Probe Modifier

Probe holds a reference to a source image and a sampling position (normalized X/Y or pixels). On each evaluation, it reads the pixel value (luma, R, G, B, A) at that position and uses it as the value of the target parameter.

It turns visual data into parametric values. Example: an audio waveform rendered as an image and sampled by Probe = visual sync with the audio without a dedicated audio node.

• Right-click a parameter → Modify With → Probe.
• Then connect the source image to the new Probe node.

Source image

Any image in the Fusion flow. Often a still or animated image that holds the information you want to use.

Sample Position

2D position to sample. Normalized 0-1 or in pixels. Animate it to sample different pixels over time.

Sample Channel

Channel to read: Red, Green, Blue, Alpha, Luminance. Depending on what you want to extract.

Output

A 0-1 scalar value (or more in float). Feeds the target parameter directly.

Glow driven by the luminance of an area:
Beauty image → Glow.Strength → Probe (Sample = specific position
                                      in the beauty,
                                      Channel = Luminance)
   = stronger Glow where the image is brighter

Audio waveform sync:
AudioWaveformVisualizer → Probe (Sample X = time-based,
                                Channel = Luminance)
                       → Scale.Y of an object
   = the object pulses to the rhythm of the waveform

Color picking from a reference:
ReferenceImage → Probe (sample center, Channel = R)
              → Color Corrector.Red
   = color dynamically linked to the reference

Mask modulation:
MaskGenerator → Probe (multiple positions)
             → Drives several parameters at once

Image (input)

Image to sample. Connect it from any image node in the flow.

Sample Position X / Y (defaults: 0.5 / 0.5)

Normalized 0-1 position (center = 0.5, 0.5). Can be animated to sample different spots.

Sample Channel (default: Luminance)

Source channel: R / G / B / A / Luma. Luminance to measure brightness, R/G/B for color picking, A to test for presence.

Sample Size (default: 1)

Size of the sampled area. 1 = a single pixel, >1 = average over an area (more stable against noise).

Sample Mode (default: Single Pixel)

Single = one point, Average = average of the area, Min / Max as needed.

• Audio-driven animation sync - render an audio waveform as an image and sample it to drive the Scale, Glow or Position of visual objects.
• Dynamic color picker - extract the color of a point in a reference image and spread it through the comp.
• Trigger based on a visual event - when the luminance of an area crosses a threshold → turn on an effect (with Probe + a threshold Expression).
• Reacting to scene luminance - a Glow / Bloom whose intensity follows the average brightness of the frame = a dynamic HDR-like effect.
• Driving from an external generator - Fast Noise rendered as an image, then Probe = a curve generated procedurally by the noise.
• Multiple parameters from a map - a "control map" image sampled by several Probes at different positions = centralized multi-parameter control.

• Sample Size for stability - on a noisy image, a single pixel gives an unstable reading. Sample Size 3-5 = a more stable average.
• Normalized position - 0-1 is resolution-independent. Better than pixels for a resizable comp.
• Animate the Position to scan - Sample Position animated linearly from 0 to 1 = sweep the whole image over time.
• Float > 1.0 is possible - on HDR images, the value can exceed 1. The target parameter gets the raw value; clamp it with Expression if needed.
• Combine it with Expression - Probe reads, Expression transforms. E.g. n1 * 100 where n1 = the Probe output, to map 0-1 to 0-100.
• Moderate performance - Probe reads an image on each evaluation. Cache the source image if it is heavy to compute.
• For multiple samples, create several Probe modifiers (one per target parameter) rather than a single shared one.