Seedance2.0 Real Human: FLF2V
Upload a reference image of a real person for identity verification and generate a personalized video with first and last frame control using Seedance2.0 FLF2V.
概要
Seedance2.0 Real Human: FLF2V is a ComfyUI workflow that verifies a real person’s identity once and then generates a personalized video where both the first and last frames are constrained by your chosen references. It uses ByteDanceCreateImageAsset to upload and verify a reference photo, returning an Asset or Access ID and a Group ID to keep all assets for the same person linked. Video generation is handled by the ByteDance2FirstLastFrameNode, which calls the Seedance 2.0 FLF2V model to condition the opening and closing frames while smoothly interpolating motion in between.
入力するもの
- ワークフローを読み込み中
ComfyUI で実行できます。 このワークフローはこのページでは実行できません。すべてが読み込まれた状態でComfy Cloudで開くか、ご自身のComfyUI用にダウンロードしてください。
Cloudで試すこのワークフローにはまだサンプル出力がありません。
よくある質問
ByteDanceCreateImageAsset returns an Access or Asset ID that represents a verified reference you can reuse instead of reuploading the same image. It also creates a Group ID after the first successful verification. Save the Group ID and use it when uploading additional assets of the same person so the backend recognizes them as one identity.
Not for the same frame. For the first frame, use either an Access ID or an image input, but not both at once. The same rule applies to the last frame. You can, however, use an Access ID for one frame and an image for the other.
No. Verify once to get a Group ID. For future assets of the same person, supply that Group ID in ByteDanceCreateImageAsset. You can then reference existing Access IDs or create new ones under the same Group ID as needed.
Use high-quality, front-facing images with good lighting and minimal occlusions for your first and last frames. Keep resolution and aspect ratio consistent with your target output, tune duration and FPS in ByteDance2FirstLastFrameNode, and iterate using PreviewAny before exporting with SaveVideo.