A weird thing I notice only with certain videos...

Sticky Fingaz wrote on 12/8/2003, 10:44 AM
I am capturing from Super VHS to DV, via a Sony VXD-2000 Camcorder, using Super VHS cables. I noticed that in some videos, almost always ones with a lot of saturation, the edges of saturated areas are very "zig zaggy". It's hard to explain, but instead of the color red just blooming off of something, it blooms off but with a lot of noticable line artifacts (I hope this makes sense). Think of an 8 bit NES game when I say blocky graphics. Nothing to do with MPEG2 compression, because I notice it in it's original DV format too.

Since this does NOT show up with my ATI video card when I capture it, I am going to point the finger at something else, but I do not know what. I can usually make this problem nearly go away, but it requires an INSANE amount of desaturation, almost to the point of a black and white video, which I don't want to do. I think it may have to do with pixels being stretched or stuffed not in their original manner, since when I view the DV I get from the camcorder, although it says 720x480, it looks way more "square" than my 720x480 video I get from my ATI card.

Any help would be really appreciated.

Comments

Spot|DSE wrote on 12/8/2003, 11:10 AM
from the hip, First place I'd look is a bent pin or bad cable if you are getting oversaturated colors.
John_Cline wrote on 12/8/2003, 11:19 AM
Yes, check the cable. But if the cable is OK, then it is probably an basically unavoidable artifact of the 4:1:1 color sampling used by DV codecs. The problem is most likely to occur on highly saturated reds and is most noticable on areas where there are diagonal areas of different colors. (Like a red triangle on a black background.) I'll bet the artifacts you're seeing look like stair-steps that are vertical lines 8 pixels wide. This is a color space conversion issue and, other than desaturating the color a bit using a proc-amp before the input of the VX-2000 camcorder, there isn't much that can be done about it once it have been captured to DV.

By the way, S-VHS is a tape format, the cable you are using is called "S-Video," which essentially means "separated video." The luminance and chrominance information are carried on separate wires in the cable. It does produce higher quality video than a composite cable, which combine the chroma and luma on one wire which than has to be electronically separated after the input of the destination device.

John
johnmeyer wrote on 12/8/2003, 11:59 AM
Don't know how to stop the problem with DV capture. To get rid of it, however, use the Secondary Color Corrector. Select something that is "blooming red." Play around with your selection until you get a broad range of red colors. Then, use the saturation control to lower the saturation of just red. I do this all the time, and it is almost a standard filter that I use on any VHS (and sometimes SVHS) tape.

If you want, you can use my settings as a starting point. I wish there were an easy way to pass along the presets here in this forum. Lacking that, here are the settings for my "VHS Red Correction" preset for the Secondary Color Corrector:

Rorate Hue: 0
Saturation: 0.816
Gamma: 1.000
Gain: 1.000
Offset: 0.0
Angle: 0.0
Magnitude: 0.00
Alpha 1.00
Show mask: unchecked
Limit Saturation: checked
Low: 61.0
High: 96.2
Smooth: 48.1
Limit Luminance: checked
Luminance Low: 62.8
Luminance High: 127.5
Luminance Smooth: 83.1

Limit Hue: checked
Limit Hue center: 100.0
Limit Hue Width: 28.6
Limit Hue Smooth: 0.0
Vince Denali wrote on 12/8/2003, 12:39 PM
A number of fine suggestions (thanks to johnmeyer for the color correction recipe) were already posted. Since I'm a vx2000 user, I've noticed that I have similar problems with white/yellow (tape or wood) on saturated blue (gym exercise mats or blue walls). The result is purpleish edge around the high contrast transition. Some of the problem may be the Sharpness settings if you're using a Custom Preset. This filter "rings" and may be causing some patterns that are not coded well by DV. While I can't confirm if the Sharpness filter is used on the A/V input, consider setting your Sharpness filter to the middle, checking the pattern, and perhaps setting it to the left most settings.

farss wrote on 12/8/2003, 1:05 PM
Firstly the VX-2000 may not be the ideal A/D for going from analogue tape to DV. I say may as I'm not 100% certain what its capabilities are. If it's lacking the sort of preprocessing that say a D8 camera has then this will not be helping matters.
I doubt though that that will have much impact on the DV artifacts. A proc amp before the A/D will let you adjust things to minimise the effect. I've seen one from Krammer that seemed reasonably priced.
If shooting in the camera it seems you can control this to some extent by optically adding blur using filters. There is an obvious downside doing that but maybe the result is more suitable for some subjects.
Softcorps wrote on 12/8/2003, 3:54 PM
The sharpness control on the VX2000 has no effect on the A/V inputs. The A/D converters in the VX2000 are actually pretty good.

James
rebel44 wrote on 12/8/2003, 9:06 PM
Not all VHS tape are in 640X480. The top of line was 525 line, but most likely the film was in 320 or 400 line resolution. When vieing on monitor at 720X480 the line effect show when picture in in motion(or camera in motion). With digital image stabilization the image can not be processed correctly due to nature of digital signal. With optical (analog) image stabilization the motion of camera and object do not show that much and allmost no sharp edge lines.
I have ATI card and capture analog signal from old VHS the picture res is not that great, but no edge lines because old camera used analog image stabilization. Unfortunatly the new cameras with optical image stabilization are 3 times more expensive that digital.Analog process signal faster that digital.
John_Cline wrote on 12/8/2003, 10:20 PM
Rebel44,

No disrespect, but you have "scan lines" and "lines of resolution" completely confused. There is no such thing as 640x480 in analog video. Measuring the resolution in terms of pixels only applies to digital video. Analog NTSC video has 525 horizontal scan lines, of which only about 486 contain active picture information. DV only encodes 480 of them. "Lines of resolution" refers to the ability to resolve individual vertical lines. This has absolutely nothing to do with the number of scan lines. Regular old VHS or 8mm videotape has the analog bandwidth to resolve just over 200 lines, S-VHS and Hi8 can resolve around 400, DV can resolve just over 500. Standard "over the air" NTSC broadcast television can resolve about 330 lines. Some higher end "standard definition" televisions are capable of resolving 800 or more horizontal lines of resolution, although there is no consumer tape or disc format that can deliver 800 lines. Nevertheless, all NTSC videotape formats, analog or digital, have essentially the same number of horizontal scan lines, around 480.

Your description of optical vs. digital image stabilization is flawed, too.

Once again, I mean no disrespect, but you have some learning to do to clear up your misconceptions. I'll bet you can find out more about these two subjects (and many others) at Adam Wilt's web site.

www.adamwilt.com

John
Spot|DSE wrote on 12/8/2003, 11:17 PM
John, you are always welcome to upload a 'preset' in the form of a veg file to the Sundance site. It would be very welcome.
BrianStanding wrote on 12/9/2003, 9:57 AM
I wonder if the original S-VHS footage is out of NTSC spec? (i.e., overexposed).

What happens if you put a Vegas scope on the captured footage? What kind of levels are you getting on the artifacted reds?