the frames around it.

I just captured 90 min. into 10 GB this very day. Reality is reality. "Tell me win a hose can fly I try use it" - some kind of retort. Maybe you should learn english before you try insulting other people.

When I do AVI captures at 352x240, my CPU usage is only about 20%. Don't know what the hell is wrong with your computer/capture setup, but whatever it is, it sounds like user error. Probably forgot to actually select a compressor in VirtualDub. And when I do realtime MPEG captures at 720x480, my PIII 800 drops about 20% of the frames, unless I do I-Frame only. Certainly not acceptable.

Finally, if you want this discussion to be helpful to anyone, primarily to person who asked the question originally, you either need to practice your english, not get into such a rage over a friendly discussion and slow down when you type, or you're just a complete idiot, in which case your input is not very trustworthy. Because quite frankly, your entire post is barely readable.

[Nelson37]

With Duron 800, 256 RAM, DMA66 40GIG and an ATI 128PRO, it is getting very close between mpeg-1 and avi, as I lose audio synch at about 1 minute with AVI, with an occassional belch of dropped frames, versus slightly lower visual quality and a few dropped frames with mpeg-1, files easier to manage, plus I'm still tweaking for quality.
Speaking of which, Vdub or any other capture software doesn't recognize my ATI and the native mpeg-2 mode doesn't work with anything, any ideas?

[kinneera]

You should download the latest video drivers for your card and the latest multimedia center from ATI. Also make sure you have DirectX 8.0a and the DirectShow fix for DX8 from Microsoft. With all of those properly installed, you shouldn't have any problems with video capture software recognizing your card. Also, if the AIW is a second video card in your system, you may need to make it the primary video device in BIOS to get some programs to use it properly.

[SHS]

kinneera that was not insulting if think it was well I'am sorry.
No matter what you say the Dazzle DVC2 or WinTV-PVR is far better option plus the time I save to do other thing.
How long do it take you do one 2 hour movie??, For me I capture one 2 hour movie, edit some junk out, burn all with in 3hours and I only use Womble Multimedia Mpeg2vcr.
Huffyuv blow in quality not clean nor as clear as RGB24 (Uncompressed RGB).
Want pass long you so called setting in VirtualDub I try them how to off this site and min other first a Custom 352x240,YUY2 then back to Compressed selete Huffyuv v2.1.1 and try mess around with config in Huffyuv dailog & it still come out looking worst then RGB24 (Uncompressed RGB).
I was just jokeing about the CPU usage I know if they got a Fast 500+ CPU they should have to much problem.
I can sure it not a user error being I have ben using computers for 20yrs now.
So what if I'am world great speller.

[kinneera]

Of course Huffyuv isn't as clean and clear as totally uncompressed RGB captures, but then no codec that employs compression is. That doesn't mean that it isn't a high quality codec in the whole scheme of compression-utilizing codecs. More importantly, though, it that it is going to be FAR better quality than real-time _MPEG_ capture, because the amount of compression being applied to generate the MPEG file is certainly much higher than that employed by Huffyuv. And again, this goes back to the point I make about the difference between realtime and post-capture MPEG encoding.

So really, it boils down to how much disk space you have (sorry, I can't dedicate an extra 12GB to just one hour of video for miniscule quality improvements), and how much processing time you're willing to dedicate. I sure like the results of post-capture MPEG encoding far better than realtime. It is especially a joke to try to do realtime variable bitrate MPEG encoding. If it only takes you five hours to process a 2 hour movie, you are clearly sacrificing some quality, but if you are happy with the result, that is all that REALLY matters.

