AI VIDEO
Das Essen wächst aus der Karte.
Eine geschlossene Speisekarte auf dunklem Marmor, eine Hand klappt sie auf — und die flach gedruckten Gerichte heben sich als winzige Reliefs aus dem Papier. Fünf Shots, jeder einzeln gerendert, jeder mit einem eigenen Start- und Endbild aus ChatGPT. Gemini Omni 1.1 Flash auf Artlist findet nur noch den Weg zwischen zwei Bildern, die es nicht erfinden muss.
- ChatGPT
- Artlist
- 45 Min
- Fortgeschritten
ERGEBNIS
Das ist dabei rausgekommen.
- 5Shots, jeder einzeln gerendert
- 9Bilder aus ChatGPT als Frames
- 2–3 cmmaximale Popup-Höhe im Prompt
- 21sfertige Laufzeit, kein Übergang
-
Die Karte liegt noch0:00
-
Aufgeklappt — alles noch flach0:03
-
Die Margherita bekommt Tiefe0:07
-
Parmesan fällt auf die Pasta0:11
-
Zwei Kugeln im Glasschälchen0:15
-
Rückfahrt auf die ganze Karte0:21
DAS BRAUCHST DU
Womit das gebaut ist.
ChatGPT
Artlist
-
Shot 1 · zu → offenBild 1 + Bild 2
-
Shot 2 · flach → ReliefBild 3 + Bild 4
-
Shot 3 · flach → ReliefBild 5 + Bild 6
-
Shot 4 · flach → ReliefBild 7 + Bild 8
-
Shot 5 · StartframeBild 9
ANLEITUNG
So baust du es nach.
-
Neun Bilder zuerst, Videos später
Kein einziger Clip entsteht aus Text allein. Erst baust du in ChatGPT die Standbilder — pro Shot ein Startframe und ein Endframe. Danach hat das Videomodell nur noch eine Aufgabe: den Weg zwischen zwei Bildern finden, die es nicht mehr erfinden muss. Neun Bilder ergeben fünf Shots. Bild 1 und 2 gehören zum Aufklappen, dann je ein Paar pro Gericht, Bild 9 startet den Schluss.
-
Die Karte ist ein Gegenstand, kein Layout
Bild 1 beschreibt kein Menü-Design, sondern ein Objekt auf einem Tisch: dunkles, ledriges Material, dünne Goldkanten, Kerze daneben, Weinglas, Marmor, eine Hand kurz vorm Zugreifen. Was in Bild 1 steht — Licht, Tisch, Hintergrund, Material — wiederholst du in jedem weiteren Bild wortgleich. Der Restaurantname auf meiner Karte steht in keinem Prompt: Den hat ChatGPT beim ersten Bild erfunden. Damit er auf allen weiteren Bildern gleich bleibt, muss Bild 1 als Referenz mitlaufen.
-
Pro Gericht ein Paar: flach und Relief
Bild 3 zeigt die Margherita als reine Drucksache — detailliert, aber körperlos, mit dem ausdrücklichen Verbot no 3D pizza, no plate, no floating object. Bild 4 zeigt exakt dasselbe noch einmal: gleicher Ausschnitt, gleiche Typografie, gleiches Licht, gleicher Durchmesser. Geändert hat sich eine einzige Eigenschaft, die Tiefe. Genau diese Disziplin macht den Clip dazwischen später möglich.
-
Pasta und Gelato sind derselbe Prompt
Bild 5 bis 8 sind Bild 3 und 4 mit ausgetauschtem Gericht. Du ersetzt Name, Beschreibungszeile, Preis und Zutatenliste — Truffle Tagliatelle für 15 Euro, Pistachio Gelato für 8 Euro — und lässt jeden anderen Satz unangetastet. Deshalb liegen im Creator Kit nur die Margherita-Bilder: Die anderen beiden Paare entstehen daraus in zwei Minuten, und jede neu formulierte Zeile wäre eine zusätzliche Chance auf Drift.
