post
After Effects compositing
Capture technique
Definition
Adobe After Effects used as the primary compositing and finishing environment for stop-motion post-production: painting out in-frame rig hardware, keying puppets or set elements shot against a green or blue backing, and assembling separately captured depth layers into a single frame.
Overview
Every stop-motion shot is compromised by its own making: the tie-down bolts, support rods and armature mounts that hold a puppet in a pose are visible in the raw frame and have to be removed before the shot can be cut into a finished film. Content-Aware Fill for video, introduced in After Effects CC 2019, replaced the older frame-by-frame Clone Stamp and Pen Tool paint-out workflow: an artist masks the rig hardware across the shot, and the tool analyses the surrounding and adjacent frames -- wherever the same background is visible without the rig in the way -- to synthesise clean replacement pixels for every frame in the range, instead of requiring a hand-painted patch on each one individually.
Where a puppet or set element is shot against a green or blue backing rather than a built set -- typically an element meant to be combined with other layers later -- Keylight pulls the matte, Key Cleaner tightens the edge detail, and Advanced Spill Suppressor removes the backing colour that bleeds onto the subject's silhouette. It is the same core keying toolchain used in live-action VFX, applied here to elements photographed frame by frame rather than at a continuous video rate.
The third role is assembly rather than cleanup. Multiplane compositing recombines a foreground puppet layer, a mid-ground set layer and a background layer -- each captured as its own pass, sometimes at a different physical scale or distance from camera -- into one frame with parallax and depth cues that a single flat shot could not produce practically on a miniature stage.
After Effects sits directly downstream of capture: Dragonframe exports a shot as a DNG or TIFF image sequence, which After Effects imports as a composition rather than as a video file, preserving full frame-by-frame resolution and colour data. From there, Camera RAW processing evens out exposure drift across a long shot, and tools such as Neat Video reduce the frame-to-frame flicker caused by small sensor and lighting variance between individually triggered exposures -- a stop-motion-specific problem that does not exist in continuously rolling video capture.
A clean plate -- a reference frame of the bare set shot with no puppet or rig present, ideally bracketing the animated pass at both ends -- is what Content-Aware Fill for video actually draws its replacement pixels from when the surrounding frames themselves don't fully clear the rig. Shooting one is a set discipline, not a compositing step: it has to match the shot's lighting and exposure exactly, and it is cheapest when planned before the animated pass rather than improvised afterward. Where no clean plate covers the exact pixels a rig occupies -- a moving reflection, a highlight the rig crosses -- the fallback is manual clone-stamp work sampling texture and light from elsewhere in the same frame, the same technique tie-down planning is meant to make unnecessary in the first place.
Dragonframe's own capture menu has started formalising that discipline rather than leaving it to crew habit: its development branch (checked 2026-07-13) adds Capture | Shoot Clean Pass..., a dedicated command for shooting the bare-set reference directly alongside the animated take (see Dragonframe). The plate it produces is exactly the input this Content-Aware Fill pass needs -- the tool-level step and the compositing step are two ends of the same pipeline.
Aardman's Early Man (2018) supplies a named-production account of the same discipline from the clay side of the craft. VFX supervisor Howard Jones described rig and puncture-wound removal as the studio's baseline compositing pass -- "the simplest stuff we do" -- while drawing a line between clearing the rig itself and preserving the surface irregularities, the animator's own fingerprints and tool marks, that give clay-motion its handmade texture rather than smoothing them away. That distinction has a direct Content-Aware Fill consequence: the rig mask has to isolate the hardware precisely rather than cover the surrounding clay surface, or the fix erases the same handmade evidence the compositor is meant to keep.
Adobe has since automated part of the masking step that rig removal depends on. AI-powered Object Matte, added in the April 2026 release (version 26.2), isolates and tracks a subject with single-click selection instead of a hand-drawn mask, and the September 2026 release (version 26.5) added disk caching so a shot's propagated Object Matte results survive closing the project -- useful on a long stop-motion sequence where recomputing an automatic matte across thousands of individually exposed frames would otherwise cost time on every session. The June 2026 release (version 26.3) separately sped up the Mask Tracker up to 5x for the manual case, with tracking startable directly from the Timeline panel. Both remain isolation aids upstream of the same Content-Aware-Fill-or-Keylight decision above: a faster or more automatic matte still has to be precise enough not to erase the handmade surface detail Early Man's compositors were careful to preserve.
Workflow
- Import the Dragonframe-exported DNG or TIFF sequence for the shot as an After Effects composition, preserving native frame resolution.
- Identify and mask any in-frame rig hardware -- tie-down bolts, support rods, armature mounts -- across its visible duration.
- Run Content-Aware Fill for video on the masked region, letting the tool draw replacement pixels from reference frames where the same background is unobstructed, then review and hand-correct any frames where the fill misreads the motion.
- For elements shot on green or blue, key with Keylight and refine the edge and spill with Key Cleaner and Advanced Spill Suppressor before compositing onto the intended background.
- Stack foreground, mid-ground and background passes for any multiplane shot, aligning parallax by eye or with tracked reference points.
- Apply Camera RAW exposure levelling and flicker reduction (e.g. Neat Video) across the full shot before handing the composite to edit.
- For an automatic isolation pass, try AI-powered Object Matte first and enable disk caching on long shots so the result persists across sessions; fall back to manual masking and the accelerated Mask Tracker (After Effects 26.3+) wherever the automatic matte misreads rig hardware against a similarly coloured puppet or set surface.