2600 and 5200 palettes compared

[Thomas Jentzsch]

I recently checked the 5200 palette as provided by the Atari800Win (default) and compared it with z26's palette. Except for the doubled number of colors I noticed quite some differences. E.g. yellow (2nd row) is much more greenish on the 2600. Also the 5200 palette is much darker overall. Is this just an emulator issue or does the 5200 really have a that much different palette?

 

Can anybody make a side by side compare on real hardware?

[vdub_bobby]

The AtariWin800 palette is off.

 

I used the exact color values from the 8-bit source for Squish 'Em in my 2600 conversion and they looked the same on real hardware (compared my 130XE to my 4-switch).

 

I assume there are no noticeable differences between a 130XE and a 5200 in color values.

Edited October 6, 2008 by vdub_bobby

[Thomas Jentzsch]

I tried to find out the correct Ballblazer colors. But some look odd, e.g. $f4 for the border color. Maybe the 5200 palette is really different.

 

I also tried the 5200 and 7800 version in MESS. Here I found that they used different values for both version. E.g. $B4/$BC for 5200 and $CC for 7800 green, $F4 for 5200 brown, $32 for 7800 brown.

Edited October 6, 2008 by Thomas Jentzsch

[SeaGtGruff]

I recently checked the 5200 palette as provided by the Atari800Win (default) and compared it with z26's palette. Except for the doubled number of colors I noticed quite some differences. E.g. yellow (2nd row) is much more greenish on the 2600. Also the 5200 palette is much darker overall. Is this just an emulator issue or does the 5200 really have a that much different palette?

 

Can anybody make a side by side compare on real hardware?

The 2600's real palette does have a greenish tinge to the yellow. All the specifications I've ever read say that hue 1 is the same as NTSC colorburst, and NTSC colorburst is definitely a greenish-yellow.

 

I don't have a 5200, but I assume its colors would be about the same as a 400/800/XL/XE? I've been trying to get some decent color palette screenshots from my 2600, 7800, and 130XE for years, but I haven't been too successful yet. The best I can do right now is record them to a VHS tape, and then play the VHS tape on my computer using a Snappy video digitizer thingy-- except I misplaced my Snappy a few years ago, and haven't found it again yet.

 

What's the best way to hook an *unmodified* 2600/7800/130XE (i.e., using a coaxial connector) up to a computer or recorder to get the best image for screen captures? If I can get it on DVD, I can play it on my computer and capture it. So I guess my best bet is to go buy a DVD recorder?

 

Michael

[SeaGtGruff]

Okay, I went out and bought a "cheap" DVD recorder, so I'll be trying to record palette screenshots from my 2600, 7800, and 130XE later tonight, then play the DVD back on my computer, capture the screenshots, and post them. Meanwhile, I'm posting some images I put together a while back showing

 

(1) a vectorscope graticule (grabbed from Wikipedia);

 

(2) a screenshot of the YIQ color space (grabbed from the freeware ColorSpace program), rotated 33 degrees (as shown by the "Phi" value in the Camera Properties window) to orient it the same way it appears on the vectorscope graticule;

 

(3) a screenshot of the YIQ color space superimposed on the vectorscope graticule, showing how the corners of the YIQ color cube fit in the little boxes on the vectorscope graticule, and with the I and Q axes from ColorSpace lined up exactly with the I and Q axes on the vectorscope graticule (the I and Q axes from ColorSpace are "backwards" or mirror images, as you can tell from the Q, because the YIQ color cube is being viewed from below or behind, or whatever-- i.e., minus 90 degrees, as shown by the "Theta" value in the Camera Properties window in picture number 2);

 

(4) the same picture as picture number 3, but without the I and Q axes from ColorSpace, so the YIQ color cube can be seen in all its glory; and

 

(5) a picture showing the color of the NTSC colorburst, which is located in the "9 o'clock" direction in the previous pictures. I selected a 1-pixel tall horizontal line from the center of the YIQ color cube to the edge of the color cube, then stretched it vertically to get picture number 5. I was working from a much larger version of the YIQ color cube than the one shown in the other pictures.

 

So now you know what color the NTSC colorburst is! The greenish tinge isn't as noticeable when the NTSC colorburst is viewed by itself this way, but the greenish tinge is much more obvious when you see it in the context of the rest of the YIQ color cube.

