ConDelta Scale (Multiple ConDelta by a Float)

The simplest node in this whole pack, and worth knowing about for exactly that reason: it multiplies a conditioning or ConDelta by a plain scalar. No addition, no baseline subtraction, no clamping, no strength range oddities - just output = input * scalar. Displayed as "ConDelta Scale (Multiple ConDelta by a Float)" in ComfyUI's node list (yes, the pack's own title has a typo - "Multiple" for "Multiply" - harmless, just don't let it confuse you into thinking it does something more elaborate).

Why you'd reach for it

Most of the interesting nodes in this pack - ConditioningAddConDelta, ApplyConDelta, the average nodes - already have a strength or a weight parameter built in, so you won't need this node for the common case of "apply a delta at some intensity." Where it earns its place is as a building block in a longer chain: pre-scaling one input before it goes into ConditioningAverageMultiple (which has no per-slot weighting of its own), halving a conditioning's magnitude before a manual subtract, or flipping a delta's direction outright by scaling with -1 instead of relying on a downstream node's strength parameter to go negative.

How it works

Multiply every value in the conditioning tensor by scalar. That's genuinely the whole operation.

The inputs and outputs that matter

• conditioning - anything: a plain conditioning, a ConDelta, an average, whatever you've got.
• scalar - default 1, step 0.01, no explicit min/max in the schema. 1 is a no-op, values above 1 amplify, values between 0 and 1 shrink, negative values flip direction.

Output is a single CONDITIONING.

How to install it

Via ComfyUI Manager: search "ComfyUI-ConDelta", install, restart. Manually:

cd ComfyUI/custom_nodes
git clone https://github.com/envy-ai/ComfyUI-ConDelta

then restart. No model files, no dependencies - it's a one-line math operation.

Common issues & troubleshooting

There's genuinely not much to troubleshoot here - it's the most predictable node in the pack. If something downstream looks wrong after inserting this, the scalar value is almost certainly the cause; check it first before assuming another node in the chain misbehaved.

Don't confuse this with ApplyConDeltaAutoScale's normalization. That node scales a delta relative to a base conditioning's own magnitude, adaptively. This node scales by a fixed number you choose, with no awareness of anything else in the graph. Reach for this when you want deliberate, repeatable control; reach for the auto-scale nodes when you want the strength to behave consistently across deltas of different sizes.

Extreme scalars produce the same failure mode as pushing any other strength parameter too far in this pack - burn, structural breakdown - once the scaled result gets added into an actual generation downstream. The node itself won't complain; the sampler will.