Announcement

**cjyo~** · 3 February 2001, 21:33

short answer: no, the Marvel G400 and 450 don't have a DV/firewire port. It's analog video input (S-Video or Composite). The 400 captures with a hardware MJPEG codec, while the 450 has a software MPEG2 codec (I believe).

If you're getting a DV camera, the best solution seems to be the PYRO:

http://www.adstech.com/products/pyro.html

As for your upgrade, the Duron 900 should be plenty. I don't know AMD, so maybe someone else can cover any compatibility issues with the Pyro or similar cards. The IBM drive is great for capture (I've got a couple myself), but you may not necessarily need the ATA controller for DV(I don't know the specs of the Motherboard).

BTW, the RT2000 is only $999...

**This_Idiot** · 4 February 2001, 10:31

The Asus A7V has an on-board Promise ATA100 HD controller. I can't tell you how well it works, since my boot drive is only ATA33, and my capture drives (a pair of IBM GX75's) are hooked up to a Fasttrak Raid controller. The mobo is based on the KT133 chipset and seems pretty solid (I'm running an 800Mhz Tbird on it).

On that particular machine I have a G200 Marvel for analog capture (and messing around with Huffy etc), and an ADS Pyro for DV capture - it all works together very nicely, although I wish I had a second G400 Marvel to replace the G200 so I could have dual-head on that machine too.

**Denman007** · 4 February 2001, 11:13

Thanks guys for your input.

I was buying the maxtor controller card now
for my gateway computer because I just bought the IBM 45Gig Drive. I don't actually have the new motherboard yet.
I do have a Sony Hi-8 camera which I'm using a Hauppauge WinTV capture card - doesn't work very well anymore. You're saying I should get a Marvel g400 to use with this camera(Hi8) and get a pyro(pro) to use with a DV - if i get 1? Both cards work with Ulead software? I'll have to check some prices on EBAY. Looks like for 2k I can get everything I need - except the time.

**cjyo~** · 4 February 2001, 14:11

Depending on the specs of your DV camera, you can also just play your hi-8 through your DV cam, which should convert it to DV for the pyro (without having to dub to DV tape). So, you might not need the Marvel in this case.

I just looked it up and the TRV11 does have:

"A/D Conversion and Pass-through"
http://www.sel.sony.com/SEL/consumer...cr-trv11.shtml

With this feature you should also be able to record from VHS or any analog source to the Pyro. So maybe you should try the Pyro and A/D passthrough, then if that doesn't work out pick up a Marvel too.

**Denman007** · 6 February 2001, 18:07

Ok guys, I have a few more questions...

1. For now I use a hi-8 camera. What is the best format to catpure my video? 320x240 is plently big for me now. I liked rendering in .rm files using video studio 3 - seem to keep the file size to a min. - but i'm having problems doing that now with VS5.0.

2. If and when I get a DV camera will my rendered files be any more compress then when
they are captured?

Thanks

Dennis

**Dr Mordrid** · 7 February 2001, 07:16

1. use the S-Video port for highest quality. What kind of problem are you having in VS5? I'm using it with no problems.

2. DV files won't render significantly smaller.

Dr. Mordrid

**Denman007** · 8 February 2001, 16:50

I can render a mpeg-1 of 15 seconds into about 3.7 megs. The rendered file plays back beautifully at 320 by 280. No problems

However, I can render a .rm file of 15 seconds into 225k. When I play it backs it is
very blury and jumpy. I use to be able to do this in VS 3.0 just fine. What's going wrong with VS5.0? It seems I can't use the same template that I use to capture in 3.0
What should I be using now? Thanks

rj · 8 February 2001, 20:08

I capture alot of television programs from an ntsc satellite (DISH Network) system.

Dr. Mordrid, I have noticed that when I use the "Brightness/Contrast" filter to increase the contrast before compression (divx), it results a compressed file that is sometimes 20% to 30% larger than the exact same data compressed without the contrast increase. This makes sense since "Brightness" should just do a level-shift and "Contrast" should expand the number of intensities represented.
Q1. Have you (or anyone else) observed a similar
linkage between contrast and compression ratio?

A similar question on deinterlacing. I have noticed that when I deinterlace the uncompressed file I get a higher compression ratio.
Q2. Am I loosing information as part of the deinterlace; or
Am I presenting the same information to the encoder in
a format that it can more efficiently compress?

In this same thread there were some questions about HuffYUV and the file size. On this topic I have another similar question. I observed that I consistently get better final compression ratios if I start with a HuffYUV compressed capture than if I start with a Matrox MJPG compressed capture. In several past threads Dr. Mordrid discussed the disadvantages of the MJPeg lossy compressions and "artifacts" it introduces.
Q3. Is the compression of these "artifacts" what causes
the MJPeg compressed capture to yield lower final
compression ratios? (The focus of Dr. Mordrid's
comments in the previous thread was video quality
and here I am asking more about video compression
ratio.)

I "feel Rajnah's pain" when he tries to capture 60 minutes of HuffYUV. I purchased a 45GB HDD thinking I would be able to capture 5+ hours of 704x480 29.970 Hz ntsc. When I switched from Matrox MJPeg to HuffYUV I found myself filling up the HDD after less than 2 hours. I've never looked back though. With the improved results using HuffYUV for the video capture I can easilly fit two hour TV programs that use to take two 650MB CD-Rs onto a single CD-R while compressing at a higher bit-rate. I just needed to spend another $200US on HDDs.

In every case I mentioned above I was using divx for my codec.
Q4. Are the variations in compression ration I mentioned
unique to divx? Or are they consistent with the TMPGEnc
MPEG-1 and MPEG-2 implementations? How about other
MPEG-4 codecs?

