---
license: apache-2.0
base_model: black-forest-labs/FLUX.2-klein-base-4B
tags:
  - flux
  - flux2
  - lora
  - text-to-image
  - style
---

# darkframe-found

Style LoRA for Found Footage. Trained on FLUX.2 Klein 4B base from Prompt Vault stills. People are framed as close-ups.

| | |
|---|---|
| Trigger | `darkframe-found` |
| Base model | `black-forest-labs/FLUX.2-klein-base-4B` |
| Rank / alpha | 32 / 32 |
| Steps | 2500 |

Start the prompt with `darkframe-found`.

## Examples

Step-2500 training samples.

![sample 0](examples/sample_0.jpg)

![sample 1](examples/sample_1.jpg)

![sample 2](examples/sample_2.jpg)

![sample 3](examples/sample_3.jpg)

![sample 4](examples/sample_4.jpg)

![sample 5](examples/sample_5.jpg)

## Fal

Use a FLUX.2 Klein 4B LoRA endpoint. This file will not load on Klein 9B or FLUX.2 dev.

```json
{
  "prompt": "darkframe-found, close-up",
  "guidance_scale": 4,
  "num_inference_steps": 28,
  "loras": [
    {
      "path": "https://huggingface.co/darkframeai/darkframe-found/resolve/main/darkframe-found.safetensors",
      "scale": 1.0
    }
  ]
}
```