Hello, I want to sort of recreate the effect that the node “Vinterp To” has in the blueprints (CPU) but in a material shader (GPU) with the use of World Position Offset and without the use of Collection Parameters.
Basically I want to create an object that follows a location, but with smoothing (not instantaneously), to do this I found the mathematical formula (smooth-step) that Vinterp/Finterp/Rinterp uses in the blueprints, the formula is:
Distance = TargetLocation - CurrentLocation
Step = DeltaTime x InterpSpeed
Movement = Distance x Step
Return Value = CurrentLocation + Movement
It takes 3 parameters: TargetLocation, CurrentLocation, Interp Speed
So the formula in one line would be like this: CurrentLocation + (TargetLocation - CurrentLocation) x (DeltaTime x InterpSpeed)
Now I have encountered a problem and it is that there is no way to access the Current Location of the object because there is no way to store that as a variable in the GPU, it is as if in every frame, the GPU starts from 0 without any persistence. I tried storing the data in render targets, but because I am not very familiar with them I didn’t found how to store the variable in just one pixel, and do that automatically with as many instances as there are in the level (each instance using it’s own pixel to store that variable without overwriting the variable for the other instances)
I’ve read somewhere that you could do this with the Custom Shader node (with HLSL) storing something like a Structured Buffer? I could be wrong but maybe there is a way to store variables persistently in the GPU without the use of render targets (which I suppose would be much slower than storing a simple vector value). This should be possible since the GPU has it’s own RAM, the VRAM! I looked for hours for the solution and didn’t find anything Thanks for your help