Commit Graph

27 Commits

Author SHA1 Message Date
Qing
ac67ddd3d6 controlnet depth add pad 2024-01-10 22:04:46 +08:00
Qing
d420ecbecd powerpaint support_strength=True 2024-01-10 21:54:35 +08:00
Qing
ccfb077a2e change mps dtype to fp32 2024-01-10 21:48:14 +08:00
Qing
c3cc7a238e fix test 2024-01-10 21:47:03 +08:00
Qing
a869982d13 change controlnet depth preprocessor 2024-01-10 21:32:05 +08:00
Qing
dfcb9017f1 set controlnet_conditioning_scale=0 when use extender 2024-01-10 21:25:34 +08:00
Qing
7545d54870 fix test 2024-01-10 21:24:59 +08:00
Qing
cbcdf3b9a2 from_single_file load_safety_checker work 2024-01-10 21:22:59 +08:00
Qing
772ef65f7b always cpu_textencoder=False if device is not cuda 2024-01-10 13:34:51 +08:00
Qing
38b6edacf0 sdxl support cpu_text_encoder 2024-01-10 13:34:11 +08:00
Qing
05a15b2e1f pass pipe_components when turn control on/off 2024-01-10 13:02:13 +08:00
Qing
db4a6f4547 optimize mem 2024-01-10 10:21:58 +08:00
Qing
fea3f2ed4e remove support_controlnet for PowerByExample 2024-01-09 22:44:01 +08:00
Qing
8ed969eec1 add enable_low_mem 2024-01-09 22:42:48 +08:00
Qing
a49c3f86d3 add enable_low_mem 2024-01-09 15:05:17 +08:00
Qing
a71c3fbe1b clean code: get_torch_dtype; mps use float16 by default 2024-01-09 13:26:20 +08:00
Qing
6f4ce66793 handle MaxRetryError 2024-01-09 11:32:54 +08:00
Qing
7c1f83e71d fix match_histograms 2024-01-09 10:31:16 +08:00
Qing
df5fd5ccf0 add back torch._C._jit_set_nvfuser_enabled(False) 2024-01-08 23:12:32 +08:00
Qing
0e6babea67 add __main__.py
https://github.com/Sanster/lama-cleaner/pull/416
2024-01-08 22:45:01 +08:00
Qing
a1a06004d8 fix lcm_lora load 2024-01-08 21:49:27 +08:00
Qing
5da47ee035 add back enable_attention_slicing for mps device 2024-01-08 21:49:18 +08:00
Qing
fd74b8556f beta2 2024-01-05 16:56:54 +08:00
Qing
f88f3cbbb3 fix import issue 2024-01-05 16:38:55 +08:00
Qing
cc06f30e9a beta1 2024-01-05 16:21:49 +08:00
Qing
a95dd10ceb fix 2024-01-05 15:31:36 +08:00
Qing
a73e2a531f rename to iopaint 2024-01-05 15:19:23 +08:00