[Anonymous65]

I have to agree that AVI capture and then compression to mpeg will give you the best results.
However, hardware and even software capture to a *high bit rate* can look very good. I have seen the Mpeg-2 Dazzle DVCII records, and they are stunning in quality! Even the ATI AIW will give excellent mpeg-2 capture *at high bitrate*. But, going directly to VCD, I have not any hardware nor software that can even come close to the quality of the time consuming AVI capture and software compression method. Direct VCD capture on the DVCII is very soft and blurry. Even hardware can't do heavy compression in real-time. Only offline software can do it at this point. As far as SVCD, it's about 50/50. Real-time hardware and even software captures can look quite good, but still, 2600 bps is a low bit rate comparred to 8 or 10 mbps need for excellent looking video done real-time.

So, for VCD and SVCD creation, go with AVI capture and software encoding afterwords. For high bit rate mpeg-2 capture, go with Dazzle or a good quality capture card like ATI or Matrox and a powerful CPU. If you want to it all, VCD, SVCD and DVD, then capture in high bit rate mpeg-2 and then encode it down to VCD. However, there are issues like editing and audio sync when encoding mpeg-2 video to VCD and SVCD.

[iparout]

well, all things considered, for someone willing to do everything (DVD,VCD,SVCD), which is the best solution ? Dazzle DVC II or ATI AIW Radeon ?

@skittelsen : are there ALWAYS audio sync problems when encoding MPEG2 to VCD ?
Let me tell you that with ATI AIW there may also be some a/v sync problems when capturing a video, because AIW used your sound card to do the sound capturing.

[D_Knife]

rpvc,
I think I might be able to help you, this post seems to have gone off the original topic a little bit.

First off Guys, there's no point argueing different codecs in this post, rpvc stated he's using a Pinnacle DC-30+, It's codec locked to it's own internal compression engine.
It spits out MJPEG AVI files of great quality.

rpvc, I also own a DC30+ and I have just added a Dazzle DVC II to my system. I intend to use the Dazzle for most day to day capture and keep the DC30+ for my high quality video work (my company work).
I found the Dazzle to be a quiet achiever, where I used to have to capture, colour correct (for me the DC30+ always needed tweaking to get the picture looking correct, maybe it's just mine) and then spend ages converting into the required format. Now I can just click and go!

Like I said, don't ditch the DC30+, it's still a fantastic card and very useful. But look at a Dazzle none the less.

And yes, about 70% time saving sounds pretty correct.

[AndrewDV]

Hi,

This is a very good topic. However, emotions should be kept personal.

From my experience with many types of capture products (including the DVCII and PVR), AVI captures using a lossless or near lossless codec provides greater flexible and quality for the following reasons:

-AVI capture is not nearly as sensitive to noise as MPEG encoders. Ex: Low LUX situations. Taking evening footage is always a hard task to do in camera work. Even with the correct aperture, exposure, etc settings, there tends to be some noise. MPEG direct captures pick up on the noise and enhance it (I'm knocking the PVR over the DVCII on this one. Both give bad results with noise, but the DVCII handles it better). AVI caps using an appropriate codec are more representative of the original quality.

-In the same category, AVI playback to tape gives superior quality. Not only is MPEG sensitive to noise, but tape is as well. Even with hardware assist (note: Dxr3 and H+ cards are excellent for MPEG playback, that is if the generated MPEG files are supported), the quality does not compare. Using Matrox's DVDMax feature is the best quality for AVIs. Using their MJPEG codec also gives outstanding color and signal boosts.

-Unless you are using a RT2x00 or similar, AVI captures are far easier to edit in NLEs. Using the DVCII, the MPEG files could only be edited with MovieStar. Using the PVR, the MPEG files are harsh to edit with anything but a special editor (note: spanned files still have an editing glitch). Unfortunatley, these editors do not have the same power as needed in Premier and MediaStudio Pro. The best solution given to me was to convert the MPEG to AVI using DVD2AVI and convert back after editing is complete. That's not what I want. I do not need yet another encoding process.

-Unless you have a RT2x000 or similar, rerendering an MPEG to a different bitrate/res is harsh. MPEG decoding is a CPU intensive process. MPEG encoding is a CPU intensive process. The 2 combined when rerendering is brutal.