-
Jeder Shot geht einzeln durch das Modell
Nicht Karte klappt auf, dann wächst die Pizza, dann die Pasta. Ein Clip, eine Verwandlung. Du lädst Startframe und Endframe hoch, schreibst den Transformationsprompt dazu und renderst vier Sekunden. Vier reichen: Die Bewegung ist langsam, die Kamera steht fast still, und je länger der Clip, desto mehr Zeit hat das Modell, sich etwas Eigenes auszudenken.
-
Der Schluss bewegt nur die Kamera
Im letzten Shot passiert auf der Karte nichts mehr. Alle drei Gerichte stehen bereits plastisch da, und der Prompt sagt genau das: keine Verwandlung, keine Größenänderung, keine Zutat, die sich bewegt — die einzige nennenswerte Bewegung ist die Kamera, die langsam zurück- und leicht hochfährt. Erlaubt sind drei Details: Dampf über der Pizza, Dampf über der Pasta, ein kalter Glanz auf dem Eis.
-
Fünf Clips hart aneinander
Am Ende liegen fünf MP4s in 1080 × 1920 vor dir. Im Schnitt kommen sie in der Reihenfolge der Handlung hintereinander, ohne Blende, ohne Übergang: 21 Sekunden. Eine Überblendung wäre hier sogar der Fehler — jeder Shot steht an einer anderen Stelle der Karte, ein harter Cut liest sich als Schnitt, eine Blende als Fehler im Material.
CREATOR KIT
Alles, was du brauchst.
Create a photorealistic cinematic vertical 9:16 scene of an elegant luxury Italian menu card resting on a dark restaurant table at night. The setting is a warm upscale Italian trattoria with candlelight, soft golden bokeh, shallow depth of field, and premium food-commercial lighting. The menu card is closed and lies naturally on the table. It is made of textured dark charcoal leather-like material with elegant thin gold borders and refined premium detailing. A realistic human hand is just about to pick up the menu from the table. Show only the hand and part of the forearm. No face visible. The table includes subtle premium restaurant details: a warm candle glow, a wine glass, soft reflections on the tabletop, subtle cutlery, and a luxurious cozy Italian restaurant atmosphere. The composition should feel like the opening frame of a high-end restaurant commercial. Camera: close cinematic perspective, slightly above table level, macro-lens feeling, shallow depth of field, natural realistic perspective. Lighting: warm tungsten, soft candlelight, rich shadows, golden highlights. Extremely photorealistic. No artificial CGI appearance. No floating objects. No text overlays. Vertical 9:16.
Create a photorealistic cinematic vertical 9:16 scene matching the exact same Italian restaurant environment, table, lighting, menu design, and visual style. The luxury Italian menu card is now being naturally opened by a person's hands above the table. The menu opens toward the camera, clearly revealing the inside pages. Show only the hands and forearms. No face visible. Inside the menu are three elegant food sections: Margherita — €12 Truffle Tagliatelle — €15 Pistachio Gelato — €8 IMPORTANT: At this stage, all food illustrations are still completely flat printed illustrations integrated into the menu surface. There are NO large 3D dishes yet. The food artwork should look premium and realistic but remain physically flat like high-quality menu printing. The menu has: dark charcoal textured material, refined thin gold borders, elegant serif typography, luxury Italian restaurant design. Keep the menu layout clean and readable. Warm candlelit trattoria atmosphere. Soft golden bokeh. Shallow depth of field. Premium cinematic food-commercial aesthetic. The inside of the menu should be clearly visible because the next shots will move closer to individual dishes. Extremely photorealistic. No floating dishes. No separate plates. No exaggerated popup elements. Vertical 9:16.
