Video generation API
Seedance
ByteDance Seedance 2.5, 2.0, and 1.5 Pro video generation — state-of-the-art cinematic quality for T2V, I2V (with start/end keyframe support), and multi-asset reference-to-video up to 30 seconds.
Provider: ByteDance17 models
Available models
| Model key | Display name | Capability | Atlas Cloud slug |
|---|---|---|---|
seedance_2_5_t2v | Seedance 2.5 T2V | text-to-video (up to 30s, 4k char prompt) | bytedance/seedance-2.5/text-to-video |
seedance_2_5_i2v | Seedance 2.5 I2V | image-to-video (start & optional end frame) | bytedance/seedance-2.5/image-to-video |
seedance_2_5_ref2vid | Seedance 2.5 Ref2Vid | reference-to-video (multi image/video/audio refs) | bytedance/seedance-2.5/reference-to-video |
seedance_2_0_t2v | Seedance 2.0 T2V | text-to-video | bytedance/seedance-2.0/text-to-video |
seedance_2_0_i2v | Seedance 2.0 I2V | image-to-video | bytedance/seedance-2.0/image-to-video |
seedance_2_0_ref2vid | Seedance 2.0 Ref2Vid | reference-to-video | bytedance/seedance-2.0/reference-to-video |
seedance_2_0_fast_t2v | Seedance 2.0 Fast T2V | text-to-video | bytedance/seedance-2.0-fast/text-to-video |
seedance_2_0_fast_i2v | Seedance 2.0 Fast I2V | image-to-video | bytedance/seedance-2.0-fast/image-to-video |
seedance_2_0_fast_ref2vid | Seedance 2.0 Fast Ref2Vid | reference-to-video | bytedance/seedance-2.0-fast/reference-to-video |
seedance_2_0_mini_t2v | Seedance 2.0 Mini T2V | text-to-video | bytedance/seedance-2.0-mini/text-to-video |
seedance_2_0_mini_i2v | Seedance 2.0 Mini I2V | image-to-video | bytedance/seedance-2.0-mini/image-to-video |
seedance_2_0_mini_ref2vid | Seedance 2.0 Mini Ref2Vid | reference-to-video | bytedance/seedance-2.0-mini/reference-to-video |
seedance_v1_5_pro_t2v | Seedance 1.5 Pro T2V | text-to-video | bytedance/seedance-v1.5-pro/text-to-video |
seedance_v1_5_pro_i2v | Seedance 1.5 Pro I2V | image-to-video | bytedance/seedance-v1.5-pro/image-to-video |
seedance_v1_5_pro_t2v_fast | Seedance 1.5 Pro T2V Fast | text-to-video | bytedance/seedance-v1.5-pro/text-to-video-fast |
seedance_v1_5_pro_i2v_fast | Seedance 1.5 Pro I2V Fast | image-to-video | bytedance/seedance-v1.5-pro/image-to-video-fast |
seedance_v1_5_pro_i2v_spicy | Seedance 1.5 Pro I2V Spicy | image-to-video | bytedance/seedance-v1.5-pro/image-to-video-spicy |
Endpoint
POST /api/generateRequest body
| Parameter | Type | Required | Description |
|---|---|---|---|
modelKey | string | ✓ | One of the model keys listed above |
prompt | string | ✓ | Text description of the image to generate |
inputUrl | string | Input image URL (for edit/i2i models) | |
aspectRatio | string | "1:1", "16:9", "9:16", etc. | |
imageQuality | string | "Low", "Medium", "High" | |
styleRef | string | string[] | Style reference image URL(s) | |
charRef | string | string[] | Character reference image URL(s) | |
brandId | string | Brand DNA ID for brand-consistent generation | |
duration | number | Video length in seconds (5, 10, 15, 20, 25, 30 for Seedance 2.5; 5, 10 for Seedance 2.0) | |
endImageUrl | string | End frame image URL for keyframe interpolation (Seedance 2.5 I2V & Seedance 2.0 I2V) | |
refImageUrls | string[] | Array of reference image URLs for character/style consistency (Ref2Vid models) | |
refVideoUrls | string[] | Array of reference video URLs for motion/style reference (Seedance 2.5 Ref2Vid) | |
generateAudio | boolean | Auto-generate matching synchronous audio |
Code example
curl -X POST https://api.oyefilmy.com/api/generate \
-H "Authorization: Bearer of_live_YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"modelKey": "seedance_2_5_t2v",
"prompt": "A slow-motion cinematic shot of a dancer spinning under golden hour light, silk fabric flowing"
}'Response
{
"success": true,
"predictionId": "pred_abc123...",
"status": "processing",
"model": "bytedance/seedance-2.5/text-to-video",
"creditCost": 5
}Seedance 2.5 is the flagship model with video generation up to 30s and extended 4,000-character prompt budget. Seedance 2.0 includes Fast (~2x faster) and Mini (cost-efficient) variants. Ref2Vid supports passing reference images and videos to maintain character and visual style consistency across clips.