A character render can win a presentation before anyone knows how the character will share a scene with a person. That gap matters when a brand commissions production through a separate agency, fabrication supplier and post team: each may deliver its own approved component without delivering a convincing exchange.
Before choosing a puppet, CG, generative AI or a hybrid, define one interaction the campaign cannot do without. Then ask how the proposed production process will make it happen—and change it.
There is a timely reason to revisit this. In its October 6, 2026 announcement, Plymouth Rock described practical effects, including a seven-foot animatronic bear, in its latest campaign. The rollout began October 5 in Massachusetts and New Jersey. That establishes a choice on one project, not a market-wide retreat from digital production or evidence of a cost advantage.
When the character looks right but the scene feels wrong
Consider a fictional travel-accessories film. A small invented character proudly occupies the space where a traveler needs to pack. The traveler offers it a place beside an organizer; the character notices, hesitates, moves, and watches the packing continue. This is a proposed scene, not an ONCE client case. The real organizer's dimensions, capacity and use would need approval; the gag must not imply magical compression.
A technically impressive character could still move before noticing the invitation. The traveler could reach into the scene without waiting for a response. Neither problem necessarily calls for a better surface texture. The missing element is a readable reason for each action.
A useful production reference comes from SCPS Unlimited's Pringleleo case page. It describes design, fabrication and puppeteering developed together, and a character designed for controlled posing rather than mechanical animation. This is the maker's account, not an independent assessment of effectiveness. Its useful lesson is narrower: the production design was connected to the performance requirement, rather than treating maximum articulation as the objective.
The ONCE Scene-Exchange Brief
Our proposed method starts with four short statements, not a technology preference:
- Cue: What does the person do that the character can notice?
- Response: Where does the character look, what follows, and which pause matters?
- Handoff: In the completed scene, what visible response allows the person to continue?
- Payoff: What does that exchange make clear, and why does the brand belong in it?
For the fictional packing scene, the invitation is the cue. Looking, hesitating and moving form the response. The traveler waits for the space to clear before packing. The payoff is a change from competing for space to making room—not an unverified claim about packing capacity.
Keep the finish rough, but preserve the relationship
A rehearsal can omit final materials or a finished environment. It should not omit the relevant scale, eyelines, spacing and action order.
For a puppet proposal, an appropriately sized stand-in and performer can explore the exchange. For CG, a rough character and live-action reference can establish timing. For a generative proposal, ask the team to demonstrate the passage through the workflow it actually proposes to use, including how the character and human footage will be obtained and joined.
Label what has not been demonstrated. A frontal greeting does not establish a complex handover, a turn behind a person or interaction with a real product. Equally, a campaign that only needs a greeting should not be charged for developing a chase sequence nobody has commissioned. The brief is a production diagnostic, not an audience study or proof of future sales. Physical interaction and machinery require separate, competent safety planning.
Compare the same revision
After the first passage, request one plausible change: the traveler's invitation arrives later, but the character must still notice it before moving.
Ask each team what that change entails. Does it require another rehearsal and shoot, animation adjustments and compositing, or regeneration, selection and reassembly? Which approved elements remain usable? Who needs to participate?
Do not judge a rough puppet against a polished digital finish. Compare completion of the same dramatic task and the work needed to revise it. None of these routes is declared the cheapest in advance; the supplier must scope its actual process.
Commission continuity, not just an object or a file
Practical production does not remove post-production. In its March 11, 2026 LNER project account, UNIT describes a physical Eleanor puppet shot against green screen, followed by compositing and puppeteer removal. That is evidence of a hybrid workflow, not a general cost comparison.
For an overseas brand working with production partners in China, this makes the purchasing question more precise. A fabrication quote can answer how the object will be made; it does not, by itself, answer who will develop the performance, direct the exchange or finish the composite. Keep those responsibilities visible when comparing proposals.
For a recurring character, retain approved proportions and eyelines, playable action examples, movements it should not perform, voice and timing references, and the relevant production versions. Retain rejected examples with reasons too: “nods before listening” is more useful to a future team than a folder marked “bad takes.” Agree separately what can be reused or adapted; possession of a model, puppet or voice file is not a complete usage arrangement.
For ONCE, the strategic opportunity is to build a repeatable way of translating personality into playable actions and explaining revision dependencies. That is our business analysis, not a claim about existing client outcomes or asset valuation.
If the team cannot explain what the character contributes beyond looking appealing, revisit the script before commissioning a full character system. Start with one exchange worth watching. Let that determine how much character development the campaign actually needs.
Sources checked October 18, 2026. The SCPS case page does not display a publication date. This article draws on written production accounts, not a shot-by-shot viewing of the films. The cover is an original AI-generated concept scene, not a photograph of these projects or an ONCE client shoot.
