Note: Bookmark-style intro on aogl.cn — not the vendor’s official docs. Features and pricing are on their site.
Gemini is my go-to Google chat surface for fast image look-dev and brainstorming copy variants before they land in data/articles/. It sits beside Claude in the tools directory because the products overlap on text chat but diverge on multimodal defaults and Google account integration.
How I use it on this site
- Hero stills — exploratory PNGs under
original/role-girl/,original/character/, and similar folders (watermarks kept as provenance, not removed here). - Palette checks — quick “does this read as the same outfit?” questions before I commit a contact sheet.
- Headline variants — three title options for changelog entries; I pick one and edit manually.
Workflow I actually follow
- Start from
gemini.google.comlinked below—signed into the Google account I am willing to use for experiments. - State constraints: aspect ratio, wardrobe anchors, and “no logo” when generating art.
- Download to
original/<project>/with dated filenames; never overwrite canonical demo assets. - Write an honest article paragraph about what shipped vs what stayed in the look-dev pile.
- For API or Vertex details, leave this page and read Google’s developer docs.
Gemini vs Claude (personal split)
Claude gets long English rewrites and careful code explanations. Gemini gets rapid visuals and Google-ecosystem questions. Neither replaces the other in my bookmarks—I list both so I do not pretend one tab solves every task.
Official news on aogl.cn
When DeepMind publishes model posts, synced summaries may appear in briefs. Those pages link out; they are not substitutes for release notes.
Not for…
- Claiming generated stills are official brand assets.
- Bulk automated publishing without human review (violates how this site is maintained).
- Decisions that require licensed professional advice.
Examples on this site that used generative look-dev: role-girl card study, Velmora fiction planet, all tool guides.
Last checked: 2026-06-18.