61b3f15f8f
Fix lowvram mode not working with unCLIP and Revision code.
11 months ago
d0165d819a
Fix SVD lowvram mode.
11 months ago
a252963f95
--disable-smart-memory now unloads everything like it did originally.
11 months ago
36a7953142
Greatly improve lowvram sampling speed by getting rid of accelerate.
...
Let me know if this breaks anything.
11 months ago
261bcbb0d9
A few missing comfy ops in the VAE.
11 months ago
9a7619b72d
Fix regression with inpaint model.
11 months ago
571ea8cdcc
Fix SAG not working with cfg 1.0
11 months ago
8cf1daa108
Fix SDXL area composition sometimes not using the right pooled output.
11 months ago
2258f85159
Support stable zero 123 model.
...
To use it use the ImageOnlyCheckpointLoader to load the checkpoint and
the new Stable_Zero123 node.
11 months ago
2f9d6a97ec
Add --deterministic option to make pytorch use deterministic algorithms.
11 months ago
e45d920ae3
Don't resize clip vision image when the size is already good.
11 months ago
13e6d5366e
Switch clip vision to manual cast.
...
Make it use the same dtype as the text encoder.
11 months ago
719fa0866f
Set clip vision model in eval mode so it works without inference mode.
11 months ago
574363a8a6
Implement Perp-Neg
11 months ago
a5056cfb1f
Remove useless code.
11 months ago
329c571993
Improve code legibility.
11 months ago
6c5990f7db
Fix cfg being calculated more than once if sampler_cfg_function.
11 months ago
ba04a87d10
Refactor and improve the sag node.
...
Moved all the sag related code to comfy_extras/nodes_sag.py
11 months ago
6761233e9d
Implement Self-Attention Guidance ( #2201 )
...
* First SAG test
* need to put extra options on the model instead of patcher
* no errors and results seem not-broken
* Use @ashen-uncensored formula, which works better!!!
* Fix a crash when using weird resolutions. Remove an unnecessary UNet call
* Improve comments, optimize memory in blur routine
* SAG works with sampler_cfg_function
11 months ago
b454a67bb9
Support segmind vega model.
11 months ago
824e4935f5
Add dtype parameter to VAE object.
11 months ago
32b7e7e769
Add manual cast to controlnet.
11 months ago
3152023fbc
Use inference dtype for unet memory usage estimation.
11 months ago
77755ab8db
Refactor comfy.ops
...
comfy.ops -> comfy.ops.disable_weight_init
This should make it more clear what they actually do.
Some unused code has also been removed.
11 months ago
b0aab1e4ea
Add an option --fp16-unet to force using fp16 for the unet.
11 months ago
ba07cb748e
Use faster manual cast for fp8 in unet.
11 months ago
57926635e8
Switch text encoder to manual cast.
...
Use fp16 text encoder weights for CPU inference to lower memory usage.
11 months ago
340177e6e8
Disable non blocking on mps.
11 months ago
614b7e731f
Implement GLora.
11 months ago
cb63e230b4
Make lora code a bit cleaner.
11 months ago
174eba8e95
Use own clip vision model implementation.
11 months ago
97015b6b38
Cleanup.
12 months ago
a4ec54a40d
Add linear_start and linear_end to model_config.sampling_settings
12 months ago
9ac0b487ac
Make --gpu-only put intermediate values in GPU memory instead of cpu.
12 months ago
efb704c758
Support attention masking in CLIP implementation.
12 months ago
fbdb14d4c4
Cleaner CLIP text encoder implementation.
...
Use a simple CLIP model implementation instead of the one from
transformers.
This will allow some interesting things that would too hackish to implement
using the transformers implementation.
12 months ago
2db86b4676
Slightly faster lora applying.
12 months ago
1bbd65ab30
Missed this one.
12 months ago
9b655d4fd7
Fix memory issue with control loras.
12 months ago
26b1c0a771
Fix control lora on fp8.
12 months ago
be3468ddd5
Less useless downcasting.
12 months ago
ca82ade765
Use .itemsize to get dtype size for fp8.
12 months ago
31b0f6f3d8
UNET weights can now be stored in fp8.
...
--fp8_e4m3fn-unet and --fp8_e5m2-unet are the two different formats
supported by pytorch.
12 months ago
af365e4dd1
All the unet ops with weights are now handled by comfy.ops
12 months ago
61a123a1e0
A different way of handling multiple images passed to SVD.
...
Previously when a list of 3 images [0, 1, 2] was used for a 6 frame video
they were concated like this:
[0, 1, 2, 0, 1, 2]
now they are concated like this:
[0, 0, 1, 1, 2, 2]
12 months ago
c97be4db91
Support SD2.1 turbo checkpoint.
12 months ago
983ebc5792
Use smart model management for VAE to decrease latency.
12 months ago
c45d1b9b67
Add a function to load a unet from a state dict.
12 months ago
f30b992b18
.sigma and .timestep now return tensors on the same device as the input.
12 months ago
13fdee6abf
Try to free memory for both cond+uncond before inference.
12 months ago
comfyanonymous
Hari
Rafie Walker