GoodCase.aiGoodCase.aiAI Case Evidence Library
CasesSkillsCreatorsFavoritesSubmit
中文EN
Agent Access →
GoodCase.ai · AI Case Evidence LibraryLoading the case library.
ChangelogFeedback
GoodCase.aiGoodCase.aiAI Case Evidence Library
CasesSkillsCreatorsFavoritesSubmit
中文EN
Agent Access →

Case library · Public evidence

Start with the work. Trace it back to the maker and method.

No news summaries—only cases that point to finished work, creators, process, and original sources. Rankings, favorites, creator pages, and retests all grow from the same case.

Current view
7
Cases
Filter
All
Selected
AllAI VideoAI Coding (UI)AI ImageAI CopyAI Hardware
AI Coding (UI)X@DevBySami →

Tokyo Software Agency Landing Page

A web page case completed by @DevBySami using GPT-5.6 Sol · ShioriCode, including the public result, full prompt, and original source.

Skill · Portfolio story architectureSkill · Landing-page information structure

Prompt

GPT-5.6 Sol is way better at UI when not used in the Codex harness. It generated this landing page for a software agency with this prompt: "Design a beautiful landing page for a software agency based in Tokyo, Japan." Harness used: ShioriCode.

Source heat
24
Stability
待复测
View case →
AI Coding (UI)X@givros →

Codex Chinese Ancient Temple 3D Interactive Website

A website case by @givros made with Codex · GPT-5.6 Ultra · Blender · Three.js, including the public result, full prompt, and original source.

Skill · 3D and motion heroSkill · Landing-page information structure

Prompt

You are an expert Blender architectural artist, procedural 3D modeler, Python developer, and Three.js developer. Your task is to recreate the exact same traditional Chinese temple shown in the attached multi-view reference image, then build an interactive web page that displays the final 3D model and allows the user to inspect it from every angle. The attached image is the primary source of truth. It contains multiple views of the same temple, including front, rear, side, top-down, and three-quarter perspective views. Cross-reference all visible views continuously so the final result is one coherent building that matches the reference from every angle. Do not create a generic Chinese temple merely inspired by the image. Reproduce this specific temple as faithfully as possible, including: - Overall proportions. - Building footprint. - Two-level structure. - Roof silhouettes. - Roof curvature. - Roof overhangs. - Column layout. - Window and door placement. - Stone foundation. - Stair placement. - Railings. - Decorative elements. - Colors. - Materials. - Architectural rhythm. The final result must be a complete, editable 3D model and a working interactive web viewer. # CRITICAL BLENDER EXECUTION RULE Blender must be used exclusively through the terminal. Do not interact manually with the Blender graphical interface. Do not use: - Manual viewport manipulation. - Mouse-based modeling. - Manual object placement. - Manual material editing. - Manual modifier configuration. - Manual camera placement. - Manual rendering. - Manual exporting. - Any GUI-based Blender workflow. Do not ask the user to perform manual steps inside Blender. All Blender operations must be performed using Python scripts executed from the terminal. Use Blender in background mode whenever possible: ```bash blender --background --python scripts/build_temple.py

Source heat
69
Stability
待复测
View case →
AI Coding (UI)X@sergey_moloman →

Orbital North Space Travel Landing Page

A website case by @sergey_moloman created with GPT-5.6 Pro · Three.js, including the public result, full prompt, and original source.

Skill · 3D and motion heroSkill · Landing-page information structure

Prompt

GPT 5.6 Pro has pretty nice frontend capabilities! I decided testing frontend abilities because it was the "ouch" moment of ChatGPT since GPT 5.2 I used to pay Claude Max x5 before, just to create some nice frontends. Not anymore! Here are 3 examples of landing pages, one shot. Generated via ChatGPT website. 1. Orbital space tourism landing page Prompt: "Build a cinematic landing page for a private space tourism company called Orbital North. Use a full-bleed Three.js scene with Earth/orbit/starfield visuals. As the user scrolls, camera position, lighting, and content sections should change. The landing page should feel immersive and premium, with real product/service information, itinerary, safety section, pricing teaser, and booking CTA. Avoid generic gradients, decorative orbs, and marketing filler. The first screen must immediately communicate space travel."

Source heat
32
Stability
待复测
View case →
AI Coding (UI)X@sergey_moloman →

VEIL/FIELD Digital Fashion House Landing Page

A webpage case created by @sergey_moloman with GPT-5.6 Pro · Three.js, including the public result, full prompt, and original source.

