fabrication
Duplicate puppets and asset replication for parallel stages
Stop-motion puppet maintenance and on-set repair
Definition
Duplicate puppets are camera-matched copies of an approved stop-motion character built so more than one stage can shoot that character at the same time and so a damaged or serviced puppet does not halt production. The useful production unit is therefore not just the single “hero” puppet but a controlled family of copies whose scale, mechanics, surface, costume and interchangeable parts remain close enough that shots cut together as one character.
Overview
In a large stop-motion production, duplication is primarily a **throughput system**. A single perfect puppet cannot feed dozens of shooting units if several sequences need the same character at once. Focus Features’ production account for *ParaNorman* describes 52 shooting units and identifies the first completed, approved puppet of each character as the “hero” from which duplicates are then reproduced. Norman alone existed as 28 full-body puppets. The reason is not spectacle: parallel stages turn one character design into a shared production resource, so fabrication must provide enough matched copies for the schedule rather than treating the approved puppet as a one-off sculpture.
Replication also changes what “matching” means. The copies must preserve the properties the camera and animator actually depend on: overall scale, silhouette, joint range, neutral posture, face-part fit, costume proportions, hair volume and visible finish. *The Boxtrolls* provides the same pattern at a different production scale: Animation World Network reports 28 Eggs puppets and roughly 15–20 copies for each main Boxtroll, within a 190-puppet production. The practical constraint is that a stop-motion character cannot be duplicated by software. Every additional shooting unit needs a physical asset occupying real stage space, and every copy can accumulate wear differently. Duplication therefore connects directly to maintenance, continuity photography and part tracking.
Aardman’s *Turtle Journey* shows why copies are not automatically identical merely because they came from one design. The studio built three copies of each turtle character, and their hand-painted shell patterns had to be replicated across those copies. That is a useful boundary: some variation is an unavoidable result of hand fabrication, but variation that changes the apparent identity of the character can become a continuity problem when shots intercut. Productions therefore need a camera-approved reference state and a repeatable finishing process, not simply repeated moulds or repeated digital files. The same logic applies to costumes and props; the *ParaNorman* production kept multiple costumes because thousands of handling events would otherwise make one outfit a single point of failure.
Duplication has limits, and those limits should be planned rather than hidden. Aardman’s *Farmageddon* directors describe a production with 35 shooting units and emphasize that stop-motion cannot simply add more animators indefinitely because there are only so many stages and physical resources. They also note that 3D printing is useful when multiples of a prop must match. That suggests the correct hierarchy: duplicate the assets that constrain parallel work, use reproducible processes where dimensional matching matters, and reserve scarce hand-finishing effort for the surfaces the camera reads. The goal is not perfect industrial sameness. It is to make several physical copies behave as one screen character while keeping enough redundancy that one repair, one reset or one occupied stage does not stop the schedule.
Weekly episodic production applies the same throughput logic to volume rather than to one recurring character. Stoopid Buddy Stoodios’ ‘’Robot Chicken’’ — which needs a fresh cast of gag characters every sketch rather than a single continuing hero — solves this by duplicating the puppet body, not the finished character: head of puppet fabrication Caroline Kastelic has described nine stock body types, molded and cast repeatedly from standing molds, that are then customized with a sculpted head, costume and paint into whatever character a given sketch needs. Animation World Network reports roughly 100 puppets built per episode, around 2,000 across a season, produced on staggered six-week fabrication cycles that overlap several episodes at once so the schedule never depends on any single finished puppet. It is close to the inverse of ParaNorman’s hero-and-copies model: instead of duplicating one approved design to feed many stages, the stock body is the reusable asset, and the one-off work is concentrated in the parts the camera actually needs to read as a new character (see Mould-making and casting for stop-motion puppets).
Commercial and branded stop-motion work sits at the small end of the same replication range documented above. A+C Animation Studios -- a UK studio fabricating puppets for shorter-turnaround branded films and campaigns (Peperami, LEGO, a BBC Sport title sequence) alongside longer-form IP -- publishes its own duplicate-count guidance for clients budgeting a build: three to six duplicates per hero puppet, plus separate "service spares" (interchangeable hands, heads and costume sets) sized to the character's screen time and the amount of physical stunt work the shots require. That is roughly an order of magnitude below ParaNorman's 28 Norman puppets or Robot Chicken's stock-body volume, which fits the shorter shoot and smaller crew a commercial spot runs on -- fewer parallel stages need feeding, so fewer copies earn their fabrication cost, and the studio treats "spares" for interchangeable parts as a distinct, cheaper line from a full hero duplicate.
At the far end of the range sits the production with no duplicates at all, and it shows the rule most clearly. *Tennis, Oranges*, Sean Pecknold's self-funded short made at Sing-Sing in Los Angeles, had one set of puppets and one shooting stage. Pecknold says that meant only one set could be worked on at a time, "and because we only had one set of puppets, that's the best we could do" (Dragonframe, 2026-10-20). The two puppets limited the production as much as the stage did: even with a second stage there would have been nothing to put on it. A one-off hero puppet is therefore a scheduling decision as well as a budget one. It commits the production to shooting in sequence, and makes every repair a pause for the whole film.
The economics of a duplicate run sit in the gap between the first copy and the rest. LAIKA head of animation Brad Schiff, interviewed at TIFF about *Wildwood* (Exclaim!, 2026-10-21), estimates that the first puppet of a character takes about a year to create, moving from concept art and 2D drawing through clay and digital sculpts, moulds, an armature and a silicone body; later duplicates take less time because that engineering is already solved. The film then ran to 36 Prue puppets and 30 Curtis puppets for its two young leads. Read beside *ParaNorman*'s 28 Normans, the figure shows the pattern holding at a larger scale: the year is spent once on the approved hero, and every later copy is bought for schedule capacity, not design.
Workflow
- 1. Approve one hero/reference puppet under the real camera and lighting before multiplying it. Record the dimensions, neutral pose, joint limits, face-part fit, surface finish, hair and costume state that later copies must preserve.
- 2. Derive the required copy count from the shooting schedule. Identify how many stages may need the character simultaneously, then add realistic capacity for maintenance, repair and reset rather than assuming every copy is always available.
- 3. Separate repeatable geometry from hand-finished appearance. Use moulds, jigs, measured armature parts, printed components or other controlled processes where interchangeability matters, while defining visual reference checks for paint, hair, fabric and other hand-made layers that can drift between copies.
- 4. Give each copy a stable identity and track where it is assigned. A production with many stages should know which puppet, costume and replacement-part set is on which unit so wear, repairs and continuity differences can be traced rather than discovered in editorial.
- 5. Compare duplicates against the hero/reference and against one another before they enter a shot. Check scale, silhouette, joint behavior, costume position, hair volume and camera-visible finish; correct meaningful drift before animation rather than during the take.
- 6. Feed maintenance history back into later copies. If a joint, skin area, costume closure or paint treatment fails repeatedly, update subsequent duplicates or spares instead of reproducing the same weakness simply because it existed on the first approved puppet.
- 7. Retire or reassign a copy when it no longer matches the approved performance envelope. Redundancy only protects the schedule when the available copies remain trustworthy enough to intercut.
- 8. Budget the hero and the duplicates as different costs. The first approved puppet carries the design and engineering time, which LAIKA's Brad Schiff puts at about a year per character on *Wildwood*. Later copies reuse the solved moulds, armature and finish, so the copy count can be set by stage demand rather than by the cost of the first build.