Learn how to use FLUX, ComfyUI, and SDXL to create AI characters for movies, children's books, and virtual influencers or other projects!

Discover techniques for generating AI character sheets, training custom Loras, and achieving consistent results across multiple images and characters.

Perfect for creators looking to elevate their AI art and storytelling capabilities.

1 Install ComfyUI

1. Download ComfyUI: https://github.com/comfyanonymous/ComfyUI

2. Extract the folder to the location where you want to install ComfyUI.

3. install git https://git-scm.com/download/win

   • standalone version

4. Download ComfyUI Manager: https://github.com/ltdrdata/ComfyUI-Manager

5. Extract it, put it into your ComfyUI directory and run it.

6. After it’s done you can run ComfyUI by clicking run_nvidia_gpu

To help you with the installation, take a look at this short guide from MDMZ:  https://www.youtube.com/watch?v=E_D7y0YjE88

Congratulations, you installed ComfyUI! 🥳 To use my workflows you’ll need some extra models and extensions, but it is super easy to install them!

In the next step I’ll show you where to get and put the models. If you’re having issues check if you can just install them via the ComfyUI Manager. Just search for them and when you click “Install” they will be put in the right directory.

2 Character Sheets with SDXL

Get the workflow here 👉 https://www.patreon.com/posts/free-workflows-113743435

Drag and drop the .json into your ComfyUI Interface.

Go to ComfyUI Manager > Install missing custom nodes > install all of them > Restart ComfyUI

If you have problems with the Flux workflow, make sure you are using the latest version of the workflow! (v05)

Download the Models:

Download these models and put them in the corresponding folders:

1. Checkpoint: https://civitai.com/models/293331/wildcardx-xl-turbo

   📁 ComfyUI_windows_portable\ComfyUI\models\checkpoints

   You can also use any other SDXL checkpoints for all the workflows. I recommend Turbo or lightning ones though for faster generation time especially for the video workflows. You can find more models on CivitAI.com. Check the example images for the settings and match them in your KSampler!

2. SDXL CONTROLNET: https://huggingface.co/thibaud/controlnet-openpose-sdxl-1.0/tree/main

   (OpenPoseXL2.safetensors)

   and: https://huggingface.co/TheMistoAI/MistoLine/tree/main

   (mistoLine_rank256.safetensors) <- For the Advanced version!

   📁 ComfyUI_windows_portable\ComfyUI\models\controlnet\sdxl

3. UPSCALE MODEL: https://openmodeldb.info/models/4x-ClearRealityV1

   📁 \ComfyUI_windows_portable\ComfyUI\models\upscale_models

3 Create Character Sheets with FLUX

Get the workflow here 👉https://www.patreon.com/posts/free-workflows-113743435

Check out the official FLUX Installation guide for ComfyUI here:  https://comfyanonymous.github.io/ComfyUI_examples/flux/

1. Drag and drop the .json into your ComfyUI Interface.

2. Go to ComfyUI Manager > Install missing custom nodes > install all of them > Restart ComfyUI

   This should install “ComfyUI-eesahesNodes” from ComfyUI manager (contains InstantXFluxUnionControlNetLoader)

3. MODEL https://huggingface.co/black-forest-labs/FLUX.1-dev/tree/main

   flux1-dev.safetensors or https://huggingface.co/black-forest-labs/FLUX.1-schnell/tree/main

   📁 ComfyUI_windows_portable\ComfyUI\models\unet

4. CLIP  https://huggingface.co/comfyanonymous/flux_text_encoders/blob/main/clip_l.safetensors

   and

   https://huggingface.co/comfyanonymous/flux_text_encoders/blob/main/t5xxl_fp16.safetensors

   📁 ComfyUI_windows_portable\ComfyUI\models\clip

5. VAE https://huggingface.co/black-forest-labs/FLUX.1-schnell/blob/main/ae.safetensors

   📁 ComfyUI_windows_portable\ComfyUI\models\vae

6. CONTROLNET MODEL: https://huggingface.co/InstantX/FLUX.1-dev-Controlnet-Union/tree/main

You can get the models via the ComfyUI Manager:

You can also use the ShakkerLabs models, which are meant to be newer and improved versions of InstantX. Though in my testing, I actually got better results with InstantX.

https://huggingface.co/Shakker-Labs/FLUX.1-dev-ControlNet-Union-Pro/tree/main

1. UPSCALE MODEL: https://openmodeldb.info/models/4x-ClearRealityV1

📁 \ComfyUI_windows_portable\ComfyUI\models\upscale_models

4 Training a LORA

For Training I recommend using FluxGym via Pinokio.

You can also train Loras online via Replicate for example. Cost 2-5 Dollars for one Lora.