camera
High-resolution RAW capture and video assist in stop motion
Capture technique / Dragonframe
Definition
High-resolution capture in stop motion separates the image used to animate from the image used to finish the shot. A low-resolution live-view or video-assist sequence is kept in sync with a full-resolution still sequence—often camera RAW—so the animator can judge movement quickly while post-production receives the highest-quality photographed frames, with scene/take/frame identity preserved across both records.
Overview
A stop-motion capture system has two different jobs that should not be confused. The animator needs immediate feedback: a responsive image that can be onion-skinned, toggled, scrubbed and played while the puppet is still reachable. The final picture needs something else: the highest-quality still the camera can record, with enough tonal and colour information for keying, compositing, grading and archival. Dragonframe's current camera-support documentation makes this split explicit. During a DSLR or mirrorless shoot it maintains two synchronized image sequences: low-resolution frames derived from live view for animation, and a high-quality still sequence—potentially camera RAW—for final output. The useful production principle is therefore not “animate on the RAW files.” It is to keep the fast working record and the finishing record tied to the same frame identity.
The difference matters because a frame can look acceptable in video assist and still fail as a final exposure. Dragonframe's own support notes warn that the two sequences are genuinely separate: if a hand enters the set after the video-assist frame but before the high-resolution capture completes, the animator's playback can look clean while the exported high-resolution frame contains the hand. That is a workflow lesson, not just a software quirk. Capture is not finished when the preview updates; the shot is safe only when the high-resolution exposure has completed and the system is ready for the next pose. Reviewing selected high-resolution frames during a long take is therefore part of photography, even when animation itself is judged from the faster assist stream.
A production example shows why the full-resolution record can carry information that JPEG-style preview frames cannot. Cinematographer Brent Yontz described photographing the stop-motion portion of *Nanuq* on a Canon Rebel XSi in RAW at 4272 × 1788. In his account, the additional highlight latitude helped on a film dominated by blue moonlight, while the extra captured information produced cleaner keys and composites; the RAW material was converted to DNG for archiving, composited at native resolution in After Effects, then converted to DPX for the film-out path and down-converted for HD distribution. Those are production-specific choices rather than a universal prescription, but they show the architectural value of retaining an original high-quality still record independently of the animator's viewing proxy.
The same separation scales beyond one camera station. Tom Livingstone's 2026 *Convergence* study of virtual-production R&D inside Aardman describes Dragonframe as a system that organizes data at the source while it captures animation, lighting states and motion-control information, carrying that structured record from the production floor into post. Dragonframe can also trigger local scripts after a high-resolution capture completes; its own documentation lists frame backup and RAW colour processing as intended uses. The important principle is durability of correspondence: scene, take, frame, exposure/pass and original camera file should stay traceable from capture through post. Fast video assist protects animation judgment; high-resolution originals protect image judgment. Treating one as a disposable substitute for the other weakens the pipeline in opposite directions.
Workflow
- Choose the final capture format before shooting. Where the camera and post pipeline support it, retain the camera's highest-quality still or RAW file rather than relying only on the video-assist image. Confirm storage, naming and downstream software support with a short test sequence.
- Set up the capture station so live view/video assist and the high-resolution still record stay synchronized. Verify that scene, take, frame and exposure/pass identifiers remain consistent across both file sets.
- Animate from the responsive assist sequence: onion skin, toggle, scrub and play it at the intended frame rate. Do not mistake the appearance of the assist frame for confirmation that the high-resolution exposure has finished.
- Wait for the capture-complete indication before touching the set again. During a long take, inspect representative high-resolution frames for focus, exposure, clipping, contamination, dead pixels or other faults that may not be visible in the assist stream.
- Preserve the original high-resolution sequence through post. If RAW files are converted to DNG, TIFF, EXR, DPX or another working format, keep the conversion reproducible and retain the original frame identity so a final pixel can be traced back to its camera capture.
- Automate repetitive integrity work where useful. A capture-complete hook can copy new high-resolution frames to a second location, launch an approved RAW-processing step, or notify production tracking without interrupting the animator's posing workflow.