*****************

On a somewhat different topic,
Q5. Is anyone aware of a MPEG-4 based codec I can LEGALLY
purchase TODAY that will work with AVI_IO, VirtualDub, and
Media Player 6.x?

I would gladly pay $$$ to indulge my hobby (i.e., recording broadcast televison shows/movies to 650MB CD-Rs) with a clear conscience. MPEG-1/MPEG-2 won't squeeze the programs I am interested in into a single sub-650 MB file. Microsoft has the technology and will "give" it to me but... as far as I can tell, the ba****ds want to lock me into their proprietary "Windows Media Technologies 7". Everything else I have found appears to be vaperware.....

rj · 8 February 2001, 20:55

Aaaaargh!!!!

I ment to post that last reply to the "Capture NOW the NittyGritty" thread...

Rookie mistake. (Check the number of postings on these replys :-)

What I ment to say in my reply to this thread was.... I LOVE MY A7V!!!

I have read a fair amount of critism on VIA chip set mobo's on this forum and as a general rule they make sense but... I have had great luck with mine.

That said, I want to make the following comments about setting up your A7V system.

The Promise driver is a MUST UPDATE item. The bug they fixed with the new release (IT IS VERY ANNOYING) is that once you have installed the origional driver, 98se takes forever to boot. The new driver fixes that.

I don't recommend using the VIA drivers from the CD either. Download the AGP and the USB drivers and load them. I had problems with the 4-in-1 driver and I prefer just installing the two I mentioned (for win98se systems).

Links to these drivers can be found in an A7V FAQ under motherboards on www.asus.com.

My system is actually quite modest (a 700 MHz Duron, 128MB RAM, an old 16GB maxtor 5400rpm ATA66 boot drive, and a new IBM 45GB 7200rpm ATA100 (I wish I had two or three more of those...). Compared to some of the A** kicking systems I've seen mentioned on this forum I am almost embarrased to admit to the Duron. Anyway, with AVI_IO a can capture 704x480 29.97Hz NTSC with never a problem.

The one persistant issue I have had is a frame drop of exactly 2 per minute. Exactly 2 per minute.

One of these days I am going to yank the cheep Creative Ensoniq AudioPCI sound card out and put different sound card in. After reading all the discussion on this forum about dropped frames, I am pretty much convinced that the Ensoniq card is the source of my frame-drop problem.

Then again, I might just get out my old digikey catalog and order me a good 10-15ppm oscillator fix the drift. All those perfectly good surface-mount PCB rework equipment at work has gotta be good for something.

Then again, Dr. Mordrid made a comment one time about performing some procedure he referred to as a "soundcardotomy" (sp?). I'm not sure what exactly he means by that but he IS the Doc and his advice really sounds solid in a case like mine....

**Ghydda** · 9 February 2001, 09:03

To rj:

Q1. Have you (or anyone else) observed a similar linkage between contrast and compression ratio?

A1. Yes, the DivX/ MS MPEG4 codec does compress harder with darker video material.

Q2. Am I loosing information as part of the deinterlace; or Am I presenting the same information to the encoder in a format that it can more efficiently compress?

A2. When deinterlacing, the two fields are spliced together into on frame. Depending on the deinterlace algoritm used, the video material is manipulated (interpolated, blended) into a new video stream. Thus the video material has chanced.
To my knowledge, the DivX codec cannot support two-field based video material, and any attempt to compress such material will join the two fields into one frame and compress the result. Thus giving the codec a really hard time of compressing any moving objects in the video stream.
This will lead to lower video quality on the interlaced video, if the same bandwith is used on both the interlacet and the deinterlacet video.

Q3. Is the compression of these "artifacts" what causes the MJPEG compressed capture to yield lower final compression ratios?

A3. Quite possibly. The artifacts make the video more "noisy", thus demanding a higher bandwidth when compressed, to achive the same quality level compared to noise free video material.

Q4. Are the variations in compression ration I mentioned unique to divx? Or are they consistent with the TMPGEnc MPEG-1 and MPEG-2 implementations? How about other MPEG-4 codecs?

A4. The variations you've seen is consistent to every lossy codec I know of, regardless of what technique they use to get the job done. The more noise free the source material is, the better result with the same bandwith is achived. This is also why every TV-capture I make is run throug a series of filters prior to compression. This cleanes up the video in such a degree that both high compression and good quality can be achived.

I hope this isn't a too far-out reply to your troubles. If anybody has different points of view or thinks I'm dead wrong here, please state so, couz all that I know about video editing comes from "learning by doing". And of course I can have made some wrong assumptions on the things I experience.

Ghydda

------------------
2+2=5 - but only for extremly large values of 2.

**Dr Mordrid** · 9 February 2001, 11:02

rj,

I've noticed a small difference, but not that large of one. Perhaps it was a high motion subject matter?

You get more compression with delacing because you're throwing away half your data. Delacing most times means dropping a whole field in each frame.

HuffYUV sure does give better results both in terms of size and quality. MJPeg does its own version of reducing the color count (quantization) vs. MPEG which can cause interesting effects when transcoding from one to the other. When you start with a "pure source" like HuffYUV a lot of comprimises this forces are avoided.

As far as relative file size goes this can depend on the source material's contrast, motion properties (low motion compresses higher because more frames are similar) and other factors.

Most MPEG's behave similarly in terms of percentage of compression. MPEG-4 is slightly more optimized though.

Buy MPEG-4? Why? You can get Microsofts for free in the On-Demand Producer package you can DL from their site. Also the OpenSouce OpenDIVX codec is in public beta.

Dr. Mordrid

[This message has been edited by Dr Mordrid (edited 09 February 2001).]

Announcement

Newbie Needs Help

Newbie Needs Help

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment