capture card/screenshot advice

Using the phrases “high resolution” and “old-school console games” in the same sentence is slightly oxymoronic. 😉

For example, the SNES has a max resolution of 256×240. If your capture device is at least that resolution, you should be able to extract close to 100% of the original pixels (except the actual color values will vary due to the analog video signal). Also, I would use the lowest capture resolution you can and (for stills) use a program like IrfanView to scale it up, if needed.

What consoles, specifically, are you using, and how are you connecting them to the capture device? I would assume that S-Video is the best the capture card can do, without some hacking, so you’re going to be limited by that interface no matter what card you use. I don’t know of any capture cards that do PAL/NTSC RGB out of the box.

You might try getting the least amount of cable possible between the console and the capture device, i.e. use something like a gender-changer that’s basically two connectors back-to-back (except the SNES has a proprietary cable you’d have to hack, but you get the idea).

For consoles with RGB output, you could use one of these:

http://www.converters.tv/products/cga_to_vga/480.html

Hooked to one of these:

http://www.pixelsmart.com/vga.html

(Although, based on the PixelSmart description, you may be able to get by with it alone…)

One more thing: When console video is involved, this site is your friend 🙂

EDIT: Kind of expensive, but here’s an all-in-one solution: http://www.pluggedin.tv/sweetspot/features/index.html (found via: http://shmups.system11.org/viewtopic.php?f=6&t=27662&sid=033a977182bdd950dfa5f7f4994fabdc)

Okay, now you have to define “close-up” 😛

If you want a picture that’s bigger than, say 256×240 from an SNES, you’re just going to be “making” pixels which simply aren’t there. There are dozens of algorithms for “intelligently” making those extra pixels, but there’s always going to be some kind of picture degradation.

Also, most or all of said algorithms are going to produce extra poor results from an NTSC composite signal, because of the decreased color bandwidth. And it gets even worse if your capture card uses lossy compression on-board.

Commodity S-Video cables for most of the consoles in your list are actually not that expensive. Plus, the one for SNES will also work on N64 and GC, and it’s the same for PS1/2/3; so it’s like getting six cables for the price of two! 😀 I’ve even seen one cable that does all three Nintendo consoles, all three Sony consoles, and Xbox (and 360 if it’s backwards compatible)!

Also, about the “print screen” thing: I’m not sure about a USB one, but the ISA/PCI/AGP capture cards I’ve used overlay the video directly on the video card’s output, so it doesn’t even exist on the level at which “print screen” operates.

Doesn’t the adapter have a Twain driver or something for capturing stills? If not, you can use Media Player Classic or VLC to take stills from the video feed.

But seriously, why can’t you just use a screen-shot from an emulator (except in the obvious cases)?

There are many reasons you can’t get a perfect image from a composite source (or even SVideo, though it’s much better quality).

For one, the image is RGB in the console, but NTSC uses a whole different color space. It describes colors in terms of their hue (basically the wavelength; think “Roy G. Biv” ;-)), saturation (how “pure” the color is) and brightness. The conversion between color spaces is subject to precision losses both ways.

Also, since the color TV system was tacked on to the existing B&W TV standard, called RS-170, most of the bandwidth of the transmission line (the yellow-ended cable from your console) is devoted to just the brightness. This is why colors can shift and smear while the basic “gray” part of the picture (the brightness) stays the same. SVideo solves this by putting the “luma” (brightness) and “chroma” (color; hue+saturation) on separate (hopefully shielded) wires. Component video is the next step up. It splits the chroma into two separate signals, for a total of three. It has about three times the bandwidth of composite and is about as good as NTSC gets; unless you count the balanced cables they (used to) use in TV stations and the like.

Finally, since all of this image data is represented as a voltage on the aforementioned wire, any electrical noise that changes that voltage affects the image – hence my recommendation to use very short cables.

Before I got a digital camera, I used to use a camcorder and capture card to take photos. If my subject was motionless (e.g. eBay photos) I would capture multiple stills of it and then use software to average them together. Since the image I wanted was the same, but the noise changed each frame, this cleaned up the image quite a bit. I still had to fix the histogram, sharpness, etc., of course, but it was a big boost in quality.

That brings up another thing: don’t hesitate to try post-processing your images. It can be especially effective on the older consoles, since you know what the original should be like (e.g. the NES is limited to 16 colors).

I hope you can get something to fit your needs.