Flux LoRAs Text-to-Image (Krea Dev)
A minimal Flux text-to-image graph with two LoRA slots. It ships with an Arcane style LoRA and a pixel art LoRA you can swap for your own.
- L40S
- Text to image
- Free template
How it works
The graph is intentionally small, with 13 native nodes and no custom node packages. UNETLoader loads flux1-krea-dev, DualCLIPLoader loads clip_l and t5xxl_fp16, and VAELoader loads ae.safetensors.
Two LoraLoaderModelOnly nodes are chained after the model. The files shipped are ARCANE_STYLE_FADOO-FLUX.safetensors and Pixelart_-_FLUX.safetensors. Because they are model-only loaders, they change the diffusion model and leave the text encoder alone.
The positive prompt goes through CLIPTextEncode and FluxGuidance. The negative prompt goes through a second CLIPTextEncode, with FluxDisableGuidance applied to it. EmptyLatentImage sets the canvas, KSampler denoises, VAEDecode decodes and SaveImage writes the file.
The example prompt is a retro game-over screen with King Kong fighting Godzilla in pixel art. The example negative prompt lists cartoon, 3D render, blurry and similar terms.
Tips
- Add the LoRA trigger words to your prompt if the LoRA needs them, such as pixelart for the pixel art file.
- Lower one LoRA strength to zero to test a single style, or use both at partial strength to blend them.
- Write a negative prompt that excludes the style you do not want, as the example does.
- Keep your own LoRAs in the same order and compare outputs with a fixed seed.
What you can use it for
- Trying a style LoRA before using it in a larger pipeline
- Pixel art game screens and sprites
- Arcane-style character art
- Testing your own trained Flux LoRA
Run it in three steps
Open the template
Click Run on ComfyDeploy. The workflow, its custom nodes and models are set up for you on a cloud machine.
Add your inputs
Provide text prompt and negative prompt in the playground or the ComfyUI editor.
Run, share or deploy
Run it on an L40S, share it with your team as a playground, or deploy it as an API endpoint.
Frequently asked questions
What GPU does the Flux LoRA workflow need?
The recommended GPU is the L40S. ComfyDeploy provisions it for you, so there is nothing to install.
Can I use my own LoRA?
Yes. Both LoRA slots use LoraLoaderModelOnly. Replace ARCANE_STYLE_FADOO-FLUX or Pixelart_-_FLUX with your own file and adjust the strength.
How do I run the Flux LoRA workflow as an API?
Deploy the workflow on ComfyDeploy and call the generated endpoint. The ComfyUIDeployExternal input and output nodes in the graph become the request and response fields.
