README + images
Browse files- .gitignore +1 -0
- README.md +73 -0
- example1.webp +0 -0
- example2.webp +0 -0
- example3.webp +0 -0
- example4.webp +0 -0
- example5.webp +0 -0
.gitignore
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
.DS_Store
|
README.md
CHANGED
@@ -1,3 +1,76 @@
|
|
1 |
---
|
2 |
license: mit
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: mit
|
3 |
+
base_model: stabilityai/stable-diffusion-xl-base-1.0
|
4 |
+
tags:
|
5 |
+
- stable-diffusion
|
6 |
+
- stable-diffusion-diffusers
|
7 |
+
- text-to-image
|
8 |
+
- diffusers
|
9 |
+
- lora
|
10 |
+
inference: true
|
11 |
---
|
12 |
+
|
13 |
+
# sdxl-ugly-sonic-lora
|
14 |
+
|
15 |
+
A LoRA for SDXL 1.0 Base which generates Ugly Sonic, using `sonic the hedgehog` as the trigger keywords.
|
16 |
+
|
17 |
+
## Usage
|
18 |
+
|
19 |
+
The LoRA can be loaded using `load_lora_weights` like any other LoRA in `diffusers`:
|
20 |
+
|
21 |
+
```py
|
22 |
+
import torch
|
23 |
+
from diffusers import DiffusionPipeline, AutoencoderKL
|
24 |
+
|
25 |
+
vae = AutoencoderKL.from_pretrained(
|
26 |
+
"madebyollin/sdxl-vae-fp16-fix",
|
27 |
+
torch_dtype=torch.float16
|
28 |
+
)
|
29 |
+
base = DiffusionPipeline.from_pretrained(
|
30 |
+
"stabilityai/stable-diffusion-xl-base-1.0",
|
31 |
+
vae=vae,
|
32 |
+
torch_dtype=torch.float16,
|
33 |
+
variant="fp16",
|
34 |
+
use_safetensors=True
|
35 |
+
)
|
36 |
+
|
37 |
+
base.load_lora_weights("minimaxir/sdxl-ugly-sonic-lora")
|
38 |
+
|
39 |
+
_ = base.to("cuda")
|
40 |
+
```
|
41 |
+
|
42 |
+
During image generation, use `sonic the hedgehog` in the prompt.
|
43 |
+
|
44 |
+
## Examples
|
45 |
+
|
46 |
+
For all generations, the negative prompt used is `blurry, low quality`.
|
47 |
+
|
48 |
+
`a close up of sonic the hedgehog (smiling for the camera with a toothy grin)++++, hyperrealistic photo for national geographic` (cfg = 13)
|
49 |
+
|
50 |
+
![](img/example1.webp)
|
51 |
+
|
52 |
+
`sonic the hedgehog relaxing on a couch, renaissance painting, (oil on canvas, aged, worn)++++` (cfg = 13)
|
53 |
+
|
54 |
+
![](img/example2.webp)
|
55 |
+
|
56 |
+
`a profile of sonic the hedgehog sitting at a desk deep in thought, (pixel art)++++, award-winning photo for vanity fair` (cfg = 13)
|
57 |
+
|
58 |
+
![](img/example3.webp)
|
59 |
+
|
60 |
+
`anatomical diagram of sonic the hedgehog, (highly detailed)++++` (cfg = 13)
|
61 |
+
|
62 |
+
![](img/example4.webp)
|
63 |
+
|
64 |
+
`sonic the hedgehog (eating at McDonald's)++, Ukiyo-e, minimalistic vector art` (cfg = 13)
|
65 |
+
|
66 |
+
![](img/example5.webp)
|
67 |
+
|
68 |
+
## Methodology
|
69 |
+
|
70 |
+
This LoRA was trained on frame-by-frame analysis of the [original 1080p trailer](https://www.youtube.com/watch?v=4mW9FE5ILJs) featuring "Ugly Sonic". Square-crops of Ugly Sonic were extracted and AI-upscaled to 1080p.
|
71 |
+
|
72 |
+
The use of `sonic the hedgehog` as the trigger keywords ensures that you won't generate the _other_ hedgehog by accident.
|
73 |
+
|
74 |
+
## Notes
|
75 |
+
|
76 |
+
- The CGI style of Ugly Sonic may overpower other style prompts. Therefore, you should weight any style prompts much higher.
|
example1.webp
ADDED
example2.webp
ADDED
example3.webp
ADDED
example4.webp
ADDED
example5.webp
ADDED