Flux with LLM

This Model allows you, to use the FLUX base model in ComfyUI as follows:

• Choose between img2img and txt2img generation

• Use LLM conditioning (requires ollama and its models / nodes)

• Use a modifier for prompts using LLM

You can use many combinations of it.

You can specify different resolutions for img2img and txt2img. The main setting uses a definition of the size in MPixels and the Aspect Ratio, the other values will be generated automatically.

You can specify a custom prompt, which will be used, if LLM is not selected.

You can specify a seed value. Open the seed node in the image input group to change the settings.

The workflow is based on "ComfyUI: Flux with LLM, 5x Upscale (Workflow Tutorial)".

I've optimized it for better readability (at least for me), shorter logic and combined main settings in one group. I left away the upscaling part.

My main motivation to do this model was the learning experience, how to work in ComfyUI.

Links for Models:

Flux.1 [dev]: https://huggingface.co/black-forest-l...

Flux.1 [schnell]: https://huggingface.co/black-forest-l...

t5xxl: https://huggingface.co/comfyanonymous...

ControlAltAI Nodes: https://github.com/gseth/ControlAltAI...

CivtiAI LoRA Used:

https://civitai.com/models/562866?mod...

https://civitai.com/models/633553?mod...

Ollama:

llama3.1: https://ollama.com/library/llama3.1

llava-llama3: https://ollama.com/library/llava-llama3

llava (alternate vision model): https://ollama.com/library/llava