richard@gryphon.UUCP (11/20/87)
In article <1391@atkins.munsell.UUCP> Guy in cast writes: >In article <1871@cadovax.UUCP> keithd@cadovax.UUCP (Keith Doyle) writes: >> >>No, 4k is definately NOT ENOUGH. I don't know just how much *is* enough, >>but it ain't 4096. > >Let me say it again. A very good facsimile display of a true color image can >be obtained using only 4096 colors. Period. Mach Banding? Add random noise. >Make the display wider and taller, not deeper. Use a better monitor. Kevin old sport, you show me an image that trasitions from (0, 0, 0) to (0, 0, 15) SMOOTHLY, without any wierd dithering artifacts, and I'll believe you when you say 4096 colors is enough. I havnt been able to acheive this however... I think the question is not really the number of colors, but the width of the dacs. 4096 shades of blue would be just fine thank you, but 15 shades of blue is just too course. Yes, we've tried mach banding, and random noise and dithering (hmm, maybe should have tried 'time domain dithering' :-) and the conclusion is we'd like 'colors that are closer together'; there is just too much of a difference between (0, 0, 7) and (0, 0, 8). I've seen the Eikonix stuff at shows, it's real nice. So how do you do it, eh, Kevin :-) >Kevin McBride, the guy in the brace Shark cheese. -- Richard J. Sexton INTERNET: richard@gryphon.CTS.COM UUCP: {hplabs!hp-sdd, sdcsvax, ihnp4, nosc}!crash!gryphon!richard "It's too dark to put the keys in my ignition..."
rokicki@rocky.STANFORD.EDU (Tomas Rokicki) (11/23/87)
What would happen if we went to 6-bit DAC's? That would give us
64 shades of each, probably enough at NTSC frequencies. We would
also need to add an eight-bit low-res mode, and extend HAM to modify
all six bits.
It would also be nice to have a 256-color palette for those eight
bit planes.
This all could be done completely compatibly with existing software.
The current hardware registers would write into the most significant
four bits of the 6-bit DAC's, and clear the low two. The current
color system calls would do what they now do, only just on the top
four bits. A new set of color calls would be supplied for the six
bit DAC's. The only problems:
- custom chip space to hold 256 18-bit palette registers
- more time to reload palette with copper between screens
Everything else, I believe, would stay the same. Imagine those
ray-traced pictures then!klm@munsell.UUCP (Kevin [Being Weird Isn't Enough] McBride) (11/24/87)
Put on your leather gloves and read your lookup tables backwards whilst letting the blood of 6 software engineers.... Huh? Sorry. Wrong group again. Oh piss. In article <2348@gryphon.CTS.COM> richard@gryphon.CTS.COM (Richard Sexton) writes: >Look. I want more, okay? Hey, fine with me. You'll find that I'm a real agreeable SOB most of the time. You can use as many bitplanes as you want, as long as you don't insist on using them in *my* back yard... >Kevin old sport, you show me an image that trasitions from (0, 0, 0) >to (0, 0, 15) SMOOTHLY, without any wierd dithering artifacts, and I'll >believe you when you say 4096 colors is enough. >I havnt been able to acheive this however... That's because you're not looking past the obvious. You think that 12 bits means 4 bits each of Red, Green, and Blue. Wrong, Godzilla Breath! :-) :-) 12 bits means you got 4096 colors. If you've got a palette of 16 million colors then you've got a lot of freedom. The hard part is to pick out the *right* 4096 colors. It has something to do with your eyeball's non-linear response to light of varying wavelengths. For example, your eyes aren't all that sensitive to large changes in yellow density at high luminance, so you only need a few shades of yellow to keep you covered. Likewise with various shades of blue. What you need is a monitor that has sufficient flicker-free resolution that at normal viewing distance the dots are smaller than the point spread of the optics in your eyeballs. With that requirement satisfied, you can create the illusion of a continuous tone image with only 4096 colors. The current amiga display is not good enough to do that trick. >... >and the conclusion is we'd like 'colors that are closer together'; there >is just too much of a difference between (0, 0, 7) and (0, 0, 8). Ok, so just *make* them be closer together. I don't mind. Really. >I've seen the Eikonix stuff at shows, it's real nice. So how do you do >it, eh, Kevin :-) Wouldn't you like to know? :-) Hey, if you've seen our stuff, then you should be satisfied with the challenge that you set forth above. The display you saw was only 4096 colors. I'm just not allowed to say *which* 4096 colors we use, or how we determined them. That's for us to know and the competition to figure out. Disclaimer: The above is the result of working weird hours on a weird project. It is the product of a now totally deranged mind and should in no way be construed as the opinion of the Eastman Kodak Co. or it's subsidiaries. -- Kevin McBride, the guy in the brace // | It's the end of the world Eikonix - A Kodak Co. // | as we know it, Billerica, MA \\ // Amiga | {encore,adelie}!munsell!klm \X/ Rules! | And I feel fine...
richard@gryphon.CTS.COM (Richard Sexton) (11/26/87)
Oh, 4096 out of 16 million. Right. Sure, that'll do.
All I was saying was that quantizing colour into 4 bit DACS
was limiting.....
--
Richard J. Sexton
INTERNET: richard@gryphon.CTS.COM
UUCP: {hplabs!hp-sdd, sdcsvax, ihnp4, nosc}!crash!gryphon!richard
"It's too dark to put the keys in my ignition..."