keropi wrote:

I hope that all this image searching was quite some time ago, I have documented the composite parts for the Small Wonder here: How to add CVBS-output to an ATI SMALL WONDER that omits it (instructions inside)

Yea, was in preparation of 8088 MPH, so somewhere around 2014-2015.
The ATi Small Wonder convinced me to get a real IBM CGA card, because apparently clones like the ATi were not compatible enough, with both timing and colours.

😀
I was always wondering if the colors can be adjusted with different resistor values but I guess it's way more complicated than that, right? is it mainly the graphic chip's fault?

keropi wrote:

I was always wondering if the colors can be adjusted with different resistor values but I guess it's way more complicated than that, right? is it mainly the graphic chip's fault?

The thing with the ATi Small Wonder is that the colours are correct in some modes (basically the standard RGBI modes, like 40col text mode and 4 colour 320x200), but not in others.
The colours in NTSC are decoded relative to a carrier. This carrier is generated by the colorburst signal. The NTSC circuit in the monitor/TV will re-sync to that colorburst every frame.
So the timing of that colorburst relative to the pixel data is very important. If the colorburst is too early or too late, your pixels will be out of sync. This is what happens on the ATi Small Wonder. In some modes the timing is just incorrect, which results in about 180 degrees phase difference, which swaps the colours around.
So if you would want to fix that, you'd have to make a fix that depends on the mode of the video card. Either that, or have a manual switch.
Since it is 180 degrees, perhaps it would work to just invert the colorburst signal.

I had no idea it was OK in some modes and not in others - certainly complicates things for an easy fix...
Composite colors was also the reason I got a real IBM CGA as well for my XT system. Even lucked out and got a ntsc 1084 monitor - perfect for the system!

Scali wrote:

Since it is 180 degrees, perhaps it would work to just invert the colorburst signal.

I guess that would work for "artifact-only" colors which are produced by rearranging white/grey pixels, as in the common mode 6 stuff, but then the 'solid' CGA colors would be inverted as well since they're independent of the pixel clock.

VileRancour wrote:

I guess that would work for "artifact-only" colors which are produced by rearranging white/grey pixels, as in the common mode 6 stuff, but then the 'solid' CGA colors would be inverted as well since they're independent of the pixel clock.

You may be right.
My idea was to make the 'inverter' engage only in mode 6, because that mode is broken. The other modes are fine on the ATi.
Could be done with a manual switch, because the ATi has an integrated circuit, so it would probably be very difficult to determine what mode it is in from the PCB.
Then all the 'regular' CGA composite modes would work.
But I have no idea what would happen with the enhanced colour modes from 8088 MPH. Then again I wouldn't expect those to work anyway, given how even the relatively subtle differences of old vs new CGA throw those colours off.

Some new info on bitplanes and raycasting.
KK has also made a raycaster on Amiga 500, and held a talk:
https://scenesat.com/videoarchive/138
Starts at about 1h35.

Scali wrote on 2019-08-26, 08:53:

You may be right. My idea was to make the 'inverter' engage only in mode 6, because that mode is broken. The other modes are fin […]
Show full quote

VileRancour wrote:

I guess that would work for "artifact-only" colors which are produced by rearranging white/grey pixels, as in the common mode 6 stuff, but then the 'solid' CGA colors would be inverted as well since they're independent of the pixel clock.

You may be right.
My idea was to make the 'inverter' engage only in mode 6, because that mode is broken. The other modes are fine on the ATi.
Could be done with a manual switch, because the ATi has an integrated circuit, so it would probably be very difficult to determine what mode it is in from the PCB.
Then all the 'regular' CGA composite modes would work.
But I have no idea what would happen with the enhanced colour modes from 8088 MPH. Then again I wouldn't expect those to work anyway, given how even the relatively subtle differences of old vs new CGA throw those colours off.

It might actually be much simpler than that.
Apparently, the "modulated" color carrier for the CVBS signal is just another digital signal from the main chip and – even better – it is routed through position three of the DIP switch, which turns composite color on and off.
So essentially, one would only have to switch that off and route the signal through a 1-bit inverter, instead. – And then never touch the switch, again!

The actual mod might be as simple as this:

(The dotted lines mean that you can use either pad. They are connected by traces on the back side.)