Log inRegister an accountBrowse CSDbHelp & documentationFacts & StatisticsThe forumsAvailable RSS-feeds on CSDbSupport CSDb Commodore 64 Scene Database
You are not logged in 
CSDb User Forums


Forums > C64 Pixeling > Joe's 136 colors technique
2012-03-03 07:26
Joodas

Registered: Oct 2011
Posts: 6
Joe's 136 colors technique

Hi all,

I am looking for information about Joe's 136 colors technique. Any links will be appreciated.

Thank you for help
2012-03-05 15:28
Jailbird

Registered: Dec 2001
Posts: 1536
About what?!?
2012-03-05 15:44
Fresh
Account closed

Registered: Jan 2005
Posts: 85
I guess he means this:
Quote:

User Comment
Submitted by Joe on 3 April 2010
Don't forget the rule of those 136 colors which derives from the existing 16 if dithered as Crosshatched or Lined,
which then looks rather ok. At least to my own experience as a small thumb on the screen.
Though, the interlace is a bit tough as is with all horizontal lines of interleaved colors.
I'd guess on the real hardware and telly the thing works very well?!

And obviously enough on the real BIG THING ;D


Source: Daring Darling (unfinished)

Dunno what they were talking about though.
2012-03-05 15:58
Jailbird

Registered: Dec 2001
Posts: 1536
Everyone on that comment-thread sounds like they were drunk or/and high. So I'd assume that Joe's 136 color rule comes from a mighty fine LSD trip. A local dealer might help you further on your quest. You're welcome, joodas.
2012-03-05 16:29
Conrad
Account closed

Registered: Nov 2006
Posts: 715
xD... I haven't had a good laugh on CSDB for a long time! Thanks, Jailbird! :)
2012-03-05 20:19
bepp

Registered: Jun 2010
Posts: 252
:-D
2012-03-05 21:13
Joe

Registered: Apr 2002
Posts: 163
Well, first of all you should have in mind never to think when choosing color. A dark color could easily be set to a bright one, without anti aliasing. There are some artists which actually bother about what hue to pitch the surface, no-no-no! That's so old fashioned.

You are absolutely unrestricted to any convention. And heck, even if you use an original setup you might end up with some pleasant accidents you cant control, such as chromatic aberration or blind outs or dark outs, it's absolutely magic. Yes!

Even if one shots the eye of any diagonal hatching, scale differences of our thumbnail artwork on any web page would be so reassuring that we are dealing with a universe of colors! Imagine your tiny winy images on any smart phone, fantastic!

I love the joke though Jailbird, it certainly is no rocket science, I believe both me, Carrion, JHM and a quite few others managed to try it out. It does work occasionally. Sometimes it looks awkward on the projected screen and definitely on the real set. It's not as smooth as the old every second line glitch Lars is a master of.
2012-03-05 21:16
Groepaz
Account closed

Registered: Dec 2001
Posts: 9100
sounds to me like you are still on that trip - good stuff :o)
2012-03-05 21:17
Joe

Registered: Apr 2002
Posts: 163
;P
2012-03-05 21:27
Jailbird

Registered: Dec 2001
Posts: 1536
Ah, now we get it! So, I've been further even more decided to use 136 colors even go need to do look more as anyone can. Can you really be far even as decided half as much to use go wish for that? My guess is that when one really been far even as decided once to use thumbnail artwork even go want, it is then that he has really been far even as decided to use even go want to do look more like. Well, all in all it's just common sense.
2012-03-06 13:41
Stainless Steel

Registered: Mar 2003
Posts: 964
You evil googletranslators.
2012-03-06 14:13
Skate

Registered: Jul 2003
Posts: 474
it's just mixing colors like checkers

1: Color 1
2: Color 2

12121212
21212121
12121212
21212121
12121212
21212121
12121212
21212121


and you have a mixed color. 16x16 there are 256 color combinations. but!

