Colors NTSC vs PAL

[Faicuai]

FYI:

 

1. These are the correct NTSC color versions of DigDug (yes, there are SEVERAL ones). Comparing it to an Arcade (Mame) version, also with messed-up color reproduction (and saturation out of control) is not the best reference. Notice the nice, blue bar at the top, the EARTHY / balanced colors of the older version, the cleaner / more solid "blue" bar at the top, and the changes in color reproduction during 1983 edition (all meant for GTIA) :

 

 

2. Summer Games' Flag screen, when correctly or accurately displayed in NTSC, shows with OFF colors.

 

3. Here is Star Raiders with CORRECT shield colors, faithful to Doug's original design:

 

 

4. Here's APX's Color Alignment tool, with Rainbow and Color Primaries:

 

 

FYI, I have not modified, tinkered-with or custom-adjusted (for this games) any of the resulting settings from the above procedure.

Edited October 2, 2014 by Faicuai

[Faicuai]

And even more, here, from some classical ones (including "Pooyie"), and all the way to uber-successful contemporary ones:

 

(please, note that these samples are not aspect-ratio compensated, not Gamma-calibrated, and Bright/contrast point is set to optimize tonal separation on the high-area of the Atari tonal scale, in order to maximize detail, thus looking a bit "dark", to some viewers):

 

 

 

Anyone here (especially game designers / authors) is free to jump in and point out any Color-reproduction deficiency. I would more than happy to make any corrections, as once we are inspecting this, we will move on into a final, simple procedure for calibrating the actual HW (with all its defects, color / timing drifts, uncertainty on Hue-Step, all of that crap considered).

Edited October 3, 2014 by Faicuai

[firestorm]

Kingdom come

 

 

Heretical but correct with Atari Basic/Logo Manuals and Service Manuals and screenshots from Atari catalogues link below.

http://www.atarimania.com/game-atari-400-800-xl-xe-kingdom_2798.html

 

 

 

 

 

Conservative, don't touch your TV hue control attitude

[firestorm]

Based on Atari 400/800 Field Service Manual, there is simple way to adjust color pot on NTSC Atari computer without using SALT cartridge. Before you try this please keep your Atari switched on to warm up for 15 minutes.

 

SETCOLOR 2,1,8 REM CHANGES BACKGROUND COLOR TO 1, PRESS RETURN

SETCOLOR 4,15,8 REM CHANGES FRAME COLOR TO 15, PRESS RETURN

 

Then adjust Atari color pot to make background and frame around the screen of the same color as possible.

[firestorm]

Found another proof of Atari NTSC correct colors, this time it's Atari GTIA internal confident document. Link below.

 

http://www.retromicro.com/files/atari/8bit/gtia.pdf

 

And Atari CGIA, page 50-51.

 

http://www.retromicro.com/files/atari/8bit/cgia.pdf

Edited October 4, 2014 by firestorm

[firestorm]

CGIA document is real breakthrough! Holy Grail of my quest for Atari NTSC colors!

[firestorm]

1 Atari CGIA document states that reference color 1 is 76.5° from RGB Red it will be easy to define it in Altirra using RGB color wheel.

 

2 Next color is different from reference in phase by 24° (Altirra hue step 24°)

 

3 Colors are described in similar maner as in Basic Manual :1 Gold, 2 Orange, 3 Red Orange and so on.. to get these colors on screen you have 2 adjust phase -33 on your monitor.

It makes sense because RGB Red witch is 76,5° minus 33° give us 43° of Golden Rod .

 

4 To adjust color pot on your Atari Color 15 has to look like this with Phase/Hue/Color Balance set to 0

 

or like that with Phase -33

 

5 Reference Color looks like this with setting Phase/Hue/Color Balance 0

 

or with Phase -33 (Gold) like that

 

6 On Atari NTSC you can't change reference color by adjusting color pot.

 

This is RGB Wheel used by me so you can practice.

http://www.rapidtables.com/web/color/color-wheel.htm

Edited October 4, 2014 by firestorm

[Faicuai]

Ok, folks, went back to my "bench", and decided to:

 

1. Fire up my two (2) 800XL (their serial #s are 7,000 units apart, same socketed Mobo, with nice RED-packed resistors on controller ports, the best keyboard ever made for it, thick-plate shields, exact video mods (C54 out, Chroma-line enabled), and both with Ultimate with upgraded CPLDs. In other words, seemingly IDENTICAL.

