{"id":14887,"date":"2026-10-10T16:30:22","date_gmt":"2026-10-10T08:30:22","guid":{"rendered":"https:\/\/imastudio.com\/?p=14887"},"modified":"2026-10-13T10:58:35","modified_gmt":"2026-10-13T02:58:35","slug":"ai-bullet-time-seedance-2-5-prompt","status":"publish","type":"post","link":"https:\/\/imastudio.com\/ja\/blog\/ai-bullet-time-seedance-2-5-prompt","title":{"rendered":"AI Bullet Time Product Video: A Seedance 2.5 Prompt"},"content":{"rendered":"

Create a Video with Seedance 2.5<\/a><\/p>\n

This AI bullet time product video starts with a bag opening: everything inside explodes into the air, hangs there for a moment, then drops onto the table as a perfect still life. Below is the full prompt behind the 12-second video, made with \u30b7\u30fc\u30c9\u30c0\u30f3\u30b9 2.5<\/a> and two reference images.<\/p>\n

One prompt, two reference images, one continuous 12-second shot. Turn the sound on.<\/figcaption><\/figure>\n

Below you will find the full prompt, a line-by-line look at what each part does, and a second version for different products in a different setting.<\/p>\n

What the bullet time effect means for a product ad<\/h2>\n

The bullet time effect makes action look frozen or extremely slow while the camera keeps moving around it. In AI video, you create it with words: tell the model to slow time, hold every object in place, and keep the camera moving.<\/p>\n

Classic bullet time freezes a moment while the camera moves around it. In a product ad, that effect turns an ordinary reveal into a spectacle: products burst out, hang in the air, and the camera glides among them before everything lands.<\/p>\n

This prompt breaks the idea into five beats across 12 seconds: the trigger, the explosion, a moment of zero gravity, gravity returning, and a final still-life landing. Each beat gets its own block of instructions, which is why the prompt is long and why it works as a template.<\/p>\n

The full prompt<\/h2>\n

Use it with Seedance 2.5 in image-to-video mode. Upload two reference images: a photo of your model and a flat-lay of your products. It is a timestamp prompt: each block starts with a time range, so the model knows what should happen at each moment of the 12 seconds. We set the video to 12 seconds, 720p, with audio on, and ran the prompt exactly as written below.<\/p>\n

