Head to head: Bytedance Seedance V1.5 Pro Image To Video vs Luma Ray 3.2 Image to Video
This one is close on aggregate, but the split decision tells you exactly where each model lives. Seedance wins when the brief demands harder camera choreography and stricter prompt obedience; Luma still has the edge when atmosphere and intimate mood need to carry the shot.
By RuntimeWire · Published
The scores are tight — 15.8 to 15.3 — but the outcome isn’t mysterious. Bytedance Seedance V1.5 Pro Image To Video takes the matchup because it was better at the harder kind of compliance: not just generating a pretty scene, but actually honoring a very specific shot design, subject description, and motion brief.
That shows up clearly in Hail-Chase Sprint, where Seedance was the stronger, more disciplined model. It delivered the lean runner, the yellow case, the rain-slick levee/canal environment, and — crucially — the requested backward-tracking-to-side arc composition with convincing speed and spray under blue-gray storm light. Luma’s version had solid storm mood and even nice lightning, but it drifted from the character spec and never really locked into the stabilized close chase-shot grammar the prompt asked for.
Luma’s win in Lantern Keeper in Mist is real, and it’s earned. It better captured the intimate front-right tracking feel, the predawn indigo palette, the dense mist, and the visible warm lantern glow while keeping the character reasonably stable. Seedance looked sharper, but sharpness wasn’t the assignment; it missed the close-up mood for too much of the clip, introduced prompt mismatches like inconsistent coat details and a forehead mark, and was late getting the lantern into the visual logic of the scene.
So the editorial read is simple: Luma Ray 3.2 is the more naturally atmospheric model, but Seedance V1.5 Pro is the one I trust more when the prompt gets exacting. In a head-to-head, that matters more. Pretty vibes are common; reliable execution is rarer.
Final call: Bytedance Seedance V1.5 Pro Image To Video wins.
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 15.8 to Luma Ray 3.2 Image to Video's 15.3.
1. Hail-Chase Sprint
A 7-second continuous shot of a lean storm chaser named Elian Voss sprinting flat-out along the narrow concrete crown of a rain-slick levee beside a churning flood canal, clutching a yellow instrument case as pea-size hail begins to bounce around him; the camera races backward just ahead of his face and chest on a stabilized rig, then subtly arcs to his left while keeping pace, so his pumping stride, flapping jacket cords, and flying spray create a fierce sense of momentum, with realistic motion blur in the background reeds and distant lightning towers; cold blue-gray late-afternoon thunderstorm light, gusty crosswind, urgent high-adrenaline mood, 16:9
Winner: Bytedance Seedance V1.5 Pro Image To Video — Model A matches the prompt more closely with a lean runner, yellow case, rain-slick levee/canal setting, strong backward-tracking-to-side arc composition, and convincing speed/spray under blue-gray storm light. Model B has good storm mood and lightning, but the subject is less consistent with the described character and the framing/motion feel less like the specified stabilized close chase shot.
2. Lantern Keeper in Mist
A 9-second continuous shot of a young lighthouse keeper named Maris Vale walking steadily along a narrow cliff path above a heaving dark sea, carrying a dented brass storm lantern close to her chest; her appearance must remain perfectly unchanged throughout: short copper hair tucked behind the right ear, pale freckled face, moss-green waxed coat with two ivory buttons missing at the collar, charcoal wool scarf, left eyebrow notch, and a tiny silver gull pin on the coat lapel; the camera performs an intimate close-up tracking move from her front-right side, drifting backward at shoulder height as she advances through dense rolling fog and salt spray, with the lantern casting warm amber light across her features while the predawn sky stays cool indigo, creating a hushed, eerie, unwavering mood with no flicker, morphing, or wardrobe changes, 16:9
Winner: Luma Ray 3.2 Image to Video — Model B better matches the prompt’s intimate front-right tracking feel, predawn indigo palette, dense mist, and visible warm lantern glow while keeping the character fairly stable. Model A is sharper and detailed, but it misses the close-up mood for much of the clip, delays the lantern presence, and shows some prompt mismatches like inconsistent coat details and a forehead mark.
See every prompt and the full side-by-side outputs in the interactive Head-to-Head.