Still the same animation but today with a different set of control parameters.
https://www.youtube.com/watch?v=4UOGd5uaCe0&feature=youtu.be
I imagine this inverted with growing circles and with correct palette to be like traveling through worm hole , nice work
waw, 3D animation
No problem to invert the direction: https://www.youtube.com/watch?v=FapUhM05mzo&feature=youtu.be
Using palettes is no problem as well, since we have the 12 bit ColorFromPalette we can use them without loosing color resolution. It just becomes tricky again when combining multiple layers with multiple palettes - if the layers are just added it might cause “white outs” or at least lots of desaturated colors, if the RGB values are averaged properly it causes a loss of color resolution again. For fast running animation both is no problem. With very slow movements it´s visible as unsteady edges.
white-outs could be avoided with using that new fadetocolor stuff I believe 
Please elaborate on your idea, @Nikolay_Hristov . The problem I see is this: If we have 2 layers and want to combine them using blend or nscale or whatever we need to scale the RGB values down to 50% first and then adding them. That means we have just 7 bits left now to describe the RGB values of the scaled layer. The range is now 0-127 and not 0-255 anymore. We loose resolution. With more layers it becomes even worse. 4 layers leave us only 6 bits for each RGB value. I´m very interested in ideas how to overcome this problem when doing multi-layer animations.
I though instead of using blend use fadeusingcolor layer -one using color mask from layer two , but not sure how this gonna render :)) . Is it possible to use CHSV -mixing hues than RGB values there should saturation / brightness calculated separately
Yes, it would work to use one layer as a “colored alpha mask” for another one. But still we handle 8 bit data and that gives poor results if many math operations are applied. To maintain high detail and resolution the original data should have 16 bits and be scaled down just at the very end after all required math is performed. Don´t get me wrong here - I´m a bit obsessed with animation quality at the moment. For 2 layers and kind of fast moving animations the FastLED functions are perfectly fine and very convenient. So play with it, there are many ways to combine RGB and HSV manipulation and often it´s hard to imagine before how it will look at the end. It´s worth to play with it and try many different ideas and approaches.
I think , that the solution is simple in concept , not sure if simple in implementation ,
The ideas is to work in layers of higher resolution , and perform all combinations needed in that parallel " world" , once it is finished , only then you scale down to the needed physical resolution , by doing this , you would always achieve the highest possible resolution
Of course the definition of an interpolation algorithm will be important , but independent of the previous step .
Makes sense ?
Absolutely, it´s called “oversampling”. It´s cool, but expensive. 2x oversampling means 4x the calaculations amount. 3x oversampling results already in 9x the amount of data that needs to be processed. That helps to get the max out of the physical matrix resolution, but it doesn´t help with the limited color depth. Btw. this is what I´m doing with the Sinoids - I render them basically in a 32 bit space (!) but map it down to a 16x16… That allows these ultra smooth transitions and shifts.