Skip to content

The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface. The fastest local inference engine in the world.

Commits on master

Refactor calc_cond_uncond_batch into calc_cond_batch.
comfyanonymous committed 2 years ago
e6482fb
--force-fp16 is no longer necessary on Mac.
comfyanonymous committed 2 years ago
1306464
IP2P model loading support.
comfyanonymous committed 2 years ago
575acb6
Add log to debug custom nodes that hang when imported.
comfyanonymous committed 2 years ago
96b4c75
Cleanup to support different types of inpaint models.
comfyanonymous committed 2 years ago
94a5a67
5d8898c
Support SDXS 0.9
comfyanonymous committed 2 years ago
327ca13
Make step on sharpen node smaller.
comfyanonymous committed 2 years ago
8ae1e4d
Add Tesla GPUs to cuda malloc blacklist.
comfyanonymous committed 2 years ago
2f93b91
Fix test.
comfyanonymous committed 2 years ago
c967392
Increase the max resolution.
comfyanonymous committed 2 years ago
11838e6
dora_scale support for lora file.
comfyanonymous committed 2 years ago
ae77590
c6de09b
Move cleanup_models to improve performance.
comfyanonymous committed 2 years ago
6a32c06
a28a9dc
Add inverse noise scaling function.
comfyanonymous committed 2 years ago
0624838
Fix regression with lcm not working with batches.
comfyanonymous committed 2 years ago
5d875d7
Fix regression with model merging.
comfyanonymous committed 2 years ago
4b9005e
Don't unload model weights for non weight patches.
comfyanonymous committed 2 years ago
c18a203
Make LCM sampler use the model noise scaling function.
comfyanonymous committed 2 years ago
150a3e9
Remove code that should be useless now.
comfyanonymous committed 2 years ago
0c55f16
SV3D support.
comfyanonymous committed 2 years ago
40e124c
Fix neg scale step.
comfyanonymous committed 2 years ago
0b78213
b1a16d4
cacb022
Increase image batch nodes maximum values.
comfyanonymous committed 2 years ago
d3406d8
d7897ff
f2fe635