raymarch | Unity project for building ray | Game Engine library

I know there are a lot of resources on this, but none of them have worked for me.
Some are: webgl readpixels is always returning 0,0,0,0,
and this one: https://stackoverflow.com/questions/44869599/readpixels-from-webgl-canvas\ as well as this one: Read pixels from a WebGL texture
but none of them have been either helpful or successful.

The goal: Render an offscreen canvas with a WebGL shader, then use that as a texture in a separate WebGL shader.

Notes:

• For these WebGL shaders, I'm using a generic vertex shader used for pixel shaders, specifically, a raytracer/raymarcher. This is: attribute vec2 a_position; void main() { gl_Position = vec4(a_position.xy, 0.0, 1.0); }. This vertex shader is inputted two triangles that cover the screen, so basically the fragment shader is doing all the work.

Problem: In order to get the image data off of the offscreen canvas, I've tried these methods:

1. The WebGL gl.readPixels function

I am follow a guide on ray marching by The Art of Code, and I tried to implement my own ray marcher without looking at the video. I was wondering if I could get some help with my code. My code is below. The confusion it at line 86, which is at the bottom of the code and marked with a comment.

Normalize should not be needed here since the ray_direction is normalized in the RayMarch function. Therefore if the normalize is removed from line 86 nothing should change, yet this is not the case. What is going on here?

This is my first shader so it's probably some dumb tiny error. I'm following a tutorial on YouTube: https://www.youtube.com/watch?v=S8AWd66hoCo&t=83s&pbjreload=10

I'm stuck with the distance function (for a sphere) and simply can't get it to work without some weird error going with my for loop. Everything before that worked perfectly. I rewrote the for-loop but get this error over and over.

Thank you in advance!

(Unity 2019.2.19f1)

ERRORS:

I’ve written a Fragment shader in GLSL, using shader toy. Link : https://www.shadertoy.com/view/wtGSzy

most of it works, but when I enable texture lookups in the reflection function, the performance drops from 60FPS to 5~FPS.

The code in question is on lines 173 - 176

I'm running into a bug with my shaders. For the vertex:

I recently read this paper about raymarching clouds (careful it´s an PDF, in case you dont want that: http://www.diva-portal.org/smash/get/diva2:1223894/FULLTEXT01.pdf) where the author goes on about optimizing (page 22ff.) the algorithm via reprojection. He states that by raymarching only 1/16th of all pixels per frame (the selected pixel hopping around in a 4x4 grid) and reprojecting the rest, he got about 10 times a performance increase.

I now tried implementing this as well in Unreal Engine 4 (custom HLSL shader) and I got the raymarching as well as the reprojection working now. However I´m stuck at actually only running the raymarching on the necessary pixels. As far as I´m aware with any branching in HLSL both sides of the branch will be calculated and one will be thrown away. I therefore can´t do something like this pseudo code in the pixel shader: if(!PixelReprojection) { return 0;} else { return Raymarch(...); } as it will calculate the Raymarch even for pixels that are getting reprojected.

I don´t see any other way though to archieving this... Is there any kind of branching in HLSL that allows this? It can´t be static as the pixels subjected to raymarching and reprojecting change every frame. I´m really curious on any ideas how the author could have achieved this tenfold increase in performance as he is writing the code on a GPU too, as far as I´m aware.

I´d greatly appreciate any kind of input here.

Regards, foodius

I'm trying to set up a pipeline to do some raymarching-rendering.

At first I set up a vertex and a geometry shader to take just 1 arbitrary float, and make a quad so I can use just the fragment shader, and input data via passing it through all shaders or uniforms.

But then I came across the compute shader, and found some tutorials but all just did nearly the same, making a quad and rendering compute-shaders output to it, I think that's pretty stupid, if you have the possibility to render how you want with the compute shader and still do tricks to get your result.

After some further research i found the function 'glFramebufferImage2D' and as far as I understood it, it attaches an ImageTexture (the one I wrote to with my compute shader in my case) to the Framebuffer (the buffer thats displayed as far as i understood it). So i dont have to do the quad generating trick. But the code I wrote just shows a black screen. Did i get something wrong? or did i missed something in my code?

This is my code: (I'm not worrying about warnings and detaching shader-programs yet. I just wanted to test the concept for now.)

main.rs

I'm working on a RayMarching Program(HomeWork) and I want it to go faster so I use the GPU with the ALEA extension. I have a problem because I can't Use The class camera In the parallel for (GPU). Thanks for your help.

I already tried to change the tag of the class and creating them inside the Parallel for

I'm raymarching Signed Distance Fields in CUDA and the scene I'm rendering contains thousands of spheres (spheres have their location stored in device buffer, so my SDF function iterates through all of the spheres for each pixel).

Currently, I'm computing distance to sphere surface as:

sqrtf( dot( pos - sphere_center, pos - sphere_center ) ) - sphere_radius

With the sqrt() function, the rendering took about 250ms for my scene. However, when I removed the call to sqrt() and left just dot( pos - sphere_center, pos - sphere_center ) - sphere_radius, the rendering time dropped to 17ms (and rendering black image).

The sqrt() function seems to be the bottleneck so I want to ask if there is a way I can improve my rendering time (either by using different formula that does not use square root or different rendering approach)?

I'm already using -use-fast-math.

Edit: I've tried formula suggested by Nico Schertler, but it didn't work in my renderer. Link to M(n)WE on Shadertoy.