sJitter

sJitter exists to remove machine-perfect regularity from the Shape system without rasterizing early. It receives an existing shape or array, modifies its geometry, and still outputs Shape data.

Blackmagic explicitly defines the data path: Input1 is a required orange Shape input. It is not an image input, a mask, or a regular RGBA cable, so MediaIn is not the correct source for sJitter.

A common production path is sEllipse / sRectangle / another shape node → sGrid or sDuplicate → sJitter → sRender. The branch remains vector Shape data until sRender converts it into pixels.

sJitter is most often used on arrays built by sGrid or sDuplicate. In that context, individual elements can receive different random offsets, sizes, or rotations instead of the entire array moving identically.

The node also works on a single shape. That is especially useful for Point Jitter, where vector control points can be perturbed to create a distressed or wobbly contour.

Jitter Mode controls temporal behavior. Fixed keeps the generated variation static unless you animate the controls yourself with keyframes or modifiers.

Random enables automatic animation of the jitter parameters inside their range sliders. If every range remains neutral, changing only the mode to Random does not create a visible effect.

Fixed and Random are therefore not low/high strength modes. Their main distinction is whether the randomized state stays static or changes automatically over time.

Random Seed chooses a particular pseudo-random state or sequence. The captured Controls screenshot shows 356; that is evidence for the saved UI state, not a claim about a universal factory default.

Reseed provides a fast way to choose another random seed without typing a value manually. It is useful while auditioning arrangements; once a layout is approved, lock the seed and stop reseeding accidentally.

Shape X Offset and Shape Y Offset are position ranges. On an array, values are randomized within the specified intervals, so different shapes no longer sit on perfectly regular coordinates.

The captured screenshot shows both Shape Offset ranges at 0.0–0.0. That is the neutral state in this capture, so no positional variation is present there.

Shape X Size and Shape Y Size define horizontal and vertical scale ranges for individual shapes. The captured 1.0–1.0 ranges preserve the incoming size.

Different X Size and Y Size ranges can alter proportions as well as overall size. That matters for circles, icons, and logo-like geometry where distortion may be undesirable.

Shape Rotate randomizes the rotation of each shape in an array. A perfect circle can hide the effect visually, so use a rectangle, star, line, or another asymmetric shape when diagnosing this control.

Point X Offset and Point Y Offset do not translate the whole object. They work on vector control points and distort the contour itself, producing distressed, hand-drawn, or wobbly geometry.

Shape Jitter and Point Jitter solve different problems. Shape Jitter varies each element's transform while preserving its internal geometry; Point Jitter changes the geometry itself.

Do not raise both blocks together during diagnosis. Prove Offset/Size/Rotate on an array first, return to a known baseline, then add Point Jitter separately.

sGrid before sJitter is useful when a two-dimensional matrix should remain recognizable but imperfect. sDuplicate is useful when copies already follow a more complex transform pattern and each copy needs extra variation.

When the design must remain recognizable, start with narrow ranges. Jitter can quickly destroy spacing and silhouette when offset, scale, rotation, and points all vary at once.

Always judge Random Mode in playback rather than on one still. Auto-animation can look harmless in one frame yet produce excessive twitch, flicker, or distracting silhouette changes over time.

Fixed Mode is better for repeatable static layout variants: choose a seed, approve the composition, and animate only the controls that truly need motion.

Do not promise that one numeric seed will produce an identical pattern across every Resolve version. For templates that move between builds, keep a visual reference and perform a live regression check.

sRender is a separate pipeline boundary. It converts Shape vectors into a raster image, so edge quality, motion blur, and final alpha should be judged after sRender rather than inferred from Shape data.

A production QC pass for sJitter should verify Shape → sJitter → sRender topology first, then Mode, seed, and one range at a time, followed by 100% raster inspection and playback for Random animation.