The Makeup Explosion\uff5cBeauty in Zero Gravity\n\nVisual Style\n\nARRI Alexa 35 cinematic realism, 35mm subtle film grain, premium European and American luxury beauty campaign, physically accurate lighting, realistic skin texture, glass reflections, metallic surfaces, and high-speed photography. Sophisticated editorial color grading, tactile material details, natural motion blur. Absolutely no CGI animation aesthetics.\n\nReference & Continuity Rules\n\nUse the uploaded reference images strictly. Preserve the same female model, facial identity, hairstyle, makeup, outfit, makeup bag, and every cosmetic product throughout the entire video. All product packaging, colors, logos, proportions, and textures must remain identical to the references.\n\nUse only the exact products shown in the uploaded images. Do not add, invent, substitute, or duplicate any products. The original product inventory must remain consistent throughout the video. Do not generate additional cosmetics or accessories, even if they would normally belong in a makeup collection.\n\nOne continuous shot from beginning to end. Zero cuts, zero jump cuts, zero hidden transitions, zero teleportation. A single physical camera moves continuously through one coherent three-dimensional environment.\n\nScene\n\nA stylish young Western woman sits at a beautifully styled vanity in a naturally lit bedroom. She casually opens a makeup bag placed on the vanity. Inside are only the cosmetic products visible in the uploaded reference images.\n\nThe environment feels like a real luxury beauty editorial shot on location, not a computer-generated showroom. Soft morning sunlight streams through a nearby window, creating delicate highlights on the products and surrounding surfaces.\n\n0\u20132.0s \u2014 The Trigger\n\nThe camera begins in a close medium shot, slightly above the vanity, slowly pushing toward the woman's hands.\n\nShe unzips the makeup bag and opens it naturally.\n\nThe instant the bag opens fully, a sudden burst of physical momentum sends the referenced products upward and outward, as if the compressed contents have been released under an unexpected force.\n\nThe woman instinctively leans back in surprise, her hair moving slightly with the motion.\n\nAll products must emerge from the actual opening of the bag and must match the uploaded references exactly. Nothing appears from nowhere. The bag remains on the vanity.\n\n2.0\u20134.0s \u2014 The Explosion\n\nThe referenced products burst into the air in different directions, tumbling and rotating according to their individual shapes, masses, and momentum.\n\nEach product retains its original appearance, packaging, color, and logo. A few tiny particles may drift naturally through the sunlight.\n\nThe camera accelerates forward and rises between the floating products, passing close to individual items and weaving through the arrangement.\n\nThe camera must physically navigate the scene. Objects never move toward the camera simply to create a close-up. No additional products appear.\n\n4.0\u20137.5s \u2014 Beauty in Zero Gravity\n\nAt the peak of the explosion, time dramatically slows to approximately 2\u20133% of normal speed.\n\nEvery product freezes at its exact three-dimensional position, preserving its orientation, distance, and trajectory at the moment of deceleration.\n\nOnly the camera continues moving.\n\nThe camera performs a continuous, elegant orbit through the frozen arrangement, passing within inches of individual products. Focus shifts naturally between their original packaging, surface textures, and reflective details.\n\nThe movement feels like an expensive high-speed beauty commercial, with precise spatial depth, realistic optical perspective, and controlled shallow depth of field.\n\nNo object changes position during this sequence. No product rotates independently. No object is duplicated, deformed, or attracted toward the camera.\n\n7.5\u201310.0s \u2014 Gravity Returns\n\nTime gradually returns to normal.\n\nEvery object resumes its original motion from the exact position and orientation where it was frozen. Gravity takes effect naturally.\n\nThe products descend toward the vanity, colliding gently with the tabletop and one another. They bounce, roll, or slide according to their physical properties.\n\nEvery movement must be driven by gravity, inertia, friction, and realistic collisions.\n\nAs the products descend, the camera smoothly lowers with them, following the action in one uninterrupted movement.\n\nOnly the original referenced products may appear throughout the sequence. No new objects or duplicates may emerge during collisions.\n\n10.0\u201312.0s \u2014 The Perfect Landing\n\nThe last products settle onto the vanity.\n\nThrough their natural trajectories, rolling motion, collisions, and friction, the referenced products end up in an unexpectedly beautiful composition around the open makeup bag.\n\nThe final arrangement resembles a premium beauty campaign still life. The products are naturally distributed across the tabletop, with a visually balanced composition that remains physically believable.\n\nThe products do not fly into formation, snap into place, or arrange themselves through invisible forces.\n\nThe woman looks down at the vanity, pleasantly surprised.\n\nThe camera completes its continuous movement and settles into a refined low-angle product composition. Soft sunlight illuminates the products, revealing their original materials, textures, and packaging.\n\nEnd on a visually striking luxury beauty tableau.\n\nFinal Frame: The woman, open makeup bag, and only the original products from the uploaded reference images are visible in one cohesive composition. Every product remains recognizable and identical to its reference image.\n\nAbsolute Restrictions\n\nNo additional products, no invented cosmetics, no product duplication, no product morphing, no changing packaging, no changing logos, no missing or substituted products, no cuts, no jump cuts, no hidden transitions, no camera teleportation, no impossible camera movement, no floating without physical cause, no automatic product alignment, no magnetic attraction, no distorted hands, no face changes, no outfit changes, no excessive particles, no magical sparkles, no neon effects, no artificial CGI sheen, no cartoon physics, no text overlays, no subtitles, no watermarks, no background music, no dialogue, no voice-over.\n\nNatural environmental and object sound effects are permitted: zipper opening, cosmetic packaging clicks, subtle air movement, light tabletop impacts, and products rolling to a stop.<\/code><\/pre>\n
