Geez video capture devices are meh and shit at once.
Just for the record, both my NES and SMS are run through a VCR first, then to the computer. If my TV had Video/Audio output, then I'd have no need for the VCR. I don't see any rainbow artifacting on the TV (or in the screencap for that matter), so perhaps it's the VCR's fault?
single frame capture
same but I applied some horizontal filter in PS to clean up the image (remove some of the ringing and such artifacts from the crappy cable)
blended mode from Dscaler:
same as above but with filter to smooth ringing artifacts
As you can see from the 1st and 3rd pic, a lot depends on what the capture card software is doing. Many employ software filters (some controllable by the app, some not/fixed). Power Video (or something like that) always applied software filtering to the incoming signal. Looking at the first image, there's definitely artifacting in the ca[ture card process that's not happening on a real TV (being it HD or SD CRT). The checker board pattern appears to be a product of sampling the chroma signal. I'm not sure why it's not smoothed out in a single frame (but it is in a two frame blend mode).
There's also something else happening in the capture card shots. There's a low pass filter being applied across the Y channel (and seemingly no way to turn it off that I can find). Just look at these shots in comparison.
Here's a snap of my SDTV:
And on my HDTV:
And the original raw shot:
Note: Both my SD and HD set are much cleaner/sharper than I can show by taking pics of the screen, unfortunately. My HD set looks extremely close to a raw emulation shot (perfectly square edged pixels) - with the exception of some colors. Pretty incredible feat for composite output. Though I strongly suspect the PCE's special composite 'filter' mode was designed to work in conjunction with better comb filters, to reproduce/pull back out the original detail (two frame process though).
EDIT: Oh, btw. This is PCE game running in 7.16mhz dot clock (higher horizontal res than Genesis 320x224 mode). And yet still manages to show such detail.
Last edited by tomaitheous; 12-16-2010 at 03:22 AM.
I meant the weird off colors seen at the edges of black outlines on the cloud (in SMB3) or the blocks/bricks on the SMS screenshot.
Somewhat akin to CGA composite artifacts in 40 column/320x200 mode, though far less extreme than 80 column/640 width.
http://en.wikipedia.org/wiki/Color_G...color_monitors
also seen here on a Tandy 1000 via composite:
http://www.youtube.com/watch?v=v4HYiM4DpwE
That perfectly describes all the funky colors in Apple][ games actually.
Still, is this color thing a problem to you? Or is it more like "other than that, those screencaps look pretty good"?
Yeah, the Apple II relies on color artifacting to blend 280 down to 140 pixels across with usually 4 colors (and 2 "palettes" -technically you could have 6 artifact colors on-screen, but there's limitations), or for models with the double highres, they use 560 down to 140 with 16 colors in the same manner.
The CoCo sometimes uses it, some CGA games offered special composite modes, Atari 8-bit rarely used it in the 320x192 mode (it already had a proper 4 color 160x192 mode with a nice 128 color palette, so it was less of an issue), and a few others.
The Genesis, SMS, NES, and many others couldn't use such artifacts due mainly to the resolutions not directly corresponding to NTSC color (3.58 MHz -with the other cases using 7.16 MHz or 14.32 MHz pixel clocks -so they perfectly doubled or quadrupled the display -though technically, 10.74 MHz would be exactly 3x, so that might work too). Machines using inconsistent composite encoders or improved encoding to reduce artifacts would also not have such "perfect" consistent artifact possibilities to use either. (and all would fail in PAL due to the different color carrier)
I was just surprised to see those artifacts... I have no idea where they're related to the capture card or not (they seem to come from the consoles -and the capture card is simply accurately displaying them), but I'm not sure.Still, is this color thing a problem to you? Or is it more like "other than that, those screencaps look pretty good"?
I haven't seen that sort of thing on those consoles before, and usually those resolutions (5.37 MHz) tend not to result in those sort of artifacts in composite anyway. (and if the SMS uses the same CXA1145 video encoder as the Genesis -which I thought it did- that woudl be really odd as the Genesis lacks artifacts of that sort in the 5.37 MHz res mode -256x240)
RF sometimes has ghosting that looks vaguely similar, but I don't think any of the cases I picked out were RF. (and it doesn't look right for RF ghosting either ans that's usually 1-directional, not on both sides of a line)
I just tested 10.74mhz dot clock a couple of days ago. On the PCE (as I don't have access to it for any other at the moment). Unfortunately (or fortunately, depends on how you see it), you can see full Luma res on the PCE. There's no filtering before and after. And the cross talk filter does nothing to effect this (on or off). You can see full pixels even for max white to pure black single pixel patterns. Surprised it one way to put it, when I saw that. I tested this both on my SD set and my HD (crt) set.-though technically, 10.74 MHz would be exactly 3x, so that might work too
There are currently 1 users browsing this thread. (0 members and 1 guests)