Create a cinematic vertical 9:16 commercial-style shot using the provided START FRAME and END FRAME as strict visual references. A person naturally picks up the luxury Italian menu card from the restaurant table and gently opens it toward the camera. The movement should feel completely natural, elegant and physically realistic. The hand first grips the menu from the table. The card lifts slightly. The second hand naturally supports it. The menu then opens smoothly toward the camera. CAMERA: The camera performs a subtle premium commercial follow movement as the menu is lifted. As the card opens, the camera gently pushes closer toward the inside pages. The final composition should naturally arrive at the provided END FRAME. Keep the camera movement slow, smooth and controlled. No dramatic orbit. No sudden zoom. No camera shake. No cuts. ENVIRONMENT: Preserve the exact warm candlelit Italian trattoria atmosphere. Keep: the table, candlelight, golden bokeh, wine glass, restaurant background, menu design, lighting, and visual composition consistent. HANDS: Hands must remain anatomically realistic. Natural finger movement. Natural grip. No extra fingers. No distorted hands. MENU: The menu remains physically realistic. The card bends only naturally at the fold. No warping. No morphing. No changing typography. No changing menu layout. The dishes inside the menu remain FLAT during this shot. Do not create 3D food yet. FINAL MOMENT: End with the menu fully opened and the viewer looking clearly at the interior menu pages. The ending should naturally prepare for a macro close-up of the Margherita section. Premium Italian restaurant commercial. Photorealistic. Elegant controlled movement.
Create a photorealistic cinematic macro close-up in vertical 9:16 of the inside of the same luxury Italian menu card. Focus specifically on the Margherita section. Preserve exactly the same menu design: dark charcoal textured surface, thin elegant gold borders, refined serif typography, warm Italian trattoria environment. Readable text: Margherita €12 The Margherita pizza is ONLY a flat printed illustration integrated directly into the menu surface. It has absolutely no physical depth yet. The illustration should still look detailed and premium: Neapolitan-style pizza, tomato sauce, mozzarella, basil, blistered crust. But everything remains completely flat like premium menu printing. Camera: cinematic macro close-up, slight three-quarter viewing angle, very shallow depth of field. Warm golden trattoria lighting. Luxury food-commercial photography. No 3D pizza. No floating object. No plate. No separate food object. Vertical 9:16.
Create a photorealistic cinematic macro close-up in vertical 9:16 matching exactly the same menu card, composition, typography, lighting and camera angle as the reference image. Focus on the Margherita section. Readable text: Margherita €12 The previously flat printed pizza has now become a small realistic three-dimensional popup relief emerging directly from the menu surface. IMPORTANT: The pizza remains physically attached to the menu. It does not float. It does not sit on a separate plate. It does not become a full-size pizza. It retains exactly the same diameter and location as the original printed illustration. It only gains physical depth. Maximum projection from the menu surface: approximately 2 to 3 centimeters. The result resembles an extremely sophisticated luxury popup book where printed food has become photorealistic. FOOD: realistic blistered Neapolitan crust, glossy tomato sauce, soft melted mozzarella, fresh basil, subtle olive oil highlights, extremely detailed food texture. The food must feel miniature. Warm golden Italian trattoria lighting. Shallow depth of field. Premium macro food advertisement. No floating food. No wooden shelf. No separate plate. No oversized pizza. Vertical 9:16.
