Head to head: Bytedance Seedance V1.5 Pro Image To Video vs Gemini Omni Flash 1.2 Reference to Video
Bytedance Seedance V1.5 Pro Image To Video vs Gemini Omni Flash 1.2 Reference to Video
By Ryan Merket · Published · Updated
Eight image-to-video trials put camera movement, street-level follow shots, liquid bursts and night-market action under close review. The matchup turns on whether a model can make the requested action read clearly while keeping its scene coherent.
Gemini Omni Flash 1.2 Reference to Video takes the saved result, **31.5 to 23.0**, winning all eight task outcomes with no ties. The editorial assessment is a **clear result on this sample**, not a claim about which model is generally better. The record includes two successful repeated trials per task type. The biggest separation came in action legibility. Gemini more convincingly showed balloons progressing from puncture to expanding water sheet and droplets, and made the curbside bag or bottle burst easier to follow in the night-market scenes. Seedance often delivered attractive liquid, glass and food detail, but its balloons could remain sphere-like and its bottle or bag action was less clear. Camera control was a relative weakness for both: Gemini conveyed a more continuous orbit through background parallax and changing bowl views, but neither ramen clip convincingly completed the requested roughly 180-degree move. In the scooter tests, Seedance sometimes had livelier weaving or a more apparent shift toward a side-follow; Gemini’s overall judged advantage came from steadier framing and stronger preservation of the requested setting and shot. **Verdict: Gemini Omni Flash 1.2 Reference to Video wins this saved task sample, 31.5–23.0, with eight task wins to Seedance’s zero.** That is a verdict on these prompts and outputs only, not a general capability ranking.
Camera motion control
A single smooth orbit (roughly 180 degrees) around a steaming bowl of ramen on a wooden table, the camera circling at a constant speed and height in one continuous move, warm restaurant light, 16:9.
OpenAI: GPT-6 Luna: Model B more clearly conveys a continuous orbit through the shifting background and bowl angle, while also showing the requested steam and warm restaurant setting. Model A has attractive, consistent food detail, but the steam is not apparent and the orbit is less clearly expressed. (Second judge pass, order swapped — scores are the average of both: Model B shows a clear, continuous orbit around the steaming bowl, with the changing view of the toppings and restaurant background supporting the cam | Anthropic: Claude Fable 5.1: Model B clearly executes a wide orbit with the…
Neon scooter follow
A single continuous 16:9 shot on a drizzly evening in the wholesale flower district follows a cobalt-blue cargo scooter ridden by a florist in a mustard rain cape as she weaves past stacked peony buckets, puddle reflections, and a reversing produce truck; the camera glides backward at her speed, then subtly cranes sideways around a steam vent to keep her centered from the waist up and tack-sharp while background storefronts, passing umbrellas, and a blinking crosswalk scramble with parallax, lit by magenta pharmacy signage and warm shop fluorescents, with an urgent but playful mood.
OpenAI: GPT-6 Luna: Model A better captures the playful, urgent flower-delivery ride, with peonies aboard, a clear weave past buckets and a truck, and lively parallax as the view shifts toward a side-follow. Model B has stronger steam-vent and pharmacy details and steadier framing, but feels more like a straight, centered approach and shows less of the florist’s cargo and weaving. (Second judge pass, order swapped — scores are the average of both: Model B better preserves the centered, front-facing follow shot and | Anthropic: Claude Fable 5.1: Model B delivers a photoreal, continuous backwar…
Fluid & particle dynamics
Cinematic slow-motion of a water balloon bursting, the sheet of water expanding and droplets scattering through the air in convincing detail against a dark background, hard side light, 16:9.
OpenAI: GPT-6 Luna: Model B clearly shows a water balloon being pierced, followed by a detailed expanding sheet and scattering droplets against a dark, side-lit background. Model A has attractive lighting and droplets, but the persistent glass-like sphere and crack pattern make the burst feel less convincing and less dynamically coherent. (Second judge pass, order swapped — scores are the average of both: Model B clearly progresses from a punctured water balloon to a convincing expanding water sheet and then scatte | Anthropic: Claude Fable 5.1: Model A shows a glassy sphere with crack-like l…
Night market glass burst
A single continuous 16:9 slow-motion shot in a crowded night market captures the instant a paper bag of fizzy tamarind soda bottles slips from a bicycle basket and one bottle bursts against the curb; the camera performs a low, gentle arc-in from sidewalk level, holding the exploding bottle near center as amber liquid, glass splinters, and tiny lime wedges bloom outward in exquisitely crisp fluid detail, illuminated by alternating teal LED stall lights and a harsh sodium streetlamp, with bystanders’ legs and scooter wheels drifting through the background in dreamy motion and a tense, mesmerizing mood.
OpenAI: GPT-6 Luna: Model B more clearly shows the paper bag tipping at the curb, followed by a splash and a believable aftermath in the night market. Model A has especially crisp liquid and glass detail, but the bag and curb action are less evident and the bottle’s continuity is less convincing. (Second judge pass, order swapped — scores are the average of both: Model B more clearly conveys the paper bag tipping into a curbside splash, with the night-market crowd, teal lighting, and scooter-wheel background matchi | Anthropic: Claude Fable 5.1: Model B closely follows the prompt with a bag s…
Matchup powered by OpenRouter.