Update README.md
Browse files
README.md
CHANGED
@@ -9,4 +9,35 @@ tags:
|
|
9 |
- lora
|
10 |
- image-generation
|
11 |
- diffusers
|
12 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
9 |
- lora
|
10 |
- image-generation
|
11 |
- diffusers
|
12 |
+
---
|
13 |
+
|
14 |
+
LoRA checkpoint aiming to improve [FLUX.1 [schnell]](https://huggingface.co/black-forest-labs/FLUX.1-schnell) photorealism.
|
15 |
+
|
16 |
+
**The weights are provided as `.safetensors` file**, so they are suited for every use-cases, including but not limited to:
|
17 |
+
- SD-WebUI
|
18 |
+
- ComfyUI
|
19 |
+
- InvokeAI
|
20 |
+
- Forge
|
21 |
+
- ...
|
22 |
+
|
23 |
+
# Training details
|
24 |
+
This checkpoint has been trained with a fairly small dataset of 100+ synthetic images generated and upscaled using [Runtime44](https://runtime44.com)
|
25 |
+
All images have been manually captionned **without** any specific trigger word.
|
26 |
+
|
27 |
+
Training was done using [kohya-ss scripts](https://github.com/kohya-ss/sd-scripts) on a local machine with 16Gb of VRAM and took ~1h
|
28 |
+
|
29 |
+
|
30 |
+
|
31 |
+
# Usage
|
32 |
+
|
33 |
+
## With [🧨Diffusers](https://huggingface.co/docs/diffusers/en/index) library
|
34 |
+
|
35 |
+
```python
|
36 |
+
from diffusers import AutoPipelineForText2Image
|
37 |
+
import torch
|
38 |
+
|
39 |
+
pipeline = AutoPipelineForText2Image.from_pretrained('black-forest-labs/FLUX.1-schnell', torch_dtype=torch.bfloat16).to('cuda')
|
40 |
+
pipeline.load_lora_weights('hugovntr/flux-schnell-realism', weight_name='schnell-realism_v1')
|
41 |
+
image = pipeline('Moody kitchen at dusk, warm golden ...').images[0]
|
42 |
+
image.save("output.png")
|
43 |
+
```
|