Skip to content
Back to skills

Flux Hf Lora

ASecurity

FLUX HF LoRA. Generate images using LoRA models with FLUX HF LoRA. Triggers: flux, image generation, lora

  • 29 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added June 2, 2026
documentationbashapidocumentation

Works with

  • api

Security analysis

A100/100

Scanned June 2, 2026

npx -y skills add awesome-genmedia/skills --skill flux-hf-lora --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Flux Hf Lora?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Flux Hf Lora
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/awesome-genmedia-flux-hf-lora/badge)](https://www.skillsdirectory.com/skills/awesome-genmedia-flux-hf-lora)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: flux-hf-lora
description: "FLUX HF LoRA. Generate images using LoRA models with FLUX HF LoRA. Triggers: flux, image generation, lora"
allowed-tools: Bash(curl *), WebFetch
---

# FLUX HF LoRA

Generate images using LoRA models with FLUX HF LoRA. Supports multiple aspect ratios (1:1, 16:9, 21:9, 3:2, 2:3, 4:5, 5:4, 3:4, 4:3, 9:16, 9:21), reproducible results via seed.

## Quick Start

> Requires an each::labs API key. Get one at [eachlabs.ai](https://eachlabs.ai).

```bash
curl -X POST https://api.eachlabs.ai/v1/prediction \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "model": "flux-hf-lora",
    "version": "0.0.1",
    "input": {
      "prompt": "A photorealistic portrait of a woman in a garden",
      "hf_lora": "alvdansen/frosting_lane_flux"
    }
  }'
```

## Parameters

| Parameter | Type | Default | Description |
|-----------|------|---------|-------------|
| `aspect_ratio` | string | 1:1 | An enumeration.. Options: `1:1`, `16:9`, `21:9`, `3:2`, `2:3`, `4:5`, `5:4`, `3:4`, `4:3`, `9:16`, `9:21` |
| `disable_safety_checker` | boolean | True | Disable safety checker for generated images. This feature is only available thro |
| `guidance_scale` | number | 3.5 | Guidance scale for the diffusion process |
| `hf_lora` | string |  | HF, Replicate, CivitAI, or URL to a LoRA. Ex: alvdansen/frosting_lane_flux |
| `image` | string |  | Input image for image to image mode. The aspect ratio of your output will match |
| `lora_scale` | number | 0.8 | Scale for the LoRA weights |
| `num_inference_steps` | integer | 28 | Number of inference steps |
| `num_outputs` | integer | 1 | Number of images to output. |
| `output_format` | string | webp | An enumeration.. Options: `webp`, `jpg`, `png` |
| `output_quality` | integer | 80 | Quality when saving the output images, from 0 to 100. 100 is best quality, 0 is |
| `prompt` | string |  | Prompt for generated image |
| `prompt_strength` | number | 0.8 | Prompt strength (or denoising strength) when using image to image. 1.0 correspon |
| `seed` | integer |  | Random seed. Set for reproducible generation |

## Examples

**Generate with LoRA:**
```bash
curl -X POST https://api.eachlabs.ai/v1/prediction \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "model": "flux-hf-lora",
    "version": "0.0.1",
    "input": {
      "prompt": "A photorealistic portrait of a woman in a garden",
      "hf_lora": "alvdansen/frosting_lane_flux"
    }
  }'
```

**LoRA with custom settings:**
```bash
curl -X POST https://api.eachlabs.ai/v1/prediction \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "model": "flux-hf-lora",
    "version": "0.0.1",
    "input": {
      "prompt": "A fantasy landscape with crystalline structures",
      "hf_lora": "example/fantasy-lora",
      "lora_scale": 0.8,
      "guidance_scale": 4.0,
      "seed": 42
    }
  }'
```

## Documentation

- [each::labs Docs](https://docs.eachlabs.ai)
- [API Reference](https://docs.eachlabs.ai/api/overview)

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.