 

Michael

 

 

 

 

 

[SeaGtGruff]

Darn, I seem to have the worst luck with these things. The DVD recorder works great, except I had to connect my 7800 to the coaxial connection of my VCR, then connect my VCR to the composite connection of my DVD recorder, because the DVD recorder doesn't a coaxial input connection. I don't know if running it through the VCR is the reason, but the picture was crappy-- the colors kept flickering out completely, leaving the picture black and white-- and when the colors *were* there, they tended to show a lot of "noise." Last year I bought a WinTV-PVR for basically the same reason-- to record my 2600/7800/130XE-- and *that* turned out to be even worse than what I got tonight with the DVD recorder.

 

Anyway, I recorded a minute of video, then played it back on my computer in Windows Media Center. I was able to grab four still frames from it with the colors showing, although the colors vary a bit from frame to frame. Still, I think you can get a pretty good idea of the "real" colors by looking at all four images to compare the variations for any given hue. And the colors were much brighter on the TV than they are on the computer, presumably because TV has a higher gamma. I'm also including some pictures where I bumped up the brightness 50%, and bumped up the contrast 20%, which I think looks a lot closer to what the TV showed-- but still not perfect. By the way, I keep my monitor at 50% contrast, 75% brightness, so that may also affect how the colors look to me on my monitor.

 

It's interesting that hue 1 does look pretty consistent, and is pretty close to the color of the NTSC colorburst that I posted.

 

These are from the 7800, using the Cuttle Cart 2. Tomorrow night I'll try the 2600, using the Krokodile Cartridge. And then later I'll try the 130XE. I guess the 7800's color palette is essentially the same whether it's running in 2600 mode or 7800 mode, aside from 7800 mode having all 256 colors versus the 128 colors of the 2600?

 

Michael

 

 

 

 

 

 

 

 

[Thomas Jentzsch]

Thanks for your effort. It seems that not only Atari800Win's palette is off, but MESS's palette too. Most noticable for me are $1x and $Ex/$Fx, which are way more greenish in your shots. The z26 palette seems pretty close.

 

BTW: Your pictures only displays 128 colors. Is that due to the test program used?

[SeaGtGruff]

BTW: Your pictures only displays 128 colors. Is that due to the test program used?

Yes, it's a 2600 program that I wrote in batari Basic, posted a long while back in another thread. I ran it on my 7800 with my Cuttle Cart 2. If someone has a 7800 program to display all 256 colors, I'll load it on the Cuttle Cart 2 if you post it.

 

Michael

[SeaGtGruff]

$Fx, which are way more greenish in your shots.

Actually, some of the frames of the recording I made show $Fx as a definite brown or tan, others look more greenish. That's why I posted multiple frames-- the colors varied a bit from frame to frame. I'm thinking of taking several frames, copying the rows for a given hue, and pasting them together, so I can get a good idea of the range in variation for that hue, then try to take the one that seems closest to the average for that hue.

 

Michael

[Rybags]

This has been discussed in several or more threads before.

 

And, once you're done, you have the PAL headache to add even more confusion.

 

In theory, all the 8-bit computers and 5200 should have the same palette, but in practice Atari stuffed around with the video circuitry with every new release, so you have differences all over the place there as well.

 

Re the default emulator palettes. They're not at all accurate, and seem to have been done to cater for appearance rather than accuracy.

[ClausB]

Don't forget the color delay pot. IIRC, it adjusts the spacing between hues. In the color wheel above, you have 360° of hue and the Atari has 15 hue levels. To space them uniformly around the wheel, the hues should be 24° apart.

 

If the spacing is too big, you get repeated hues between the top end and low end. That appears to be the case with the emulators' hues shown above. There are repeated yellows, greens, and even oranges at the bottoms of the images.

 

Was there ever a standard for adjusting those pots?

[+Mitch]

BTW: Your pictures only displays 128 colors. Is that due to the test program used?

Yes, it's a 2600 program that I wrote in batari Basic, posted a long while back in another thread. I ran it on my 7800 with my Cuttle Cart 2. If someone has a 7800 program to display all 256 colors, I'll load it on the Cuttle Cart 2 if you post it.

