mesh-avatar-studio
Turn one illustration into an animated 2D mesh avatar with a coding agent and a local editor
翻译准备中
尚未检测到更新。GitHub 发布版本后会显示在这里。
个人 GitHub 账号下公开仓库的所有者可以登录并验证所有权,自行管理介绍,无需管理员批准。
登录并生成验证文件,在 30 分钟内将其提交到此仓库的默认分支。其他分支或分叉仓库中的文件无效。然后在下方检查文件,即可无需管理员批准直接管理项目。请保留文件以维持管理权限。
如需登录,将前往登录页面。登录后请回到此处,再次点击所选的验证按钮。
在项目文件夹中运行 /nomorevibe verify,即可验证所有权并自行管理介绍。
一分钟完成验证正在加载完整 README。 以下为已保存的摘录。
Browser tool that designs and exports App Store Connect marketing assets for iPhone, iPad, Mac, Apple Watch, plus product page headers, search results creatives, and dual-iPhone layouts.
Upload PNG/JPEG/WebP screenshots → design a consistent set (caption, fonts, templates, backgrounds, frames) → export individual PNGs or a ZIP of every selected size. All composition runs client-side.查看完整 README ↗Browser tool that designs and exports App Store Connect marketing assets for iPhone, iPad, Mac, Apple Watch, plus product page headers, search results creatives, dual-iPhone layouts (all iPhone ASC sizes), and an MIT foldable iPhone Duo frame.
Upload PNG/JPEG/WebP screenshots → guided flow (upload → devices → style → AI review → export) → export individual PNGs or a ZIP. Composition runs client-side. Optional AI (Vercel AI Gateway + AI SDK, with direct xAI fallback) calls only from Next.js route handlers.
localStorage).With device frames on:
Covered by unit tests in src/lib/geometry.test.ts.
First-class across all iPhone ASC sizes (6.9″ / 6.5″ / 6.3″ / 6.1″, portrait + landscape), plus the Dual platform target. Variants: side-by-side, overlapping, tilted. Each device uses its own screenshot from the ordered set.
Added from bunlongheng/frames (MIT): closed, open-portrait, and open-landscape PNG bezels under public/frames/iphone-duo/.
Why this source: Hoverify and Screeny Duo PNGs disallow redistribution as an asset pack; jal-co/iphone-duo ships Apple’s GLB (not openly licensed for bezel redistribution). bunlongheng/frames is MIT and already used for MacBook.
Source: Screenshot specifications
Source: Creative assets specifications
| Placement | Aspect | Resolution used here |
|---|---|---|
| Product page header | 21:9 | 3840×1646 |
| Universal creative | 16:9 | 5244×2950 |
| Search results | 3:2 | 3840×2560 (Apple max) |
Images must be JPEG/JPG/PNG without alpha; this tool always encodes opaque RGB PNGs.
Real PNG bezels from open community sources (see NOTICE):
jonnyjackson26/device-frames-media)bunlongheng/frames)bunlongheng/frames)npm install
cp .env.example .env.local # optional AI keys
npm run dev
Open http://localhost:43123.
npm run test
npm run lint
npm run build
npm start
main is what Vercel builds.computerjunges — production app-store-screenshots-computerjunges.vercel.app. Pushes to main deploy production; PRs get preview deployments.| Variable | Required | Default | Purpose |
|---|---|---|---|
AI_GATEWAY_API_KEY | No* | (unset) | Vercel AI Gateway key (preferred for local + preview). |
VERCEL_OIDC_TOKEN | No* | (auto on Vercel) | OIDC auth for Gateway on Vercel deployments. |
XAI_API_KEY | No* | (unset) | Direct xAI fallback when Gateway creds are absent. |
XAI_RATE_LIMIT_PER_IP_PER_MINUTE | No | 5 | Max AI route calls per client IP per minute |
XAI_RATE_LIMIT_PER_IP_PER_DAY | No | 30 | Max AI route calls per client IP per UTC day |
XAI_RATE_LIMIT_GLOBAL_PER_DAY | No | 200 | Max AI route calls across all IPs per UTC day |
XAI_MAX_PROMPT_LENGTH | No | 500 | Max characters for image-generation prompts |
XAI_VISION_MAX_EDGE | No | 768 | Longest edge (px) for vision payloads |
XAI_VISION_MAX_BASE64_CHARS | No | 900000 | Server reject threshold for vision payloads |
*AI UI enables when any of Gateway key, OIDC, or XAI_API_KEY is present. If none are set, the rest of the app works and AI panels show exactly what to configure.
| Feature | Gateway model id | Notes |
|---|---|---|
| Background / graphic generation | spacexai/grok-imagine-image | ~$0.02 / image (xAI Imagine) |
| Vision / review / layout / copy | spacexai/grok-build-0.1 | Cheapest vision-capable Grok text model |
Integration: Vercel AI SDK (ai + plain provider/model strings). Server routes:
GET /api/ai/status — enabled flag, provider path, models, caps (never returns keys)POST /api/ai/generate-image — background / graphicPOST /api/ai/analyze — style suggestions from a screenshotPOST /api/ai/copy — multilingual headline/caption packsPOST /api/ai/layout — auto-pick best device layoutPOST /api/ai/review — scorecard + one-click fixesRate limits are in-memory per server instance. For multi-instance production, replace with Redis/KV.
Listed in the UI. Includes freeform canvas dragging, per-shot caption overrides, app-preview video, In-App Event creatives, cloud preset sync, and additional official marketing frame packs.