EnglishModels OverviewVeo 3.1 (Video)

Veo 3.1 Model

Overview

Veo 3.1 (Google DeepMind) is integrated as Veo 3.1 Fast in StoryFlow.

Capabilities

  • Fast Generation: Quicker results compared to Sora.
  • First/Last Frame Interpolation: Unique feature. Specify both the first and last frames, and Veo generates smooth transitions.

Inputs

  • Text (Prompt): Describe video content.
  • Reference Image:
    • Upload 1–2 images.
    • 1 image: used as the first frame.
    • 2 images: used as first and last frames, even if they differ greatly.

Parameters

ParameterTypeDefaultAllowedWhat it does
generationModestringFIRST_LAST_FRAMESFIRST_LAST_FRAMES, REFERENCE_2_VIDEOControls how reference images are interpreted. FIRST_LAST_FRAMES uses 1–2 images as start/end frames. REFERENCE_2_VIDEO uses 1–3 images as references.
aspectRatiostring16:916:9, 9:16Sets output aspect ratio.
resolutionstring720p720p, 1080pSets output resolution. 1080p is only available for 16:9.

Tips & Constraints

  • First/Last Frames: Connect 1–2 images; use 16:9 for 1080p.
  • Reference Mode: Only supported by veo3_fast; requires 16:9 and 1–3 images.
  • 1080p Retrieval: May take longer. If the provider falls back, you may receive 720p instead.