Timeline for OpenGL Vertex Shader "joint matrix * weight" multiplication performance
Current License: CC BY-SA 4.0
9 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Aug 17, 2021 at 9:18 | vote | accept | Enes Altınkaya | ||
| Aug 15, 2021 at 9:23 | answer | added | Enes Altınkaya | timeline score: 1 | |
| Aug 13, 2021 at 14:50 | history | edited | Enes Altınkaya | CC BY-SA 4.0 |
deleted 12 characters in body
|
| Aug 13, 2021 at 13:52 | comment | added | Enes Altınkaya | Thanks for the suggestions, I updated the question. | |
| Aug 13, 2021 at 13:51 | history | edited | Enes Altınkaya | CC BY-SA 4.0 |
added 1415 characters in body
|
| Aug 13, 2021 at 11:55 | comment | added | DMGregory♦ | Did you try multiplying the four joint matrices by your position vector, then summing the resulting vectors instead? It's a similar cost in terms of arithmetic, but your temporaries are 4 floats instead of 16, which can make more efficient use of the hardware. Also, the order of your skin matrices and the model matrix looks a bit strange - have you considered pre-multiplying the model matrix into the joint matrix uniforms you pass to the shader, so you do those multiplications once per bone instead of once per vertex? | |
| Aug 13, 2021 at 9:25 | history | edited | Enes Altınkaya | CC BY-SA 4.0 |
edited title
|
| Aug 13, 2021 at 8:42 | review | First posts | |||
| Aug 13, 2021 at 18:59 | |||||
| Aug 13, 2021 at 8:40 | history | asked | Enes Altınkaya | CC BY-SA 4.0 |