Request: PAL mixing test program

[groepaz]

After the vic20 color measurement for odd/even lines someone did a while ago... i somehow forgot to verify we are actually using the right colors for odd and even lines. Since this is a subtle effect, another test program is needed. One of you guys (mike?) can probably knock this up much quicker than i can....

What is needed is a program that shows all "mixed" colors that can be created by alternating colors every other scanline, at least for the colors with close (or the same) luminance. And then those in both possible variations of how the colors can be applied to the scanlines (ABABABAB vs BABABABA)

The c64 testprogram is here: https://sourceforge.net/p/vice-emu/code ... n/paltest/ - the wanted bit is the thing in the top/left, it shows all luma pairs, and then the possible mix colors.

Thanks in advance!

[Mike]

Let's see what I can come up with in a collaborative session of MG Text Edit with FCBPaint. Stay tuned ...

[Mike]

What is needed is a program that shows all "mixed" colors that can be created by alternating colors every other scanline, at least for the colors with close (or the same) luminance. And then those in both possible variations of how the colors can be applied to the scanlines (ABABABAB vs BABABABA)

Here we go (download). Load and run 'mix.03.prg' with at least +16K RAM expansion to get this:



Lots of pastel colours.

The left column of the mixed colours contains the pure colours "LRLRLRLR" in consecutive scanlines, the right column of the mixed colours contains the pure colours "RLRLRLRL".

For those interested, the path went:

MG Text Edit -> mix.00.prg (hand editing)
mix.00.prg -> mg2fcb -> mix.01.fcb
mix.01.fcb -> fcbpaint -> mix.02.fcb (hand editing)
mix.02.fcb -> fcbpaint -> mix.03.prg (PRG export)
mix.03.prg -> VICE, IrfanView -> mix.04.png (ordinary screenshot)

Greetings,

Michael

[groepaz]

NICE! could you provide a screenshot from the real thing too?

What i want to find out right now is if VICE uses the right "palettes" for odd and even lines. That can be seen with those color pairs that produce significantly different mixed color when the odd/even lines are swapped around (eg light green vs light purple in your screenshot).

[Mike]

[...] could you provide a screenshot from the real thing too?

That's not possible for me at the moment, due to RL issues.

I'd suppose it would anyhow be useful to obtain a bigger sample of VIC-20, CRT and camera (shot) combos - to see if there's a significant variation in those mixing differences. There might exist VIC-I samples for which those differences between left and right column are rather small to non-existent, for example.

(... having a VIC-20 connected to a current flat screen display, say with a S-Video to HDMI converter, probably isn't called for either, as that converter very likely will impose its own 'ideas' how the colours have to turn out ...)

[groepaz]

Ah too bad.... someone else perhaps? Mine has +8k only ...

[mathop]

I don't have any memory expansion for my PAL VIC20 but I may be able to hack together something that displays something similar on the unexpanded machine.

[Mike]

In the meantime, people who actually own the necessary equipment shouldn't be dissuaded from taking a picture and posting it here.

[mathop]

Program for unexpanded VIC20 at http://www.michiel.boland.org/vic20mix.zip

Output from VICE:

Output from a real machine (do not have a CRT, well, not one that displays colors)

detail view:

Captures made by two separate capture cards

this one appears to be a bit desaturated

This appears to be a bit oversaturated, so the truth is probably somewhere in the middle.
The same card but with the comb filter disabled (i.e. no color averaging of subsequent lines)

[groepaz]

Cool, thanks, i added them to the testrepo. Also the odd/even lines seem to be correct in VICE. The colors themselves are still a bit off here and there... *sigh*

[necronom]

I can take a photo of it running on my CRT Sony Trinitron if that is any use. Hopefully I can get a decent picture. I'll try to do one in the next few hours.

[groepaz]

It would indeed be nice to get some pictures from different people... if only to confirm (or not) that the colors vary a bit between individual chips

[tlr]

Cool, thanks, i added them to the testrepo. Also the odd/even lines seem to be correct in VICE. The colors themselves are still a bit off here and there... *sigh*

My experience is that colors look slightly different depending on which type they are. Background colors are slightly off from the same border colors on the machines I have seen. Might be just due to some kind of AEC like effect superimposed when graphics is being redered as opposed to border.

[necronom]

Here are two pics.
From my capture source, so it's probably the best one:

From a camera pointed at my CRT, so it's hard to get it looking right:

[JonBrawn]

My experience is that colors look slightly different depending on which type they are. Background colors are slightly off from the same border colors on the machines I have seen. Might be just due to some kind of AEC like effect superimposed when graphics is being redered as opposed to border.

Rest assured, I shall not be deliberately trying to replicate that in my FPGA... having said which, I also haven't intentionally tried to implement the infamous jail bars, and yet they turned up to the party uninvited. What is "AEC"?