MAJORFailure Reference

Luma Dream Machine Motion Failure — Refund Guide

Technical Classification

Temporal Coherence Degradation

Temporal Coherence Degradation on Luma occurs when the temporal attention window between frames partially collapses, causing the model to either freeze on a single frame for multiple time-steps or to drift motion direction without prompt change. This is particularly common past the 4-second mark on Dream Machine — the longer the clip, the more attention windows have an opportunity to degenerate.

How to identify this failure

✕Mid-clip frame freeze for 200-500ms
✕Motion direction changes without prompt change
✕Motion stutter (forward, back, forward) in a clip that should be unidirectional
✕Final second of a 5-second clip noticeably degraded vs first second

Real generation examples

Prompt used

"A skateboarder rolling forward down a street"

Failure observed @ 0:03

Skateboarder freezes at 0:03, resumes at 0:04 in a slightly different position

Documentation strength

If you need to escalate

HIGH — Luma support responses temporal-coherence failures on paid Dream Machine output with timestamps.

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 Temporal Coherence Degradation 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 Temporal Coherence Degradation 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 "Temporal Coherence Degradation". 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 Luma support response credits for motion stutter?

Yes — temporal coherence failures are refundable on paid output. Cite "Temporal Coherence Degradation" with the timestamp.

Why does Luma stutter or freeze mid-clip?

Temporal attention windows between frames partially collapse, causing the model to repeat or freeze on a single frame for multiple decoder steps.

How do I reduce Luma motion failures?

Keep clips under 4 seconds where possible. Avoid complex multi-direction motion in a single prompt. AVA pre-flights long-duration motion prompts.

Catch it before you generate

AVA scores this failure mode against your prompt in real time

Free Chrome extension. Analyzes your prompt as you type, flags failure-prone patterns specific to this model, and tells you what to rewrite — before you commit credits to a generation that will fail.

Install free All 105 failure modes

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.

Claim $50 founders