 

Michael

 

7800 program: http://www.atari7800.org/bin/ntsc/Color.zip

 

Mitch

[Thomas Jentzsch]

Don't forget the color delay pot. IIRC, it adjusts the spacing between hues. In the color wheel above, you have 360° of hue and the Atari has 15 hue levels. To space them uniformly around the wheel, the hues should be 24° apart.

 

If the spacing is too big, you get repeated hues between the top end and low end. That appears to be the case with the emulators' hues shown above. There are repeated yellows, greens, and even ranges at the bottoms of the images.

I suppose the repeated colors at the buttom of the MESS palette screen are a strong indicator for a setting way above 24°.

[SeaGtGruff]

In the color wheel above, you have 360° of hue and the Atari has 15 hue levels. To space them uniformly around the wheel, the hues should be 24° apart.

 

If the spacing is too big, you get repeated hues between the top end and low end. That appears to be the case with the emulators' hues shown above. There are repeated yellows, greens, and even ranges at the bottoms of the images.

I suppose the repeated colors at the buttom of the MESS palette screen are a strong indicator for a setting way above 24°.

Not necessarily "way above," because a small additional difference between each hue adds up. For example, hue 15 would have the following phase value for each of the following steps between adjacent hues:

 

If 24 degrees, then hue 15 = phase +009 degrees.

If 25 degrees, then hue 15 = phase +023 degrees.

If 26 degrees, then hue 15 = phase +037 degrees.

If 27 degrees, then hue 15 = phase +051 degrees.

If 28 degrees, then hue 15 = phase +065 degrees.

 

This assumes that hue 1 = NTSC colorburst = phase +033 degrees. I've used the phase values given by the ColorSpace program. For each degree that's added to the step, the phase value of hue 15 increases by 14 degrees. By rotating the YIQ color cube to those phase angles and picking the color at the edge of the cube in the 9 o'clock direction, I come up with the following colors:

 

 

Although some of the frames I captured showed a greenish tinge to hue 15, others appeared to be on the orange side of yellow, rather than on the green side of yellow. It looks to me like hue 15 may be between +51 and +65 degrees, perhaps closer to +65 degrees, which would suggest a step value between 27 and 28 degrees, possibly closer to 28 degrees.

 

Michael

 

Edit: But it's kind of hard to tell from just one hue. The thing to do would be to build a color wheel for all 15 hues, each wheel based on a different step value. Of course, the actual step value is probably *not* an integer, but using integer step values would be a place to start.

 

--------------------------------------------------
|Hue|+024.000|+025.000|+026.000|+027.000|+028.000|
|---|--------|--------|--------|--------|--------|
|001|+033.000|+033.000|+033.000|+033.000|+033.000|
|002|+057.000|+058.000|+059.000|+060.000|+061.000|
|003|+081.000|+083.000|+085.000|+087.000|+089.000|
|004|+105.000|+108.000|+111.000|+114.000|+117.000|
|005|+129.000|+133.000|+137.000|+141.000|+145.000|
|006|+153.000|+158.000|+163.000|+168.000|+173.000|
|007|+177.000|-177.000|-171.000|-165.000|-159.000|
|008|-159.000|-152.000|-145.000|-138.000|-131.000|
|009|-135.000|-127.000|-119.000|-111.000|-103.000|
|010|-111.000|-102.000|-093.000|-084.000|-075.000|
|011|-087.000|-077.000|-067.000|-057.000|-047.000|
|012|-063.000|-052.000|-041.000|-030.000|-019.000|
|013|-039.000|-027.000|-015.000|-003.000|+009.000|
|014|-015.000|-002.000|+011.000|+024.000|+037.000|
|015|+009.000|+023.000|+037.000|+051.000|+065.000|
--------------------------------------------------

This table uses phase values between -180 and +180, because that's the range that the ColorSpace program uses.

Edited October 8, 2008 by SeaGtGruff

[Thomas Jentzsch]

 

Although some of the frames I captured showed a greenish tinge to hue 15, others appeared to be on the orange side of yellow, rather than on the green side of yellow. It looks to me like hue 15 may be between +51 and +65 degrees, perhaps closer to +65 degrees, which would suggest a step value between 27 and 28 degrees, possibly closer to 28 degrees.

