Comfyui Workflow (Image Generation)
the minified version of iGEN II
required Custom Nodes:
installed frontend is currently: 1.37.11
if you need help installing nunchaku or how to use the nodes check the readme files in the repo readme folder at github.
llama-cpp-python is needed for the loading_method: GGUF (llama-cpp-python) in the Smart LML Node
nunchaku wheels:
nunchaku model files:
Wildcards
Description
requires ComfyUI_Eclipse Version: 2.4.28
added a new node to use the prompt data from loaded images: pipe io generation data (gated),
changed smartloader+ v1/v2 and its pipe out node to populate configure_sampler
load image from folder has a new seed_input input to freeze the current loaded image until the seed changes e.g. in mode -4 (shuffle)
load image (metadata pipe) has a delete button now to cleanup comfys input folder from within the workflow
fix load image from folder/read prompt from files re-execution
what changed
both load image groups are able to load / use the prompt data if available, selectable with a fast muter (sampler settings, pos prompt, neg prompt and seed)
that is why the model loader also have to populate it's sampler settings because they have a higher prio because of model specific settings (when saved or enabled)
in the inital render group everything is collected and assigned (if not none)
to prevent errors both prompt inputs from the image are using dummies in case nothing is provided to keep the flow flowing
fixed the save behavior of negative prompt even when it's option is set to no
the negative prompts are collected inside the prompt group, together with "Input Prompts" disable the negative prompt option in the load image groups if you don't want to use the negative prompt from the loaded image.