Veo vs Kling vs Runway vs Seedance vs Wan: Prompt Formats Compared

Five video models make up most creators' toolkits right now: Veo, Kling, Runway, Seedance and Wan. They can all turn a still into a moving shot, but they don't read prompts the same way. They differ on clip length, aspect ratio, audio, negative prompts and how they handle timing, and a prompt written for one often underperforms on another.

Here's how they compare, and how to write the same shot for each.

The specs below were checked in September 2026 against each platform's own guides. Video models update often, so check again before you rely on a limit.

The shot we'll translate

Start frame: a woman in a red wool coat on a rainy neon street at night. Action: she looks up from her phone, then turns toward the camera and smiles. Camera: slow push-in. Sound: rain, distant traffic, one soft line of dialogue.

In image to video, the start frame already sets the look, so every version below is about motion and sound. Our image to video guide explains why.

A woman in a red wool coat looking down at her phone on a rainy neon-lit street at night — The start frame for the shot: she is still looking at her phone, before any motion. Every prompt below starts from this one image. Made with GPT Image from a prompt built with Frame Juice Image settings: Photograph, Medium Shot, Eye Level, 50mm, f/1.8, Neon and Urban, Cinestill 800T, Rain.

The prompt behind this image:

sharp subject: a woman in a red wool coat looking down at her phone on a rainy street at night, blurred background: neon signs and wet pavement reflecting coloured light, real photograph captured with a camera, medium shot, waist-up framing, eye level camera angle, neutral natural perspective, 50mm standard prime lens, mirrors human eye perspective, neutral compression, timeless rendering, f/1.8 wide aperture, beautiful bokeh, smooth background separation, urban neon lighting, mixed artificial light sources, vibrant color casts, night city atmosphere, Cinestill 800T tungsten cinema film stock, iconic red halation glow, cinematic night atmosphere, warm tungsten color, falling rain, wet glossy surfaces, puddle reflections, rain droplets, professional photography, highly detailed, photorealistic.

Veo

Lengths: 4, 6 or 8 seconds. Ratios: 16:9 and 9:16. Audio: yes, native.

How it reads prompts: camera first, then subject, action, context and style. Dialogue goes in quotes. Sound effects and ambience can sit on their own lines, for example SFX: and Ambient noise:. Timed beats can be written as timestamps like [00:00-00:04]. There's a separate negative prompt field in Flow and Vertex AI.

Best at: cinematic realism with synchronised sound and dialogue.

Slow push-in. The woman looks up from her phone, then turns toward the camera and smiles. She says softly: "You came back." SFX: rain on the pavement. Ambient noise: distant traffic.

A woman in a red wool coat, closer to the camera, turned toward it and smiling on a rainy neon street — The last beat of the same shot: the push-in has brought us closer, she has turned and smiles. This is the moment the dialogue line lands, so it is where Veo's native sound matters most. Made with GPT Image from a prompt built with Frame Juice Image and Motion settings: Cinema, Cinematic Drama, Dolly on Track, Push-In, 85mm, 180° Natural, Falling Rain, Medium Close-Up, f/1.8, Cinestill 800T.

The prompt behind this image:

sharp subject: still frame from the end of the shot, a woman in a red wool coat has turned toward the camera and smiles softly, her phone lowered, blurred background: a rainy neon street at night, cinematic motion picture still, 24fps filmic look, cinematic drama pacing, restrained motivated camera work, camera on a dolly track, perfectly smooth linear travel, the camera pushes in toward the subject, 85mm lens, flattering compression with a soft background, 180-degree shutter, natural cinematic motion blur, rain falling steadily, droplets splashing on surfaces, medium close-up, head and shoulders framing, f/1.8 wide aperture, beautiful bokeh, smooth background separation, Cinestill 800T tungsten cinema film stock, iconic red halation glow, cinematic night atmosphere, warm tungsten color, professional photography, highly detailed, photorealistic.

Kling

Lengths: 3 to 15 seconds. Ratios: 16:9, 1:1 and 9:16. Audio: yes.