S: Solid Color (Original C64 Color)
X: Mixed Color
M: Mirrored Color (same color with inverted checker pattern)

+ 0123456789ABCDEF

0 SMMMMMMMMMMMMMMM
1 XSMMMMMMMMMMMMMM
2 XXSMMMMMMMMMMMMM
3 XXXSMMMMMMMMMMMM
4 XXXXSMMMMMMMMMMM
5 XXXXXSMMMMMMMMMM
6 XXXXXXSMMMMMMMMM
7 XXXXXXXSMMMMMMMM
8 XXXXXXXXSMMMMMMM
9 XXXXXXXXXSMMMMMM
A XXXXXXXXXXSMMMMM
B XXXXXXXXXXXSMMMM
C XXXXXXXXXXXXSMMM
D XXXXXXXXXXXXXSMM
E XXXXXXXXXXXXXXSM
F XXXXXXXXXXXXXXXS


Number of Solid Colors = 16
Number of Mixed Colors = 120
Number of Mirrored Colors = 120

Since mirrored colors are duplicates, there are 120 + 16 = 136 colors.
2012-03-06 14:28
Graham
Account closed

Registered: Dec 2002
Posts: 990
There's also true color mixing when using different colors of with the same luminance:

AAAAAAAAAAAAAAAA
BBBBBBBBBBBBBBBB
AAAAAAAAAAAAAAAA
BBBBBBBBBBBBBBBB
AAAAAAAAAAAAAAAA
BBBBBBBBBBBBBBBB
AAAAAAAAAAAAAAAA
BBBBBBBBBBBBBBBB


which results in a slightly different color than:

BBBBBBBBBBBBBBBB
AAAAAAAAAAAAAAAA
BBBBBBBBBBBBBBBB
AAAAAAAAAAAAAAAA
BBBBBBBBBBBBBBBB
AAAAAAAAAAAAAAAA
BBBBBBBBBBBBBBBB
AAAAAAAAAAAAAAAA


Although using the same colors :)
2012-03-06 14:33
Skate

Registered: Jul 2003
Posts: 474
My favorite picture using this technique is Mermaid's Self Portrait with Unicorns .
2012-03-06 18:25
Graham
Account closed

Registered: Dec 2002
Posts: 990
And Oh Noes, They Be Stealing My Bench
2012-03-06 21:37
redcrab

Registered: Aug 2011
Posts: 25
The downside with crosshatching is that you tend to get big color-differences depending on where on the screen you have it, resulting in horizontal stripes, often in blue-ish and yellow (when watching on a tube television at least)

"Waiting for inspiration" by Carrion (very nice picture otherwise) looks like a zebra. -At least on my tv's.

Do anyone know WHY it is like that?
2012-03-06 22:18
Groepaz
Account closed

Registered: Dec 2001
Posts: 9100
"The downside with crosshatching is that you tend to get big color-differences depending on where on the screen you have it, resulting in horizontal stripes, often in blue-ish and yellow"
its related to the effect graham mentioned - also in a checkerboard pattern PAL effects apply.
2012-03-07 15:04
Perplex

Registered: Feb 2009
Posts: 219
Here's a photo of a TV with a C64 connected, demonstrating the difference between odd/even mixing:



Rows 1 and 4 contain the plain colours. Row 2 is mixed with colours from row 1 on even lines and colours from row 4 on odd lines. Row 3 is mixed with colours from row 1 on odd lines and colours from row 4 on even lines.
2012-03-07 17:53
redcrab

Registered: Aug 2011
Posts: 25


But I'm still puzzled by that hires thing. Here is a photo of a TV showing checkerboard patterns with light gray and white. The upper one starts with a white pixel in the upper left corner, and the lower one is inverted. (gray in upper left)

As you can see it alternates between blue and yellow for every character row. It is in fact exact every char even though the camera seems to bleed a bit too. And the lower one does the same, but the opposite way. This is not logic to me.. But I might just be stupid :)

(this pattern may also be familiar to some from the sky in mr sids version of canabalt)
2012-03-07 18:41
Graham
Account closed

Registered: Dec 2002
Posts: 990
The vertical green/red bars effect only happens when using composite. If you use S-Video, that effect will not happen.

It's caused when the composite signal is seperated into a luma and a chroma signal again by using a low/high-pass filter with a cutoff frequency ~4 MHz. Since the filter doesn't work perfect, parts of the hires luma pixel pattern will still leak over to the chroma signal.

~8 MHz pixel clock with alternating pixels = ~4 MHz frequency. Remember: Chroma color carrier has 4.43 MHz.

