|
Post by spenoza on Aug 21, 2022 19:20:31 GMT
For folks who are interested, here’s where the 3 to 1 compression claim comes from as well as an explanation of how RNC was critical to generating FMV on Sonic 3D Blast. youtu.be/IehwV2K60r8
|
|
|
Post by elmer on Aug 21, 2022 20:17:26 GMT
For folks who are interested, here’s where the 3 to 1 compression claim comes from as well as an explanation of how RNC was critical to generating FMV on Sonic 3D Blast. Thanks, interesting to watch, but honestly the thing that I liked the most was the use of raster interrupts to create the perception of a higher resolution. The 3:1 claim was obviously true for that usage, but look at the graphics for the video, and you can see why ... there are large areas in each frame that are either a simple flat color, or a simple dither pattern. That's not usually the case with in-game graphics.
|
|
|
Post by turboxray on Aug 22, 2022 15:59:53 GMT
For folks who are interested, here’s where the 3 to 1 compression claim comes from as well as an explanation of how RNC was critical to generating FMV on Sonic 3D Blast. Thanks, interesting to watch, but honestly the thing that I liked the most was the use of raster interrupts to create the perception of a higher resolution. The 3:1 claim was obviously true for that usage, but look at the graphics for the video, and you can see why ... there are large areas in each frame that are either a simple flat color, or a simple dither pattern. That's not usually the case with in-game graphics. Heh.. I've been touting 'hblur' line offset for PCE since like 2006 and people are like "yeah-yeah whatever". He says it gives the perception of increased resolution, but it doesn't (well it does.. in a distorted sort of way, if you pause it). At 60fps, it looks pretty solid. And if you want to see what the effect looks like, just have a screen shot in native resolution- and set a motion-blur filter to 1 pixel at 0 degrees. It's pretty convenient way to visual the effect, but you'll see that small single pixel detail turns in to double pixel wide detail - 50% reduction in detail. Edge transitions though don't look "double wide pixels" though because of the averaging, so the perception loss in detail is dynamic. I'd say the perception average is around ~70-75% of the source resolution, but of course edge transitions are now smoother (ala more colors on the transitions). And if line dithering is applied you get deliberate/controlled increased color. It's also nice in that you don't have to cycle between two pics to get the average (this trick is basically a variation of that). Anyway, the effect works great on consoles, but not so much on emulators that drop a frame here and there (looks jarring when a frame drop occurs). My PCE color converting app has line dithering as an option solely for this purpose. I chose line dithering not for compression reasons though, but because it minimizes the "flicker" since the difference is now just XOR pattern offset every other frame (it will never show as line-dithering regardless of the frame). If you used XOR pattern dither as the source, then every frame will show as line-dither - which is more noticeable if you're fast switching between two 'images'. I've used it for stills (works nice for images that have "glow" or "bright" light sources), but I've also used this line offset hblur in my SuperGrafx demos, where the translucent light causes a blur on just the illuminated areas (of the BG layers) but not the rest. On a related note to line blur techniques, Mortal Kombat intro video on the PAL MegaCD does a line-averaging approach in a different way; they alternate between two lines.
|
|
|
Post by elmer on Aug 22, 2022 20:23:18 GMT
Heh.. I've been touting 'hblur' line offset for PCE since like 2006 and people are like "yeah-yeah whatever". If what you're talking about is using the +/- 1 X offsets in order to help blur single-pixel dither patterns, then I'm probably also someone that's in the "yeah-yeah whatever" camp ... but perhaps I'm missing out on something. What I found clever about what the Sonic programmers were doing is using the +/- 1 offset to help obscure the low vertical resolution, and the way that it allowed them to use vertical lines to create the dither pattern, which would have dramatically improved the compression ratio. If you were choosing to delibrately apply that blurring as an "effect" on a full-resolution screen, then I'm far less convinced that I'd perceive the results as positive ... but that sounds like something that you have done far, far more research into than I have. As for Mortal Kombat and switching between 2 images at 60Hz in order to create the impression of more colors ... yeah, that was done on quite a few of the old computers, especially the Atari 8-bit IIRC.
|
|