EasyControlnet (Advanced)

This node and easy controlnetLoader++ do the same job - load a ControlNet checkpoint, feed it a conditioning image, apply it, hand back conditioning ready for your sampler - and they're close enough in field layout that it's worth just naming the one difference plainly: this one has no union_type dropdown. If you're loading a plain, single-purpose ControlNet (a dedicated canny model, a dedicated depth model, anything that isn't a multi-condition union checkpoint), that field would do nothing useful anyway, and this is the leaner version without it.

When to reach for this one instead

If you're on a modern union ControlNet - the kind that bundles canny, depth, pose, and a handful of other conditions into one checkpoint - use easy controlnetLoader++ instead, since you'll want that union-type selector to tell the model which condition your image actually represents. If you're loading anything else - an older single-condition SD 1.5 model, one of BFL's dedicated Depth or Canny tools for Flux, a purpose-built model that only ever does one job - this node covers exactly the same ground with one fewer field to think about.

The inputs that matter

pipe, image, and control_net_name are required, identical to the ++ version. control_net optionally takes an already-loaded model instead of loading fresh by name. strength, start_percent, and end_percent are the same ControlNet fundamentals as anywhere else - how hard it pushes, and across which part of the denoising schedule. scale_soft_weights applies that same gentler, layer-scaled influence curve instead of a flat strength multiplier, useful when a ControlNet is overpowering your prompt at a strength that would otherwise look about right.

Outputs are pipe, positive, and negative, wired the same way as the ++ variant - this node's output plugs into your sampler exactly the same way.

Installing it

ComfyUI Manager: search ComfyUI Easy Use, install, restart. By hand:

cd ComfyUI/custom_nodes
git clone https://github.com/yolain/ComfyUI-Easy-Use

Windows: double-click install.bat. Linux/Mac: run pip install -r requirements.txt yourself. Restart ComfyUI. The ControlNet checkpoint itself goes in models/controlnet - a separate download this node doesn't fetch for you.

Where people get stuck

Because this node and the ++ version look nearly identical at a glance, the most common mistake is picking this one for a union model and then wondering why results are inconsistent or condition-confused - without union_type, there's no way to tell a union checkpoint which condition group your image belongs to, so it's left guessing internally or defaulting in a way that may not match your intent. If that's your situation, switch to easy controlnetLoader++ rather than trying to work around it here. Beyond that, the usual ControlNet troubleshooting applies: strength too high relative to what your specific checkpoint's card recommends tends to fight the prompt, and start_percent/end_percent narrowed to only the early portion of the schedule is the standard move when you want the ControlNet to lock composition without dictating every later detail.