Shoulders-up shot of a man in a hard hat on a building site who points over his shoulder at the structure behind him and says: "The second floor goes up next week." Spoken audio with natural lip sync, wind and site noise underneath, bright overcast daylight.

Talking Head & Lip Sync · default settings, one run, 5 s, 720p, 16:9 · Updated September 20, 2026

Not run yet
MiniMax FastH3 V2
Not run yet
P-Video 2 Pro
Not run yet
Wan 3.0
Not run yet
Wan 3.0 Prime
FastH3 V2
4.6videomodels.ai score
  • Prompt adherence4.5
  • Motion & consistency4.5
  • Camera & framing4.5
  • Quality & detail4.5
  • Look & lighting4.5
  • Realism & physics4.5
  • Text rendering—
  • Audio5.0

Clear over-the-shoulder pointing gesture with matching spoken line.

Seedance 2.5
3.4videomodels.ai score
  • Prompt adherence3.0
  • Motion & consistency3.5
  • Camera & framing4.0
  • Quality & detail3.5
  • Look & lighting3.5
  • Realism & physics3.5
  • Text rendering—
  • Audio—

The delayed pointing gesture is clear, but the required audio is absent.

Seedance 2.0
3.5videomodels.ai score
  • Prompt adherence3.0
  • Motion & consistency4.0
  • Camera & framing3.5
  • Quality & detail3.5
  • Look & lighting3.5
  • Realism & physics3.5
  • Text rendering—
  • Audio—

The thumb-over-shoulder gesture reads, but speech and audio are absent.

Seedance 2.0 Fast
3.9videomodels.ai score
  • Prompt adherence3.5
  • Motion & consistency4.0
  • Camera & framing4.0
  • Quality & detail4.0
  • Look & lighting4.0
  • Realism & physics4.0
  • Text rendering—
  • Audio—

The pointing and visible talking work well, but the required audio is absent.

Kling 3.0 Turbo
3.4videomodels.ai score
  • Prompt adherence2.5
  • Motion & consistency4.0
  • Camera & framing3.5
  • Quality & detail4.0
  • Look & lighting4.0
  • Realism & physics4.0
  • Text rendering—
  • Audio2.0

A clean, realistic backward point and site setting are undercut by absent speech.

Veo 3.1
3.9videomodels.ai score
  • Prompt adherence3.5
  • Motion & consistency4.0
  • Camera & framing4.5
  • Quality & detail4.0
  • Look & lighting4.0
  • Realism & physics4.0
  • Text rendering—
  • Audio—

The over-shoulder point and visible talking are strong, but the required audio is absent.

Veo 3.1 Fast
3.9videomodels.ai score
  • Prompt adherence3.5
  • Motion & consistency4.0
  • Camera & framing4.5
  • Quality & detail4.0
  • Look & lighting4.0
  • Realism & physics4.0
  • Text rendering—
  • Audio—

The presenter gesture and lip movement are convincing, but the required audio is absent.

Happy Horse 1.1
3.0videomodels.ai score
  • Prompt adherence2.5
  • Motion & consistency3.5
  • Camera & framing3.0
  • Quality & detail3.5
  • Look & lighting3.5
  • Realism & physics3.5
  • Text rendering—
  • Audio2.0

The gesture and site are clear, but the required spoken audio is absent.

MiniMax H3
3.4videomodels.ai score
  • Prompt adherence2.5
  • Motion & consistency4.0
  • Camera & framing3.5
  • Quality & detail4.0
  • Look & lighting4.0
  • Realism & physics4.0
  • Text rendering—
  • Audio2.0

A stable realistic site presentation with a clear point, but the dialogue is absent.

Gemini Omni Flash
4.1videomodels.ai score
  • Prompt adherence3.5
  • Motion & consistency4.0
  • Camera & framing4.5
  • Quality & detail4.0
  • Look & lighting3.5
  • Realism & physics4.5
  • Text rendering—
  • Audio5.0

Speech is exact, but the backward gesture is a weak fist rather than a clear point.

Wan 2.7
3.4videomodels.ai score
  • Prompt adherence3.0
  • Motion & consistency4.0
  • Camera & framing3.5
  • Quality & detail4.0
  • Look & lighting3.5
  • Realism & physics4.0
  • Text rendering—
  • Audio2.0

The pointing action reads clearly, but the required speech is absent.