Use the provided START FRAME and END FRAME as strict spatial references. This shot contains ONE transformation only. The menu card itself never changes. The printed Margherita illustration remains permanently attached to its exact original position on the menu card. IMPORTANT: The pizza must NOT detach from the menu. The pizza must NOT fly toward the camera. The pizza must NOT become a full-size pizza. The pizza must NOT appear as a separate object. The pizza must NOT increase in diameter. Instead, the existing flat printed pizza illustration slowly develops physical depth directly from the menu surface. TRANSFORMATION: At the beginning, the pizza is completely flat. First, the printed crust subtly begins to rise a few millimeters from the surface. Then the tomato sauce and mozzarella gradually gain shallow physical volume. The entire existing illustration slowly extrudes outward PERPENDICULAR to the menu surface. The pizza remains anchored to exactly the same center point throughout the entire transformation. Its diameter NEVER changes. Its position NEVER changes. Its orientation NEVER changes. It only gains DEPTH. Maximum final projection: approximately 2 to 3 centimeters. The transformation should feel like a premium physical popup book becoming photorealistic. FOOD DETAIL: The crust gradually becomes freshly baked and slightly blistered. The tomato sauce becomes moist and glossy. The mozzarella develops subtle melted texture. At the end, two or three tiny fresh basil leaves gently fall from just above the pizza and naturally settle onto the miniature popup. Add extremely subtle steam. CAMERA: Very slow cinematic macro push-in only. Almost locked camera. No orbit. No dramatic zoom. No camera repositioning. The transformation itself is the hero. LIGHTING: Warm premium Italian trattoria lighting. Soft candlelight. Shallow depth of field. Luxury food-commercial cinematography. STRICTLY AVOID: No flying pizza. No floating pizza. No separate pizza. No separate plate. No wooden platform. No giant food. No morphing. No crossfade. No teleportation. No changing menu. No changing typography. No warped text. No ingredient explosion. FINAL FRAME: End precisely matching the provided END FRAME.
Use the provided START FRAME and END FRAME as strict spatial references. This shot contains ONE transformation only. The menu card itself never changes. The printed Truffle Tagliatelle illustration remains permanently attached to its exact original location. The dish must NOT detach from the card. It must NOT fly forward. It must NOT become a large separate plate. It must NOT increase in size. TRANSFORMATION: Begin with the pasta completely flat and printed into the menu. The printed plate first develops only a few millimeters of shallow physical depth. Then the existing printed tagliatelle ribbons gradually become three-dimensional. The ribbons gently gain volume directly from their existing printed shapes. They remain compact and physically anchored to the menu. Do not make noodles fly through the air. Do not generate pasta from outside the menu. The existing illustration itself becomes dimensional. Roasted mushrooms subtly gain depth and realism. At the end, two or three very thin black truffle slices gently drift downward from immediately above the miniature pasta and settle naturally onto it. A very small amount of finely grated parmesan may fall gently onto the dish. Keep all ingredient movement subtle and elegant. Maximum popup depth: approximately 2 to 3 centimeters. CAMERA: Very slow cinematic macro push-in. Subtle natural rack focus toward the pasta. Almost locked composition. No orbit. No dramatic zoom. No camera shake. LIGHTING: Warm candlelit Italian trattoria. Luxury food-commercial lighting. Soft golden highlights. Shallow depth of field. STRICTLY AVOID: No flying noodles. No pasta explosion. No separate plate. No giant food. No floating dish. No teleporting ingredients. No morphing. No changing menu design. No changing text. No warped typography. End precisely matching the provided END FRAME.
Use the provided START FRAME and END FRAME as strict spatial references. This shot contains ONE transformation only. The menu card never changes. The printed Pistachio Gelato illustration remains permanently attached to its exact original position. The dessert must NOT detach. It must NOT float. It must NOT fly toward the camera. It must NOT increase in size. It must NOT become a separate bowl. TRANSFORMATION: Begin with the dessert completely flat and printed into the menu. The existing printed bowl slowly gains a few millimeters of physical depth. It becomes a shallow popup relief directly connected to the menu surface. Then the existing printed gelato scoops gradually develop volume. The scoops softly rise outward from the card while remaining in exactly the same position. They become realistic, creamy and cold. The transformation should look like the printed illustration itself becoming physically three-dimensional. Maximum popup depth: approximately 2 to 3 centimeters. At the very end, a small number of chopped pistachio pieces gently fall from immediately above the dessert. They bounce very subtly and settle naturally onto the gelato. Keep this motion restrained and premium. CAMERA: Very slow macro push-in. Subtle rack focus. Almost locked camera. No orbit. No dramatic zoom. No sudden movement. LIGHTING: Warm elegant Italian trattoria. Soft golden candlelight. Luxury dessert-commercial cinematography. Very shallow depth of field. STRICTLY AVOID: No floating bowl. No giant gelato. No flying dessert. No ingredient explosion. No separate plate. No teleportation. No morphing. No changing menu. No changing typography. No warped text. End precisely matching the provided END FRAME.
