Change README to show how to use it with diffusers

#4
Files changed (1) hide show
  1. README.md +53 -2
README.md CHANGED
@@ -13,9 +13,60 @@ This model is NOT the 19.2M images Characters Model on TrinArt, but an improved
13
 
14
  このモデルはTrinArtのキャラクターズモデル(1920万枚再学習モデル)ではありません! とりんさまAIボットのモデルの改良版です。このモデルはオリジナルのSD v1.4モデルのアートスタイルをできる限り残したまま、アニメ・マンガ方向に調整することを意図しています。
15
 
16
- A Diffusers port of the pre-trained model is available at https://huggingface.co/ayan4m1/trinart_diffusers_v2
17
 
18
- (by ayan4m1 - thanks!)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
19
 
20
 
21
  ## Stable Diffusion TrinArt/Trin-sama AI finetune v2
 
13
 
14
  このモデルはTrinArtのキャラクターズモデル(1920万枚再学習モデル)ではありません! とりんさまAIボットのモデルの改良版です。このモデルはオリジナルのSD v1.4モデルのアートスタイルをできる限り残したまま、アニメ・マンガ方向に調整することを意図しています。
15
 
16
+ ## Diffusers
17
 
18
+ The model has been ported to `diffusers` by [ayan4m1](https://huggingface.co/ayan4m1] (thanks!)
19
+ and can easily be run from one of the branches:
20
+ - `revision="diffusers-60k"` for the checkpoint trained on 60,000 steps,
21
+ - `revision="diffusers-95k"` for the checkpoint trained on 95,000 steps,
22
+ - `revision="diffusers-115k"` for the checkpoint trained on 115,000 steps.
23
+
24
+ For more information, please have a look at [the "Three flavors" section](#three-flavors).
25
+
26
+ ### Example Text2Image
27
+
28
+ ```python
29
+ # !pip install diffusers==0.3.0
30
+ from diffusers import StableDiffusionPipeline
31
+
32
+ # using the 60,000 steps checkpoint
33
+ pipe = StableDiffusionPipeline.from_pretrained("naclbit/trinart_stable_diffusion_v2", revision="diffusers-60k")
34
+ pipe.to("cuda")
35
+
36
+ image = pipe("A magical dragon flying in front of the Himalaya in manga style").images[0]
37
+ image
38
+ ```
39
+
40
+ ![dragon](https://huggingface.co/datasets/patrickvonplaten/images/resolve/main/a_magical_dragon_himalaya.png)
41
+
42
+ If you want to run the pipeline faster or on a different hardware, please have a look at the [optimization docs](https://huggingface.co/docs/diffusers/optimization/fp16).
43
+
44
+ ### Example Image2Image
45
+
46
+ ```python
47
+ # !pip install diffusers==0.3.0
48
+ from diffusers import StableDiffusionImg2ImgPipeline
49
+ import requests
50
+ from PIL import Image
51
+ from io import BytesIO
52
+
53
+ url = "https://scitechdaily.com/images/Dog-Park.jpg"
54
+
55
+ response = requests.get(url)
56
+ init_image = Image.open(BytesIO(response.content)).convert("RGB")
57
+ init_image = init_image.resize((768, 512))
58
+
59
+ # using the 115,000 steps checkpoint
60
+ pipe = StableDiffusionImg2ImgPipeline.from_pretrained("naclbit/trinart_stable_diffusion_v2", revision="diffusers-115k")
61
+ pipe.to("cuda")
62
+
63
+ images = pipe(prompt="Manga drawing of Brad Pitt", init_image=init_image, strength=0.75, guidance_scale=7.5).images
64
+ image
65
+ ```
66
+
67
+ ![brad_pitt](https://huggingface.co/datasets/patrickvonplaten/images/blob/main/manga_man.png)
68
+
69
+ If you want to run the pipeline faster or on a different hardware, please have a look at the [optimization docs](https://huggingface.co/docs/diffusers/optimization/fp16).
70
 
71
 
72
  ## Stable Diffusion TrinArt/Trin-sama AI finetune v2