The reason why the pattern is fixed to character sizes is simple. Both, pixel clock and color carrier clock derive from the same master clock: 17.734475 MHz. Divided by 4 it's exactly the PAL color carrier frequency of 4.43361875 MHz, divided by 18 it's the C64 system clock of 0.985 MHz. C64 system clock * 8 is the VIC2 pixel clock of 7.882 MHz. So there is an integer ratio between color carrier and pixel clock of 16/9. :)
2012-03-07 18:57
Monte Carlos

Registered: Jun 2004
Posts: 266
I give it the name "Geos Pattern"
2012-03-07 20:28
Wile Coyote
Account closed

Registered: Mar 2004
Posts: 580
Quote: Here's a photo of a TV with a C64 connected, demonstrating the difference between odd/even mixing:



Rows 1 and 4 contain the plain colours. Row 2 is mixed with colours from row 1 on even lines and colours from row 4 on odd lines. Row 3 is mixed with colours from row 1 on odd lines and colours from row 4 on even lines.


It has been known for a long time, the method of placing one multi colour pixel above another multi colour pixel to produce a new colour. Green and Pink being one of the best combos to produce Beige.

(maybe someone already has) It would be interesting to see a converter that produces an image based on the 28 colours, and outputs and image (using FLI) with pixels that are 2x2 hires pixels / 1x2 multi colour pixels.
2012-03-07 20:38
Graham
Account closed

Registered: Dec 2002
Posts: 990
To my knowledge used the first time in 46 Colour Demo/Radwar. Another early demo using that effect is Goatbeard/Oneway (12 sprites part).
2012-03-08 09:20
ChristopherJam

Registered: Aug 2004
Posts: 933
Redcrab - it's worse than that. Power cycle your c64 and display the same pattern, and there's a good chance the situation will be reversed, (possibly even scrolled by a partial character width? I can't remember now).

It's all to do with the relative phase of the colour signal and the dot clock, and the circuit that produces them doesn't appear to be consistently initialised :(
2012-03-08 12:58
Monte Carlos

Registered: Jun 2004
Posts: 266
I like that color scheme very much, although it is not complete, as the colors 5,10,12 and 14 can be mixed in any combination. If this would be taken account for, it would be a perfect color scheme.
The disadvantage of this technique is, that it must be calibrated, because some TVs displays mix color 1 when mixing abababab and mix color 2 when mixing babababa and other TVs display the exchanged mix colors.
Also, as seen in "Courtesy of soviet", y soft scrolling causes the mix colors to alternate dependent on the y shift.
If y shift is changed continously, the mix colors flicker.
2012-03-08 13:28
algorithm

Registered: May 2002
Posts: 692
Quote: It has been known for a long time, the method of placing one multi colour pixel above another multi colour pixel to produce a new colour. Green and Pink being one of the best combos to produce Beige.

(maybe someone already has) It would be interesting to see a converter that produces an image based on the 28 colours, and outputs and image (using FLI) with pixels that are 2x2 hires pixels / 1x2 multi colour pixels.


Its also less straightforward than that, as the next line color is determined by the previous line color, moving onto the other 2x2 will result in the first line in the next 2x2 section etc, having a different color based on the color on the previous line. Have a look at the Algotecher demo for some examples in MUCSU-FLI (non interlaced) and MR-FLI (interlaced) which utilises the pal blending to create additional colors. Some additional examples in the MUCSU-FLI mode are in the demo 'introducing MUCSU-FLI'

Furthermore, there is the blending that occurs on the horizontal plane as well. Taking these all to account would require a huge lot of computations (unless opting for the suboptimial method)

MonteCarlos, I also believe if the image is scrolled, the issue with the color blends changing can be rectified by scrolling it in even amount eg 2,4,8

ChristopherJam, that is interesting in regards to the initialisation theory, perhaps this may also be linked to the VSP crashes that occur sometimes when powering on the c64.

2012-03-08 14:00
Monte Carlos

Registered: Jun 2004
Posts: 266
Yeah, scrolling in even steps doesn't change any color.
I wonder what's that "cool" demo from dez. 1985, radwar cites in their release in the intro?
2012-03-08 15:44
Joe

Registered: Apr 2002
Posts: 163
Joodas got an email last saturday, three hours after the opening of this thread. I shared some thoughts on the subject and I can expand a little here:

The actual vista of more colors is of course secondary and more a optimistic will, if one wants, of reading images. I'd rather want to express the possibility of a working method of adding layers on top of each other. Instead of thinking about hatching areas to mold volumes, which is rather the method of an additive color principle (oil or digital means of giving hues, tones and depth): Whereas one thinks of flat representations of objects concerned about ditheringstyle and technique.

