Head to head: Bytedance Seedance V1.5 Pro Image To Video vs Happy Horse
This matchup wasn’t close once motion direction, subject fidelity, and scene logic were put under pressure. Bytedance Seedance V1.5 Pro Image To Video can produce attractive frames, but Happy Horse was the one that actually followed the briefs.
By RuntimeWire · Published
Happy Horse wins because it understands what these prompts are asking the camera and the scene to do, not just how to make them look. The aggregate gap — 16.7 to 11.4 — reflects a model that was more reliable on motion, staging, and prompt adherence across both tests.
In Awning Shadow Noodles, Happy Horse nailed the shot design: a continuous left-to-right push along the stall, a believable lighting drift under the awning from warm to cooler and back, and the cramped Taipei alley feel the prompt depended on. Seedance had respectable noodle handling and decent prop presence, but the scene read more static and interior-facing, which undercut the whole point of the moving street-side setup.
The bigger miss for Seedance came in Midnight Okonomiyaki Pass. Happy Horse delivered an actual okonomiyaki on a griddle, with the handheld move evolving from close-up cooking detail into a wider plating context, plus coherent steam, fluorescent light, and late-night bar claustrophobia. Seedance looked like it wandered into a different dish entirely — more teppanyaki skewers and vegetables than okonomiyaki — and once the core subject is wrong, nice image quality doesn’t rescue the result.
What separates the two here is simple: Happy Horse preserved the identity of the food, the logic of the camera move, and the environmental cues that make these prompts specific rather than interchangeable. Seedance showed flashes of polish, but too often it treated the prompt as aesthetic suggestion instead of instruction.
Final call: Happy Horse is the clear winner. If you care about prompt fidelity and shot coherence in image-to-video food scenes, it’s the model that actually delivers the assignment.
How they were tested
We ran 2 fresh video tasks, generated on the fly for this matchup so neither model could prepare in advance, and had gpt-5.4 score each one. Bytedance Seedance V1.5 Pro Image To Video scored 11.4 to Happy Horse's 16.7.
1. Awning Shadow Noodles
A short 16:9 video clip in a tiny Taipei alley noodle stall at 11:47 a.m.: a cook in a faded teal apron snaps a nest of hand-pulled noodles into a hammered steel pot while the camera makes a slow shoulder-height push from left to right past jars of red chili oil and chipped porcelain spoons; midway through the shot, a gust shifts the striped orange awning and a bank of sunlight that had been warming the steam and the cook’s forearms slides away, turning the scene noticeably cooler and dimmer before brightening again in a soft, believable transition across the pot, counter, and rising vapor; the noodles whip, the broth splashes, and the mood changes from bustling warmth to brief overcast calm and back, all in one continuous shot with no cuts.
Winner: Happy Horse — Model B better matches the prompt’s continuous left-to-right stall-side push, believable awning-driven lighting shift from warm to cooler and back, and the tiny Taipei alley atmosphere. Model A has solid noodle action and props, but the framing feels more static/interior and the lighting transition is less clearly expressed across the scene.
2. Midnight Okonomiyaki Pass
A short 16:9 video clip in a cramped Osaka basement snack bar just after midnight: in one unbroken energetic handheld documentary-style shot, the camera starts tight on a spatula chopping cabbage into an okonomiyaki on a blackened griddle, then arcs around the cook’s right shoulder and drifts backward as she flips the pancake, squeezes a zigzag of brown sauce, dusts bonito flakes that flutter in the heat, and slides the finished round onto a blue-rimmed plate for a waiting customer; the fluorescent ceiling light is steady and slightly green, steam fogs the lens edge for a second, stools scrape, the cook keeps moving without pause, and the mood is sweaty, intimate, and triumphant with no scene cuts, hard transitions, or jumps in time or place.
Winner: Happy Horse — Model B matches the prompt much better: it shows an actual okonomiyaki on a griddle with handheld progression from close-up to wider plating context, coherent cooking actions, and the cramped midnight bar atmosphere with fluorescent lighting and steam. Model A looks more like teppanyaki skewers/vegetables than an okonomiyaki, and its action sequence and subject fidelity are notably off despite decent image quality.
See every prompt and the full side-by-side outputs in the interactive Head-to-Head.