Grok Imagine Video
3.7videomodels.ai score
  • Prompt adherence4.0
  • Motion & consistency3.0
  • Camera & framing3.5
  • Quality & detail3.5
  • Look & lighting4.0
  • Realism & physics3.5
  • Text rendering—
  • Audio5.0

The action and exact line are present, with softer dynamic motion and warmer-than-overcast light.

P-Video 2
3.6videomodels.ai score
  • Prompt adherence3.0
  • Motion & consistency3.5
  • Camera & framing4.0
  • Quality & detail4.5
  • Look & lighting4.5
  • Realism & physics4.0
  • Text rendering—
  • Audio2.0

The close construction portrait looks excellent, though speech audio is absent and the hand softens.

Runway Gen-4.5
3.4videomodels.ai score
  • Prompt adherence3.0
  • Motion & consistency3.5
  • Camera & framing4.0
  • Quality & detail4.0
  • Look & lighting4.0
  • Realism & physics4.0
  • Text rendering—
  • Audio2.0

A believable pointing performance is undermined by the final rear-facing continuity jump.

Hailuo 2.3
3.5videomodels.ai score
  • Prompt adherence2.5
  • Motion & consistency4.0
  • Camera & framing4.5
  • Quality & detail4.0
  • Look & lighting4.0
  • Realism & physics4.0
  • Text rendering—
  • Audio2.0

Clear sustained over-shoulder pointing, but the required spoken line is absent.

Kling 2.6
3.4videomodels.ai score
  • Prompt adherence2.5
  • Motion & consistency4.0
  • Camera & framing3.5
  • Quality & detail4.0
  • Look & lighting4.0
  • Realism & physics4.0
  • Text rendering—
  • Audio2.0

The turn and sustained point are coherent, but no required dialogue is present.

FLUX 3
4.2videomodels.ai score
  • Prompt adherence4.5
  • Motion & consistency4.0
  • Camera & framing3.5
  • Quality & detail4.0
  • Look & lighting4.0
  • Realism & physics4.0
  • Text rendering—
  • Audio5.0

The spoken line, gesture, PPE and construction setting all land cleanly.

PixVerse V5
3.3videomodels.ai score
  • Prompt adherence3.0
  • Motion & consistency3.0
  • Camera & framing3.5
  • Quality & detail4.0
  • Look & lighting4.0
  • Realism & physics4.0
  • Text rendering—
  • Audio2.0

The sustained point and site look are good, but the action is somewhat jumpy and too wide.

Seedance 1.5 Pro
3.9videomodels.ai score
  • Prompt adherence3.5
  • Motion & consistency4.0
  • Camera & framing4.5
  • Quality & detail4.0
  • Look & lighting4.0
  • Realism & physics4.0
  • Text rendering—
  • Audio—

The gesture and visible speech are strong, but no required audio is present.

FastH3
4.1videomodels.ai score
  • Prompt adherence4.0
  • Motion & consistency4.0
  • Camera & framing3.5
  • Quality & detail4.5
  • Look & lighting4.5
  • Realism & physics4.0
  • Text rendering—
  • Audio4.5

It includes the target line, but substantial unrelated speech precedes it.

Cosmos 3 Super
3.4videomodels.ai score
  • Prompt adherence3.0
  • Motion & consistency3.0
  • Camera & framing3.5
  • Quality & detail3.5
  • Look & lighting4.0
  • Realism & physics3.5
  • Text rendering—
  • Audio4.5

The construction gesture is clear, but the spoken content is unrelated to the requested line.

Lance
3.4videomodels.ai score
  • Prompt adherence2.5
  • Motion & consistency4.0
  • Camera & framing4.0
  • Quality & detail4.0
  • Look & lighting4.0
  • Realism & physics4.0
  • Text rendering—
  • Audio2.0

The clean pointing shot is too wide and has no required spoken audio.

Wan T2V
3.1videomodels.ai score
  • Prompt adherence3.0
  • Motion & consistency3.5
  • Camera & framing4.0
  • Quality & detail3.0
  • Look & lighting2.5
  • Realism & physics3.5
  • Text rendering—
  • Audio2.0

The gesture is clear, but garbled overlays and missing speech hurt the result.

HunyuanVideo 1.5
2.9videomodels.ai score
  • Prompt adherence2.5
  • Motion & consistency3.0
  • Camera & framing3.5
  • Quality & detail3.0
  • Look & lighting3.5
  • Realism & physics3.0
  • Text rendering—
  • Audio2.0

A plausible site gesture, but arm ghosting and absent speech weaken the result.

