Update README.md
Browse files
README.md
CHANGED
@@ -18,7 +18,7 @@ The original Perturbed-Attention Guidance for unconditional models and SD1.5 by
|
|
18 |
Also there is an extra implementation of the Perturbed-Attention Guidance (PAG) on Stable Diffusion XL (SDXL) for the 🧨 diffusers library by [multimodalart](https://huggingface.co/multimodalart)
|
19 |
[code](https://huggingface.co/multimodalart/sdxl_perturbed_attention_guidance) / [demo](https://huggingface.co/spaces/multimodalart/perturbed-attention-guidance-sdxl)
|
20 |
|
21 |
-
This repository is just a simple implementation of the Perturbed-Attention Guidance (PAG) on Stable Diffusion XL (SDXL) for the 🧨 diffusers library to "
|
22 |
|
23 |
|
24 |
## Quickstart
|
@@ -37,28 +37,28 @@ pipe = pipe.to(device)
|
|
37 |
```
|
38 |
|
39 |
Unconditional sampling with PAG:
|
40 |
-
![image/jpeg](
|
41 |
|
42 |
```py
|
43 |
output = pipe(
|
44 |
"",
|
45 |
image=init_image,
|
46 |
-
strength=0.
|
47 |
-
num_inference_steps=
|
48 |
guidance_scale=0.0,
|
49 |
-
pag_scale=
|
50 |
pag_applied_layers=['mid']
|
51 |
).images
|
52 |
```
|
53 |
|
54 |
Sampling with PAG and CFG:
|
55 |
-
![image/jpeg](
|
56 |
```py
|
57 |
output = pipe(
|
58 |
-
"
|
59 |
image=init_image,
|
60 |
-
strength=0.
|
61 |
-
num_inference_steps=
|
62 |
guidance_scale=4.0,
|
63 |
pag_scale=3.0,
|
64 |
pag_applied_layers=['mid']
|
|
|
18 |
Also there is an extra implementation of the Perturbed-Attention Guidance (PAG) on Stable Diffusion XL (SDXL) for the 🧨 diffusers library by [multimodalart](https://huggingface.co/multimodalart)
|
19 |
[code](https://huggingface.co/multimodalart/sdxl_perturbed_attention_guidance) / [demo](https://huggingface.co/spaces/multimodalart/perturbed-attention-guidance-sdxl)
|
20 |
|
21 |
+
This repository is just a simple implementation of the Perturbed-Attention Guidance (PAG) on Stable Diffusion XL (SDXL) for the 🧨 diffusers library to "image-to-image".
|
22 |
|
23 |
|
24 |
## Quickstart
|
|
|
37 |
```
|
38 |
|
39 |
Unconditional sampling with PAG:
|
40 |
+
![image/jpeg](example 1.png)
|
41 |
|
42 |
```py
|
43 |
output = pipe(
|
44 |
"",
|
45 |
image=init_image,
|
46 |
+
strength=0.6,
|
47 |
+
num_inference_steps=40,
|
48 |
guidance_scale=0.0,
|
49 |
+
pag_scale=4.0,
|
50 |
pag_applied_layers=['mid']
|
51 |
).images
|
52 |
```
|
53 |
|
54 |
Sampling with PAG and CFG:
|
55 |
+
![image/jpeg](example 2.png)
|
56 |
```py
|
57 |
output = pipe(
|
58 |
+
"A man with hoodie on is looking at sky, photo",
|
59 |
image=init_image,
|
60 |
+
strength=0.6,
|
61 |
+
num_inference_steps=40,
|
62 |
guidance_scale=4.0,
|
63 |
pag_scale=3.0,
|
64 |
pag_applied_layers=['mid']
|