Thomas Jentzsch Posted October 14, 2008 Author Share Posted October 14, 2008 Your palette is actually fine. Atari would generate palette similar to yours with the TV set's saturation turned to the maximum. Take the saturation variable into consideration while computing. So less saturation with higher lumas, right? The question is: How much less? Quote Link to comment Share on other sites More sharing options...
Thomas Jentzsch Posted October 14, 2008 Author Share Posted October 14, 2008 Another try. This time I converted YIQ into HSV, limited S and V to [0..1] and finally converted HSV to RGB. | H| R G B| R G B| R G B| R G B| R G B| R G B| R G B| R G B| +--+-----------+-----------+-----------+-----------+-----------+-----------+-----------+-----------+ | 1| 44 50 0| 76 86 0|108 123 0|140 159 0|172 196 0|204 232 0|224 255 0|224 255 0| | 2| 84 22 0|121 44 0|157 75 0|194 114 0|230 161 0|255 198 0|255 198 0|255 198 0| | 3|141 35 0|177 55 0|214 80 0|250 109 0|255 114 0|255 115 2|255 129 28|255 141 48| | 4|166 0 8|203 0 10|239 0 12|255 34 44|255 60 69|255 80 88|255 97 104|255 110 117| | 5|156 0 103|192 0 126|229 0 150|255 16 173|255 45 183|255 67 191|255 86 197|255 101 202| | 6|100 0 187|143 0 224|186 0 255|190 14 255|197 40 255|202 62 255|207 79 255|211 93 255| | 7| 87 0 255| 87 0 255| 87 0 255|104 25 255|117 46 255|129 63 255|138 77 255|146 90 255| | 8| 1 0 255| 1 0 255| 27 26 255| 48 48 255| 66 65 255| 81 80 255| 93 93 255|104 104 255| | 9| 0 79 238| 0 84 255| 0 84 255| 2 86 255| 26 102 255| 46 115 255| 63 126 255| 77 136 255| |10| 0 98 143| 0 123 180| 0 148 216| 0 173 253| 0 175 255| 23 182 255| 46 189 255| 65 195 255| |11| 0 82 19| 0 118 41| 0 155 69| 0 191 106| 0 227 150| 15 255 193| 44 255 200| 67 255 206| |12| 0 95 5| 0 132 9| 0 168 15| 0 205 22| 1 241 31| 35 255 64| 60 255 86| 81 255 104| |13| 37 88 0| 52 125 0| 67 161 0| 82 198 0| 98 234 0|106 255 0|107 255 0|122 255 27| |14| 48 62 0| 76 98 0|104 135 0|131 171 0|159 208 0|187 244 0|196 255 0|196 255 0| |15| 61 13 0| 97 33 0|134 62 0|170 100 0|207 148 0|243 205 0|255 226 0|255 226 0| This one is much better for the brighter values. But the dark values for hues 7 to 9 (blue) show some problems now. Limiting helps, but it seems to require improvements. Though I remember that on some palettes I have seen, blue looked similar. Quote Link to comment Share on other sites More sharing options...
Kr0tki Posted October 15, 2008 Share Posted October 15, 2008 So less saturation with higher lumas, right? The question is: How much less? No, less saturation for all luminances. Multiply the I and Q variables by a value around 0.2 to 0.3 before computing the RGB values. Converting to YUV won't change much. Quote Link to comment Share on other sites More sharing options...
Thomas Jentzsch Posted October 15, 2008 Author Share Posted October 15, 2008 No, less saturation for all luminances. Multiply the I and Q variables by a value around 0.2 to 0.3 before computing the RGB values. That makes sense. IIRC I read somewhere that NTSC signals are limited for I and Q too. Converting to YUV won't change much. That's the part I do not understand. YUV is similar to YIQ, but not identical. The standard for NTSC was changed for YIQ to YUV ~30 years ago. How would an older TV understand the different format? And what was used for the 2600 signal generation? YIQ I suppose, right? Quote Link to comment Share on other sites More sharing options...