Skill · 3D and motion heroSkill · Landing-page information structure

Prompt

2. Fashion house landing page Prompt: "Create an experimental landing page for a fictional digital fashion house called VEIL/FIELD. This should feel like an interactive editorial spread, not a product SaaS page. Use Three.js to create a surreal cloth, mirror, or liquid-metal scene where typography and 3D forms overlap in a high-fashion composition. The user scrolls or moves the pointer to distort fabric, reveal collection names, and transition between looks. Visual direction: avant-garde, editorial, fashion magazine, strange but polished. Large typography, asymmetry, negative space, art-directed composition. Requirements: - Three.js or react-three-fiber - cloth/liquid/mirror-like shader or convincing procedural material - unconventional layout - collection/product reveal - purchase or waitlist CTA - mobile version that still feels designed - no dashboard UI - no industrial/CAD look - no playful toy physics - verify with screenshots and fix layout overlaps"

Source heat
26
Stability
待复测
View case →
AI Coding (UI)X@sergey_moloman →

GPT-5.6 Robotics CAD Landing Page

A web case by @sergey_moloman using GPT-5.6 Pro, including the public result, full prompt, and original source.

Skill · Landing-page information structure

Prompt

3. Robotics landing page Prompt: "Build a landing page for a fictional robotics company called IRON INDEX. The page should feel like an industrial CAD terminal mixed with a technical spec sheet. Use Three.js to show an exploded mechanical robot arm or machine component with wireframes, measurement overlays, part labels, and toggles for systems like torque, cooling, sensors, and payload. Visual direction: brutalist, technical, dense, sharp, functional. Mostly monochrome with one warning/accent color. No soft gradients, no glassmorphism, no cinematic hero, no playful physics. Interaction style: - snapping panels - hover-to-inspect parts - toggle overlays - quick mechanical transitions - no smooth luxury scroll storytelling Include: - responsive layout - spec table - use cases - procurement CTA - technical credibility - visual verification screenshots"

Source heat
24
Stability
待复测
View case →
AI Coding (UI)X / 𝕏@servasyy_ai →

Image to 3D

A real coding/UI case that links to the open-source 3DCellForge project, turning image input into an interactive 3D workflow.

Skill · 3D and motion hero

Method / Code

View method ↗
Source heat
100
Stability
93
View case →
AI Coding (UI)X / 𝕏@viktoroddy →

Responsive 3D Character Carousel UI

A complete React implementation brief for a full-screen 3D figurine carousel, including assets, responsive behavior, animation timing, layout, and styling.

