Clay animation and claymation AI video prompts in Seedance require a different directorial vocabulary than live-action or 2D animation. The model needs to know three things simultaneously: what the plasticine characters look like (surface texture, proportions, visible fingerprints), what physical laws govern the miniature world (squash-and-stretch on landing, glossy sauce behavior, stop-motion micro-jitter), and what the camera is doing frame-by-frame through that handcrafted space. Getting even one of these wrong produces output that reads as generic 3D animation rather than the tactile, imperfect, physically-grounded aesthetic that makes claymation distinctive.
The five Seedance clay animation prompts below cover five different structural problems in claymation direction. The dim sum chase sequence shows how specifying a behavioral rule for the antagonist — compresses and lunges, always just misses — produces a repeatable comedic engine across six shots. The Tim Burton-inspired horror town shows how a no-cut single-shot constraint forces the model to plan environment changes without editorial breaks. The Western kitchen comedy demonstrates how writing the script as a causality chain (stunt generates next stunt's setup material) produces an escalating sequence that keeps building past its apparent resolution. The clay talk show shows how borrowing a recognizable real-world format as the container lets the absurdist clay content carry all the weight. And the clay diorama charge sequence shows how rendering a human experience as a hardware specification — charging port, battery percentage — transforms a universal emotion into a physical object the miniature camera can observe.
1. The plasticine chase — behavioral rule as repeating comedic engine across six shots
See the full prompt on scenic.sh →
"the front constantly opening into a huge soft mouth... the creature closely follows, knocking over a soup noodle stand, broth spilling widely across the ground — always just missing."
Why this works: At 147 likes, this prompt's most transferable technique is not its shot list — it is the behavioral rule established in the monster's character description: "constantly opening into a huge soft mouth" and "lunging forward but always just missing." This rule fires automatically across every shot without re-instruction. In Shot 1, the creature misses. In Shot 2, the broth-slicked ground creates a new hazard, but the creature still misses. In Shot 4, it almost reaches the girl's back "but still fails to touch." The dramatic engine (pursuit + near-miss) never requires the prompt to say "miss again" — the behavioral rule carries the structure.
The squash-and-stretch physics are encoded per-shot at the action level, not as a global style note. Shot 3: "slides briefly then immediately compresses and rebounds." Shot 4: "slams into the roof, flattens on impact, then instantly rebounds." Each landing event specifies the compression and the subsequent rebound — the two phases of plasticine squash-and-stretch are named individually at each collision. A global "use squash-and-stretch" note at the top of the prompt would have produced inconsistent application; per-shot physics at the action level ensures the model knows exactly when and how the material deforms.
The Chinese street environment is specified as a single deep spatial axis (dim sum shop → vendor area → lantern alley → roof drying area → teahouse facade → balcony) with a connecting physical feature in each zone (cloth rope, soup noodle stand, rooftop gap). Each zone provides one stunt resource. The spatial sequence is the plot: the girl moves through resources, the creature triggers them, and the sequence ends at the balcony — a terminus with no further resources, which is why the last shot can be a near-bite cliffhanger.
The takeaway: establish a behavioral rule for your clay antagonist in the character description, not in individual shot notes. "Always just missing" is a rule that propagates across the entire chase without re-instruction. Specify squash-and-stretch at the action level per landing event ("compresses on landing and instantly launches," "flattens on impact, then instantly rebounds") rather than as a global style instruction. Map each environment zone to one stunt resource and sequence the chase as movement through those resources.
2. The no-cut single-shot claymation — environment as psychological actor in 15 continuous seconds
See the full prompt on scenic.sh →
"Gigantic human fingers briefly appear far above the miniature town, adjusting one rooftop like a stop-motion animator between frames. The characters stare upward in existential horror."
Why this works: At 32 likes, the structural engine of this prompt is the "15-second continuous single-shot. No cuts." constraint at the top, which forces the model to plan the entire sequence as one unbroken take rather than as cuts between scenes. In a multi-shot sequence, atmosphere can be established in Scene 1 and dropped in Scene 3; in a continuous shot, every atmospheric element must be introduced and maintained through camera motion alone. The walls bending, streetlights wobbling, and windows sliding all happen within the same locked environment — the town's physical fabric shifts, but the camera's relationship to the scene does not.
The environment's psychological responsiveness — walls that bend when characters become anxious, a town that distorts as tension builds — is the claymation-specific technique at the center of this prompt. In stop-motion animation, the animator's hands are always present between frames; objects move because someone moved them. This prompt turns that physical fact into narrative content. The "gigantic human fingers briefly appearing far above the miniature town, slightly out of focus, adjusting one rooftop like a stop-motion animator between frames" is not a horror reveal — it is the physical reality of the medium made visible as the punchline to the gothic figure's question: "Do you ever feel like someone is moving us?"
