Scale Image Values

This is the smallest node in the pack by a wide margin - one input, one output, zero parameters to configure. Its job, going by its own description: take an IMAGE with values in any range at all - "depth in meters, HDR linear RGB, arbitrary scalars broadcast across 3 channels, etc." - and hand back something scaled to whatever range the node considers usable. That's a real need in this specific pack, more than most: SHARP passes depth in meters, raw Gaussian attribute channels, and other genuinely non-pixel data through ComfyUI's IMAGE socket, which was designed for 0–1 color data. A Preview Image node, or anything downstream expecting a normal image, needs that data remapped into something displayable first - that's this node's whole reason to exist.

What it actually does - and what it doesn't tell you

Worth being upfront here: because there are no configurable inputs at all beyond the image itself, this article can't tell you exactly what scaling rule it applies - whether it's a min-max normalize to 0–1, a fixed divide, a clamp, or something else. There's no min/max/mode field in the node's own schema to hint at it, and guessing at the internals rather than reading them off the interface isn't something to do here. What you can rely on is the shape of the contract: any-range image data in, one image out, no way to steer how the remapping happens from inside the node itself.

The inputs and outputs that matter

• image (required, IMAGE) - any value range. The node's own tooltip explicitly calls out depth-in-meters and HDR linear RGB as example use cases, plus "arbitrary scalars broadcast across 3 channels" - a description that reads like it was written with this pack's own depth and Gaussian-attribute outputs specifically in mind.

Output: image (IMAGE) - the scaled result.

Installing it

Part of ComfyUI-Sharp: Manager search "Sharp" (install highest, try nightly if that build acts up), Manager → Install via Git URL with https://github.com/PozzettiAndrea/ComfyUI-Sharp.git, or manual clone into custom_nodes + pip install -r requirements.txt --upgrade + python install.py, restart.

Common issues

Given there's genuinely nothing to tune here, there's not much to troubleshoot either - if the output doesn't look the way you expected, there's no setting on this node to adjust. Two practical options if the automatic scaling isn't giving you what you need: pre-scale the data yourself before it reaches this node, or reach for a node with actual min/max/clamp controls if your use case needs more precision than a fixed rule can offer. This is a convenience utility for a quick preview, not a general-purpose remapping tool - treat it that way and it does its job fine.