That would result into repeating yellow/orange, no? 26° for me.

 

But shouldn't it be exactly 24° (15*24°=360°)? Else, what reference do we have?

 

Edit: But it's kind of hard to tell from just one hue. The thing to do would be to build a color wheel for all 15 hues, each wheel based on a different step value. Of course, the actual step value is probably *not* an integer, but using integer step values would be a place to start.

 

--------------------------------------------------
|Hue|+024.000|+025.000|+026.000|+027.000|+028.000|
|---|--------|--------|--------|--------|--------|
|001|+033.000|+033.000|+033.000|+033.000|+033.000|
|002|+057.000|+058.000|+059.000|+060.000|+061.000|
|003|+081.000|+083.000|+085.000|+087.000|+089.000|
|004|+105.000|+108.000|+111.000|+114.000|+117.000|
|005|+129.000|+133.000|+137.000|+141.000|+145.000|
|006|+153.000|+158.000|+163.000|+168.000|+173.000|
|007|+177.000|-177.000|-171.000|-165.000|-159.000|
|008|-159.000|-152.000|-145.000|-138.000|-131.000|
|009|-135.000|-127.000|-119.000|-111.000|-103.000|
|010|-111.000|-102.000|-093.000|-084.000|-075.000|
|011|-087.000|-077.000|-067.000|-057.000|-047.000|
|012|-063.000|-052.000|-041.000|-030.000|-019.000|
|013|-039.000|-027.000|-015.000|-003.000|+009.000|
|014|-015.000|-002.000|+011.000|+024.000|+037.000|
|015|+009.000|+023.000|+037.000|+051.000|+065.000|
--------------------------------------------------

This table uses phase values between -180 and +180, because that's the range that the ColorSpace program uses.

Can you convert that table into colors?

Edited October 8, 2008 by Thomas Jentzsch

[SeaGtGruff]

 

Although some of the frames I captured showed a greenish tinge to hue 15, others appeared to be on the orange side of yellow, rather than on the green side of yellow. It looks to me like hue 15 may be between +51 and +65 degrees, perhaps closer to +65 degrees, which would suggest a step value between 27 and 28 degrees, possibly closer to 28 degrees.

That would result into repeating yellow/orange, no? 26° for me.

 

But shouldn't it be exactly 24° (15*24°=360°)? Else, what reference do we have?

I found a page that claims it's 27 degrees, but it isn't an "official" reference-- it's where someone was trying to recreate the 400/800/XL/XE palette using RGB formulas. There was another page-- which I couldn't find last night-- that looks to be more "official," and if I remember correctly, it gave different step values-- yes, value*s*. As I recall, it gave two different step values-- which one was used depended on some voltage setting, if I remember correctly.

 

As far as whether it's an exact integer or not, my understanding is that it's a *time* delay (but I may be wrong about that, since I don't fully understand it), so we're dealing with computer chip timings, which don't need to conform to nice integers, seeing as they deal in tiny fractions of a second. And I don't know if a fixed timing delay step would correlate to a fixed color phase step.

 

Anyway, I tried using some color picking utilities to see what the hue values were for 1 and 15, and it looked to me like something between 27 and slightly more than 27.7 degrees looked correct. I'm going to create some actual colors tables for some different color phase steps.

 

Michael

 

PS: Also, when I used the color picker to check the hue values (HVL, or HVS) for a given Atari hue, they varied depending on the luminance value!

Edited October 8, 2008 by SeaGtGruff

[Rybags]

Actually, I think the colour trimmer only has an effect on the timing of the waveform, so just causes all of the colours to shift left/right within the colour wheel.

 

It's a known "issue" that Atari 8 bit computer colours become less saturated as the luma value increases - just look at a 256 colour display, the higher values become very washed out.

[Thomas Jentzsch]

It's a known "issue" that Atari 8 bit computer colours become less saturated as the luma value increases - just look at a 256 colour display, the higher values become very washed out.

That's true for the 2600 too (see first post).

[vdub_bobby]

Don't forget the color delay pot. IIRC, it adjusts the spacing between hues. In the color wheel above, you have 360° of hue and the Atari has 15 hue levels. To space them uniformly around the wheel, the hues should be 24° apart.

