Ati Radeon 64 DDR What are the best settings for capture fro

[foleydav]

Ati Radeon 64 DDR What are the best settings for capture from VHS?

[Sulik]

For resolution, MPEG2 352x480 or 480x480, IBBP (4P/2B), deinterlace, as much motion estimation as your cpu can handle.
The quality depends mostly on the bitrate: 4Mbps would be a minimum, 6Mbps preferably.

[mpack]

<TABLE BORDER=0 ALIGN=CENTER WIDTH=85%><TR><TD><font size=-1>Quote:</font><HR size=1 color=black></TD></TR><TR><TD><FONT SIZE=-1><BLOCKQUOTE>
On 2001-12-19 21:34:42, foleydav wrote:
Ati Radeon 64 DDR What are the best settings for capture from VHS?
</BLOCKQUOTE></FONT></TD></TR><TR><TD><HR size=1 color=black></TD></TR></TABLE>

You don't mention whether you use PAL or NTSC, or whether your intended target is VCD or SVCD (I assume its one of those, since that is what this site is all about!).

If your computer is fast enough, *and* you have the disk space, then you should capture using a lossless codec such a Huffyuv - this preserves as much of the information from the source as possible. Cap resolution should match the target (VCD/SVCD) resolution, or be an integer multiple or fraction of it, for easy scaling (you don't have to scale both dimensions by the same amount - for SVCD it is best to preserve the full vertical resolution, which preserves interlaced fields).

If you do a lossless cap you can then apply VDub filters to remove noise, and frameserve into TMPGenc for MPEG encoding.

[wcb4]

<TABLE BORDER=0 ALIGN=CENTER WIDTH=85%><TR><TD><font size=-1>Quote:</font><HR size=1 color=black></TD></TR><TR><TD><FONT SIZE=-1><BLOCKQUOTE>
Cap resolution should match the target (VCD/SVCD) resolution, or be an integer multiple or fraction of it, for easy scaling (you don't have to scale both dimensions by the same amount - for SVCD it is best to preserve the full vertical resolution, which preserves interlaced fields).
</BLOCKQUOTE></FONT></TD></TR><TR><TD><HR size=1 color=black></TD></TR></TABLE>

if you have a machine with the horsepower, capture at a size that is larger that the target, and NOT a multiple of the intended resolution.... this is an old photoshop trick for scaling things up and down when you need to smooth a picture.... by scaling in multiples, the interpolation is usually nearest neighbor and linear interpolation, by giving it a non-multiple or non-fraction part, you cause it to do better interpolation to scale the image, often resulting in a smoother image without having to run and noise removal filter.... That being said, here's what I do.....

1.) use MMC to cap to MPEG2, bitrate up around 6, i frame only 640x480 [the 480 is a multiple or 240, but 640 is not a multiple of 352, so the scaling down eliminates the need to use noise reduction most of the time](your choise weather to let MMC to the de-interlace or not)
2.) use dvd2avi to create my d2v and mpa file (audio is already mpeg audio)
3.) use tmpgenc to open the d2v and mpa files to encode...I do the scaling there.
4.) Sometimes when capping via the composite input(from the vcr, the cable box runs in via svideo), I get a line of junk at either the top or bottom of the file, in which case I use tmpgenc to clip 3 lines from top and bottom combined (depending on how far I have to clip to get rid of the junk line, sometimes all 3 at the top, sometimes 2 at the top and one at bottom) and 2 from each side (just to maintain the aspect ratio, since you capture in 4:3, I make the final output a 4:3 picture because I let the scaling be full screen (keep aspect ration) in tmpgenc.


I get very nice captures using this technique. I have captured from TV and some very good quality VHS sources and most of the folks I hve shown them to have thought they were watching VHS. This has given me the best captures using MMC... I have not dones and captures using vdub, because for some reason vdub seems to want to keep bombing out on my machine

[Sulik]

Why would you do I-frames only if you can do IBBP ?
I-frame only would be only if you have a very old cpu (< 400MHz).
IBBP with zero motion estimation will always give you better results than I-frame only.

