Skip to content

Seedance (via Fal.ai)

ByteDance Seedance v1, v1.5, and v2.0 models hosted on Fal.ai.

Quick Example

from tarash.tarash_gateway import generate_video
from tarash.tarash_gateway.models import VideoGenerationConfig, VideoGenerationRequest

config = VideoGenerationConfig(
    provider="fal",
    model="bytedance/seedance-2.0/text-to-video",
    api_key="YOUR_FAL_KEY",
)

request = VideoGenerationRequest(
    prompt="A panda eating bamboo in a lush forest",
    duration_seconds=5,
)

response = generate_video(config, request)
print(response.video)

Supported Models

Model prefix Duration Image-to-Video Notes
fal-ai/bytedance/seedance 2s–12s ✅ Seedance v1/v1.5; reference-to-video
bytedance/seedance-2.0 4s–15s ✅ Seedance v2.0; text-to-video and image-to-video

Parameters

Seedance v2.0 (bytedance/seedance-2.0/*)

Parameter Required Notes
prompt ✅
duration_seconds — 4–15 seconds
resolution — 480p, 720p, 1080p
aspect_ratio — 16:9, 9:16, 1:1, 21:9, 4:3, 3:4
generate_audio — Synchronized audio generation
seed —
image_list (reference) — Starting frame for image-to-video
image_list (last_frame) — Ending frame for image-to-video
extra_params.end_user_id — User identifier

Seedance v1/v1.5 (fal-ai/bytedance/seedance/*)

Parameter Required Notes
prompt ✅
duration_seconds — 2–12 seconds
image_list (reference) — Image-to-video / reference-to-video