All rendered colors are at 50℅ saturation, left to right X axis are the hues, pretty easy to spot the critical hues by the peaks, and from bottom up on the Y axis is 0 to 100% luminance. The white region is where RGB values would otherwise wash out, so as a result not all luminance values are rendered.
If I set the saturation to 0, then the entire graph would be greyscale bottom up, and if I set the saturation to say 95%, the curved line graph you see here would be narrowed down to a basically curved representation of very vibrant colors. If I set the saturation to 100%, you'd be lucky if it rendered 6 pixels at the peaks of the hues, as everything else would be washed out.
I do have other graphs at other saturation levels from the same algorithms I wrote to render this, but damn if they ain't buried in years old personal archives..
Edit: I'm sure you won't get the same exact luminance/brightness values from conventional color processing algorithms, as I'm measuring the color values differently, experimentally, and most important, visually.
I never gave this experimental color space a formal name, but my working codename for it was RealHSL. It really is just a different perspective of the RGB cube, adjusted by luminance weights, in polar coordinates, with the center axis being the 0% saturation greyscales.
It becomes a lot easier to see if you make all the pure white pixels transparent and wrap it around a cylinder in Blender or whatever 3D modeling software, then rotate the cylinder around and you'll see the cylinder edges are sliced at the faces of the RGB cube.
I think you're just mapping/recreating the CIELAB system in RGB then.
For apparent color, the Lab* coordinates are a lot easier to work with, and they already do exactly what you're attempting to do with the polar coordinates (since LCh is essentially the same coordinate system but with polar coordinates).
This is a Winamp AVS Visualization I made in 2011 apparently, but ended up just as one of my study projects that led to Color Painter, before the AI boom gave me a programmer's crash..
Oh, I've got my reasons that I didn't pursue this all that much beyond 2014, I came up with a whole new color model altogether. Not a color space, an entire color model with its own processing system, using function mappers to map entire regions of images into full luminance scale photographic color gradients.
In my new system, every single gradient, the entire gradient is considered and treated as an individual color 'object', and any one pinpoint RGB is considered a solid color. My newer software is called Color Painter, and I call the process itself NICE ART (Natural Interface for Computers and Electronics Abstract Rendering Technology).
Let's say I wanna take a photo of an iceberg, and render it in fire colors instead. All I gotta do is load the photo, type 'orange fire' into the color search box, it loads a photographically correct gradient colorspace of real world fire, then I paint some of that loaded fire color over some areas of the iceberg, then hit Adjust, and tada, it's done.
I've since also put that project aside due to the rapid advances in AI managing to more or less make my Color Painter obsolete by today's expected standards, but since I loathe AI, I still use my prototype software occasionally.
Do you mean this is the 25th to 75th percentile of saturation?
What luminance or L* have you picked?
All rendered colors are at 50℅ saturation, left to right X axis are the hues, pretty easy to spot the critical hues by the peaks, and from bottom up on the Y axis is 0 to 100% luminance. The white region is where RGB values would otherwise wash out, so as a result not all luminance values are rendered.
If I set the saturation to 0, then the entire graph would be greyscale bottom up, and if I set the saturation to say 95%, the curved line graph you see here would be narrowed down to a basically curved representation of very vibrant colors. If I set the saturation to 100%, you'd be lucky if it rendered 6 pixels at the peaks of the hues, as everything else would be washed out.
I do have other graphs at other saturation levels from the same algorithms I wrote to render this, but damn if they ain't buried in years old personal archives..
Edit: I'm sure you won't get the same exact luminance/brightness values from conventional color processing algorithms, as I'm measuring the color values differently, experimentally, and most important, visually.
Which color space are you working in? Are you using something larger than sRGB?
I think this task would be a lot easier in CIELAB.
I never gave this experimental color space a formal name, but my working codename for it was RealHSL. It really is just a different perspective of the RGB cube, adjusted by luminance weights, in polar coordinates, with the center axis being the 0% saturation greyscales.
It becomes a lot easier to see if you make all the pure white pixels transparent and wrap it around a cylinder in Blender or whatever 3D modeling software, then rotate the cylinder around and you'll see the cylinder edges are sliced at the faces of the RGB cube.
I think you're just mapping/recreating the CIELAB system in RGB then.
For apparent color, the Lab* coordinates are a lot easier to work with, and they already do exactly what you're attempting to do with the polar coordinates (since LCh is essentially the same coordinate system but with polar coordinates).
I just found this from my other NICE ART account, I figured Google would have deleted that account by now given how long I haven't logged in..
https://www.youtube.com/watch?v=7HWUtNXpUFU
This is a Winamp AVS Visualization I made in 2011 apparently, but ended up just as one of my study projects that led to Color Painter, before the AI boom gave me a programmer's crash..
Oh, I've got my reasons that I didn't pursue this all that much beyond 2014, I came up with a whole new color model altogether. Not a color space, an entire color model with its own processing system, using function mappers to map entire regions of images into full luminance scale photographic color gradients.
In my new system, every single gradient, the entire gradient is considered and treated as an individual color 'object', and any one pinpoint RGB is considered a solid color. My newer software is called Color Painter, and I call the process itself NICE ART (Natural Interface for Computers and Electronics Abstract Rendering Technology).
Let's say I wanna take a photo of an iceberg, and render it in fire colors instead. All I gotta do is load the photo, type 'orange fire' into the color search box, it loads a photographically correct gradient colorspace of real world fire, then I paint some of that loaded fire color over some areas of the iceberg, then hit Adjust, and tada, it's done.
I've since also put that project aside due to the rapid advances in AI managing to more or less make my Color Painter obsolete by today's expected standards, but since I loathe AI, I still use my prototype software occasionally.
For 2014 that sounds pretty cool, but I agree, technology has continued its march onward. For better or worse we'll see.