C64->A8 Pallete colours conversion?

[José Pereira]

Anyone has a table with the correspondant (colour,luminance) ofthe C64 colours on the A8 Palette?

Thanks.

José Pereira.

[STE'86]

these are my interpretation of the c64 colours converted to a 256 colour atari palette in indexed colour mode.

 

they can maybe be tweaked a bit, but apart from the green being a bit dark by comparison to the c64 this conversion of my wizball screen which uses all the colours seemed to work out fairly well

[José Pereira]

Thanks.

 

But I was asking for the C64 0->F to the correspondant A8 00->FF even numbers of the 128 colours Pallete.

Edited December 14, 2012 by José Pereira

[emkay]

The chosen palette values ARE almost correct, as they fit into the possible colour spread/range.

 

In 256 colours you get a slight better contrast value.

[Rybags]

There are near matches for practically all, only Black is the same.

 

C64 has slightly higher saturation. C64 white luma value should be brighter than luma $E on Atari.

Most C64s have 5 bits luma control where 32 levels would be possible although only 9 are used.

Phase angle of available colours won't necessarily match on both machines. C64 colours have pairs at exact 180 degree interval, Atari don't.

[Synthpopalooza]

Here's an interesting question:

 

Using PAL blending or XL paint mode, is it possible to duplicate those 16 commodore colors on the Atari? That is, finding the correct 8 color settings through color blending, to get the Commodore palette.

[emkay]

Here's an interesting question:

 

Using PAL blending or XL paint mode, is it possible to duplicate those 16 commodore colors on the Atari? That is, finding the correct 8 color settings through color blending, to get the Commodore palette.

 

No need for that. As there is no "real C64" palette. While on the A8 we exactly have clear 3 different palette types, C64 has 5 different types? The resulting picture depends on the displaying device which was back then very different....

What's to do is more to be careful when adopting the colours, to keep the colour range inside the colour room.

I tried to explain that in the Rastaconverter threads already....

[Creature XL]

Here's an interesting question:

 

Using PAL blending or XL paint mode, is it possible to duplicate those 16 commodore colors on the Atari? That is, finding the correct 8 color settings through color blending, to get the Commodore palette.

 

Was the first thing I did when the mode was introduced.

However, at that point of time I used it with 4 grays and 4 colors. So I couldn't get far as the VIC-2 has 7 UV-phase angles and with 4 colors you only get 4

So, only chance is using color in the "luma lines".

[Rybags]

Pal blending results in lower saturation. To duplicate C64 colours more exactly would require higher saturation in many cases so it probably wouldn't work.

[Creature XL]

these are my interpretation of the c64 colours converted to a 256 colour atari palette in indexed colour mode.

 

they can maybe be tweaked a bit, but apart from the green being a bit dark by comparison to the c64 this conversion of my wizball screen which uses all the colours seemed to work out fairly well

 

piped this pic thru RasterConverter. At a distance of 4.3 it already looks great. I hope to get closer to 0.0

Its running for 6h (104M evaluations)

[Lazarus]

Most C64s have 5 bits luma control where 32 levels would be possible although only 9 are used.

There is no 5 bit DAC for luma control but some resistor array which happens to produce 7 levels of luma + black + white.

 

While on the A8 we exactly have clear 3 different palette types, C64 has 5 different types?

C64 has two types: old palette with 5 lumas (only revision 1 VICs) and 9 lumas.

[José Pereira]

 

 

piped this pic thru RasterConverter. At a distance of 4.3 it already looks great. I hope to get closer to 0.0

Its running for 6h (104M evaluations)

So when could we expect to see the picture, by night?

If so you can then start converting Wizball to A8. It's perfect for you as it can be the same as you have on Har'em just stars as 2Missiles like I have in Delta.

Here you have all to start:

http://www.atariage.com/forums/topic/197750-if-anyone-is-interested-or-at-least-talk-about-this-one

[emkay]

 

C64 has two types: old palette with 5 lumas (only revision 1 VICs) and 9 lumas.

 

Then it's a nifty task by the gfx designers to subtle "enhance" the look of a picture in emulations, using up to 11 different lumas? As long as you watch it on a B/W-TV , things were better, having different colours at different lumas, for sure.

[Rybags]

11? Where do you get that from?

 

9 lumas includes black and white. The earlier linked article from the Pal blending thread has the info http://www.pepto.de/projects/colorvic/

[emkay]

11? Where do you get that from?

 

9 lumas includes black and white. The earlier linked article from the Pal blending thread has the info http://www.pepto.de/projects/colorvic/

 

Actually, I converted some of them already. The images thermselves were based on up to 12 different lumas. Possibly it's some "artefacts" from drawing the picture in a PC tool and converting it to the C64 format. It just nothing special, as TV-Sets always vary...

 

[emkay]

It's just a different kind, whether you have different colours in different lumas, or you can have 16 colours in 16 lumas (or even 8 lumas)

 

 

Original PAL Atari with a C1084 Monitor. The magmatic shine through the grey stones. Particular Red need the different lumas on the A8 to shine better.