How it reads prompts: scene, character, action, camera, audio and style, in that order. Dialogue works well written as Name (tone): line. Timed steps can be written as 0–2 s: lines. It has a negative prompt field, but works best with a short list of around six terms.

Best at: longer clips, strong motion and physical action, flexible lengths.

Rainy neon street at night. The woman. 0–2 s: she looks up from her phone. 2–5 s: she turns toward the camera and smiles. Slow push-in. Woman (soft, warm): "You came back." Rain and distant traffic.

Runway

Lengths: 2 to 10 seconds. Ratios: 16:9, 9:16, 1:1, 4:3, 3:4 and 21:9. Audio: no native audio.

How it reads prompts: in image to video, describe motion only and let the image set the look. Describe the subject in simple, general words. There's no negative prompt, so describe what you want rather than what you don't. Sequence actions with first, then, finally.

Best at: clean, controlled motion from a strong image, and the widest choice of ratios.

Slow push-in. First the woman looks up from her phone, then she turns toward the camera, finally she smiles. Rain keeps falling.

Seedance

Lengths: 4 to 15 seconds. Ratios: 21:9, 16:9, 4:3, 1:1, 3:4 and 9:16. Audio: yes.

How it reads prompts: time-coded beats in brackets, like [0-3s], work well. Keep dialogue short, roughly 12 words for a 10-second clip. No negative prompt.

Best at: precise timing, multi-beat action and cinematic ratios.

[0-3s] Slow push-in, the woman looks up from her phone. [3-6s] She turns toward the camera and smiles, saying softly: "You came back." Rain and distant traffic throughout.

One shot, five formats

Frame Juice Motion Pro writes the same shot in each of these formats, and its settings line tells you what to set in the tool itself: length, ratio, audio and which image goes where.

See Frame Juice Motion

Wan

Lengths: 5, 10 or 15 seconds. Ratios: 16:9, 4:3, 1:1, 3:4 and 9:16. Audio: yes.

How it reads prompts: positive and negative parts are split, including in ComfyUI workflows. Keep the positive prompt under about 800 characters. Sequence actions with first, then, finally.

Best at: open workflows, local and ComfyUI pipelines, and good results at lower cost.

Positive: Slow push-in. First the woman looks up from her phone, then she turns toward the camera and smiles. Rain falls, neon reflections ripple on the wet pavement.

Negative: face morphing, extra fingers, flicker, text, watermark.

At a glance

Longest clips: Kling and Seedance (up to 15 s), and Wan (15 s in fixed steps).
Most ratios: Runway and Seedance, both including 21:9.
No native audio: Runway.
No negative prompt: Runway and Seedance.
Timing: Veo and Seedance have their own timestamp formats, Kling uses second ranges, Runway and Wan follow first, then, finally.

Length, ratio and audio are settings in each platform's interface, not words in the prompt. Our five checks before rendering cover what to look at before you spend credits.

Hub platforms

Tools like ElevenLabs' image and video playground host several of these models in one place. That's convenient, but the prompt still needs the format of the model you pick there, not a generic one. Some models also aren't available in every region.

Wet city street at night lit by neon signs, traffic lights and car headlights — Neon Night Film, an image look in Frame Juice Image. A rainy neon street like this makes a strong start frame for any of the five models.

Which should you use?

For dialogue-driven cinematic scenes, Veo. For longer, physical, action-heavy clips, Kling. For controlled motion from a strong image in unusual ratios, Runway. For precisely timed multi-beat shots, Seedance. For open, local or budget workflows, Wan.

The honest answer for most creators: two of them, and the habit of writing each prompt in the format that model expects. The camera words themselves carry over between all five, so our camera moves vocabulary works wherever you render.

Veo vs Kling vs Runway vs Seedance vs Wan: Prompt Formats Compared

The shot we'll translate

Veo

Kling

Runway

Seedance

Wan

At a glance

Hub platforms

Which should you use?

More articles

Image to Video Prompts: What to Write When the Image Sets the Look

Camera moves for AI video: a working vocabulary

Five checks before you spend credits on AI video