Nodes/ComfyUI-CozyGen/CozyGen Float Input
ComfyUI Node

CozyGen Float Input

A slider for CFG, denoise, and LoRA weights — on your phone

By gsusgg·Created about a year ago·Updated 11 months ago· 24
CozyGen Float Input
    • FLOAT
    param_nameFloat Parameter
    priority10
    default_value1.00
    min_value0.00
    max_value1024.00
    step0.10
    add_randomize_togglefalse

    CozyGen Float Input is the decimal sibling of CozyGen Int Input: one of the four "Static" nodes the CozyGen author broke out of the original all-in-one dynamic node so values would actually survive loading a workflow back into ComfyUI. Where the dynamic node guessed at types and stored everything in your browser cache, this one is typed FLOAT in the graph, which means what you set is what comes back when you reopen the workflow.

    It exists to expose one continuous number to the /cozygen web page - and for a control surface like this, a float slider is where the real fun is.

    The inputs that matter

    • param_name - the label on the web page.
    • default_value (1.0) - the starting point.
    • min_value (0) and max_value (1024) - bound the slider. The default max of 1024 is generous; tighten it to whatever range makes sense for the parameter you're exposing.
    • step (0.1) - how finely the slider moves.
    • add_randomize_toggle - adds a Randomize checkbox on the page.

    Output is a single FLOAT. The natural targets are the numbers people like to fiddle with from a phone: CFG scale, denoise strength, LoRA weight, or any strength multiplier.

    Where you'd actually use it

    The classic setup is exposing CFG scale or denoise on your img2img workflow - the two knobs you actually turn while iterating. Wire it into a KSampler's cfg or a VAEEncode/denoise path, and the page shows a slider instead of forcing you into ComfyUI's desktop graph. Combined with a randomize toggle on a seed, that's a complete "iterate from the couch" rig.

    Install

    Same as every CozyGen node. ComfyUI Manager (search "CozyGen"), or:

    cd ComfyUI/custom_nodes
    git clone https://github.com/gsusgg/ComfyUI_CozyGen
    cd ComfyUI_CozyGen
    pip install -r requirements.txt
    

    Restart ComfyUI. Add --listen to reach it from your network, and export the finished workflow with API into the pack's workflows/ folder. Then the slider shows up at http://<your-comfyui-ip>:8188/cozygen.

    A note on taste

    Don't expose every float in your graph - a page full of sliders stops being usable. Pick the two or three knobs you genuinely tweak per image and leave the rest as defaults inside the workflow. That's what the priority field is for too: 0 pushes a field to the top, so put CFG first and let the less-used ones sink.

    Thin node, clear job: one float, one slider, one wire. If that's what your phone workflow needs, this is the one to reach for.

    CategoryCozyGen/Static

    Inputs (7)

    NameTypeDefaultDescription
    param_nameSTRINGFloat Parameter
    priorityINT10
    default_valueFLOAT1.00
    min_valueFLOAT0.00
    max_valueFLOAT1024.00
    stepFLOAT0.10
    add_randomize_toggleBOOLEANfalse

    Outputs (1)

    NameTypeDescription
    FLOATFLOAT