Blender Input Integer

Every workflow eventually needs an integer you want to fiddle with - steps, CFG, batch size, which frame of a video sequence to render. BlenderInputInt turns that number into a proper slider/spinner on the panel the ComfyUI-Blender add-on generates inside Blender, so you can dial it from your 3D viewport instead of tabbing back to a browser. It's the integer sibling of BlenderInputFloat, and the only one in the pack with a couple of Blender-specific party tricks.

Mechanically it's a subclass of ComfyUI's built-in Int primitive node. In a plain ComfyUI graph it outputs whatever value you give it, and the rest of the inputs are the recipe for the control the Blender add-on renders. That split - dumb pass-through in ComfyUI, real UI in Blender - is the pack's whole design.

Inputs and outputs

The ones you'll actually set:

• value - the integer the node outputs. This is the live value, i.e. whatever the panel currently holds.
• default - starting value when the workflow loads.
• min / max - slider bounds in the add-on. You want a steps control? Set min 1, max 60. Don't leave these at the default ±2-billion range unless you enjoy a slider with no resolution.
• step - how much the spinner increments per click.

Then the Blender-specific pair, both booleans: camera_width and camera_height. Flip one on and the add-on shows a button that grabs the current camera's render resolution and writes the width or height straight into this field. This is the killer feature for image-to-image work in Blender: your output should match the camera, and this is the zero-typing way to keep it honest.

order does what it does on every input node - positions the control in the panel, lower first. There's an optional group input for nesting it inside a Blender Input Group box.

Output is a single INT that feeds anything wanting an integer downstream: KSampler steps, latent batch size, or an image size for the Blender Output Save Image node.

Installing it

Same routine as the whole pack. ComfyUI side, Manager → search "ComfyUI-Blender", or:

cd ComfyUI/custom_nodes
git clone https://github.com/alexisrolland/ComfyUI-Blender

Restart ComfyUI. No extra Python dependencies, but the README is blunt that you need a recent ComfyUI - the pack uses the custom-node schema v3 and won't load on older builds. The Blender add-on is a separate ZIP from the latest release, installed via Edit > Preferences > Add-ons > Install from Disk.

Where people get burned

The camera_width/camera_height buttons only appear if you export the workflow in API format and import it into the add-on - and they read the active camera, so if your scene has no camera or the wrong one active, the button writes nonsense. Two other habits worth adopting: name the node so the label reads sensibly (the title becomes the label), and when you run the workflow once in ComfyUI to confirm it executes before exporting, remember that the node's min/max sliders in Blender obey the values you hard-coded in ComfyUI - widen them there if your panel slider feels unresponsive. Running Blender and ComfyUI on one GPU, keep an eye on VRAM; the number field is light, the generation behind it is not.