The Tim Burton character design vocabulary ("oversized glasses," "pale gothic figure in a black suit," "enormous eyes," "fingerprints visible in the clay, fabric textures, tiny imperfections") grounds the existential content in tactile material specificity. A character described as "clay figure" produces a generic result; a character whose clay surface has "fingerprints visible" and whose clothes have "real fabric textures" signals the hand-made material register that distinguishes claymation from CGI.
The takeaway: use the no-cut single-shot constraint to force continuous environmental planning rather than scene-by-scene construction. In a 15-second single shot, the environment must carry all the atmospheric shifts without cuts — write those shifts as physical events (walls bending, windows sliding) not as new scenes. Specify the claymation material register per character (fingerprints in clay, fabric textures, tiny imperfections) to separate the output from generic 3D animation. Use the medium's physical reality (the animator's hands exist between frames) as narrative content.
3. The prop causality chain — each stunt generates the next stunt's hazard
See the full prompt on scenic.sh →
"she slides on spilled gravy, swings from tavern signs... just as she thinks she has won, the lid rattles open and dozens of tiny armored biscuit-bandit dumplings climb out in formation."
Why this works: At 6 likes but with a fully articulated structure, this prompt is written as a physical causality chain where every hazard is generated by the previous stunt's byproduct. The cauldron "sprays boiling steam and broth" — and that broth becomes the "spilled gravy" the girl slides on in the next beat. The market stalls it "flips" become the surfaces she "swings from" using tavern signs. The food inside becomes the "food projectiles" she launches. Each destructive action by the antagonist produces the next scene's interaction resource for the protagonist.
This causality chain structure removes the need to invent discrete scene transitions. In a typical action comedy, you write Scene 1 (kitchen), Scene 2 (street market), Scene 3 (rope bridge) as separate set pieces connected by cuts. In a causality chain prompt, the cauldron's path through the frontier town generates each new location's destruction and the debris that becomes the girl's next tool — the scene transitions are embedded in the cauldron's momentum rather than specified as separate scene headers.
The resolution beat — the girl wedges the cauldron under the arch — is written as an apparent conclusion that the prompt immediately undercuts. "Just as she finally wedges the cauldron... and thinks she has won, the lid rattles open and dozens of tiny armored biscuit-bandit dumplings climb out in formation" extends the causality chain one step past its apparent end. The resolution generates a new problem. This escalation-past-resolution structure is the comedic engine: the chain never truly closes.
The claymation-specific descriptors ("colorful Western-style stop-motion plasticine animation, with dusty textures, glossy sauces, bouncy squash-and-stretch motion, visible hand-shaped clay details, and dense handcrafted frontier town environments") are stacked in a single list rather than distributed through the scene descriptions. This concentrates all material specification in one declaration that applies globally, freeing the scene-by-scene text to carry only the physical action.
The takeaway: write the clay action comedy as a causality chain where each stunt generates the next stunt's resource. Cauldron broth → spilled gravy → girl slides. Market stall wreckage → swinging surface → girl launches food. The chain removes the need for explicit scene transitions. Add one escalation beat past the apparent resolution — the moment when the antagonist opens to reveal a new threat. Stack all claymation material descriptors in a single global list and let the scene descriptions carry only the physical action.
4. The clay talk show — borrowing a recognizable format as the claymation container
See the full prompt on scenic.sh →
"EINSTEIN: 'In my professional opinion… stupidity is a fundamental force of the universe.'... The CRT TV flickers ON — static, then an orange cat fighting suited hyenas in kung-fu chaos."
Why this works: At 2 likes but with precise structural engineering, this prompt uses format borrowing as its primary technique — the familiar late-night talk show format provides the entire staging, character dynamic, and pacing logic without needing to be invented from scratch. The recording studio, two chairs, a host and guest, a CRT TV prop, and an audience laugh track are all conventions of the format that the model already knows how to render. The claymation registers on top of that format scaffold, not beneath it — the model knows where the host sits, how the camera cuts between host and guest, and how the audience reacts.
The three-beat escalation structure (setup → TV activation → punchline) is built into the talk show format but amplified by the impossible TV content. Beat 1: Einstein says something absurdly confident. Beat 2: the CRT TV activates mid-dialogue, showing orange-cat-vs-hyena kung-fu chaos. Beat 3: Einstein watches and treats the impossible footage as confirmation of his theory. The TV activation is the structural pivot — it moves from talk show into absurdist territory without requiring a new location or a scene transition. The TV is already established in the staging.
The dialogue-first character specification ("dry tone" for the host, "oversized head, messy white hair, wrinkled suit, holding a book titled 'AI Physics'" for Einstein) tells the model that these are clay puppets playing typed roles, not photorealistic people. The hybrid "claymation + semi-realistic" style tag clarifies that the clay material quality is not full Aardman-style plasticine — it is a middle register where the character proportions are exaggerated (oversized head) but the surface has realistic detail ("realistic clay puppet feel"). This precision prevents the model from defaulting to full cartoon-clay or full photorealism.
