simple latent upscale with no face fix or any other nodes. i think everything in this version is vanilla minus a few chroma specific nodes like the t5tokenizer and flow shift. feel free to swap to the full clip as well as i'm using a compressed version of that in this workflow. for locations of the clips used to generate this image: <a target="_blank" rel="ugc" href="https://huggingface.co/camenduru/FLUX.1-dev/blob/main/clip_l.safetensors">https://huggingface.co/camenduru/FLUX.1-dev/blob/main/clip_l.safetensors</a> <a target="_blank" rel="ugc" href="https://huggingface.co/city96/t5-v1_1-xxl-encoder-gguf/tree/main">https://huggingface.co/city96/t5-v1_1-xxl-encoder-gguf/tree/main</a> these both go inside of ../ComfyUI/models/clip

Files with SHA256: 76962d1a278624be2498ea669efa21032cf79c5c86595569b927ddbad9244cbe