Hasn't supercat (or?) posted several times that, since getting 24 degree spacing exactly is hard and requires special equipment, that mostly they adjusted the pot so that 15==1? Or some such?

 

And I'll repeat again that based on comparing a 2600 to a 130XE to AtariWin800: 2600 color values == 130XE color values != AtariWin800 color values.

 

Where Stella, z26, MESS, Atari++, etc. fit in I have no idea.

 

 

EDIT: Here's exactly what I dealt with:

What Squish 'Em looks like in AtariWin800:

What Squish 'Em (my port, but exact same color values) looks like in Stella:

And in z26:

 

Real hardware, both 2600 and 130XE, look like the Stella/z26 screenshots. Probably closer to z26 IIRC.

Edited October 8, 2008 by vdub_bobby

[ClausB]

Actually, I think the colour trimmer only has an effect on the timing of the waveform, so just causes all of the colours to shift left/right within the colour wheel.

Are you sure?

[SeaGtGruff]

Real hardware, both 2600 and 130XE, look like the Stella/z26 screenshots. Probably closer to z26 IIRC.

I think the z26 palette is just about the closest of the emulator palettes, at least as far as the hues go-- but the colors don't get bright enough; there should be a greater range in contrast from the lowest luminance to the highest luminance.

 

Michael

[SeaGtGruff]

Actually, I think the colour trimmer only has an effect on the timing of the waveform, so just causes all of the colours to shift left/right within the colour wheel.

Are you sure?

I don't know about what the hardware does, but when I look at the colors in my screenshots using the ColorPic utility, the hues reported by ColorPic definitely change as I move the cursor left and right along a given Atari hue line. In contrast, if I use the Color Space program to rotate the YIQ color cube, capture a phase angle vector, and move the cursor back and forth, the hue reported by ColorPic is fairly constant.

 

Michael

[SeaGtGruff]

I haven't tried recording a palette display on my 2600 or 130XE yet-- or the 256-color display on my 7800-- but I spent the night working with Color Spaces, ColorPic, and Paint to see what a 27-degree step looks like.

 

First, here's some interesting information I'd found a while back. It deals with the GTIA chip (NTSC version), but should be fairly relevant to the TIA as well-- I think. Unfortunately, the page has disappeared from the web, but I'd saved it on my computer. The juicy stuff is in a table, and I've rearranged it into more of a list format. There were some comments added to the original text, which are mostly in square brackets, and I've added some comments of my own in curly brackets.

 

http://homepage.ntlworld.com/kryten_droid/Atari/800XL/atari_hw/gtia.htm
{Sadly, this web page is no longer available!}

GTIA
CO14805 (NTSC)
[added comments to original text shown in light purple]
{MSR's comments are shown in curly brackets.}
{All comments in square brackets are in light purple.}

9. Electrical Parameters
[interesting but not too relevant to logic]

c. Dynamic Operating Characteristics

Parameter: COLOR DELAY LINE OUTPUT
NOTE: Output is OPEN DRAIN and pull-up affects leading edge delay

Note: 1 {?}

Signal type: ATE OSC

Symbol: TDD {In "TDD," the "DD" is subscripted.}

OUTPUT DELAY TIME with VDEL (pin 17) = 7.0V and
WITH COLOR = C3..0
{In "VDEL," the "DEL" is subscripted.}
{"C3..0" refers to the four bits of the color nybble or hue value.}

C3..0: 0000 (no color out)
C3..0: 0001 {hue 1}	  Max: 167	 Unit: ns
C3..0: 0010 {hue 2}	  Max: 188	 Unit: ns
C3..0: 0011 {hue 3}	  Max: 209	 Unit: ns
C3..0: 0100 {hue 4}	  Max: 230	 Unit: ns
C3..0: 0101 {hue 5}	  Max: 251	 Unit: ns
C3..0: 0110 {hue 6}	  Max: 272	 Unit: ns
C3..0: 0111 {hue 7}	  Max: 293	 Unit: ns
C3..0: 1000 {hue 8}	  Max: 314	 Unit: ns
C3..0: 1001 {hue 9}	  Max: 335	 Unit: ns
C3..0: 1010 {hue 10}	 Max: 356	 Unit: ns
C3..0: 1011 {hue 11}	 Max: 377	 Unit: ns
C3..0: 1100 {hue 12}	 Max: 398	 Unit: ns
C3..0: 1101 {hue 13}	 Max: 419	 Unit: ns
C3..0: 1110 {hue 14}	 Max: 440	 Unit: ns
C3..0: 1111 {hue 15}	 Max: 461	 Unit: ns
{"ns" means "nanosecond," or 0.000000001 seconds (i.e., one-billionth of
a second).}

