---
tags:
- text-to-image
- lora
- diffusers
- template:diffusion-lora
widget:
- text: 'Super Paint, An abstract painting of a red basket filled with yellow french fries. The basket is made of red paper, and the basket is attached to a light blue wall. The yellow fries are arranged in a wavy pattern, and there is a yellow letter M on the red paper. There is a shadow of the basket on the left side of the painting, and on the right side, there are black lines that form the letters V and X.'
  output:
    url: images/1.png
- text: 'Super Paint, A painting of a glass jar with a green lid. The jar is filled with pickles. The pickles are yellow and green. There is a green stripe on the lid of the jar. There are black letters on the jar that are black. The jars are sitting on a light colored surface. Theres a shadow on the left side of the jars. The wall behind the jar is white.'
  output:
    url: images/2.png
- text: 'Super Paint, A painting of two cookies on a plate. The cookies are brown with black dots on them. The plate is white with a red stripe on it. There is a white bucket next to the cookies. The bucket has a blue rim around it. The wall behind the bucket is dark blue. The table under the plate is a light green color.'
  output:
    url: images/3.png
- text: 'Super Paint, An abstract painting of two pieces of bread on a light blue background. The left piece of bread has jam on it, while the right piece has peanut butter on it. The peanut butter is in the shape of a heart. The jam is a vibrant red color, and there are small white dots on the top of the jam. The painting is done in a simple, cartoon style.'
  output:
    url: images/4.png
base_model: black-forest-labs/FLUX.1-dev
instance_prompt: Super Paint
license: creativeml-openrail-m
---
![adasd.png](https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/E7EycN1526lVTk-E8LMwS.png)

<Gallery />

# Model description for Flux-Super-Paint-LoRA

Image Processing Parameters 

| Parameter                 | Value  | Parameter                 | Value  |
|---------------------------|--------|---------------------------|--------|
| LR Scheduler              | constant | Noise Offset              | 0.03   |
| Optimizer                 | AdamW  | Multires Noise Discount   | 0.1    |
| Network Dim               | 64     | Multires Noise Iterations | 10     |
| Network Alpha             | 32     | Repeat & Steps           | 26 & 3400 |
| Epoch                     | 25   | Save Every N Epochs       | 1    |

    Labeling: florence2-en(natural language & English)
    
    Total Images Used for Training : 36

## Best Dimensions & Inference

| **Dimensions** | **Aspect Ratio** | **Recommendation**       |
|-----------------|------------------|---------------------------|
| 1280 x 832      | 3:2              | Best                     |
| 1024 x 1024     | 1:1              | Default                  |

### Inference Range

- **Recommended Inference Steps:** 30–35

## Setting Up
```python
import torch
from pipelines import DiffusionPipeline

base_model = "black-forest-labs/FLUX.1-dev"
pipe = DiffusionPipeline.from_pretrained(base_model, torch_dtype=torch.bfloat16)

lora_repo = "strangerzonehf/Flux-Super-Paint-LoRA"
trigger_word = "Super Paint"  
pipe.load_lora_weights(lora_repo)

device = torch.device("cuda")
pipe.to(device)
```
## Trigger words

You should use `Super Paint` to trigger the image generation.

## Download model

Weights for this model are available in Safetensors format.

[Download](/strangerzonehf/Flux-Super-Paint-LoRA/tree/main) them in the Files & versions tab.