Kr0tki Posted October 15, 2008 Share Posted October 15, 2008 (edited) IIRC I read somewhere that NTSC signals are limited for I and Q too. Indeed they are; here are the limits given. YUV is similar to YIQ, but not identical. YUV is essentially the same colourspace as YIQ, only with the coordinate system rotated 33°. So by converting to YUV and rotating your initial angle of 150° by 33° (don't remember which direction though), you would achieve the same result. The standard for NTSC was changed for YIQ to YUV ~30 years ago. How would an older TV understand the different format? It is not exactly different format; it's only interpretation of colours that changes. I believe that displaying YIQ-generated signal on a YUV TV-set (and vice versa) would show good display with bad colours. Which then would be overcome by turning the TV's hue knob by 33°. (Yeah I know, knobs on TVs are rare these days, but anyway ) And what was used for the 2600 signal generation? YIQ I suppose, right? If you HAVE to turn the hue knob by 33° while switching the TV set from the 2600 to a TV programme, then the two signals code colours differently. (Actually, that's my speculation, but it sounds logical. ) Edited October 15, 2008 by Kr0tki Quote Link to comment Share on other sites More sharing options...
Thomas Jentzsch Posted October 15, 2008 Author Share Posted October 15, 2008 IIRC I read somewhere that NTSC signals are limited for I and Q too. Indeed they are; here are the limits given. You mean the limits for I (~0.6) and Q (~0.52)? I am already using them. Still colors become oversaturated. If you HAVE to turn the hue knob by 33° while switching the TV set from the 2600 to a TV programme, then the two signals code colours differently. (Actually, that's my speculation, but it sounds logical. ) Fortunately I live in a PAL country. So I cannot test that myself. I suppose modern NTSC may be able to adjust automatically to the 33° change, but maybe someone has an old NTSC TV and can confirm your speculation. Quote Link to comment Share on other sites More sharing options...
Kr0tki Posted October 16, 2008 Share Posted October 16, 2008 (edited) You mean the limits for I (~0.6) and Q (~0.52)? I am already using them. Still colors become oversaturated. And they should. Just like setting TV's saturation to the max, but you don't do it normally, do you? Edited October 16, 2008 by Kr0tki Quote Link to comment Share on other sites More sharing options...
Thomas Jentzsch Posted October 16, 2008 Author Share Posted October 16, 2008 You mean the limits for I (~0.6) and Q (~0.52)? I am already using them. Still colors become oversaturated. And they should. Just like setting TV's saturation to the max, but you don't do it normally, do you? Ah, I understand. I'll divide the value by ~2 and try again. Quote Link to comment Share on other sites More sharing options...
Thomas Jentzsch Posted October 17, 2008 Author Share Posted October 17, 2008 (edited) Ok, I think I can calculate palettes pretty accurate now. Attached you find 3 examples. The phase angles are: - 360/14.5 = ~24.8 (hue 15 is in the middle between hue 1 and 14) - 360/14 = ~25.7 (hue 15 = hue 1) - 360/13.5 = ~26.7 (hue 15 is in the middle between hue 1 and 2) Y = 0 .. 0.625, I = -0.5957/2 .. +0.5957/2, Q = -0.5226 .. +0.5226, saturation <= 1.0. Now we only have to decide which angle is correct. Edited October 17, 2008 by Thomas Jentzsch Quote Link to comment Share on other sites More sharing options...
Kr0tki Posted October 17, 2008 Share Posted October 17, 2008 I can't see the white colour on any of these palettes... Quote Link to comment Share on other sites More sharing options...
Thomas Jentzsch Posted October 17, 2008 Author Share Posted October 17, 2008 I can't see the white colour on any of these palettes... Too dark? Is that one better? Quote Link to comment Share on other sites More sharing options...