Parameter: COLOR DELAY LINE OUTPUT

Note: 1

Signal type: ATE OSC

Symbol: TDD

OUTPUT DELAY TIME with VDEL (pin 17) = 5.0V and
WITH COLOR = C3..0

C3..0: 0000 (no color out)
C3..0: 0001 {hue 1}	  Typ: 190	 Max: 190	 Unit: ns
C3..0: 0010 {hue 2}	  Typ: 209	 Max: 225	 Unit: ns
C3..0: 0011 {hue 3}	  Typ: 227	 Max: 260	 Unit: ns
C3..0: 0100 {hue 4}	  Typ: 246	 Max: 295	 Unit: ns
C3..0: 0101 {hue 5}	  Typ: 264	 Max: 330	 Unit: ns
C3..0: 0110 {hue 6}	  Typ: 283	 Max: 365	 Unit: ns
C3..0: 0111 {hue 7}	  Typ: 302	 Max: 400	 Unit: ns
C3..0: 1000 {hue 8}	  Typ: 320	 Max: 435	 Unit: ns
C3..0: 1001 {hue 9}	  Typ: 339	 Max: 470	 Unit: ns
C3..0: 1010 {hue 10}	 Typ: 358	 Max: 505	 Unit: ns
C3..0: 1011 {hue 11}	 Typ: 376	 Max: 540	 Unit: ns
C3..0: 1100 {hue 12}	 Typ: 395	 Max: 575	 Unit: ns
C3..0: 1101 {hue 13}	 Typ: 413	 Max: 610	 Unit: ns
C3..0: 1110 {hue 14}	 Typ: 432	 Max: 645	 Unit: ns
C3..0: 1111 {hue 15}	 Typ: 451	 Max: 680	 Unit: ns
{The "Typ" or "Typical" values are in light purple}
{to indicate they were added to the original text.}

Parameter: COLOR SELECT DELAY

Note: 1

Signal type: ATE OSC

Symbol: TDD

Max: 610

Unit: ns

[VDEL is a voltage that controls the colour delay line step size:
either (146 + n * 21) ns or (155 + n * 35) ns.
Since the master crystal oscillator runs with a period of about 279 ns,
this delay cannot be clocked digital logic.
There may be some analogue delay line circuit within the GTIA.
The apparent design intention is that there are 15 colours equally
spaced in 24° steps around the 360° of possible phase shifts.
One fifteenth of the colour carrier period is 18.62433862 ns for NTSC.]

{The NTSC color subcarrier frequency is 227.5 * 262.5 * 60 * 1000 / 1001
Hertz, or 3579545.45_45 Hertz (where "_45" indicates "repeating 45"), so
one NTSC color subcarrier cycle is 0.000000279365079365_079365 seconds,
or "about 279 ns"-- 279.365079_365079 ns, to be exact. Assuming that the
color delay step value between each hue is 21 ns (as indicated by the
first set of max values), we can convert that ns value to phase degrees
by dividing by 279.365079_365079 ns and then multiplying by 360 degrees,
which gives us 27.06136_36 degrees for the step value. The second set of
max values indicate a step value of 35 ns, or 45.10227_27 degrees, which
seems obviously incorrect. However, the typical values seem to indicate
a step value of about 18.6428571_428571 ns-- the given values increment
by either 19 ns or 18 ns, but (451 - 190) / 14 = 18.6428571_428571--
which gives us about 24.023863_63 degrees for the step value. To keep it
simple, we'll round these to 27 degrees when VDEL = 7.0V, or 24 degrees
when VDEL = 5.0V. Of course, this refers to the GTIA chip, and I don't
know how the GTIA's VDEL pin line compares to the TIA's DEL pin line.}

Here's another bit of information from another page:

 

http://www.xmission.com/~trevin/atari/video_notes.html

Notes on Video Output

...