Create a photorealistic cinematic macro close-up in vertical 9:16 of the fully opened luxury Italian menu card inside a warm upscale Italian trattoria at night. The menu card already contains all three completed miniature 3D popup dishes: Margherita — €12 Truffle Tagliatelle — €15 Pistachio Gelato — €8 IMPORTANT: All three dishes are already fully transformed. They remain physically attached to their exact positions on the menu card. Each dish projects only approximately 2 to 3 centimeters from the menu surface. Nothing floats. Nothing appears as a separate full-size plate. The opened menu fills most of the frame. Use a relatively close camera position. The three-dimensional nature of the food should be clearly visible. The Margherita and Tagliatelle sections are especially prominent while all three dishes remain readable. Preserve: dark charcoal textured menu material, thin elegant gold borders, refined serif typography, warm candlelight, soft golden restaurant bokeh. The popup dishes are extremely photorealistic miniature food. Camera: close three-quarter perspective, premium macro commercial photography, very shallow depth of field. No hands. No people. No floating elements. No motion blur. Vertical 9:16.
Create a photorealistic cinematic vertical 9:16 hero shot of the same fully opened luxury Italian menu card inside the same warm upscale Italian trattoria at night. Match exactly the same: menu design, food design, lighting, restaurant environment, dish positions, typography, and popup geometry from the reference image. Show all three completed miniature 3D popup dishes: Margherita — €12 Truffle Tagliatelle — €15 Pistachio Gelato — €8 All dishes remain physically attached to the menu card and project only approximately 2 to 3 centimeters from its surface. Nothing floats. Nothing becomes full-size. The camera is now positioned slightly farther away than in the reference close-up. Reveal the complete opened menu card as a clean premium hero composition. The entire menu should fit comfortably inside the vertical frame with subtle surrounding restaurant environment visible. Warm candlelight. Soft golden bokeh. Dark elegant table. Luxury Italian trattoria atmosphere. Premium restaurant advertising photography. Photorealistic materials. Shallow depth of field, but enough depth that the complete menu remains readable. No people. No hands. No additional dishes. No floating objects. Vertical 9:16.
Create a short cinematic vertical 9:16 end reveal shot using the provided START FRAME and END FRAME as strict visual references. The opened luxury Italian menu card is already fully transformed. All three miniature popup dishes are already present and remain exactly as shown: Margherita, Truffle Tagliatelle, Pistachio Gelato. IMPORTANT: Nothing on the menu transforms during this shot. The dishes remain completely fixed to their exact positions on the menu card. Do not change: the dishes, their size, their shape, their position, the menu layout, the typography, the prices, the lighting, or the restaurant environment. THE ONLY MEANINGFUL MOVEMENT IS THE CAMERA. CAMERA: Begin exactly from the closer composition shown in the START FRAME. Perform one extremely smooth slow cinematic dolly-out / pull-back. The camera gently moves backward and slightly upward, revealing more of the opened menu card. Maintain approximately the same three-quarter viewing angle. Create subtle natural parallax between the raised miniature dishes, the menu surface and the softly blurred restaurant background. The movement must feel like an expensive premium food-commercial camera move. No sudden zoom. No orbit. No camera shake. No cuts. No perspective jump. The camera should naturally arrive at the wider composition shown in the END FRAME. FOOD: The three dishes remain miniature 3D popup reliefs physically attached to the menu surface. They must never float or detach. Do not turn them into full-size food. Do not create separate plates. Do not move ingredients. Only allow extremely subtle secondary detail: faint natural steam from the Margherita, faint natural steam from the Tagliatelle, subtle cold creamy sheen on the Pistachio Gelato. Keep these effects extremely restrained. ATMOSPHERE: Warm upscale Italian trattoria at night. Soft golden candlelight. Warm background bokeh. Dark elegant restaurant interior. Premium cinematic food advertising. Shallow depth of field. Realistic materials. Natural shadows. Luxurious but believable. FINAL MOMENT: End precisely on the provided END FRAME composition. The complete opened menu card is clearly visible with all three miniature popup dishes integrated into the design. Hold the final hero composition briefly and steadily. The result should feel like the final shot of a premium Italian restaurant commercial.