-Blending animations (from a product like Amorphium, Moho, or Cool3D) with MPEG video looks terrible. In general, animations have sharp edges and compression destroys it. When doing a bluescreen, you do not want the blue edges to mix in with your animation. This is due to compression. A lossless AVI codec does not cause this effect.

-Hardware MPEG encoders still require going through TMPGEnc for VCD/SVCD compliant files. This is the most dissapointing aspect. My reason for using MPEG is not tape projects and it's not to save space (especially when I have 120GB RAID0). My reason for using MPEG is create VCDs and SVCDs. If it captures in MPEG, it should not require TMPGEnc.

-Simply stated, realtime software MPEG encoding provides horrible quality.

-Realtime software AVI editors are going to be released very soon. In other words, edits involving blending, moving path, 3D transistions, etc happen immediately. Exporting the file happens nearly as quick.

OK, I really could go on about my dissapointment with MPEG encoders (poor device drivers, audio sync, non-existent tech support, disgruntled users, etc.). The money spent on these devices can be used to enhance the machine itself (resulting in better editing and quicker results).

However, if you ignore everything and still want a MPEG encoder, consider the DVCII or the Matrox RT2x00.

The RT2x00 is definitely the product for professional editing. It not only captures using the C-Cube, but can re-encode after editing using the hardware. In other words, everything happens realtime (not just capturing).

The C-Cube in the DVCII is single piped which means it can only capture using hardware assist.

Deciding on those 2 should be based on budget and need (the DVCII is around $250 while the RT2x00 is over $800).

[iparout]

well, here are my questions :

1) how much diskspace do you need to capture a 2-hour long MPEG 2 in real-time and how much diskcpace when capturing the same video in avi first and then encoding it to MPEG 2 ?

2) how much time do you need to encode a 2-hour long AVI video to VCD compatible(mpeg-1) or MPEG-2 (DVD) ?

<font size=-1>[ This Message was edited by: iparout on 2001-08-17 03:23:30 ]</font>

[kinneera]

There are no absolute answers to these questions because they depend on way too many variables. The bitrate, and thus the quality you are looking for in MPEG2, will have a large impact on how much space you need for it. High bitrates, or I-frame only captures (which are needed on a lot of mid-range to slower systems to avoid dropped frames) will require much more space.

For AVI, it depends on the compression you choose, if any, and the configuration you choose for the codec. With Huffyuv, it would probably take ~16GB to store 2 hours using the YUY2 compression at predict gradient setting. Since 2 hours most likely won't fit on one CD, however, you can sometimes capture and process in a couple of segments, so it takes less space (if disk-space is a concern).

As for encoding time, it depends on the filters you apply. With TMPGEnc on my PIII 800, I get about 2:1 time when I do CQ (VBR) at 2.5 Mbps on old home videos I captured. (i.e., a 5 min. clip takes 10 min. to encode). This is with motion search precision at "high". If you add a color filter or similar, it will take longer. And effective noise reduction filtering will double the encoding time.

[spidey]

Get a dazzle and live your life. The time saved well offsets the minute difference in quality. To be honest with you all I think that I get far better quality out of my Dazzle than my old Hauppage Win TV (w/ Huffyuv). I have recapped most of my old Vdub cap's into Mpeg caps, and they look and sound great. As far as editing an Mpeg, if it's such an issue, use the SOURCE RANGE FEATURE and re-encode it in TMPG. This is an extra step, somewhat bac tracking towards an avi dub / conversion, but they do seem to encode much quicker again as they're already in Mpeg format.

On VHS dubs I use a Dazzle DVC 1 USB, cap to Mpeg 1 3000 x 224, then I re-encode in TMPG down to VCD. Bit of redundancy and wasted time, but the results are a perfect dub. In Vdub / avi cap, Huffyuv never captured darkness well, and there was always lots of noise.

