# Imagegen triage

Built-in imagegen; two hypotheses from the same desktop/mobile screenshots.
Guide: ../../../IMAGEGEN-DESIGN-GUIDE.md.

| Observation | Decision | Reason |
| --- | --- | --- |
| B uses a buy/sell example to make the user's instruction tangible | Adopt | Visitors can explore one concrete choice instead of reading repeated descriptions |
| B groups the holding, chosen price and calendar | Adapt | Use native text and a compact price ticket; keep labels readable and selectable |
| B preserves a conditional asset mix on mobile | Adopt | Either asset or a mix is essential product meaning |
| A repeats prose across three columns and drops its result on mobile | Reject A as the final layout | The phone still loses the important outcome and becomes a long tutorial |
| Both retain the old “get paid to wait” headline | Reject | The source screenshot over-anchors the generation; waiting does not guarantee a return |
| A invents a price chart and a 1,000 USDC deposit | Reject | A fabricated market path is unnecessary to explain the user input |
| B labels outcomes “if little trades / if target fills” | Reject | Trading can occur in both directions; these conditions oversimplify the final inventory |
| B's dollar symbol replaces the quote unit | Adapt | State USDC per WETH explicitly |
| B leaves fixed-date/fee information off its phone board | Adapt | Keep fixed price/date and uncertain return visible, with fees in a native disclosure |
| Sculptural token/price/calendar objects provide visual grouping | Adapt | Build the grouping in HTML/CSS. Raster text would be small, inaccessible and impossible to switch correctly |

Chosen composition: B's worked example, simplified into a connected holding → instruction →
possible-outcome graphic. A cream price ticket gives the instruction priority inside the
existing espresso/ivory/coral identity. Native Buy/Sell radio controls change the illustrative
holding and price together, without JavaScript, wallet access, live pricing or storage.

Visible wording is deliberately edited after generation. A clear headline, one supporting
sentence, two actions and one example replace the long marketing explanation. No human
comprehension or conversion benefit is asserted from visual preference alone.