 

2. The 800XLs were connected via Y/C video (sVideo-cable) to s-Video input of Sony Wega CRT (and also on RF port).

 

3. I also fired up my 800/Incognito, with primary display being Viewsonic VP930b pro-monitor (sRGB), via DVDO iSCAN-HD video processor, and simultaneously connected RF output INTO Sony WEGA RF port.

 

4. ALL machines' pot. calibrated exactly per CPS SuperSalt RevA FD100335 (my own cart). It is a must, and for multiple reasons.

 

5. All machines' DISPLAY-output was independently calibrated with APX Color Alignment tool, together with BLUE and RED filter from Avia-II kit, in order to dependably fine-tune four key variables: Black-level, White-level, HUE and SATURATION, independently of screen and rendered color-space).

 

 

Now, here are the results:

 

1. COLOR DECODE of first color (what Atari calls "GOLD") or SuperSalt's color-bar #1 (top-to-bottom) is different on [iScanHD + LCD], than actual CRT:

• On iScan/LCD, the color comes out Rust with greenish tint.
• On CRT it shows as plain RUSTY / reddish (even when SIMULTANEOUSLY inspecting the 800's LCD and CRT/RF outputs).
• All of the above inspected at Luma values (low) pre-set on APX Color Alignment.

 

2. My Sony WEGA CRT decodes my 800XLs sVideo and RF output THE SAME (it shows first color as red-rusty).

 

3. There was a SUBSTANTIAL difference in native color-saturation levels between my 800XLs and not so large (but still different) than my 800/Incognito, particularly visible on the REDs, when seen through red-filter on CRT (and APX NTSC color pattern).

 

4. The iScan+LCD's was capable of delivering Greens and Blues pretty much out of the range of anything my CRT seemed to be capable (under equivalent Hue / Saturation adjustment).

 

5. The CRT was capable of delivering REDs that seemed beyond the reach of the iScanHD+LCD combo.

 

6. After POTs + Display-rendering calibration, I found (inspecting on APX Color Alignment tool, which shows all screen-colors in LOW luma values):

• The iScanHD+LCD combo missed Color 1 (Rust), Color 14 (Green-Orange) and Color 15 (Orange). Star Raiders shields displayed OK, per D. Neubauer specs.
• When applying the SAME color-calibration procedure mentioned above, Altirra and iScan/HD output are virtually identical (very modest differences).
• The CRT screen ended-up messing Color 6 (COBALT BLUE), Color 10 (Blue-Gray), and Color 14 (Green-Orange). Color 15 kind of passed. Because Color 10 was missed here, Star Raiders got messed up with alien-looking GREEN-color shields.
• No amount of POT messing (nor smoking :-) could eventually resolve or align these differences (I deliberately off-set the POT adjustment, off-SuperSalt-specs, and then compensated on Phase/Hue, trying as carefully as possible to reproduce "ideal" color targets).

 

So there you have the BOTTOM-LINE:

 

1. First-color decode is DIFFERENT between CRT and vide-processor/LCD display systems (even when generated from the SAME MACHINE!)
2. The above means that a few titles (that use Color 1, in any form or shape) will show as more rusty-greenish / opaque in LCD/RGB and will show as rusty-reddish in CRT.
3. Possible Color Gamut / spectrum differences when displaying / converting Atari's signal into RGB (better greens and blues) and CRT (better REDs).
4. Color response and differences between these two domains CANNOT BE ALIGNED (at this point).
5. Calibration results on ALTIRRA and real [HW+iScanHD+LCD] come very close from each other.

Edited October 5, 2014 by Faicuai

[firestorm]

Faicuai have you warm up Atari for 15 minutes?

 

Now I can have Star Riders with bluish shields on Phase -33

[Faicuai]

Faicuai have you warm up Atari for 15 minutes?