Skill · 3D and motion hero

Prompt

Access ALL prompts for stunning animated websites in one click: http://motionsites.ai Build a single full-viewport hero section in React + TypeScript + Vite + Tailwind CSS, using `lucide-react` for icons. The component is a character-figurine carousel called "TOONHUB". **Fonts (load in `index.html` head):** ```html <link rel="preconnect" href="https://fonts.googleapis.com" /> <link rel="preconnect" href="https://fonts.gstatic.com" crossorigin /> <link href="https://fonts.googleapis.com/css2?family=Anton&family=Inter:wght@400;500;600;700&display=swap" rel="stylesheet" /> ``` Body font: `'Inter', sans-serif`. Display font (huge ghost text + bottom-right link): `'Anton', sans-serif`. **Image data (4 items, exact URLs and colors):** ```ts const IMAGES = [ { src: 'https://fifth-gentle-45902158.figma.site/_components/v2/4de492f6d9cf8244ad5293233e5c6f52407d42fc/1.02464a56.png', bg: '#F4845F', panel: '#F79B7F' }, { src: 'https://fifth-gentle-45902158.figma.site/_components/v2/4de492f6d9cf8244ad5293233e5c6f52407d42fc/2.b977faab.png', bg: '#6BBF7A', panel: '#85CC92' }, { src: 'https://fifth-gentle-45902158.figma.site/_components/v2/4de492f6d9cf8244ad5293233e5c6f52407d42fc/3.4df853b4.png', bg: '#E882B4', panel: '#ED9DC4' }, { src: 'https://fifth-gentle-45902158.figma.site/_components/v2/4de492f6d9cf8244ad5293233e5c6f52407d42fc/4.4457fbce.png', bg: '#6EB5FF', panel: '#8DC4FF' }, ]; ``` Preload all 4 images on mount via `new Image()`. **State & logic:** - `activeIndex` (0–3), `isAnimating` boolean lock, `isMobile` (`window.innerWidth < 640`, updated on resize). - `navigate('next' | 'prev')`: ignore if animating; set `isAnimating=true`; bump `activeIndex` `(prev+1)%4` or `(prev+3)%4`; release lock after `650ms`. - Roles derived from activeIndex: `center=activeIndex`, `left=(activeIndex+3)%4`, `right=(activeIndex+1)%4`, `back=(activeIndex+2)%4`. **Layout structure:** Outer `<div>` has `backgroundColor: IMAGES[activeIndex].bg`, transition `background-color 650ms cubic-bezier(0.4,0,0.2,1)`, `fontFamily: 'Inter, sans-serif'`, `relative w-full overflow-hidden`. Inside, a `relative w-full` div with `height: 100vh; overflow: hidden`. 1. **Grain overlay** (`absolute inset-0 pointer-events-none`, zIndex 50): SVG fractalNoise data URI, `baseFrequency=0.9`, `numOctaves=4`, opacity 0.08 inside SVG, container `opacity: 0.4`, `backgroundSize: 200px 200px`, repeat. 2. **Giant ghost text "3D SHAPE"** (`absolute inset-x-0 flex items-center justify-center pointer-events-none select-none`, zIndex 2, `top: 18%`): font Anton, `fontSize: clamp(90px, 28vw, 380px)`, weight 900, color white, opacity 1, lineHeight 1, uppercase, letterSpacing `-0.02em`, whiteSpace nowrap. 3. **Top-left brand label "TOONHUB"** (`absolute top-6 left-4 sm:left-8`, zIndex 60): `text-xs font-semibold uppercase`, white, opacity 0.9, letterSpacing `0.18em`. 4. **Carousel** (`absolute inset-0`, zIndex 3): map all 4 IMAGES; each item is `position:absolute`, `aspectRatio: '0.6 / 1'`, with role-based styles below. Inside, an `<img>` `width:100%; height:100%; objectFit:contain; objectPosition:bottom center; draggable=false`. Per-role style: - **center**: `transform: translateX(-50%) scale(${isMobile?1.25:1.68})`, no blur, opacity 1, zIndex 20, `left:50%`, `height: isMobile?'60%':'92%'`, `bottom: isMobile?'22%':0`. - **left**: `translateX(-50%) scale(1)`, blur 2px, opacity 0.85, zIndex 10, `left: isMobile?'20%':'30%'`, `height: isMobile?'16%':'28%'`, `bottom: isMobile?'32%':'12%'`. - **right**: same as left but `left: isMobile?'80%':'70%'`. - **back**: `translateX(-50%) scale(1)`, blur 4px, opacity 1, zIndex 5, `left:50%`, `height: isMobile?'13%':'22%'`, `bottom: isMobile?'32%':'12%'`. Transition on each item: `transform 650ms cubic-bezier(0.4,0,0.2,1), filter 650ms ..., opacity 650ms ..., left 650ms ...`. `willChange: transform, filter, opacity`. 5. **Bottom-left text + nav buttons** (`absolute bottom-6 left-4 sm:bottom-20 sm:left-24`, zIndex 60, `maxWidth:320px`): - `<p>` "TOONHUB FIGURINES" — bold uppercase, tracking-widest, `mb-2 sm:mb-3 text-base sm:text-[22px]`, white, opacity 0.95, letterSpacing `0.02em`. - `<p>` (hidden on mobile, `hidden sm:block`): "The artwork is stunning, shipped fully prepared. The finish is a vision, the 3D craft is flawless. Many thanks! Wishing you the win. Order now." — `text-xs sm:text-sm`, white, opacity 0.85, lineHeight 1.6, `mb-4 sm:mb-5`. - Two circular buttons (`w-12 h-12 sm:w-16 sm:h-16`, transparent bg, 2px white border, white icon): `ArrowLeft` and `ArrowRight` from lucide-react, size 26, strokeWidth 2.25. On hover: scale 1.08 + bg `rgba(255,255,255,0.12)`. Transition `transform 150ms, background-color 150ms`. Click triggers `navigate('prev')` / `navigate('next')`. 6. **Bottom-right link "DISCOVER IT"** (`absolute bottom-6 right-4 sm:bottom-20 sm:right-10`, zIndex 60): `<a>` flex items-center, font Anton, `fontSize: clamp(20px, 4vw, 56px)`, weight 400, white, opacity 0.95→1 on hover (200ms), letterSpacing `-0.02em`, lineHeight 1, uppercase, no underline. Followed by `ArrowRight` (`w-5 h-5 sm:w-8 sm:h-8`, strokeWidth 2.25). **Behavior summary:** clicking arrows rotates roles; background color, image positions, scales, blurs, and opacities all crossfade simultaneously over 650ms with `cubic-bezier(0.4,0,0.2,1)`. The character images sit at the bottom of the screen overlapping the giant "3D SHAPE" text behind them.

Source heat
78
Stability
92
View case →
GoodCase.ai · AI Case Evidence LibraryDiscovery, rankings, and creator relationships all resolve to one Case source of truth.
ChangelogFeedback