Yan is an English-language AI prompt workshop. The site exists to give designers, illustrators, photographers, marketers and indie creators a deep, reusable model for writing AI image and video prompts — without depending on a paid SaaS, a cloud API, or a "magic 8K masterpiece" template.
What we publish
- Guides — long-form pieces on the structural fundamentals of AI prompt writing: subject formula, negative prompts, product photography, video, character consistency.
- Examples — real prompt teardowns across portrait, product poster, illustration and video shots, each with three rewrites for different scenarios.
- Articles — focused deep dives on specific parameters and platforms (Midjourney --stylize, SDXL vs Flux, LoRA, ControlNet, AI video model comparison, commercial licensing, and more).
What we build
Beyond the writing, the site ships local tools that run entirely in your browser: a structured editor (write subject / style / light / composition / parameters as separate blocks), a template library, a generator with tag-based picking, a reference-image organizer, a prompt lab, and a personal library to save your best prompts. None of these tools call a cloud model. Your prompts, references and history never leave your machine.
Who writes this
The site is edited by a small team that uses AI image and video tools daily for client work and personal projects. Every guide and teardown is written from working prompts — not theory, not parroting other tutorials. When a recipe changes (and they change often), the relevant page gets updated and stamped with a new date.
Editorial principles
We prefer explainable prompt structures over large keyword piles. When a page recommends a template, parameter or negative prompt, it should also explain when to use it, what problem it solves, and where it can fail. Model-specific pages are kept separate where possible because Midjourney, SDXL, Flux, Runway, Kling and Sora do not respond to prompts in exactly the same way.
Local-first privacy
The tools on this site are designed as browser-side helpers. They do not send your prompts to a model API. Saved prompts, history and workspace data depend on local browser storage, so exporting important work is recommended before clearing browser data or switching devices.
Corrections and updates
If you find a factual mistake, broken link, outdated parameter range or unclear workflow, contact us with the page URL and a short note. We prioritize corrections that affect user decisions, reproducibility, privacy or commercial usage.
Languages
The site is available in both English and 中文. Use the language switch in the top right to flip between versions. Most pages have direct equivalents in each language.