 

Now I can have Star Riders with bluish shields on Phase -33

Well, that would not be possible with any of my machine's *native* output, by-the-specs SuperSALT calibration, and with a simple Phase / Huse adjustment. Not even with three (3) hours of continuous testing.

 

My suggestion is that you download APX Color Alignment tool, and post here the 1-bar tests (color-per-screen, with color-names as identified by the tool) and the NTSC-pattern and Rainbow output (5-bar tests).

 

NOTE: at this point, something tells me we would need to re-evaluate SuperSalt bottom-bars color-match procedure. I have a problem seeing Color 1 and Color 15 coming out virtually the same. I do not believe that should be the case.

Edited October 5, 2014 by Faicuai

[firestorm]

Help me find APX Color Aligment tool please. Can you attach it or point where I can find it?

[Faicuai]

Help me find APX Color Aligment tool please. Can you attach it or point where I can find it?

 

Here you go:

 

AnalogComputing-38_1986-SideA.atr

 

Here's what you need to do:

 

1. Mount .ATR on SIO-to-SD or NUXX or Altirra. It will boot on its own DOS (you need BASIC!)

2. Go to Basic, and load "colorbars.bas"

3. For Full-Screen (color-by-color) dump: Select one "(1) Bar" option, and then Color Bars.

4. For NTSC Pattern: Select "(5) Bars" option, then Color Bars, and then NTSC pattern.

5. For RAINBOW Pattern: Select "(5) Bars option, then Color Bars, and then RAINBOW.

 

I think it would be an excellent starting-point, in order to evaluate (more precisely) your adjustments, and the results your 800XL + Sony Monitor are delivering.

[firestorm]

Thank you very much Faicuai

 

Are you getting color bar test and purity test in the same time right?

[thorfdbg]

First-color decode is DIFFERENT between CRT and vide-processor/LCD display systems (even when generated from the SAME MACHINE!)

This is not a miracle. The color space - not even the color conversion matrix - of NTSC (the CRT) and HDTV (the LCD) are different. That is, the color coordinates of the red, green and blue primaries of both monitors are different.

 

The NTSC color space is 601, the HDTV color space is a bit larger and based on 709 colors. The color spaces are different, hence the color appearence is different.

[Faicuai]

This is not a miracle. The color space - not even the color conversion matrix - of NTSC (the CRT) and HDTV (the LCD) are different. That is, the color coordinates of the red, green and blue primaries of both monitors are different.

The NTSC color space is 601, the HDTV color space is a bit larger and based on 709 colors. The color spaces are different, hence the color appearence is different.

Well, that is the obvious part, though. This may also explain the better green ans blues I am getting in the digital domain.

 

The REAL problems here are:

 

1. Getting Atari reference/target NTSC palette RIGHT, even on a modern CRT (which is not really happening). This is probably a sign that color DECODING changed or evolved from the days Atari's visual reference was defined or inspected.

2. Color-calibrating the system so the presence of (probably) an unrecoverable color-decode error can be minimized.

3. As a bonus, extract color reproduction as similar as possible between Color systems (CRT + LCD/video-processor).

 

We ain't stopping at the conversion, color space and gamut crap, though. And that's the goal here, hopefully, with a simple (integral) criteria so everyone can calibrate their systems, consistently.

 

I will be posting something along these lines, soon. I am working on #2, as we speak.

Edited October 5, 2014 by Faicuai

[firestorm]

Can anybody explain this? In attachment you will find page from first Analogue issue.

From 1982, times when on NTSC Atari color number 1 still was GOLD.

 

 

 

After 30 some years now we have green in this place

 

 

 

Is there anybody who still have TV made in beginning 1980's? Maybe this is a key to the mystery of Atari colors?

 

I will promise you I won't stop till I found the truth!

 

What I need is Your help.

analog_no_01 30.pdf

Edited October 6, 2014 by firestorm

[Faicuai]

Can anybody explain this? In attachment you will find page from first Analogue issue.

From 1982, times when on NTSC Atari color number 1 still was GOLD.

 

 

 

After 30 some years now we have green in this place

 

 

 

Is there anybody who still have TV made in beginning 1980's? Maybe this is a key to the mystery of Atari colors?

 

I will promise you I won't stop till I found the truth!

 

What I need is Your help.