When you select a color on the Atari, C/GTIA actually converts the hue
to a phase shift of its chrominance signal. The formula appears to be
190º - 27º × hue for hues from 1-15.

...

(This indicates that the true phase equation, at least for GTIA, is
closer to  208º -27º × hue.)

And this is what I was working on tonight:

 

Let's assume that the TIA's color phase shift step value is 27 degrees
from one hue to the next, that hue 1 is the same as colorburst, and that
colorburst has a phase angle of 33 degrees.

Note: That's actually the phase angle for the Q axis of the YIQ color
cube-- colorburst is 180 degrees-- but 33 is the phase angle we need to
plug into the Color Space program to put colorburst in the "9 o'clock"
position, so if we equate colorburst with 33 degrees, and add steps of
27 degrees to it, we get values that can be conveniently plugged into
the Color Space program, then we can simply take the color that ends up
in the "9 o'clock" position.

-----------------------------------------------------
|TIA|Color Space|	   ColorPic		|   Paint   |
|---|-----------|-----------------------|-----------|
| H |Phase Angle| H | S | V | R | G | B | H | S | L |
|---|-----------|-----------|-----------|-----------|
|H01| + 033.000 |070,255,255|213,255,000|047,240,120|
|H02| + 060.000 |049,255,255|255,208,000|033,240,120|
|H03| + 087.000 |027,255,255|255,116,000|018,240,120|
|H04| + 114.000 |353,255,255|255,000,030|235,240,120|
|H05| + 141.000 |316,255,255|255,000,189|210,240,120|
|H06| + 168.000 |282,255,255|180,000,255|188,240,120|
|H07| - 165.000 |262,255,255|093,000,255|175,240,120|
|H08| - 138.000 |243,255,255|014,000,255|162,240,120|
|H09| - 111.000 |222,255,255|000,076,255|148,240,120|
|H10| - 084.000 |198,255,255|000,178,255|132,240,120|
|H11| - 057.000 |160,255,255|000,255,169|107,240,120|
|H12| - 030.000 |123,255,255|000,255,014|082,240,120|
|H13| - 003.000 |095,255,255|108,255,000|063,240,120|
|H14| + 024.000 |076,255,255|187,255,000|051,240,120|
|H15| + 051.000 |056,255,255|255,238,000|037,240,120|
-----------------------------------------------------

I grabbed screenshots of these phase angles from Color Space, then used
Paint to cobble together a palette, and compared it to one of my 7800
screenshots, and the hues do look pretty close, although they aren't a
perfect match. I doubt that the step is 27 degrees exactly. Also, the
NTSC signal is infamous for having variations in the colors ("Never The
Same Color," as they say), and I can definitely see a lot of variations
in the colors when I advance my DVD recording frame by frame.

Here's the palette I cobbled together using Color Space, Color Pic, and Paint, displayed side-by-side with one of my 7800 screenshots. I flipped the cobbled-together palette around so the brightest values are alongside the brightest values of the 7800 screenshot:

 

 

Michael

[Kr0tki]

In this post I've sent a modification of the Atari800 2.0.3 sources, which generates the NTSC and PAL GTIA palettes based on the official GTIA reference document. The colours where computed using the YUV colourspace and the real GTIA colour timings, so I believe it is not possible to get better palettes You might find it useful.

[Thomas Jentzsch]

Here's the palette I cobbled together using Color Space, Color Pic, and Paint, displayed side-by-side with one of my 7800 screenshots. I flipped the cobbled-together palette around so the brightest values are alongside the brightest values of the 7800 screenshot:

Interesting. In your picture hue 15 is somewhere between hue 1 and 2. And hue 15 is even closer to yellow than hue 1. Other palettes show hue 15 somewhere between hue 14 and hue 1. Most games I know about use hue 1 for yellow, maybe because it is more predictable?

 

Hm...

 

Maybe supercat's theory is right. Atari set up the consoles so that hue 1 = hue 15. Maybe with the instruction of an intentional little deviation, so that they could better claim to have 128/256 colors. And depending on into which direction this deviation was introduced, we either get are more green or red tint for hue 15. Which would explain the pretty different palettes existing.

 

Does that make sense?