MAJORFailure Reference

Vidu Text Rendering Failure — Pre-Generation Risk Reference

Technical Classification

Symbolic Rendering Coherence Failure

Symbolic Rendering Coherence Failure on Vidu occurs when the model attempts to render legible text and produces non-language glyphs, hallucinated words, or character soup. This is consistent across all current video diffusion models — text is a known weakness — but Vidu has particularly high incidence on storefront signs, screen content, and book/poster text in the frame.

How to identify this failure

✕Storefront sign reads as non-words in Vidu output
✕Screen content shows hallucinated language
✕Book/poster text becomes random glyphs
✕License plates show impossible character combinations
✕Logo text spells the brand wrong

Real generation examples

Prompt used

"Streetview of a bakery storefront, 'Sweet Crumbs' on the sign, golden hour"

Failure observed

Sign read "Sweeit Crmbz" with broken kerning across all frames

Prompt used

"Laptop on desk showing a code editor with Python script"

Failure observed

Screen content was uniform glyph soup, no recognizable Python syntax

Documentation strength

If you need to escalate

HIGH — Text rendering failures on Vidu are recognized refund cases when the prompt specified the text and the output is illegible.

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 Symbolic Rendering 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 Symbolic Rendering 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 "Symbolic Rendering 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

Why is text rendering broken on all Vidu models?

Video diffusion models tokenize the image visually, not symbolically. Text is rendered as a pixel pattern that resembles letters but doesn't encode the underlying language. This is a known industry-wide limitation.

Should I avoid text in Vidu prompts?

For credit efficiency, frame text out of focus or out of frame. AVA L1 scanner flags text-heavy prompts before generation.

Does specifying the exact text in the prompt help on Vidu?

Marginally — current Vidu versions render specified text more legibly than open-ended signage, but failures remain common past 3-4 words.

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.

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