[Coco] Coco3 CMP color swatches

[Robert Gault]

Stephen H. Fischer wrote:
> Hi,
> 
> Thanks, the gif looks like what I wish my digital camera could capture off
> the TV screen.
> 
> Please describe the exact steps that you used to do the capture and the
> color resolution at each step.

Since I don't have a digital camera, I ran my program in MESS (set to 
composite mode), did a screen capture from Win98SE, pasted into Paint 
Shop Pro6, reduced the colors used for the sake of the list bandwidth, 
and saved the file as a .gif.
> 
> Somehow we lost some colors as Paint Shop Pro 8 says there are only 59
> unique colors in the gif. Should be at least 64.

That is probably the result of reducing the color count and dithering. 
The colors look the same but the image is much smaller and bandwidth 
friendly.

> 
> Also why are there dots on number 3?

Dots are an artifact, probably of the color count reduction process. 
This was not intended to be perfect, just an example.

> 
> I was planning to do what I think you did, but are we not just reading out
> the color values that (Jeff and / or John Collier) or the MESS authors 
> coded
> into their programs?

Indeed. For correct display of such a chart as displayed on a Coco one 
would need to look at a Coco. Any method of capturing and then printing 
will result in changes. That includes the best quality film you can find 
and essentially any printing process.

> 
> If you redo the capture and discover where the missing colors were lost,
> just tell me what you did, don't send another image unless I say OK.

You can get the program to play with from my web site.
http://home.att.net/~robert.gault/Coco/Downloads/Downloads.htm
Download PALETTE.BAS

> 
> I really do think that we are just recovering the coded values in the
> program and will find out that the 24 bit color numbers match those in the
> emulator listings.

Of course, but do you expect that any camera is going to exactly 
reproduce the colors from your Coco's screen? The only way you can 
determine that is with a spectrophotometer after measuring a color block 
on screen and comparing it to a color block on a photograph.

I can assure you that any film process will be more accurate than any 
current digital process. Having said that, no film is perfect in its 
color rendition.

> 
> If a digital camera could do a better job, much better than what I made
> available, then that would be very useful. TV rasters are very bad and 
> cannot be improved as the CoCo 3 is only generating one field. Frame?
> 
> But I keep coming back to, is there not a standard way to map a two bit A/D
> into a 8 bit A/D?

Of course, just proportionally expand the RGB colors; 0=0 and 3=256.

> 
> Your gif is so good that I will be printing it along with my Digital Camera
> image and comparing it to my RCA TV S-Video display and my Sony 
> KV-1311CR TV
> with RGB display.

I wouldn't bother as the .gif is not the original capture of the MESS 
display. So it is neither Coco nor accurate.

> 
> It will be my temporary replacement for page 295.
> 
> NOTE!!!!!! I will be changing the background to white and the text to 
> black.
> 
> I also will be using filters in PSP 8 to try and remove the raster lines in
> the digital picture and come up with a single 24 bit value for each Swatch.

You will just be introducing the errors that PSP will generate, on top 
of the errors inherent in your digital camera capture of your monitor's 
screen. Neither your camera nor your monitor are known quantities. For 
that matter, neither is the precision of the Coco3 DAC.

> 
> When I get the chart  ready for printing then some CoCoers can compare the
> results and report.

How can any CoCoer make the comparison except by eyeball of a chart 
printed by you against their own monitor's screen image? That will 
introduce the errors inherent in human color perception on top the 
errors mentioned above.

 ><snip>

Don't get me wrong. What you are trying to do is an interesting project. 
It is just that I don't think you can afford the equipment necessary to 
have any chance of producing quality color charts. You would get much 
better results (regards color matching at least) if you bought a 
commercial standardized color chart, selected by eye the swatches that 
matched the Coco colors, and then asked other users to use the same 
color charts with their systems.

If you think I'm nit picking, let me add that I've seen what 
professionals in the paint industry go through to develop colors to spec 
and it is not a job for amateurs. The comments I've made above don't 
address anything other than wave length and there are others that 
influence color perception; intensity, texture, reflectivity, etc.