[emkay]

This is what Rastaconverter puts out after 100M eval and some "colour room" adjustments before...

It's a bit darker , but does that usually mean worse? I don't think so.

 

So the colour value is more important to a fixed image than the brightness itself. That's why it works on any displaying device. (Just looking back to the title of the thread). After the colours were correct, the brightnesses do their additional benefits.

Edited December 16, 2012 by emkay

[Lazarus]

Then it's a nifty task by the gfx designers to subtle "enhance" the look of a picture in emulations, using up to 11 different lumas? As long as you watch it on a B/W-TV , things were better, having different colours at different lumas, for sure.

What the hell are you talking about?

[emkay]

What the hell are you talking about?

 

Seems more complex, doens't it?

It's about the brightness value of a used colour. Sometimes you get a conversion of 16 colours resulting in 15 greys. Adjusting them to the relevant parts , normally you can do that to the "known" amount of colours. But sometimes you see colours disapear even at a value above the 5 or 9 colours.

You know, on the A8 you get real 8 or 16 different luma values, and 16 different chroma values. But the mix of the colours results different on different screens. "Modern" displays tend to show blue brighter... but that doesn't mean, the A8 has more than 8 or 16 real lumas.

As it is in the eye of the observer, the colours vary in the emulations and people tend to use the colours that fit more to their project.

 

So it's nothing really special. Except the problem that people do colour-calculations where they don't really fit. Referring to the title of the thread.

Edited December 16, 2012 by emkay

[Lazarus]

C64 has 9 lumas, no matter how gfxians use the colors and no matter what display device you use.

[Rybags]

Variable perception which depends on a TV or monitor's individual quirks isn't a reliable technique.

 

That aside, look at the colourspace conversion formulas. There's always a bias towards green being the dominant component with red then blue with lesser contributions. e.g. G=0.587 R=0.229 B=0.114

 

You can try for yourself in seconds - just open Paint and do a fill with the pure primary colours.

 

On a TV - I would suspect that some colours appear brighter than others due to how they're generated. A pure primary colour will only activate one RGB triad element, other colours activate 2 or more in variable levels.

Edited December 16, 2012 by Rybags

[Lazarus]

That aside, look at the colourspace conversion formulas. There's always a bias towards green being the dominant component with red then blue with lesser contributions. e.g. G=0.587 R=0.229 B=0.114

 

No, the weighting is just there so that green is most accurate since the human eye sees green much brighter and more accurate than blue and red. They are supposed to give more accuracy to the more visible color components. Those weightings are the result of tests done with people where they should judge the brightness of red green and blue. It's an average of lot's of people. In reality, every human has a different weighting of RGB colors, I for example see blue lighter as red.

 

On a TV - I would suspect that some colours appear brighter than others due to how they're generated. A pure primary colour will only activate one RGB triad element, other colours activate 2 or more in variable levels.

 

No. As I said: The human eye has inbuilt weighting already. The YUV-weightings only try to give bandwidth according to the way the humans eye accuracy on R, G and B.

 

A PAL CRT TV has a pretty well defined gamma value, VGA CRTs have a different gamma. On TFT/LCD screens, the gamma value even depends on the screen you use and the angle you look at it. It's impossible to have a "correct" palette. Also there is a big problem: YUV/YIQ color spaces cover colors which do not exist in VGA RGB. Especially light saturated colors do not convert to VGA RGB since one or more RGB color channel will quickly reach it's maximum of 0xFF.

Edited December 19, 2012 by Lazarus

[Xuel]

Here's what I get with Vice 2.4.6 default palette to RastaConverter with default color distance (YUV) to Altirra palette:

 

C64 A8  RGB
--- --  ---
0   00  000000
1   0E  EEEEEE
2   26  B64736
3   9C  7CEBFC
4   58  C555F0
5   BA  6DDD42
6   74  213BCD
7   EE  FFF765
8   18  C77C23
9   E6  896F00
A   2A  FA8B7A
B   06  666666
C   0A  AAAAAA
D   BE  B1FF86
E   7A  87A1FF
F   0E  EEEEEE

Note that C64 1 and F both resolve to E on A8.

[Rybags]

You'd want to decrease the overall gamma then... there's no way that light grey from C64 should correspond to luma E - realistically it should be somewhere around 9-A.

[Creature XL]

Feels not accurate to use a C64 emulator's RGB values (which aren't exact to start with) and then use a A8 RGB palette to convert them to GTIA values.

This is an old thread, so I have no clue if I (or anyone) mentioned it. So I will just say it (again?).

The VIC-2 constructs colors in the YUV space (or YpbPr to be precise). GTIA does the same. So to find the closest colors the correct way seems to find the phase angle ( determined by u and v) which are closest to the once of the VIC-2.

Finding the luma values shouldn't be a problem, except when the VIC-2's values have totally different spacing.

Saturation is a said above, a problem as the C64 is a bit more saturated AFAICS.

 

I truly have the feeling I posted that somewhere before.

 

EDIT: See post #14 for a link which helps understand my babbling

Edited February 10, 2015 by Creature XL