Chessboard display in phased ChronoColour (pCC)

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

Andrew Davie

Just a quick mockup to see how a chessboard would look.
It's REALLY hard to get a decent contrast/look in a mix of just 3 colours.
Also, PF graphics make it hard to make nice-looking pieces.

I wish someone would do this in sprites; I don't have the skills.



Bomberman94

#1
I like it - the game pieces are easy to recognize.

Three colors means: two per figure and chess board has blue (and black ,,doesn't count")? Or three different colors per square?

Andrew Davie

#2
Quote from: Bomberman94 on 02 Jan 2026, 03:14 AMI like it - the game pieces are easy to recognize.

Three colors means: two per figure and chess board has blue (and black ,,doesn't count")? Or three different colors per square?


No... it's complicated.

We can only have one colour on any scanline on any frame.
So, different colours for each full line are used both in successive frames, and in successive scanlines.
The 8 available colours are formed by mixing 3 "primary" colours, which are "rotated" over lines and frames.
We can't change these 8 colours (they're a consequence of the mixing of the set 3 colours).

So that means that anything on a line must be formed from the 8 colours - but actually to be pedantic they must be formed from the 3 colours that make up those 8 colours. But not really - anything on a line is actually only ever one colour. But that colour changes over time between the 3 primary colours, in sequence.  So the line's colour goes... red... yellow... blue... red...yellow... blue. And I also change which pixels are on or off every frame, so some frames we see a pixel when the line is red, and we don't see that same pixel when the line is yellow or blue. It's multiplexing the pixels on/off as the colours 'roll' through the 3 primary colours. It's complex.

So in the chessboard we see black - the dark squares, blue - the light squares, and then "white" for the white pieces, "pink" for the shading on those white pieces, "brown" for the dark pieces, and "yellow" for the shading on those dark pieces. Or something like that.  All of those colours are formed from the 3 "primary" colours that are used to set colours on any scanline.

So, primary colours used are NTSC  0x44, 0x2A, 0x96
That is, red, yellowish, blue

Everything you see on the screen is always only one of those colours... BUT a different one every frame, and every scaline. But to be clear there are never OTHER colours on the screen, just those 3. But the rapid changing of which of those three colours gives us the mix.

It makes it very hard to chang.  You can't, for example, say "make the white brighter" -- because that would mean making ALL the colours brighter because white is all 3 mixed. If I wanted brighter "white" then I also have to have brighter blue, and yellow and red. Which means all the other mixes change too. So everything is very very much tied together as a massive interdependent mass of colours.




Andrew Davie

#3
I asked ChatGPT about optimal colours...

Quotei have three colours (any). I form 8 colours by mixing the 3 -- and on/off pixels.  Kind of multiplexing in time. One colour per TV frame, so it takes 3 frames to form the 8 "colours". I want to represent a chessboard, with one of the colours being blue for the light squares. What otehr two colours should I use to enable me to have (whitish) pieces, and darker pieces which are clearly distinguishable. Each piece must have a light/highlight on one edge.

We couldn't do this sort of thing even 5 years ago. That it understood the question is remarkable enough. But that it gave a sensible - and workable - answer is better yet...

QuoteYou are effectively doing 3-bit temporal colour synthesis (three primaries, on/off, summed by persistence), giving you 8 perceived colours. One bit is already committed to blue for light squares.

What you want the remaining two primaries to achieve:
    •    Distinct light vs dark pieces
    •    Pieces readable on blue squares
    •    Pieces readable on dark squares
    •    A highlight edge achievable by toggling one channel

Recommended choice

Use:
    •    Blue (given)
    •    Green
    •    Red

This is not arbitrary — RGB is optimal for perceptual separability under temporal multiplexing.

Resulting 8 colours (by frame accumulation)

R    G    B    Perceived
0    0    0    Black
0    0    1    Blue
0    1    0    Green
1    0    0    Red
0    1    1    Cyan
1    0    1    Magenta
1    1    0    Yellow
1    1    1    White

Practical mapping for chess

Board
    •    Light squares: Blue
    •    Dark squares: Black (absence of all three)

Pieces
    •    White pieces: White (RGB)
    •    Highlight edge: drop blue → Yellow highlight
(high contrast, reads as lighting)
    •    Black pieces: Red or Green (choose one consistently)
    •    Highlight edge: add the other channel
    •    Red piece → highlight = Yellow
    •    Green piece → highlight = Cyan

This works because:
    •    Yellow is perceptually "lit" without being white
    •    Cyan reads as a cool specular highlight
    •    Both remain distinct from blue squares
    •    Temporal flicker averages cleanly

Why not other primaries?
    •    Using non-RGB primaries (e.g. amber, purple) collapses contrast when temporally mixed
    •    Blue + two secondaries produces muddy greys
    •    RGB guarantees maximal perceptual distance with only 3 bits

Summary

Blue + Red + Green is the correct choice.
It gives you:
    •    A stable blue board
    •    True white pieces
    •    Dark pieces that are not confused with board colour
    •    Clean highlight edges using a single-bit change

This is essentially the minimum viable colour basis for readable chess under temporal multiplexing.