I believe that this simple method share similarities to subractive color (water color) where one not only got a grain of texture and new possibilities to handle surfaces as flat fields of information. It might sound silly of introducing color, light, hue and grain (that is scale and dimension) in one and the same mean. We could go on in the mere technical aspect of the dilemma or as I have understood, in a more general form problem. Not very much has happened in the field of c64 art, part from perhaps the graphic-modes and the fact that people are getting better at the craft. Leon and Lars has both showed some very fine ways of dealing with transitional spaces, just as Sit did recently. Carrion showed some potential ways of layering fields and I'm trying to get rid of dithering in the convential manner. Veto showed that fields give rise to complex situations of interdependant narratives and Nada proved the potential of a fine net of sprayed pixels.

We all give rise to different approaches towards representations regardless of the subjective design. That is the most important, I'd guess: To look forward.

As for the Zebras, here's a small demo showing the awkward stripes.
2012-03-16 21:10
WVL

Registered: Mar 2002
Posts: 847
Instead of scrolling in an even number (2/4/6/whatever) of pixels per frame, you can also scroll in an odd number of pixels, if you have 2 sets of bitmap where you reverse the mixing colors in the 2nd bitmap.

That way the colors stay the same and doesn't depend on the speed that you're scrolling with. -> idea (C) WVL ;)
2012-03-16 21:56
Digger

Registered: Mar 2005
Posts: 320
@WVL: nice idea :)

Does anybody know how to enable native PAL emulation in VICE?

1) to get checkerboard (hires) stripes (described by RedCrab and Graham)
2) to get blended colours (like on the real C64)

@RedCrab/Graham:
That's why they disappear when you turn your telly to BW.
Anyone has S-Video cable? I've made one a while ago but it didn't seem to work as Graham describes. Photo please – I don't believe! ;+)
2012-03-17 05:45
Groepaz
Account closed

Registered: Dec 2001
Posts: 9100
"1) to get checkerboard (hires) stripes (described by RedCrab and Graham)"
thats not emulated :) it'd also be quite computing intense to do so it probably wont happen anytime soon. use s-video damnit =P
2012-03-17 23:22
Copyfault

Registered: Dec 2001
Posts: 254
Quote: Instead of scrolling in an even number (2/4/6/whatever) of pixels per frame, you can also scroll in an odd number of pixels, if you have 2 sets of bitmap where you reverse the mixing colors in the 2nd bitmap.

That way the colors stay the same and doesn't depend on the speed that you're scrolling with. -> idea (C) WVL ;)


If the screen layout allows it you could even display all "background colours" using sprites -> Voila, scrolling @ 1 rasline/frame without flickering :))
2014-05-24 22:45
Jailbird

Registered: Dec 2001
Posts: 1536
You guys are overcomplicating a very simple matter. No one except Joe is thinking like this about pixelling :)
2014-05-26 04:14
JackAsser

Registered: Jun 2002
Posts: 1522
Quote: You guys are overcomplicating a very simple matter. No one except Joe is thinking like this about pixelling :)

And nobody except you (and obviously I) are reading two year old threads... ;)
2014-05-26 19:43
Pex Mahoney Tufvesson

Registered: Sep 2003
Posts: 38
JackAsser: ...to put old threads into great use, I have an idea:

Could we use the vertical hires stripes (alternating white and black pixels) running on a PAL output to determine the startup phase relationship of the VIC-II clocks?

Then we would have _the_ solution of finding out when a C64 is powered on in a VSP-safe state. :-D

---
Have a noise night!
http://mahoney.c64.org
2014-05-26 20:46
Groepaz
Account closed

Registered: Dec 2001
Posts: 9100
so now you ONLY have to find a way to determine the looks of those stripes by software =)
2014-05-26 21:33
soci

Registered: Sep 2003
Posts: 433
Coding that is easy:
01234567890ABCDEF
#################

Type in the number/letter where
it's the most green.

But it does not work very well when using a proper TV set with s-video, as there's hardly any stripes to see.
2014-05-26 21:37
Groepaz
Account closed

Registered: Dec 2001
Posts: 9100
thats not quite a software solution when it requires the user to enter the result :)
2016-04-01 20:22
Digger

Registered: Mar 2005
Posts: 320
Finally, I've created a small tool (alpha) to explore all these 136 colours:
https://dl.dropboxusercontent.com/u/254436606/c64/136-colors/in..