Kr0tki Posted October 17, 2008 Share Posted October 17, 2008 Yes, it is. Sorry if I sounded annoying. Quote Link to comment Share on other sites More sharing options...
Thomas Jentzsch Posted October 17, 2008 Author Share Posted October 17, 2008 Not at all. Quote Link to comment Share on other sites More sharing options...
ClausB Posted October 19, 2008 Share Posted October 19, 2008 The phase angles are:- 360/14.5 = ~24.8 (hue 15 is in the middle between hue 1 and 14) If hue 15 is exactly in the middle, then the phase angle would be 360°/15 = 24° That is how I used to adjust my Ataris, except for my 400, since I only had a B&W TV back then. Quote Link to comment Share on other sites More sharing options...
Thomas Jentzsch Posted October 19, 2008 Author Share Posted October 19, 2008 If hue 15 is exactly in the middle, then the phase angle would be 360°/15 = 24° Both are true. You define that hue 15 = hue 14 + 24° = hue 1 - 24° I define that hue 15 = hue 14 + 24.8°/2 = hue 1 - 24.8°/2 In both cases, hue 15 is in the middle between 1 and 14. Quote Link to comment Share on other sites More sharing options...
ClausB Posted October 19, 2008 Share Posted October 19, 2008 If hue 15 is exactly in the middle, then the phase angle would be 360°/15 = 24° Both are true. You define that hue 15 = hue 14 + 24° = hue 1 - 24° I define that hue 15 = hue 14 + 24.8°/2 = hue 1 - 24.8°/2 In both cases, hue 15 is in the middle between 1 and 14. Not quite right, because hue 15 - hue 14 is the same as hue 14 - hue 13 and as hue 13 - hue 12, etc. Therefore, hue 15 - hue 14 is much more than 24.8°/2. Quote Link to comment Share on other sites More sharing options...
Thomas Jentzsch Posted October 19, 2008 Author Share Posted October 19, 2008 Not quite right, because hue 15 - hue 14 is the same as hue 14 - hue 13 and as hue 13 - hue 12, etc. Therefore, hue 15 - hue 14 is much more than 24.8°/2. I stand corrected, you are right. hue 15 = hue 14 + 24.8° = hue 1 - 24.8°/2 Quote Link to comment Share on other sites More sharing options...
SeaGtGruff Posted October 20, 2008 Share Posted October 20, 2008 Not quite right, because hue 15 - hue 14 is the same as hue 14 - hue 13 and as hue 13 - hue 12, etc. Therefore, hue 15 - hue 14 is much more than 24.8°/2. I stand corrected, you are right. hue 15 = hue 14 + 24.8° = hue 1 - 24.8°/2 I love what you've been doing with this. The other night I searched the old stella lists and found an excellent thread about all of this ("tia hue luminance," which you had participated in), including a link to a web page with a program that Adam Wozniak did that lets you adjust the phase step and such-- very cool stuff, and the thread has some great information from Eric Ball! Michael Quote Link to comment Share on other sites More sharing options...
Thomas Jentzsch Posted October 20, 2008 Author Share Posted October 20, 2008 I had completely forgotten about this. Thanks for digging it out. Quote Link to comment Share on other sites More sharing options...
SeaGtGruff Posted October 21, 2008 Share Posted October 21, 2008 I had completely forgotten about this. Thanks for digging it out. By the way, what formulas are you using for calculating the RGB values? And are you going from YIQ to RGB, or from YUV to RGB, or...? Michael Quote Link to comment Share on other sites More sharing options...
Thomas Jentzsch Posted October 21, 2008 Author Share Posted October 21, 2008 I had completely forgotten about this. Thanks for digging it out. By the way, what formulas are you using for calculating the RGB values? And are you going from YIQ to RGB, or from YUV to RGB, or...? YIQ to RGB (unlimited) to HSV (limit S) to RGB (limiting) Pretty overcomplicated Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.