camera
Live view versus the high-resolution frame
Dragonframe / Capture technique
Definition
In digital stop-motion capture the animator works from two different pictures of the same set: a low-resolution live view (video assist) streamed from the camera for posing, onion-skinning and frame-to-frame comparison, and the full-resolution still that the camera writes when a frame is shot. The two are produced by different paths through the camera, so focus, depth of field, exposure and even framing can differ between them, and the high-resolution frame, not the live view, is the image that reaches the film.
Overview
Every frame of a digitally captured stop-motion shot exists twice before it is exposed. The animator poses the puppet against a live view: a small, fast-refreshing video image the camera streams to the capture software, which Dragonframe calls video assist and uses for onion-skinning, toggling and step-through against earlier frames. When the frame is shot, the camera writes a different picture altogether: a full-resolution still made at the shooting aperture and exposure. The craft point, and the reason this page exists, is that the two are not the same image at two sizes. They are made differently, and the gap between them is something a crew manages deliberately rather than a fault to be fixed.
Focus and depth of field show the gap most clearly. Dragonframe's own documentation explains that with an electronically controlled lens the camera holds the aperture wide open during live view, so the animator sees a shallow depth of field and a narrow focus range, while the high-resolution capture is made stopped down (the example given is f/22) with far deeper focus. Its listed remedies are telling: accept the difference, switch to a manual-aperture lens on an adapter, or turn on depth-of-field preview on Canon bodies (in Dragonframe 5 and later, in the Camera Settings section of the Cinematography workspace). Canon's paid Stop Motion Animation Firmware attacks the same problem from the camera side: on supported bodies it doubles the video-assist resolution from 960x640 to 1920x1280, adds focus peaking for RF and RF-S lenses set to manual focus, and, with Dragonframe 5, an aperture lock that keeps the lens closed down through live view and capture (see [[lens-selection-stop-motion]] and [[focus-pulling-stop-motion]]).
Exposure and framing differ too. Many Nikon bodies have no live-view exposure simulation, so the live image brightens or darkens on its own and ignores setting changes; Canon bodies simulate exposure, but a manual Nikon lens on a Canon body, Dragonframe's own recommendation for avoiding aperture flicker, can throw that simulation off, which is what the Exposure Preview Offset setting exists to correct. On a dim, stopped-down set the simulator can simply run out of range. Dragonframe's answer there is the bash light: a work light that is on for the animator and the video assist and off for the exposure, switched through a DDMX-512 or a relay, or timed with a capture delay (see [[ddmx-512-dmx-lighting-control]]). Framing is the one gap with no full fix: the documentation says plainly that the live view and the stills are not framed identically, and suggests only that a different image size, Large or Medium JPEG, may sample a section of the sensor that sits closer to the live view.
The position this book takes is that live view is an animation tool, not a cinematography tool, and a production should treat it that way. Posing, arcs and spacing are judged against video assist, because it is fast and it is what onion-skinning compares. Focus, exposure, colour and composition are judged against high-resolution test shots (see [[dragonframe-test-shots]]), which is why Dragonframe separates the Animation and Cinematography workspaces and lets the animator switch the Animation view to high-res review. The final look belongs later still: the live view from a stills camera is an 8-bit sRGB JPEG that cannot carry a LUT, so the grade is applied to the RAW files in post. On-set colour management only keeps the monitor honest (see [[dragonframe-color-management-ocio]]); it does not turn the live view into the frame.
Workflow
- Set the high-resolution capture (image size, RAW/JPEG, aperture, shutter, ISO, white balance) in the Cinematography workspace first; this is the image the film is made of.
- Take a high-resolution test shot and judge focus, exposure and composition on it, not on the live view.
- Compare the live view against that test shot. Close the gap where the camera allows it: depth-of-field preview or the Canon firmware's aperture lock for focus, Exposure Preview Offset for brightness, a different image size for framing.
- Where the gap cannot be closed, light for the animator separately: rig a bash light that is on for posing and off for the exposure, through DMX, a relay or a capture delay.
- Animate against video assist, using onion skin and toggling for spacing and arcs.
- Switch the Animation view to high-res review at intervals, and whenever lighting, focus or the camera changes, so a drift is caught in frames rather than days.