Tip: for any person in your prompt, say \u201cadult\u201d explicitly, for example \u201cyoung adult woman.\u201d<\/p>\n
What each part of the prompt does<\/h2>\n
The rules at the top<\/h3>\n
Before any action, the prompt sets three things: a visual style (a cinema-camera look, film grain, no CGI feel), continuity rules (keep the same model, outfit and products, with packaging and colors identical to the references), and one instruction that this is a single continuous shot with no cuts. These lines keep the result looking like one real take instead of a collage.<\/p>\n
The most useful paragraph is this one: \u201cUse only the exact products shown in the uploaded images. Do not add, invent, substitute, or duplicate any products.\u201d More on why below.<\/p>\n
0\u20132 seconds: The trigger<\/h3>\n
\"Three
Frames at 0.2, 1.0 and 1.8 seconds.<\/figcaption><\/figure>\n
The woman opens the bag and the contents burst out. The key line is \u201cThe products must emerge from the actual opening of the bag. Nothing appears from nowhere.\u201d In our video the products leave the bag at about the 1.8-second mark, so the explosion has a visible cause instead of items popping into frame.<\/p>\n
2\u20134 seconds: The explosion<\/h3>\n
\"Three
Frames at 2.6, 3.4 and 4.0 seconds.<\/figcaption><\/figure>\n
Here the prompt gives each product its own motion: tumbling and rotating \u201caccording to their individual shapes, masses, and momentum.\u201d It also gives the camera a job: \u201cThe camera must physically navigate the scene. Objects never move toward the camera simply to create a close-up.\u201d The result is a camera that rises between the flying products, not products that fly at the lens.<\/p>\n
4\u20137.5 seconds: Zero gravity<\/h3>\n
\"Three
Frames at 4.6, 5.0 and 6.6 seconds.<\/figcaption><\/figure>\n
This is the bullet time beat. The prompt asks time to slow to \u201capproximately 2\u20133% of normal speed,\u201d every product to freeze in place, and only the camera to keep moving. In our video the products hang in the air in soft light while the camera moves in close to individual items, such as a blush compact.<\/p>\n
Treat the slow-down as a request, not a guarantee. In our two runs, the camera moved in close to individual products in one and held a low angle in the other, and in the outdoor run the products clearly drifted more slowly than during the explosion. Generate a few takes and keep the one that matches your idea.<\/p>\n
7.5\u201310 seconds: Gravity returns<\/h3>\n
\"Three
Frames at 7.4, 8.2 and 9.0 seconds.<\/figcaption><\/figure>\n
\u201cEvery object resumes its original motion from the exact position and orientation where it was frozen\u201d keeps the fall continuous with the freeze. Products drop toward the table and collide, driven by gravity and friction instead of anything that looks animated. In our video most items have landed by about 9 seconds.<\/p>\n
10\u201312 seconds: The perfect landing<\/h3>\n
\"Three
Frames at 9.8, 10.6 and 11.95 seconds.<\/figcaption><\/figure>\n
The last beat asks for a composition that looks designed but stays \u201cphysically believable,\u201d and adds that products must not \u201cfly into formation, snap into place, or arrange themselves.\u201d The video ends with the products spread across the table, the bag behind them, and the woman smiling over the result.<\/p>\n
The restrictions at the bottom<\/h3>\n
The closing list repeats the biggest risks as hard rules: no cuts, no invented or duplicated products, no changed packaging, no text overlays, no music. It also allows natural sounds such as a zipper and light impacts, so the video still gets realistic audio without a soundtrack.<\/p>\n
What the reference images do<\/h2>\n
The two reference images shaped the result as much as any line of text.<\/p>\n
\n
\"AI-generated
Model reference<\/strong><\/figcaption><\/figure>\n
\"AI-generated
Product reference.<\/strong> Both images are AI-generated.<\/figcaption><\/figure>\n<\/div>\n
Whatever the model holds becomes the bag<\/h3>\n
We wrote \u201cmakeup bag,\u201d but the model reference shows her holding a white studded handbag, and that handbag is the bag that appears in the video. If you want a different bag, use a model photo where she is not holding one, or give the bag its own reference image.<\/p>\n
\u201cUse only the products shown\u201d keeps the inventory honest<\/h3>\n
In our first run, the prompt listed lipstick, mascara and brushes by name, but none of them were in the product image. The video included them anyway. When we replaced that list with \u201cuse only the exact products shown in the uploaded images,\u201d the extra items stopped appearing in our second run. If you name products in your prompt, name only what your reference image actually contains.<\/p>\n
Swap the products and the setting<\/h2>\n
To test whether the structure is reusable, we kept the five beats and changed the products, the bag, the setting and the light. This version uses nine outdoor gear items in a sports bag on a park picnic bench, in soft daylight.<\/p>\n