syrinx

Hi all, I'm working on a patch where I need to be able to address individual pixels precisely using their xy coordinates. I noticed what seems like weird behaviour when using the cell operator inside [jit.gl.pix @type float32].

I would expect cell to generate values that step by 1 and end in a .5. In testing, however, the values step by a value less than 1. The result is that by the middle of the texture, the integer part of the value no longer corresponds to the cell coordinate.

Using norm * dim instead generates a texture where the values step by 1 and end in .5. I'm not sure if this is intended behaviour, but I found it confusing.

comment

loadbang

message

jit.gl.texture

jit.matrix

jit.cellblock

jit.gl.pix

norm

toggle

jit.world

cell

Hi all, I'm working on a patch where I need to be able to address individual pixels precisely using their xy coordinates. I noticed what see

jit-gl-pix-type-float32-cell-values-step-by-less-than-1

I'm no jit.gl.pix jedi, but maybe adding [floor] or [ceil] after [cell] will produce the output you expect

That's actually how I ran into the issue - I wanted to use floor to get the coordinates . Because cell steps by less than 1 with @type float32, however, the values wind up being off by the middle of the texture. 

Assuming an 8x8 texture, for example, the desired coordinate values across the horizontal axis would be 1 2 3 4 5 6 7 8, but running the output of cell into floor gives 1 2 3 4 4 5 6 7.

Using norm * dim as a substitute does what I need, but I mention it here because it seems like a bug, or at minimum weird behaviour that might trip someone up.

That's actually how I ran into the issue - I wanted to use floor to get the coordinates . Because cell steps by less than 1 with @type float32, however,  the values wind up being off by the middle of the texture. 

Assuming an 8x8 texture, for example, the desired coordinate values across the horizontal axis would be 1 2 3 4 5 6 7 8,  but running the output of cell into floor gives 1 2 3 4 4 5 6 7.

Using norm * dim as a substitute does what I need, but I mention it here because it seems like a bug, or at minimum weird behaviour that might trip someone up.

jit.gl.pix @type float32 - cell values step by less than 1