SETTINGS
Die Einstellungen im Überblick.
| Plattform | Artlist, AI-Bereich |
|---|---|
| Videomodell | Gemini Omni 1.1 Flash |
| Bildmodell | ChatGPT, 941 × 1672 in 9:16 |
| Referenzen pro Clip | Startframe + Endframe |
| Clip-Rohmaterial | 1080 × 1920, 24 fps |
| Shots | 5 Stück: 4 × 4 s, 1 × 5 s |
| Bilder insgesamt | 9 Start- und Endframes |
| Ton | aus dem Modell: Restaurant, Besteck, Stimmen |
| Schnitt | 5 Clips hart aneinander, 21 Sekunden |
WORAUF ES ANKOMMT
Darauf musst du achten.
Start- und Endbild müssen dieselbe Szene sein
Gleicher Winkel, gleiche Karte, gleiche Schrift, gleiches Licht. Beim Food-Popup darf sich idealerweise nur die Tiefe des Gerichts ändern. Sobald zwischen den beiden Bildern noch etwas anderes wandert — die Kamera ein Stück, eine Kerze, die Lage der Karte im Bild — muss das Modell zwei Veränderungen gleichzeitig erzählen und löst das, indem es die schwierigere davon dramatisiert. Meistens ist das die Tiefe: Das Gericht hebt ab.
Kein einziges echtes Food-Foto als Referenz
Der Reflex ist verständlich: Man will dem Modell zeigen, wie eine gute Margherita aussieht, und legt ein Pizzafoto dazu. Genau daran scheitert der erste Versuch. Sobald ein normal großer Teller im Referenzbereich liegt, interpretiert das Modell die Aufgabe um — dann fliegt eine komplette Pizza aus der Karte heraus. Die einzigen Referenzen sind die beiden Karten-Bilder. Wie das Essen aussehen soll, steht als Text im Prompt.
Der Satz, der das Essen an der Karte hält
The existing printed illustration itself becomes three-dimensional. It remains permanently attached to the menu surface. It only gains depth. Diese drei Zeilen stehen sinngemäß in jedem Transformationsprompt, und sie tragen den ganzen Effekt. Sie beschreiben keine Handlung, sondern eine Grenze: Was da wächst, ist die vorhandene Abbildung — nichts kommt dazu, nichts löst sich ab, nichts wird größer.
Zwei bis drei Zentimeter ist eine Zahl, kein Gefühl
Klein, subtil oder miniature sind Wünsche; das Modell darf sie auslegen. Maximum projection from the menu surface: approximately 2 to 3 centimeters ist eine Kante, an der es sich messen lassen kann. Dieselbe Logik zieht sich durch alles andere in diesen Prompts: gleicher Durchmesser, gleicher Mittelpunkt, gleiche Ausrichtung, nur Tiefe. Je konkreter die Grenze, desto weniger Interpretation.
Deine Karte, die aufsteht.
Neun Bilder, fünf Clips, ein Gericht nach dem anderen. Du tauschst Küche, Karte und Preise aus — die Bildpaare und die Verbotssätze bleiben stehen.
FAQ


