The Video Generation node creates new video clips from text prompts and media references. Choose a generation mode based on the inputs you have, select a compatible model, configure the output, and click [Run]. The completed video is passed to the Video output port so it can be connected to other nodes in the workflow.
1. Generation Modes
Select a mode from the tabs at the top of the Node Panel. The node’s input ports and quick-add controls change to match the selected mode.
1.1 Text to Video
Use [Text to Video] when you want to create a complete scene from a written description without supplying a visual reference.
- Prompt input: Connect a Text node to the Prompt port, or enter the description directly in the prompt box.
- Best for: Exploring new concepts, establishing shots, stylized scenes, and clips whose composition can be described entirely with text.
1.2 Image to Video
Use [Image to Video] to animate a still image or guide the generated clip with visual references.
- Prompt input: Describe the subject’s movement, camera movement, scene changes, and any details that should remain consistent.
- Image input: Connect one or more images through the Image port, or use [Image] in the Node Panel to add a reference. The maximum number depends on the selected model.
- Best for: Animating artwork or product images, preserving a subject’s appearance, and adding controlled motion to an existing composition.
1.3 First/Last Frame to Video
Use [First/Last Frame to Video] when the clip must begin from a specific image and optionally finish on another image.
- **First Frame\*:** Connect the required starting image to define the opening composition.
- Last Frame: Connect an optional ending image to guide where the movement, camera, or transition should finish.
- Prompt input: Describe how the subject and camera should move between the supplied frames.
- Best for: Planned transitions, before-and-after shots, controlled camera moves, and sequences that must arrive at a defined final composition.
1.4 Omni to Video
Use [Omni to Video] to combine several media types in one multimodal request.
- Prompt input: Explain how the connected references should influence the generated result.
- Image inputs: Connect images as subject, character, style, or composition references.
- Video inputs: Connect videos as motion, performance, or scene references.
- Audio inputs: Connect audio files to guide timing, rhythm, speech, or sound.
- Quick-add controls: Use [Image], [Video], and [Audio] in the Node Panel to add references directly.
- Best for: Reference-rich scenes that need coordinated visuals, movement, timing, and audio.
2. Node Settings
The controls at the bottom of the Node Panel determine how the video is generated. Available values can change when you switch models or generation modes because each model supports different inputs and output options.
2.1 Choose a Model
Select the model menu to choose the generation engine. Use the table below as a starting point; Fast, Mini, Standard, and Turbo tiers are useful for iteration, while full and Pro tiers generally prioritize final quality and control.
| Model | Best for |
|---|---|
| Seedance 2.5 | 30-second audiovisual storytelling that needs precise multimodal reference control and targeted video editing. |
| Seedance 2.0 | Multimodal, cinematic scenes that combine strong motion, reference media, and synchronized audio. |
| Seedance 2.0 Fast | Faster multimodal iterations when you want to test a scene without leaving the Seedance workflow. |
| Seedance 2.0 Mini | Quick drafts and lower-cost prompt tests before committing to a final generation. |
| Kling 3.0 Pro | High-detail cinematic clips, longer or more complex action, strong prompt adherence, and polished audio. |
| Kling 3.0 Standard | General-purpose video generation with a balance of quality, speed, and credit use. |
| Kling 3.0 Turbo Pro | Faster generations that still prioritize detail and production-ready visual quality. |
| Kling 3.0 Turbo Standard | Rapid, economical drafts for testing motion, framing, and prompt direction. |
| Veo 3.1 | Cinematic realism, strong prompt adherence, and polished audiovisual scenes. |
| Veo 3.1 Fast | Quicker Veo iterations when turnaround matters more than maximum output quality. |
2.2 Model Specifications
| Model | Duration | Aspect ratio | Resolution | Generate audio | Prompt length | Maximum reference inputs |
|---|---|---|---|---|---|---|
| Seedance 2.5 | 4–30 seconds | Auto, 16:9, 9:16, 1:1, 21:9, 4:3, 3:4 | 480P, 720P | ✅ | 3–20,000 characters | 30 images; 10 videos; 10 audio files |
| Seedance 2.0 | 4–15 seconds | Auto, 16:9, 9:16, 1:1, 21:9, 4:3, 3:4 | 480P, 720P, 1080P, 4K | ✅ | Up to 20,000 characters | 9 images; 3 videos; 3 audio files |
| Seedance 2.0 Fast/Mini | 4–15 seconds | Auto, 16:9, 9:16, 1:1, 21:9, 4:3, 3:4 | 480P, 720P | ✅ | 3–20,000 characters | 9 images; 3 videos; 3 audio files |
| Kling 3.0 Pro/Standard | 3–15 seconds | 16:9, 9:16, 1:1 | Not selectable | ✅ | 3–2,500 characters | 4 images; 2 first/last frames |
| Kling 3.0 Turbo Pro/Standard | 3–15 seconds | 16:9, 9:16, 1:1 | Not selectable | ❌ | Up to 3,072 characters | 1 image |
| Veo 3.1/3.1 Fast | 4, 6, or 8 seconds | 16:9, 9:16 | 720P, 1080P | ✅ | Up to 10,000 characters | 1 image |
2.3 Model Support by Generation Mode
| Model | Text to Video | Image to Video | First/Last Frame | Omni to Video |
|---|---|---|---|---|
| Seedance 2.0/2.5 | ✅ | ✅ | ✅ | ✅ |
| Kling 3.0 Pro/Standard | ✅ | ✅ | ✅ | — |
| Kling 3.0 Turbo Pro/Standard | ✅ | ✅ | — | — |
| Veo 3.1/3.1 Fast | ✅ | ✅ | — | — |
2.4 Output Settings
Configure the output controls after selecting a model. If a value is unavailable, that model does not support it in the current generation mode.
| Setting | What it does | How to use it |
|---|---|---|
| Duration | Sets the length of the generated clip. | Drag the slider or enter an available duration. Longer videos generally require more generation time and credits. |
| Ratio | Sets the video’s frame shape. | Choose [Auto], [16:9], [9:16], [1:1], [21:9], [4:3], or [3:4] when available. [Auto] lets the model follow the reference or its default framing. |
| Resolution | Sets the output dimensions and level of visible detail. | Choose [480P], [720P], [1080P], or [4K] when supported. Higher resolutions can take longer and use more credits. |
| Generate Audio | Controls whether the model creates an accompanying audio track. | Switch it [On] for generated sound or [Off] for a silent clip. Audio generation is only available on supported models. |
| Advanced Settings > Seed | Sets the random starting value used for the generation. | Keep the same prompt, reference inputs, model, output settings, and seed for a more similar variation. Enter a different value or use the refresh control to explore a new result. |
| Credit estimate | Shows the estimated credits for the next generation. | Review the number beside the credit icon. It updates when you change the model or output settings. |
| Run | Starts the video generation with the current inputs and settings. | Check the prompt, references, and credit estimate, then click [Run]. |