Experiment · active

Typography and poster legibility

Can the model render exact short text while preserving a premium poster composition?

Runs
5
Variable
text length and hierarchy
Prompt
typography-posters-prompt v1.0

Protocol

Active round-one controlled comparison using one frozen prompt across five hosted model endpoints or quality tiers.

Frozen prompt template

CONTROL PROMPT AB-R1-TYPOGRAPHY-001. A wide 16:9 premium art-deco research poster in midnight navy, aged brass, cream, and one restrained red accent. Render exactly two lines of large centered readable text: first line MEASURE THE MAGIC and second line NOT THE MARKETING. Include a small clockwork owl emblem above the text and thin geometric fan-line ornament. Institutional museum-archive print design, crisp letterforms, balanced margins. No other letters, words, numbers, signatures, logos, or watermark.

Review rubric

  • Typography
  • Prompt Adherence
  • Composition

Notes

One unseeded generation per model. Exact visible deviations are preserved in each run's review notes.

Evidence

Recorded runs

Generated specimen r1-typography-grok-imagine r1-typography-grok-imagine
successgrok-imagine-image

R1 Typography Grok Imagine

Vision adherence score 10.0/10. Exact two-line text with no additional generated words: MEASURE THE MAGIC / NOT THE MARKETING. Strong owl emblem, art-deco ornament, palette, legibility, and margins.

Size
1280×720
Seed
not exposed
Method
first party tool
View experiment →
Generated specimen r1-typography-minimax-image-01 r1-typography-minimax-image-01
successimage-01

R1 Typography Minimax Image 01

Vision adherence score 6.0/10. The two requested headline lines are spelled correctly, but an additional third line of pseudo-text appears below, violating the exact-text and no-other-words constraints.

Size
1280×720
Seed
not exposed
Method
official api
View experiment →
Generated specimen r1-typography-openai-gpt-image-2-high r1-typography-openai-gpt-image-2-high
successgpt-image-2-high

R1 Typography Openai Gpt Image 2 High

Vision adherence score 9.6/10. Exact two-line text with no additional generated words, excellent owl emblem and art-deco construction. Slightly heavier ornamental density than requested but fully legible.

Size
1672×941
Seed
not exposed
Method
official api
View experiment →
Generated specimen r1-typography-openai-gpt-image-2-low r1-typography-openai-gpt-image-2-low
successgpt-image-2-low

R1 Typography Openai Gpt Image 2 Low

Vision adherence score 9.7/10. Exact two-line text with no additional generated words, strong clockwork owl emblem, art-deco framing, and clear hierarchy. Composition is denser than the Grok variant.

Size
1672×941
Seed
not exposed
Method
official api
View experiment →
Generated specimen r1-typography-openai-gpt-image-2-medium r1-typography-openai-gpt-image-2-medium
successgpt-image-2-medium

R1 Typography Openai Gpt Image 2 Medium

Vision adherence score 9.8/10. Exact two-line text with no additional generated words, excellent legibility, coherent mechanical owl emblem, restrained red accent, balanced margins, and strong art-deco geometry.

Size
1672×941
Seed
not exposed
Method
official api
View experiment →