EDIT Workflow - base edit workflow for the gguf model of flux klein 9b-kv, includes 4 reference image slots. you can add more if you want by duplicating the image load image node and resize node than passing the conditioning for the positive and negative through the resize node and attach the vae to it and its connected! also includes the imagescaletototalpixel node on the end at a 4.0 value to upscale the image to 2k resolution as a final result. this just provides a higher quality generation with a nice resolution to work with.
TEXT TO IMAGE Workflow - Klein 9b-kv without the reference images in the workflow. YOU CAN GENERATE 4 STEP IMAGES WITH THE MODEL so i figured why not include a workflow for that as well.
REQUIRES KJ NODES (install in comfyui manager)
FILES:
GGUF model
https://huggingface.co/QuantStack/FLUX.2-Klein-9B-KV-GGUF/tree/main
fp8 model
https://huggingface.co/black-forest-labs/FLUX.2-klein-9b-kv-fp8/tree/main
text encoder
VAE
https://huggingface.co/Comfy-Org/flux2-dev/blob/main/split_files/vae/flux2-vae.safetensors
Description
Version 2 removes the reference images to utilize the models T2I capabitilies without editing. its still Klein 9b at the end of the day but its available to generate with