Color counts per screen

For anything related to VDP (plane, color, sprite, tiles)

Moderators: BigEvilCorporation, Mask of Destiny

blargg
Newbie
Posts: 7
Joined: Sat Feb 20, 2010 6:27 am

Post by blargg » Tue Feb 23, 2010 2:26 pm

sheath wrote:BTW, I play with your NTSC filter on my RPTV at all times. Would you consider a comparison between raw emulator output color counts, the NTSC filter's, and Composite useful in any way?
Maybe for the NES one. The SNES one reduces to 13-bit RGB before filtering. The Genesis one isn't very accurate, specifically not doing the rainbow effect at all. I'd stick with video captures.

sheath
Very interested
Posts: 141
Joined: Wed Aug 15, 2007 1:44 pm
Location: Texas
Contact:

Post by sheath » Tue Feb 23, 2010 4:16 pm

Sounds good to me. I'll try to get some NES shots up in the next week. Do you have any specific games/screens you'd like to see?

sheath
Very interested
Posts: 141
Joined: Wed Aug 15, 2007 1:44 pm
Location: Texas
Contact:

Post by sheath » Fri Jun 18, 2010 3:17 pm

I am presently massively updating the pages for dozens of 16-bit games from '89-'91. I have decided to primarily stick to the attract modes for each game and to only post screenshots for now. The screenshots are PNG format from the MPEG-2 capture video and the color counts are taken to represent the relative unique colors on screen from the Composite A/V outputs of each system. This method is tested and the numbers are proven to be reliable. That is if I were to take more screenshots from the same scenes using the exact same set up the color counts vary little or not at all.

It is historically true that debate over the importance of maximum color counts affected the 16-bit consoles more than any other generation, just as it is true that game consoles and their games targeted consumers. The vast majority of consumers in the US did not have televisions with RGB or S-Video inputs by 1992. So we can assume that the Composite outputs represent a "best case" scenario for what consumers could actually see at the time.

This approach allows for the subjectivity by which games were, or were not, judged "more colorful" to be measured using standard statistical analysis. As a less pointed side effect one can see how many colors can actually be counted by modern software and subjectively compare that to how many colors are apparent to the human eye. No standard multiple exists between say 45 colors from the chip output to 61000+ colors from the Composite outputs. Based on these numbers however, one of three conclusions may be observed as more software is "polled" for Composite color counts. Each systems' game library may be shown to have roughly the same range of colors actually displayed. Some systems may be proven to generally display more colors than other systems by a consistent multiple. Each system may be shown to display totally different color count ranges with no obvious correlation to other systems.

As it stands today it appears as though the TG16, Genesis and SNES library typically output colors in the same range. Scenes in the low 1,000s represent static screens of mostly one or two colors for text and a flat color background. Whereas screenshots counted in the 100,000s are typically full bore action sequences with detailed backgrounds, more than three character sprites with simultaneous special effect sprites. Most gameplay related scenes linger in the 20,000-80,000 range. No game so far displays more than about 150,000 Composite colors by this method. Other forms of compression, and video filters boost the measurable colors significantly but uniformly. This implies that the Composite outputs of each system did in fact "normalize" the system's color outputs, but further data (not to mention other test methods) may prove otherwise.

Post Reply