My first Capturing (Pics) + Some questions

[iparout]

Hi there guys...

I have the Matrox marvel G450eTV card and I just conected my VCR to the card to get some of my VHS tapes on the Hard Disk and then burn them onto (V)CD. The VCR is connected to the Composite Input of the VCR...

I captured a video sample to see how it works and test the different capturing codecs and I noticed some *MAJOR* differences...

First off, I captured the video from the PC\VCR Matrox Utility using YUY2 and 704x576 (25 fps) -I am in a PAL region. The quality was good. Here is a screenshot :



Then I used VirualDub to capture to Uncompressed AVI just to compare... So I chose video format --> YUY2 Full (704x576) and Compression ---> No Compression (YUY2). The result was this :



You can easily notice that the Matrox Capture has more "warm" colors than VirtualDub. What was surprising is that even when I used the HuffyUV through VDub, although the quality was not inferior to the Uncompressed AVI, the colors were not as warm as in the Matrox capture either...

1) Why is there such a difference in the colors ? Have any other matrox users noticed something similar ? Could it possibly be because of the Matrox YUY2 codec or is it because of the program I use to capture (PC-VCR and VDub respectively ?

2) Can anyone tell me what is this Flying Dutchman's YUY enabling utility which is needed for Matrox cards ? Can it improve the quality of capturing through VDub (i.e. make the colors "warmer" ? - see above)

3) Which of the two captures is the best in your opinion ? Keep in mind that I want to encode it into VCD or SVCD afterwards...

4) Since I will be encoding into PAL VCD later, which has a resolution of 352x288, is it wise to capture to 704x576 and then reside to VCD specs during the VCD encoding (TMPGEnc) or should I capture directly to 352x288 ? Which of those two methods will yield the best quality ?

Thanks in advance for any answers.

[Brian Ellis]

Yia sou,

IMHO, appreciation of colour rendering, especially from commercial sources, is purely subjective. If you go to a TV dealer who has a "wall" of sets, you will never see two with exactly the same colour rendering, although most of them are pretty close. The same with cameras. Theoretically, in a studio, the colour rendering of cameras is carefully set up so that they match, but, in practice, they rarely do. If you look carefully at any studio broadcast, the highlights and middle tones are usually well matched, but the shadow tones often differ. Just look into the folds of someone wearing a grey suit to see what I mean. IMHO, don't bother about minor shade differences because it is impossible to match the gamma curve of all three primaries (which is an approximate method, anyway, because the tube phosphors do not match the eye's colour receptors) and the eye is very accommodating. Even the spectral response of individuals with full colour vision can vary, as can be shown by their variable sensitivity on the borderlines between red/IR and violet/UV.

The thing that strikes me most about your examples is the jaggies on Mr Lalas' left shoulder, but this difference will not be visible on VCD/SVCD.

If you are capturing to transcribe immediately onto VCD, I don't believe the capture resolution is important (it is with SVCD, though). In this case, capture directly with MPEG-1. However, if you do ANY editing of whatever type, capture at the highest resolution, but not in MPEG-1, and change to MPEG-1 at lower resolution, only at the end of the editing, when you are ready to burn.

Please remember, we have on this forum many who are more royalist than the king. They argue for weeks on minor points of quality which never show up in the end result, because they forget that the last link in the chain is that which is weakest, a TV set which, in itself, is a terrible compromise of video quality. Quite often, the TV actually renders invisible faults that a critical analysis of the video file will reveal are there.

That should put the cat amongst the pigeons.

[iparout]

First off, thank you very much for your input... It seems like you are right about colors and, to be honest, that's a minor thing to care about... However there is another problem now, that has to do with the brightness of the colors when captured and displayed back on TV...

Those screenshots may look good here, but when the captured video (.avi) is played with WMP and displayed on my TV through the TV-Out of my Matrox Card, the colors on my TV set are much more darker than the original VHS tape. There VHS has a much more bright image than the captured video...

Especially the video captured via PC-VCR is *EXTREMELY* dark, to the extent that you can't make out some minor details in the dark areas of the picture. The VDub capture is SLIGHTLY better, however it's still much more dark than the original footage on the VHS (For example, the sky has a cyan color in the VHS, however it is dark blue (?!?) in the captured video.)

I believe that it has nothing to do with the TV-Out but with the capturing process, however it is really annoying, especially in the areas which are originally dark in the VHS footage and in which you cannot make out almost anything in the captured video...

Does anyone have any ideas on how to fix that, or is it sth that will have to deal with ?

Thanks in advance.

[Brian Ellis]

I no longer have a Matrox video capture system installed, but I seem to remember that there is a function which allows you to adjust the brightness, contrast and colour balance on capture ???? Maybe a small tweak will suffice. Certtainly, this feature exists in many capture tools. I convert my analogue to DV now, much simpler and more reliable, zero drops, zero problems.

[mits]

Kalimera,

About colour rendering, you can "correct" the colors in your final mpg using tmpgenc's filters. Usually trial and error method (try some settings, see how they look on the tv set and when you are happy stick with them.
FD's utility is for Marvel G200/G400 and Rainbow Runner cards and it enables the hidden YUY2 capture ability, so you don't need it since you have a G450.
IMHO it is better to capture in 704*576 and then resize to 352*288 for VCD or 480*576 for SVCD. The reasons are :
1. for VCD, by resizing (say in Vdub using bicubic method) the chroma noise is reduced, also you can deinterlace (blending both fields since you have capture both of them)
2. for SVCD you definetely need both fields and a small resize will not hurt that much the quality of your final mpg.
I believe that it is better to capture in avi format, edit and then encode (e.g. vdub for applying resize and other filters, then frame serve with vdub and encode with and tmpgenc )

Good luck (kai kales prospathies)

[iparout]

Thanks for that aderfe...

Since you mentioned it, I am using billinear resizing in VDub. It is supposed to be much better than biubic, as long as you don't enlarge.