Feedback welcome :)
2016-04-02 07:57
ilesj

Registered: Jun 2012
Posts: 26
That's cool. But to what are the luminance values based on? Looking at the 'base' colors (distance 0), the colors that are luminance pairs have different values in this chart. For example brown and blue.
2016-04-02 08:10
Digger

Registered: Mar 2005
Posts: 320
@ilesj: You're right – the luminance was derived from the HTML hex colours thus not identical with C64 color luminances.

I've replaced the values as measured by Pepto (http://www.pepto.de/projects/colorvic/)

Thanks for feedback!
2016-04-02 10:22
Oswald

Registered: Apr 2002
Posts: 4400
I think a converter could consider using these 'colors', but no idea if the outcome would look good. even 8 bit colors can use dithering, and these are dithered from the beginning so.. also about 1/3 or half of them is unusable because of big luma distance imho.
2016-04-02 14:26
Digger

Registered: Mar 2005
Posts: 320
@Oswald: I agree, hence luma distance slider to find out which colours are flicker safe :)
2016-04-04 00:44
Skate

Registered: Jul 2003
Posts: 474
Very nice observation tool, thank you for that.

Something like "exclude solid colors" checkbox could be useful to see the mixed colors alone. That would be useful while using with the Luminance Diff Threshold slider.
2016-04-04 04:33
Oswald

Registered: Apr 2002
Posts: 4400
also to show the 'true' mix could be a feature a little dithered box.
2016-04-04 08:01
Groepaz
Account closed

Registered: Dec 2001
Posts: 9100
i used this 136 color palette in my own converter... it can be very useful depending on the input. and indeed, max. luma-distance slider is one of the key features :)
2016-08-11 20:43
Digger

Registered: Mar 2005
Posts: 320
The 136 colors explorer tool has been updated. I've added a hires/multi/line dither patterns (per Oswald's request). Enjoy :)
https://og2t.github.io/retro-palette-explorer/
2016-08-12 16:27
Rastah Bar

Registered: Oct 2012
Posts: 169
Quoting Skate
it's just mixing colors like checkers

1: Color 1
2: Color 2

12121212
21212121
12121212
21212121
12121212
21212121
12121212
21212121




Can you also mix more than 2 colors? For example
12121212
34343434
12121212
34343434

or

12121212
23232323
12121212
23232323
2016-08-12 16:45
Joe

Registered: Apr 2002
Posts: 163
As Color Bar suggests: I found out when changing from ordered dithering to work with lining things. Which came from the experiences of having hires-sprites expanded in the border overlaying rasters and hatches of these "surfaces". It gives one level more of say texture of textile quality and visually a small impression of something very colorful or at least processed and elaborate images.
RefreshSubscribe to this thread:

You need to be logged in to post in the forum.

Search the forum:
Search   for   in  
All times are CET.
Search CSDb
Advanced
Users Online
Shine/[S] & G★P
Mihai
hedning/G★P
Joe/WD/EXON/Nah-Kolo..
Logiker
Aomeba/Artline Desig..
Mibri/HF
DivertigO
Mr. SID
Lynx/ΤRIΛD
Guests online: 57
Top Demos
1 Unboxed  (9.7)
2 Uncensored  (9.7)
3 Coma Light 13  (9.7)
4 Edge of Disgrace  (9.6)
5 Comaland 100%  (9.6)
6 The Shores of Reflec..  (9.6)
7 Lunatico  (9.6)
8 X Marks the Spot  (9.5)
9 Wonderland XII  (9.5)
10 OMG Got Balls!  (9.5)
Top onefile Demos
1 Smile to the Sky  (9.5)
2 Instinct  (9.5)
3 Daah, Those Acid Pil..  (9.5)
4 Dawnfall V1.1  (9.4)
5 Pandemoniac Part 2 o..  (9.4)
6 Crystal Gazer  (9.4)
7 Rewind  (9.4)
8 Bad Boy  (9.4)
9 Official X2018 Report  (9.4)
10 Party Horse  (9.4)
Top Groups
1 Fossil  (9.7)
2 PriorArt  (9.6)
3 Oxyron  (9.4)
4 Booze Design  (9.4)
5 Censor Design  (9.3)
Top Fullscreen Graphicians
1 Facet  (10)
2 Joe  (9.9)
3 Veto  (9.8)
4 Mirage  (9.6)
5 Jailbird  (9.6)

Home - Disclaimer
Copyright © No Name 2001-2019
Page generated in: 0.103 sec.