[iparout]

if someone wants to stick to software SVCD or DVD encoding to achieve better quality, what capture card should he choose ? The source would mainly be VHS and TV and perhaps DVD. Should he go for a DV card with analog inputs or a card that captures AVI (compressed or not) ?

[Kool C]

real-time capture in mpeg,interesting discussion you`re into
well has anyone tried a software called DVMPEG that works
directly into any videoediting software.......read more
at daravision.com....
Or does anyone know if its better use the IEEE1394(firewire)
and go directly via the camera and maybe use the adobe premiere with avisynth plug-in..then you can probably get
a realtime capturing .
kindly regards
KoolC

[Kool C]

sorry ! wrote wrong adress for DVMPEG it is DARVISION.COM

KoolC

[Freddie100]

Quote:
-Hardware MPEG encoders still require going through TMPGEnc for VCD/SVCD compliant files. This is the most dissapointing aspect. My reason for using MPEG is not tape projects and it's not to save space (especially when I have 120GB RAID0). My reason for using MPEG is create VCDs and SVCDs. If it captures in MPEG, it should not require TMPGEnc.

The Dazzle II captures in hardware real time either DVD,SVCD and VCD. It does not require TMPGEnc. Just cut and burn with Nero.

[SHS]

Freddie100 have you ever heard of multiplexing min MPEG Hardware encode chip do not support this features.
So you have use TMPGEnc and multiplex with “Mux” it only take 1min or two to do.

[DigitalOv]

CATUREING TO AVI THEN CONVERTING IS ALOT BETTER


DIRECT MPEG CAPTURE IS LOSSY

CAtureing to AVI (MJPEG or HUFFY CODEC) then Converting to MPEG would result in a Higher Quality Video

[kinneera]

Just an observation - the ATI MMC 7.1 is capable of capturing MPEGs that are immediately ready for burning (with VCDImager, anyway). Nonetheless, I stand by my assessment that you are going to get much higher quality going through AVI first.

[thxkid]

Have been Video Capturing for over 2 years now. here's the conculsion I have came up with 1000's of hours of making family VCD's and SVCD's.
AVI to VCD or SVCD gives you the best quaility. But I keep this down to 5 minutes of less of AVI on the computer.
capturing 10 minute family tapes to computer AVI is ok, anything else is way to time consuming.
When I needed to do 60-90 or 120 min programs, AVI to MPEG is out of the question. So for the past 6 months I have been looking for a fast way to capture directly to MPEG. I tired all the MPEG capture devices under $300. None produced for me a good quality VCD, MPEG1.
Then I tried hauppauge TV PVR, but could not make the MPEG2 into a great quality SVCD, until I got the patch updates from Hauppauges German web site.
Now I produce near perfect SVCD's using this device and the playing around with the bitrate settings that the patch would provide for me. What I mean by near perfect is that my SVCD looks very very close to the original source.
Also I now use the 99mmin. CD-R's that CompUSA sell at their stores.
Yes, you do not want to be editing if you are capturing directly to MPEG1 or MPEG2, its just best if you plan to capture at very long time lengths.
If you plan to capture in AVI then edit your scenes, the best software I have found to use is Ulead Pro 6.o with the HollywoodFX Silver Plugin for pro looking digital transitions.
Step 1. Capture From source using PVR device
Step 2. demux MPEG using TMPGE beta b
Step 3. Burn SVCD using NERO 5.5
I can do A 90min. video source completed, including all these steps and printing covers and CD labels in less than 2 1/2 hours.
If you capture 90min. in AVI format and get it down to a completed SVCD or VCD expect to spend about 8-20 hours depending fast your computer is.

[kinneera]

You can do longer AVI captures if you apply a decent lossless codec like Huffyuv. Like I mentioned previously in the thread, I can fit about 100 min. into 9-10GB using Huffyuv. Of course, this still might be too big, but since a lot of us seem to have sizeable hard drives...