Andrew Davie

#4
Quote from: Andrew Daviei am doing temporal multiplexing as described, but i am also doing line-by-line multiplexing - changing from R to G to B down the lines, and of course appropriately on every frame. any name for this?

Quote from: ChatGPTThere is no widely established single term for that exact combination, but it sits at the intersection of a few known techniques. The most accurate way to name it is to compose the terms, not invent a new one.

What you are doing (precisely)

You are combining:
   1.   Temporal colour multiplexing
– colour components distributed across successive frames (field-sequential colour)
   2.   Spatial multiplexing by scanline
– different colour primaries assigned to different horizontal lines
   3.   Persistence-based colour synthesis
– the eye integrates both over time and vertically

⸻

Closest established terms

Field-sequential colour
This is the standard term for "one primary per frame" displays (e.g. early colour TV, DLP projectors).

Line-sequential colour
This term has been used historically for systems where colour information alternates by scanline (not common, but documented in early TV research).

Spatiotemporal colour multiplexing
This is the most correct descriptive term for the combination you are using. You will see it in vision science and display-tech literature.

⸻

Best concise name (recommended)

If you want a technically defensible label:

Spatiotemporally multiplexed field- and line-sequential colour

That sounds long because the technique genuinely is a hybrid.

For a shorter working term (engineering shorthand):

Line-interleaved field-sequential colour

or

Spatiotemporal RGB multiplexing

⸻

What it is not
   •   Not colour subpixel rendering (that is static spatial RGB)
   •   Not Bayer or CFA (no reconstruction filter)
   •   Not dithering (no error diffusion)
   •   Not PAL/NTSC chroma encoding

⸻

Historical relatives (for context)
   •   Early mechanical colour TV used field-sequential colour
   •   Some CRT experiments tried line-sequential colour to reduce flicker
   •   Modern DLP uses field-sequential colour only
   •   You are effectively trading vertical resolution + temporal resolution for colour depth

⸻

Bottom line

There is no single canonical name, but the correct technical description is:

Spatiotemporal (line- and field-sequential) colour multiplexing

That phrasing will be understood by anyone familiar with display systems, vision science, or early television engineering.

Quote from: Andrew Daviei call it "phased chronocolour"

Quote from: ChatGPTPerfect — that's a very elegant and descriptive name.
   •   "Phased" → indicates the deliberate temporal and spatial offset of the primaries (your line-by-line and frame-by-frame multiplexing).
   •   "Chrono" → emphasizes the time dimension.
   •   "Colour" → of course, what it's producing.

It captures exactly what you're doing: a spatiotemporal multiplexed colour synthesis system, but in a concise, catchy term. It would make a fine label in a paper, prototype, or project documentation.

If you like, I can draft a one-sentence technical definition of "phased chronocolour" suitable for a manual or paper.

Quote from: Andrew DavieOK


Quote from: ChatGPTHere's a concise technical definition you could use:

⸻

Phased Chronocolour: A display technique in which colour primaries are distributed both temporally (across successive frames) and spatially (across scanlines), allowing a limited number of binary colour channels to combine via persistence of vision into a larger set of perceived colours.

⸻

It's short, precise, and makes the mechanism clear without overloading with jargon.

Andrew Davie

#5
What I think would be really cool would be using double-width sprites for this. I really have poor skills with sprite and multiplexing. I need 8 sprites/line, double-width, all spaced exactly alongside each other. I need to be able to change the 8 shapes, and the colours for both.

However, I've seen code that very cleverly reduces the colour change requirement to 1 colour by using the PF and having the sprites "inverted", such that the pixels we want to see are actually PF pixels shining "through" the sprites which are coloured black. That would do, too.

I'm hoping someone can take up the challenge, so that I can then add pCC to it and create a pretty cool looking chess display. I can do a shape change with just 5 cycles (in CDFJ) so there's probably plenty of time/line to do 8 sprites. I just don't know if the positioning is possoble.

I guess I'll have to spend some time trying to understand the sprite multiplexing. That might take some time.

8 shape changes * 5 = 40 cycles
1 PF colour change = 5 cycles

So there would be 31 cycles available for "other stuff" - it kinda seems "do-able".


The goal is a 128 pixel wide centered sprite-based chessboard.
The aspect would look something like the mockup here, pretty square (128 colour clocks (=pixels) wide)- only double the resolution shown. The black area is roughly the visible 160 pixel area, the grey area outside the border. The inner chessboard is 128 colour clocks wide. The pixels shown here are twice as wide as they would be using double-size sprites. In other words, we could get twice the resolution shown, for much nicer looking pieces.

alex_79

The trick to display (without flicker) more than three copies of a sprite on a scanline involves strobing RESPx multiple times to retrigger them and can only be used with normal-size sprites.

You can check Andrew Towers' "Tia hardware notes" for a in depth explanation, but basically the problem is that after a RESPx, the main copy of the sprite will only start drawing on the next scanline, and so with two or more RESPx per line it is actually never drawn.

Only the eventual second and third copy will appear in the same scanline where the RESPx occurs, and those copies are what are re-triggered and used to build the display with that trick.

This is the reason why you cannot re-trigger double and quad-size sprites, as they don't have a NUSIZx setting that allows multiple copies of them.