Suggestions for higher quality image for SVCD

[puar]

I somehow managed to get my captures to work at 704x480. Unfortunately, I can't capture at anything like 352x480 because when I start capturing, all of a sudden my screen only captures a half of it (it changes the ratio). This isn't the problem at hand, but if anyone has any suggestions about that, that'd be good.

Anyway, what I've noticed, is that I actually don't see that much/if any improvement from my original 352x240 capture after converting it to MPEG2 using TMPEnc. It seems what I'm lacking is the sharpness or crispness I thought I would get with a higher resolution capture.

I'm capturing in VirtualDub at 720x480, and then converting it in TMPEnc to SVCD using it's SVCD (NTSC) template.

I'm sure there are tweaks to make it look a bit better, so any suggestions would be great.

[Phil Pugh]

There could be many reasons why you aren't seeing any difference. The usual one is the source material. If you are capturing from VCR tapes thenb you won't see much difference because (in general) you have already lost most of the resolution by going to tape.

If you are using DV tapes I would expect a very noticable difference.

Does te original tape have a noticable video noise ? This is the worst enemy of s/w transcoders like TMPGEnc - you can engage the noise limiting s/w - but the processing times do increase quite noticeably.

Phil

[Dr Mordrid]

If you're using TMPGEnc for SVCD you might want to change its default CQ rate control mode setting to CQ-VBR. You might also benefit from using its High Quality motion serarch setting if there is any action in the clip.

Dr. Mordrid

[puar]

Another question

Thanks for your input to everyone in the past two posts. It's been very helpful. I do have one more question.

I've seen in a few other posts mentioning that some people will do the following:

- capture at 704x480
- resize/compress in virtualdub to (i dunno what resolution)
- convert to mpg in TMPEnc or whatever prog of choice
- burn

My question is, does this make any difference at all versus just capturing in 704x480 and then converting/resizing directly in TMPEnc?

One more question, I get confused on the settings for Interlace, Non-Interlace, etc.

Thanks for the help.

Bobby

[Sciascia]

If you can capture in the same resolution as the final product, that is the best method. SVCD is 480*480 and VCD is 352*240. For SVCD type capture you will have to use Huff or uncompressed and avi_io to use a custom frame size (V-Dub may work as well).

If you have to resize for compression, then V-Dub is better than TMPGEnc, there are different types (bicubic, precise bilinear, etc..) and they usually are better than TMPGEnc's resizing. CCE does not resize, so you have to frameserve to it.

Interlaced/deinterlaced. VCD is MPEG1, so it is deinterlaced (like a PC monitor) and SVCD is MPEG2 and it is interlaced just like a TV. SVCD will look much better on a TV for this reason, plus the higher datarate and resolution.

[Flying dutchman]

My own experience (for VCD rather than SVCD) is that a simple 2:1 resize is better than either bicubic or bilinear resize.
The reason being that 2:1 resize uses a 2x2 non-overlapping matrix that reduces noise quite effectively by averaging 4 pixels into 1...