Releases: kwistzzqq-byte/image2-ads-studio
Release list
Image2 Ads Studio v0.2.0: Bilingual Community Edition
Image2 Ads Studio v0.2.0: Bilingual Community Edition
This release makes Image2 Ads Studio easier to understand and test for both English and Chinese users.
Image2 Ads Studio is an open-source advertising prompt agent for turning business briefs and reference images into debranded, production-ready prompts for Image2 or other image-generation tools. It is not a prompt dump: it parses the brief, retrieves advertising templates and visual recipes, runs an LLM prompt brain, checks deterministic execution details, and outputs an optimized prompt.
What's New
- Bilingual README entry: English remains the GitHub default, with a full Chinese README for Chinese communities.
- Bilingual Web UI: automatically follows browser language and supports manual
中文 / Englishswitching. - Output-language control: generated prompts can now be produced in
zh-CNoren. - Bilingual gallery: 100 public cases now include Chinese titles, Chinese briefs, Chinese optimized prompts, English prompts, source use notes, and risk notes.
- New Chinese gallery page:
examples/gallery/cases.zh-CN.md. - Stronger community validation: the export validator now checks bilingual gallery fields.
Current Open Assets
- 320 advertising business templates
- 240 visual recipes
- 100 gallery cases
- Local Web workbench
- LLM prompt brain interface
- Deterministic prompt validation
- Manual Image2 testing workflow
中文说明
这版主要解决中文用户看不懂项目入口的问题:README、案例库和 Web UI 都已经支持中英双语。现在可以用中文输入商业需求,让系统输出中文 optimized prompt;也可以切到英文,用英文 prompt 去测试海外生图工具。
项目的目标不是收集一堆 prompt,而是把“帮我做一张奶茶海报”“产品图高级一点”“门店活动要发朋友圈”这类口语化商业需求,整理成更稳定、可执行、可记录的广告作图 prompt。
Feedback Wanted
The next question is product direction: should the project prioritize a real Image2 adapter, batch SKU / ecommerce planning, or stronger before-after evaluation cases?
Image2 Ads Studio Community Edition v0.1.0
Image2 Ads Studio is an advertising prompt agent for Image2 and other image-generation tools.
This Community Edition focuses on the prompt-planning layer before real image generation:
- Local Web prompt workbench
- Core prompt agent framework
- 220 advertising business templates
- 140 visual recipes
- 100 one-image-one-optimized-prompt gallery cases
- LLM prompt brain interface
- Deterministic prompt validation
- Reference image policy
- Manual Image2 testing workflow
The project is intentionally different from a prompt dump. Prompt galleries answer "what prompt can I copy?" Image2 Ads Studio answers "given this business brief, copywriting, industry, aspect ratio, and reference image policy, what optimized advertising prompt should the agent produce?"
Core pipeline:
Brief -> Intent Parser -> Advertising Templates -> Visual Recipes -> LLM Prompt Brain -> Deterministic Checks -> Optimized Prompt
Useful links:
- Repo: https://github.com/kwistzzqq-byte/image2-ads-studio
- Gallery: https://github.com/kwistzzqq-byte/image2-ads-studio/blob/main/examples/gallery/cases.md
- Launch kit: https://github.com/kwistzzqq-byte/image2-ads-studio/blob/main/docs/launch-kit.md
- Announcement discussion: #1
- Test case discussion: #2
Current scope does not call a real image-generation API yet. The release makes the prompt planning layer open, inspectable, and useful before connecting real Image2 / OpenAI Images adapters.