MAJORFailure Reference

Runway ML Style Preset Failure — Pre-Generation Risk Reference

Technical Classification

Style Conditioning & Aesthetic Adherence Failure

Runway Gen-4 occasionally drops style conditioning signals when the prompt includes both compositional content and style modifiers. The text encoder weights subject-noun tokens above style-modifier tokens, so "1970s film grain, soft focus, sepia tone" can be silently ignored in favor of the literal subject. The output looks reasonable but doesn't match the requested aesthetic — making it unusable for branded or art-directed work.

How to identify this failure

✕Requested film stock / grain pattern absent
✕Color treatment defaulted to clean digital instead of requested look
✕Lens characteristic (anamorphic, vintage glass) ignored
✕Era-specific styling (1970s, 1990s, 2010s) ignored
✕Style transferred to wrong elements (e.g. only background, not subject)

Real generation examples

Prompt used

"Anamorphic vintage glass, 1970s Kodachrome, soft focus, woman walking through a park"

Failure observed @ 0:00

Subject correct but style cues entirely absent — output looked like clean modern digital footage

Prompt used

"Super-8 grain, washed-out colors, kid riding bike down sidewalk"

Failure observed @ 0:00

Subject correct but grain + color treatment both ignored; output rendered as clean digital

Documentation strength

If you need to escalate

HIGH — Runway support refunds when style modifiers in the prompt are demonstrably absent in the output.

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 Style Conditioning & Aesthetic Adherence 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 Style Conditioning & Aesthetic Adherence 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 "Style Conditioning & Aesthetic Adherence 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

Will Runway support escalate credits for style preset failures?

Yes — when the style modifiers in your prompt are explicit and demonstrably absent in the output. Submit the prompt text, generation ID, and a screenshot showing the missing style treatment.

Why does Runway ignore my style cues?

The text encoder weights subject-noun tokens above style-modifier tokens. When the prompt is dense with compositional content, style cues are pushed below the truncation threshold.

How do I get style cues to land in Runway?

Put style modifiers FIRST in the prompt before the subject. Use 1-2 style anchors, not 5. Avoid mixing era-style ("1970s") with technical-style ("anamorphic") in the same prompt — pick one. AVA pre-flights style adherence.

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