Happy Horse 1.0
4.4videomodels.ai score
  • Prompt adherence4.5
  • Motion & consistency4.0
  • Camera & framing4.5
  • Quality & detail4.5
  • Look & lighting4.0
  • Realism & physics4.0
  • Text rendering—
  • Audio5.0

A clean close-up with the exact line and a readable over-shoulder building gesture.

Helios 14B
2.9videomodels.ai score
  • Prompt adherence2.5
  • Motion & consistency3.0
  • Camera & framing3.5
  • Quality & detail3.5
  • Look & lighting3.0
  • Realism & physics3.5
  • Text rendering—
  • Audio2.0

The pointing action reads, but harsh backlight and absent speech miss key requirements.

Seedance 1.0 Pro Fast
3.4videomodels.ai score
  • Prompt adherence3.0
  • Motion & consistency3.5
  • Camera & framing4.0
  • Quality & detail3.5
  • Look & lighting3.5
  • Realism & physics3.0
  • Text rendering—
  • Audio—

It gestures toward the structure, but the required spoken audio is absent.

Wan 2.6
3.5videomodels.ai score
  • Prompt adherence3.0
  • Motion & consistency4.0
  • Camera & framing4.0
  • Quality & detail4.0
  • Look & lighting3.5
  • Realism & physics4.0
  • Text rendering—
  • Audio2.0

Clear site gesture, but the required spoken line is absent.

Kling 3.0 Omni
3.3videomodels.ai score
  • Prompt adherence3.0
  • Motion & consistency3.5
  • Camera & framing3.0
  • Quality & detail4.0
  • Look & lighting4.0
  • Realism & physics3.5
  • Text rendering—
  • Audio2.0

Pointing action is clear, but speech and its required audio are absent.

P-Video
3.0videomodels.ai score
  • Prompt adherence2.5
  • Motion & consistency3.0
  • Camera & framing3.0
  • Quality & detail4.0
  • Look & lighting4.0
  • Realism & physics3.5
  • Text rendering—
  • Audio2.0

Tight edge cropping and a blurred, discontinuous pointing arm weaken the take.

Kling 2.5 Turbo
3.0videomodels.ai score
  • Prompt adherence2.5
  • Motion & consistency3.5
  • Camera & framing3.0
  • Quality & detail3.5
  • Look & lighting3.5
  • Realism & physics3.5
  • Text rendering—
  • Audio2.0

The requested pointing performance is clear, but the speech track is missing.

Seedance 1.0 Pro
3.6videomodels.ai score
  • Prompt adherence3.0
  • Motion & consistency4.0
  • Camera & framing4.0
  • Quality & detail4.0
  • Look & lighting3.5
  • Realism & physics3.5
  • Text rendering—
  • Audio—

The pointing action is clear, but the required spoken audio is absent.

PixVerse V4.5
3.2videomodels.ai score
  • Prompt adherence2.5
  • Motion & consistency3.5
  • Camera & framing3.5
  • Quality & detail3.5
  • Look & lighting4.0
  • Realism & physics3.5
  • Text rendering—
  • Audio2.0

The pointing site scene works, but wide back-facing framing and garbled subtitles hurt it.

PixVerse V4
3.3videomodels.ai score
  • Prompt adherence3.0
  • Motion & consistency3.5
  • Camera & framing3.5
  • Quality & detail4.0
  • Look & lighting4.0
  • Realism & physics3.5
  • Text rendering—
  • Audio2.0

A solid close construction shot with minor hand, face, and scaffolding instability.

Hailuo 02
3.6videomodels.ai score
  • Prompt adherence3.0
  • Motion & consistency4.0
  • Camera & framing4.0
  • Quality & detail4.5
  • Look & lighting4.0
  • Realism & physics4.0
  • Text rendering—
  • Audio2.0

The visual setup is strong, but no audio track supplies the requested spoken line.

Hailuo 01 Director
3.4videomodels.ai score
  • Prompt adherence3.0
  • Motion & consistency3.5
  • Camera & framing4.0
  • Quality & detail4.0
  • Look & lighting4.0
  • Realism & physics3.5
  • Text rendering—
  • Audio2.0

Clear over-shoulder point, but the clip has no spoken audio.