01-31-2009, 11:49 PM
480i/60 and 480p/30, but no idea what it really does. Since it has a loader it may try to patch the code, or change those field flags, or could even include some game specific fixes.
I didn't mention in my previous post, but real interlaced games have two types.
First type renders two half height images (one of them shifted) nicely layed out in the memory next to eachother, one field is presented, the other is drawn, like normal double buffering. What you see is sawtooth effect but at full resolution if nothing moves. There is nothing you can do to make this type progressive.
The second type does not do double buffering but renders full height images at 60 fps (also shifts every second). The video-out still receives half height images though, there is a mixer circuit at the end that can automatically downscale it to the right size. It's a kind of full screen antialiasing effect, as suggested in one of the pdfs in the sdk, 2x vertical supersampling. The result is the same on screen as in the first case. The major difference is that we have full resolution images in memory and from then on the presentation settings are entirely controlled by GS. It would be possible to drop half of the frames (60->30 fps, stops shaking) and switch the video-out to progressive mode, that way downscaling would not happen because the output size it needs matches exactly the image in memory. Of course, in real life, things aren't so simple, the output mixer can be configured to position and alpha-blend two image sources (or one image and a color value) on screen in many ways. This trick needs to be carefully carried out to be successful, and only works for the games that run in this mode.
I didn't mention in my previous post, but real interlaced games have two types.
First type renders two half height images (one of them shifted) nicely layed out in the memory next to eachother, one field is presented, the other is drawn, like normal double buffering. What you see is sawtooth effect but at full resolution if nothing moves. There is nothing you can do to make this type progressive.
The second type does not do double buffering but renders full height images at 60 fps (also shifts every second). The video-out still receives half height images though, there is a mixer circuit at the end that can automatically downscale it to the right size. It's a kind of full screen antialiasing effect, as suggested in one of the pdfs in the sdk, 2x vertical supersampling. The result is the same on screen as in the first case. The major difference is that we have full resolution images in memory and from then on the presentation settings are entirely controlled by GS. It would be possible to drop half of the frames (60->30 fps, stops shaking) and switch the video-out to progressive mode, that way downscaling would not happen because the output size it needs matches exactly the image in memory. Of course, in real life, things aren't so simple, the output mixer can be configured to position and alpha-blend two image sources (or one image and a color value) on screen in many ways. This trick needs to be carefully carried out to be successful, and only works for the games that run in this mode.
Software rendering in complex games like shadow of the colossus, nfs, gt4, okami, any that still runs around 10 fps.