[foleydav]

Reply to wcb4: Not to sound like a newbie but I am. What does MMC stand for?

[Fred©]

MMC stands for ATI MultiMedia Center, which is now at 7.5 version...

But I am newbie on clipping with TMPEGEnc... I capture my VHS with VirtualDub cause I though it was the only freeware to crop that junk lines of top and botton video...
Now, If I can capture in MPEG realtime with MMC, with those junk lines and then convert to VCD or SVCD clipping those lines, I would like from our mate the step-by-step procedures, because I am newbie in english (i speak portuguese)

Thanks,
Frederico Porto.

[wcb4]

<TABLE BORDER=0 ALIGN=CENTER WIDTH=85%><TR><TD><font size=-1>Quote:</font><HR size=1 color=black></TD></TR><TR><TD><FONT SIZE=-1><BLOCKQUOTE>
On 2001-12-20 13:00:50, Sulik wrote:
Why would you do I-frames only if you can do IBBP ?
I-frame only would be only if you have a very old cpu (< 400MHz).
</BLOCKQUOTE></FONT></TD></TR><TR><TD><HR size=1 color=black></TD></TR></TABLE>

I don't know about you, but I have a 1.1 GHz PIII and 512 MB of ram..... I can not capture 640x480 30 fps to mpeg2 at around a 6 bitrate with anything other that I frame only without dropping frames.... I can capture at vcd resolution (352x240) but as I said, I cap larger so that the scale down gets rid of most of the noise.... If you can cap 640x480 or 720x480 at that bitrate then more power to you. I just can't do it.



I generally capture, as I said, to MPEG-2 using MMC 7.5 (this eliminates the problem with the files being too large) at 640x480, set the bitrate up around 6 and I have to use I-frame only (if you can do IBP, knock yourself out). I usually do not de-interlace during the capture, but that is just my preference.

After the capture, I open the MP2 files in dvd2avi and save the project from there. this creates a d2v file and an mpa file (the audio is already mpeg audio at that point)

open tmpgenc, load the standard vcd template and use the high quality setting for motion estimation (highest does not seem to really give significantly bette results but takes much much longer)

I then hit the advanced tab and choose my de-interlacing options from there, usually even-odd field adaptation, unless capturing animation, size is full screen, (don't keep aspect ration, you probably captured at 640x480, which is 4:3, you are outputting to vcd 352x240, which is not 4:3, let it distort it, it all works out in the end) make sure it says the source is 1:1 vga capture. You can also select clip frame here to get rid of the junk... remember that you captured in 4:3 aspect ration, clip your frame accordingly, if you clip from top and bottom, make it a multiple of 3 and make the side to side the same multiple of 4 (3h,4v or 6h,8v or 9h,12v etc) I have found that if you capture from copmposite you usually get a line of junk 2 pixels fromt he top, if capturing from tv, its 1 pixel from the top, if capturing via s-video, you usually do not get any, so I personally use the svideo out from myh cable box to the input on the card and do all my captures from tv that way).

I also generally here edit out my commercials if capturing from tv. I use the source setting and select start and end frames for the first segment, save the project, then open it again, find the next segment and speicfy a new output file name na dsave that again... keep repeating until you have all the segments.

Then use the batch encode to encode all the parts of the show... I personally do not re-assemble them, but use nero set to 0 gap and do a disc at once write this way I can skip from one section to the next very easily, but you could just as easily to re-assemble the individual mpegs into one longer mpg file and then use whatever tool you like to create chapters on your vcd..

Your mileage may vary, but I personally get very good results... If anyone has the server space and bandwidth and is willing to donate it, I would be happy to put a minute or two of my captured video up so you can see if the quality using this technique is good enough for you. I have a cable modem, but my upstream is very limitedm and a lot of folks read this board