gguf-node test pack

locate gguf from Add Node > extension dropdown menu (between 3d and api; second last option)

setup (in general)

• drag gguf file(s) to diffusion_models folder (./ComfyUI/models/diffusion_models)
• drag clip or encoder(s) to text_encoders folder (./ComfyUI/models/text_encoders)
• drag controlnet adapter(s), if any, to controlnet folder (./ComfyUI/models/controlnet)
• drag lora adapter(s), if any, to loras folder (./ComfyUI/models/loras)
• drag vae decoder(s) to vae folder (./ComfyUI/models/vae)

run it straight (no installation needed way; recommended)

• get the comfy pack with the new gguf-node (pack)
• run the .bat file in the main directory

or, for existing user (alternative method)

• you could git clone the node to your ./ComfyUI/custom_nodes (more details here)
• either navigate to ./ComfyUI/custom_nodes first or drag and drop the node clone (gguf repo) there

workflow

• drag any workflow json file to the activated browser; or
• drag any generated output file (i.e., picture, video, etc.; which contains the workflow metadata) to the activated browser

simulator

• design your own prompt; or
• generate random prompt/descriptor(s) by the simulator (might not applicable for all models)

Prompt

anime style anime girl with massive fennec ears and one big fluffy tail, she has blonde long hair blue eyes wearing a maid outfit with a long black gold leaf pattern dress, walking slowly to the front with sweetie smile, holding a fancy black forest cake with candles on top in the kitchen of an old dark Victorian mansion lit by candlelight with a bright window to the foggy forest

Prompt

drag it <metadata inside>

Prompt

drag it <metadata inside>

booster

• drag safetensors file(s) to diffusion_models folder (./ComfyUI/models/diffusion_models)
• select the safetensors model; click Queue (run); simply track the progress from console
• when it was done; the boosted safetensors fp32 will be saved to the output folder (./ComfyUI/output)

cutter (beta)

• drag safetensors file(s) to diffusion_models folder (./ComfyUI/models/diffusion_models)
• select the safetensors model; click Queue (run); simply track the progress from console
• when it was done; the half-cut safetensors fp8_e4m3fn will be saved to the output folder (./ComfyUI/output)

convertor (alpha)

• drag safetensors file(s) to diffusion_models folder (./ComfyUI/models/diffusion_models)
• select the safetensors model; click Queue (run); track the progress from console
• the converted gguf file will be saved to the output folder (./ComfyUI/output)

convertor (reverse)

• drag gguf file(s) to diffusion_models folder (./ComfyUI/models/diffusion_models)
• select the gguf model; click Queue (run); track the progress from console
• the reverse converted safetensors file will be saved to the output folder (./ComfyUI/output)

convertor (zero)

• drag safetensors file(s) to diffusion_models folder (./ComfyUI/models/diffusion_models)
• select the safetensors model; click Queue (run); track the progress from console
• the converted gguf file will be saved to the output folder (./ComfyUI/output)
• zero means no restrictions; different from alpha; any form of safetensors can be converted; pig architecture will be applied for the output gguf

latest feature: gguf vae🐷 loader is now working on gguf-node

• gguf vae is able to save memory consumption of your machine
• convert your safetensors vae to gguf vae using convertor (zero)
• then use it with the new gguf vae loader
• same as gguf clip loaders, gguf vae loader is compatible with both safetensors and gguf file(s)

disclaimer

• some models (original files) as well as part of the codes are obtained from somewhere or provided by someone else and we might not easily spot out the creator/contributor(s) behind, unless it was specified in the source; rather let it blank instead of anonymous/unnamed/unknown
• we hope we are able to make more effort to trace the source; if it is your work, do let us know; we will address it back properly and probably; thanks for everything

reference

• sd3.5, sdxl from stabilityai
• flux from black-forest-labs
• pixart from pixart-alpha
• lumina from alpha-vllm
• aura from fal
• mochi from genmo
• hyvid from tencent
• wan from wan-ai
• ltxv from lightricks
• cosmos from nvidia
• pig architecture from connector
• comfyui from comfyanonymous
• comfyui-gguf from city96
• llama.cpp from ggerganov
• llama-cpp-python from abetlen
• gguf-connector (pypi|repo)
• gguf-node (pypi|repo|pack)