Google Veo Anatomy Artifact — Pre-Generation Risk Reference
Technical Classification
Anatomical Topology & Coherence Failure
Veo's diffusion denoiser can produce non-manifold mesh geometry for human anatomy — supernumerary fingers, elbows that hinge in the wrong direction, and limbs that pass through each other. The failure mode is consistent across Veo 2 and Veo 3 and is most severe in close-up shots of hands, articulated motion, and multi-character interactions where occlusion-handling breaks down.
How to identify this failure
- ✕Six or more fingers on a single hand
- ✕Elbow or knee articulating backwards
- ✕Arm phasing through torso during motion
- ✕Fingers fusing into a paddle shape
- ✕Limb length changing between frames
Real generation examples
Prompt used
"Close-up of a chef chopping vegetables, natural light"
Failure observed @ 0:02
Right hand grew a seventh finger at 0:02; thumb articulated backwards at 0:04
Prompt used
"Two people shaking hands in an office lobby"
Failure observed @ 1:08
Hands fused into a single 9-finger mass during the handshake at 1:08
Documentation strength
If you need to escalate
VERY HIGH — Google support treats anatomical artifacts as a recognised critical failure; include the Generation ID and a timestamped failure point.
AVA is a pre-purchase prevention tool, not a post-purchase recovery tool. Platforms generally do not guarantee credit refunds for output-quality failures; goodwill credits are at each platform's discretion. The strength rating reflects how well-formed your support ticket can be, not a promised outcome.
Prevention + documentation steps
- 01
Score your prompt before you generate
Run your prompt through AVA's pre-flight scoring against the Anatomical Topology & Coherence Failure pattern. Green light = generate. Yellow/red = rewrite using the suggested fix before you commit credits.
- 02
Capture Generation ID + timestamp if it failed anyway
Find the Generation ID in the URL or share link. Note the exact time when the Anatomical Topology & Coherence Failure first appears (e.g. "failure first visible at 1.2s"). Timestamped evidence is significantly stronger than a general complaint.
- 03
Use the correct technical term in your support ticket
Describe this failure as "Anatomical Topology & Coherence Failure". This term maps to a recognised internal workflow in the support system and routes the ticket to the right team.
- 04
Submit via the correct support channel
Runway has no direct email intake. Pro+ plan: open the in-app AI Assistant (help widget bottom-right of app.runwayml.com), describe the failure with the technical term, attach evidence. Free/Standard plan: human support isn't available — your channel is Discord #community-help with @On Call - Moderators.
Frequently asked questions
Does Google Veo refund credits for anatomy artifacts?
Yes. Anatomy failures are recognised as a known critical mode. Submit a support request through your Google AI Studio billing page with the Generation ID, the technical term "Anatomical Topology & Coherence Failure", and a timestamp.
Why does Veo generate extra fingers?
Diffusion video models trained on millions of internet videos still struggle with hand topology — there is no explicit skeletal prior in the model, so finger count is statistical rather than enforced. AVA's L1 hand-risk classifier flags prompts most likely to fail before you spend credits.
Which Veo prompts are highest risk for anatomy artifacts?
Close-up hands, finger-detail tasks (typing, playing instruments, signing), multi-character handshakes/hugs, sports, and dance. Avoid these or pre-flight with AVA.
Score your prompt
Score your prompt against this failure mode in 30 seconds
Paste your prompt and the platform you intend to use. AVA returns a red/yellow/green score against this specific failure mode plus a concrete rewrite if the risk is high.
AVA Pro · founders' round
$50 for 6 months of unlimited scoring across all failure modes + personal failure-history dashboard. Locks in $13/mo grandfathered after.
Related failures across models
If you’re seeing this failure, you may also encounter these on other models:
Glyph
Text in the prompt renders as non-letters, mirrored glyphs, or unrelat…
Anatomical
Extra fingers, fused digits, impossible hand geometry in Veo close-up …
Multimodal
Veo 3 outputs silent track, mismatched ambience, or stylistically wron…
Physics
Veo output violates fundamental physics — fluid inversion, floating ob…
Motion
Veo output contains stilted, repeating, or frozen motion segments — su…
Camera-Conditioning
Veo output uses a static camera or generic camera motion instead of th…
Pick a different tool for Veo failures
Some prompt shapes will keep failing on Veo. Routing those shots to a different vendor is the cheapest fix.