Jean-Francois Charles

I've sometimes fed a float32 one-plane matrix to a jit.gl.slab. My reasoning was that it would save some data duplication to avoid a conversion to a 4-plane matrix.

It turns out that this doesn't work anymore with gl3: the one-plane matrix is then treated as a red plane.

I could just convert my matrices to 4-plane (jit.pack or jit.matrix)

But would someone point me in the right direction to massage the gl3 version of the shader in order to let the GPU use a single plane input on the internal RGB or RGBA?

Here is a simple example working in gl2, but not in gl3.

button

jit.gl.slab

toggle

loadbang

message

jit.matrix

jit.fill

multislider

jit.gl.videoplane

jit.noise

jit.gl.render

jit.window

qmetro

Hi, 
I've sometimes fed a float32 one-plane matrix to a jit.gl.slab. My reasoning was that it would save some data duplication to avoid a conversion to a 4-plane matrix.
It turns out that this doesn't work anymore with gl3: the one-plane matrix is then treated as a red plane.
I could just convert my matrices to 4-plane (jit.pack or jit.matrix)
But would someone point me in the right direction to massage the gl3 version of the shader in order to let the GPU use a single plane input on the internal RGB or RGBA?
Here is a simple example working in gl2, but not in gl3.

Hi, 
I've sometimes fed a float32 one-plane matrix to a jit.gl.slab. My reasoning was that it would save some data duplication to avoid a co

gl2-vs-gl3-single-plane-matrix-to-gl-slab

You could sample the input matrix (which gets converted to a texture in jit.gl.slab) lile this: 

vec4 myMatVal = texture(myMat, texcoord).rrrr; 

This would work in the shader I believe. 

Anyway why are you passing matrices to jit.gl.slab? 

It could be more performant to make the calculation in jit.gen if you are working with matrices. 

Passing a one-plane matrix to the GPU is not much more efficient than passing a 4 planes matrix.

You could sample the input matrix (which gets converted to a texture in jit.gl.slab) lile this: 
vec4 myMatVal = texture(myMat, texcoord).rrrr; 
This would work in the shader I believe. 
Anyway why are you passing matrices to jit.gl.slab? 
It could be more performant to make the calculation in jit.gen if you are working with matrices. 
Passing a one-plane matrix to the GPU is not much more efficient than passing a 4 planes matrix. 

(I wanted to avoid copying a 1-plane matrix into 4 planes of a 4-plane matrix.)

Great, thanks for that! You made my day!
(I wanted to avoid copying a 1-plane matrix into 4 planes of a 4-plane matrix.)


gl2 vs gl3: single plane matrix to gl.slab