 

...Well, it is hard to say... Changes on Atari color-coding? Changes on NTSC color-decode? Not easy.

 

However, I would tell you that:

 

1. We will have to deal with the green. I am not sure the extent to which its coding or decoding can be altered (nor if it really matters).
2. My units (800, and 2x800XLs) seem outputting the Altirra's equivalent of HueStart = -52, and HueStep=24 degrees. I have direct grabs (analog-to-digital) frames conversion, of entire 256-palette, and I have worked with them carefully.
3. Since for most of folks out-there, "Hue Start" is a given / fixed on their machines, I will soon be posting a simple procedure to calibrate these machines Hue, and *Saturation*, while fully reproducing 256 colors (not 240), and doing so consistently (for the most part) either on digital / LCD or analog / CRT domains (NTSC). Adjustments to Hue / Phase will not be necessary, thus being able to easily plug several machines to the same Monitor / Display, with ONLY adjustments in Saturation (which wildly varies from machine to machine).

 

In the meantime, keep hunting!

Edited October 7, 2014 by Faicuai

[dmsc]

Hi,

 

Reading your discussion, I remembered writing a color-bars demo program back in the eighties.

 

I specifically choose the colors back then so my (probably uncalibrated) 800 XL showed the best approximation to the TV station color bars :-)

 

If you run the attached ATR, you can see that I choose color 15 as green and 1 as yellow. This can be approximated in Altirra with hue-start at -52 and hus-step at 24.5.

 

 

Daniel.

colorbars.zip

[Joey Z]

 

...Well, it is hard to say... Changes on Atari color-coding? Changes on NTSC color-decode? Not easy.

 

However, I would tell you that:

 

1. We will have to deal with the green. I am not sure the extent to which its coding or decoding can be altered (nor if it really matters).
2. My units (800, and 2x800XLs) seem outputting the Altirra's equivalent of HueStart = -52, and HueStep=24 degrees. I have direct grabs (analog-to-digital) frames conversion, of entire 256-palette, and I have worked with them carefully.
3. Since for most of folks out-there, "Hue Start" is a given / fixed on their machines, I will soon be posting a simple procedure to calibrate these machines Hue, and *Saturation*, while fully reproducing 256 colors (not 240), and doing so consistently (for the most part) either on digital / LCD or analog / CRT domains (NTSC). Adjustments to Hue / Phase will not be necessary, thus being able to easily plug several machines to the same Monitor / Display, with ONLY adjustments in Saturation (which wildly varies from machine to machine).

 

In the meantime, keep hunting!

I think the issue here lies in that most analog TV's apparently did not use the proper YIQ decoding that was the NTSC specification at the time. I can see why, two extra filters, then two delay lines to correct for the delay of the analog filters... lot's of complex numbers to calculate all of that, and then the effort required to make the theoretical model work in real life...

 

Anyway, with the advent of digital TV's and DSP designs, all of that is done in software now, so it's relatively easy to do the ACTUAL YIQ demodulation. Thus, atari looks different on modern TV's and studio monitors (which both are doing real YIQ demodulation probably) vs. any other typical TV from the 80's, which was likely skipping the YIQ and using a simpler method of demodulation.

[Faicuai]

I think the issue here lies in that most analog TV's apparently did not use the proper YIQ decoding that was the NTSC specification at the time.

 

(...)

 

Anyway, with the advent of digital TV's and DSP designs, all of that is done in software now, so it's relatively easy to do the ACTUAL YIQ demodulation. Thus, atari looks different on modern TV's and studio monitors (which both are doing real YIQ demodulation probably) vs. any other typical TV from the 80's (...)

 

...Great post! (almost missed it).

 

It is precisely the page where I am sitting now: possible (and real) evidence of the above lies in comparing Altirra's output with DVDO iScan / DVI processor output (fed with Atari's analog signal). They virtually match (and this seen comparing them in the SAME VP930b Monitor).

 

The above so-close match is not possible (from the same machine) on my CRT set:

1. The chroma "delta" of Color 1 (SuperSalt first color bar, from top-to-bottom) is UNRECOVERABLE between both color systems.
2. Overall color reproduction differences can be minimized substantially, but results will be abruptly different in the deepest areas of the blues and greens (much better in digital) and the REDs (definitely stronger, richer, on CRT).