The takeaway: borrow a familiar real-world format (talk show, game show, podcast, news broadcast) as the staging scaffold for your clay content. The format provides camera positions, character placements, and pacing conventions — you only need to specify what makes your version absurdist. Use one prop as the structural pivot (the CRT TV) that transitions the scene into new territory without requiring a new location. Specify the hybrid clay register ("claymation + semi-realistic") to prevent default drift toward full cartoon or full photorealism.
5. The metaphor-as-object diorama — rendering a human experience as a hardware specification
See the full prompt on scenic.sh →
"fumbles for a socket between his shoulder blades, plugs in a thick clay cable... on his enormous belly, a miniature screen flickers on — 1%, a thin red sliver, the word FRIDAY 23:00."
Why this works: At 0 likes but with a distinctive conceptual technique, this prompt demonstrates the metaphor-as-object approach: a universal human experience (exhausted at the end of the week, needing to recharge overnight) is rendered not as an emotional description but as a hardware specification. The man does not feel tired; he has "enormous hollowed-out face, dark circles sculpted deep." He does not rest; he "reaches behind his own back, fumbles for a socket between his shoulder blades, plugs in a thick clay cable." His battery level is not implied — it reads 1%, with a timestamp and day label on a miniature belly-screen.
Each hardware detail is designed for a clay camera to observe at close range. The "hollowed-out face" is a sculptural property visible in medium shot. The "socket between the shoulder blades" requires a close-up cut behind the figure. The "belly screen with 1% battery, thin red sliver, FRIDAY 23:00" requires an extreme close-up of the clay surface display. The prompt is structured as a sequence of observable hardware states, and each state implies a corresponding shot scale — the prompt is simultaneously a narrative and a shot list.
The "hyperrealistic miniature diorama" frame does critical work. A clay figure with a charging port in a stylized cartoon world would read as whimsy; a clay figure with a charging port in a "hyperrealistic miniature diorama" with "real fabric shirt wrinkled and untucked," "real fabric blanket now draped over him somehow," and "cable taut from wall to back" reads as a documentary observation of an impossible event. The hyperrealism is the gap that creates the comedy — the camera treats the metaphor as physical fact.
The six-cut structure (stumbles in → collapses → fumbles for socket → belly screen activates → Saturday morning, unchanged → belly screen reads 12%) encodes the story entirely through hardware states. The narrative is told through battery percentages: 1% on Friday night, 12% on Saturday morning after a full night charging. No dialogue, no facial expressions that need rendering — just an observable power level that any miniature camera could read off a clay belly-screen.
The takeaway: render human experiences as hardware specifications rather than emotional descriptions. "Exhausted" becomes "enormous hollowed-out face, dark circles sculpted." "Needs to recharge" becomes "reaches for a socket between the shoulder blades." "Slept all night" becomes "cable taut from wall to back, real fabric blanket draped over him somehow." Each hardware detail should imply its own shot scale — belly-screen close-up, shoulder-socket cut-in, medium stumble-through-door. Embed the full narrative in observable state changes (1% → 12%) rather than action or dialogue — the miniature camera observes hardware, not feeling.
Seedance clay animation prompt cheat sheet
Across all five, the structural principles that make claymation Seedance prompts work:
- Behavioral rule — establish a repeating rule for the clay antagonist in the character description ("compresses and lunges, always just misses"). The rule propagates across all shots without re-instruction and functions as the sequence's comedic or dramatic engine.
- No-cut single-shot constraint — use "15-second continuous single-shot. No cuts." to force architectural planning. The environment must carry all atmospheric shifts without editorial breaks; specify them as physical events (walls bending, windows sliding). The medium's physical reality (the animator's hands exist between frames) is available as narrative content.
- Prop causality chain — write the script so each stunt generates the next stunt's hazard. Cauldron broth becomes spilled gravy; destroyed stalls become climbing surfaces. Add one escalation beat past the apparent resolution. Stack all claymation material descriptors in a single global list.
- Format borrowing — use a familiar real-world format (talk show, podcast, game show) as the staging scaffold. Specify one prop as the structural pivot that transitions the scene without requiring a new location. Declare the hybrid clay register ("claymation + semi-realistic") to anchor the material quality.
- Metaphor-as-object — render human experiences as hardware specifications observable by a miniature camera. Each hardware detail implies a shot scale. Tell the narrative through observable state changes (1% → 12% battery) rather than dialogue or described feeling. Use "hyperrealistic miniature diorama" to make the camera treat the impossible as documentary fact.
Browse the Scenic animation gallery for more clay and stop-motion examples, or see Seedance surreal prompts and Seedance comedy prompts for related techniques. Read how to write Seedance 2 prompts for the complete prompting guide.