I am not sure if there is any point on pursuing a "perfect" match with Atari's original Color target / palette.

[Faicuai]

Hi,

 

Reading your discussion, I remembered writing a color-bars demo program back in the eighties.

 

I specifically choose the colors back then so my (probably uncalibrated) 800 XL showed the best approximation to the TV station color bars :-)

 

If you run the attached ATR, you can see that I choose color 15 as green and 1 as yellow. This can be approximated in Altirra with hue-start at -52 and hus-step at 24.5.

 

ataribars.png

 

Daniel.

 

Thanks for the image!

 

It is a great utility, but Green-color is not 15. The primaries are located on 4,8 and 12 bars in SuperSalt pattern. That's where you would get decent RGB reproduction, from.

 

It would be great to be able to re-code this utility (with some expertise on Atari Graphics) so we could come up close enough to the SMTPE Pluge pattern (which will allow you, unmistakably, to nail down Black level, White level, HUE and SATURATION, perfectly, with use of Blue-filter):

 

http://www.youtube.com/watch?v=nOIRhJkrLBo

 

The narrow / sandwiched & shifted under-bars are ABSOLUTELY key.

Edited October 9, 2014 by Faicuai

[+MrFish]

I think the issue here lies in that most analog TV's apparently did not use the proper YIQ decoding that was the NTSC specification at the time. I can see why, two extra filters, then two delay lines to correct for the delay of the analog filters... lot's of complex numbers to calculate all of that, and then the effort required to make the theoretical model work in real life...

 

Anyway, with the advent of digital TV's and DSP designs, all of that is done in software now, so it's relatively easy to do the ACTUAL YIQ demodulation. Thus, atari looks different on modern TV's and studio monitors (which both are doing real YIQ demodulation probably) vs. any other typical TV from the 80's, which was likely skipping the YIQ and using a simpler method of demodulation.

 

Although... modern LCD's are not perfect either. I currently have 4 LCD's wired to a single machine (one which is KVM'ed to 2 other machines), and none of the 4 monitors -- two even being the exact same make and model -- have the same color or brightness/contrast characteristics. Granted they're all from slightly different periods (except the twins)...

 

 

The above so-close match is not possible (from the same machine) on my CRT set:

 

What kind of CRT set do you have -- out of curiosity?

Edited October 9, 2014 by MrFish

[firestorm]

Hi,

 

Reading your discussion, I remembered writing a color-bars demo program back in the eighties.

 

I specifically choose the colors back then so my (probably uncalibrated) 800 XL showed the best approximation to the TV station color bars :-)

 

If you run the attached ATR, you can see that I choose color 15 as green and 1 as yellow. This can be approximated in Altirra with hue-start at -52 and hus-step at 24.5.

 

ataribars.png

 

Daniel.

 

Thank you DMSC, do you remember in witch year you wrote this program?

[firestorm]

In 1987 in USA standard NTSC was changed from NTSC(1953) to SMPTE "C". They changed colorimetry values for prime colors (RGB) and white point.

Is this the answer I'm looking for?

 

Here you have what I found on internet.

 

 

This is vectorograph of SMPTE "C" NTSC standard. Atari color 1 is at 180° of phase.

 

I found original SMPTE document with tables of phase values for NTSC color bars.

 

And vector diagram from Texas Instrument document. Pinpointing phase angles of prime colors.

 

All of this might be very useful for Altirra.

 

Interesting fact is that prime colors are not spaced 120° from each other.

All phase angles are like in Atari CGIA document(page 50-51)

SMPTE tables.pdf

Texas Instruments.pdf

cgia.pdf

Edited October 9, 2014 by firestorm

[dmsc]

¡Hi!

 

Thank you DMSC, do you remember in witch year you wrote this program?

 

Well, I had my 800XL from 1985, and in 1990 I purchased a PC compatible, so it is probably from '88 or '89. Also, here in Chile, stations broadcasted the simple color bars at that time, without the bottom part.

 

But as I said earlier, probbly my atari was badly calibrated :-)

 

Daniel.