KoeYe Claude Opus 5.5 commited on
Commit
cca5a6d
·
1 Parent(s): c3746b4

Four more Image-to-Scene cases: Suburban Umbrella Pair, Rosie's Service Bin, Rosie's Roadside Billboard, Suburban Driveway Mailbox

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. README.md +4 -2
  2. cases.html +2 -2
  3. data/progressive-scenes.json +48 -0
  4. data/site.js +0 -0
  5. index.html +2 -2
  6. media/i2s/comp-17-rosies-service-bin-reset-v3/astra.jpg +0 -0
  7. media/i2s/comp-17-rosies-service-bin-reset-v3/deepseek.jpg +0 -0
  8. media/i2s/comp-17-rosies-service-bin-reset-v3/deepseek_full.jpg +3 -0
  9. media/i2s/comp-17-rosies-service-bin-reset-v3/fable.jpg +0 -0
  10. media/i2s/comp-17-rosies-service-bin-reset-v3/gemini.jpg +0 -0
  11. media/i2s/comp-17-rosies-service-bin-reset-v3/gemma-off.jpg +3 -0
  12. media/i2s/comp-17-rosies-service-bin-reset-v3/gemma-on.jpg +0 -0
  13. media/i2s/comp-17-rosies-service-bin-reset-v3/glm.jpg +0 -0
  14. media/i2s/comp-17-rosies-service-bin-reset-v3/grok.jpg +0 -0
  15. media/i2s/comp-17-rosies-service-bin-reset-v3/gt.jpg +3 -0
  16. media/i2s/comp-17-rosies-service-bin-reset-v3/inkling.jpg +0 -0
  17. media/i2s/comp-17-rosies-service-bin-reset-v3/muse.jpg +0 -0
  18. media/i2s/comp-17-rosies-service-bin-reset-v3/opus.jpg +0 -0
  19. media/i2s/comp-17-rosies-service-bin-reset-v3/qwen-off.jpg +0 -0
  20. media/i2s/comp-17-rosies-service-bin-reset-v3/qwen-on.jpg +0 -0
  21. media/i2s/comp-17-rosies-service-bin-reset-v3/qwen-on_full.jpg +3 -0
  22. media/i2s/comp-17-rosies-service-bin-reset-v3/ref-1.jpg +0 -0
  23. media/i2s/comp-17-rosies-service-bin-reset-v3/sol.jpg +0 -0
  24. media/i2s/comp-18-rosies-roadside-billboard-reset-v3/astra.jpg +0 -0
  25. media/i2s/comp-18-rosies-roadside-billboard-reset-v3/deepseek.jpg +0 -0
  26. media/i2s/comp-18-rosies-roadside-billboard-reset-v3/fable.jpg +0 -0
  27. media/i2s/comp-18-rosies-roadside-billboard-reset-v3/gemini.jpg +0 -0
  28. media/i2s/comp-18-rosies-roadside-billboard-reset-v3/gemini_full.jpg +3 -0
  29. media/i2s/comp-18-rosies-roadside-billboard-reset-v3/gemma-off.jpg +0 -0
  30. media/i2s/comp-18-rosies-roadside-billboard-reset-v3/gemma-on.jpg +0 -0
  31. media/i2s/comp-18-rosies-roadside-billboard-reset-v3/glm.jpg +0 -0
  32. media/i2s/comp-18-rosies-roadside-billboard-reset-v3/grok.jpg +0 -0
  33. media/i2s/comp-18-rosies-roadside-billboard-reset-v3/gt.jpg +3 -0
  34. media/i2s/comp-18-rosies-roadside-billboard-reset-v3/inkling.jpg +0 -0
  35. media/i2s/comp-18-rosies-roadside-billboard-reset-v3/inkling_full.jpg +3 -0
  36. media/i2s/comp-18-rosies-roadside-billboard-reset-v3/muse.jpg +0 -0
  37. media/i2s/comp-18-rosies-roadside-billboard-reset-v3/opus.jpg +0 -0
  38. media/i2s/comp-18-rosies-roadside-billboard-reset-v3/qwen-off.jpg +0 -0
  39. media/i2s/comp-18-rosies-roadside-billboard-reset-v3/qwen-on.jpg +0 -0
  40. media/i2s/comp-18-rosies-roadside-billboard-reset-v3/ref-1.jpg +0 -0
  41. media/i2s/comp-18-rosies-roadside-billboard-reset-v3/sol.jpg +0 -0
  42. media/i2s/comp-51-suburban-driveway-mailbox-recovery-v2/astra.jpg +0 -0
  43. media/i2s/comp-51-suburban-driveway-mailbox-recovery-v2/deepseek.jpg +0 -0
  44. media/i2s/comp-51-suburban-driveway-mailbox-recovery-v2/fable.jpg +0 -0
  45. media/i2s/comp-51-suburban-driveway-mailbox-recovery-v2/gemini.jpg +0 -0
  46. media/i2s/comp-51-suburban-driveway-mailbox-recovery-v2/gemini_full.jpg +3 -0
  47. media/i2s/comp-51-suburban-driveway-mailbox-recovery-v2/gemma-off.jpg +0 -0
  48. media/i2s/comp-51-suburban-driveway-mailbox-recovery-v2/gemma-on.jpg +0 -0
  49. media/i2s/comp-51-suburban-driveway-mailbox-recovery-v2/glm.jpg +0 -0
  50. media/i2s/comp-51-suburban-driveway-mailbox-recovery-v2/grok.jpg +0 -0
README.md CHANGED
@@ -14,7 +14,9 @@ short_description: Benchmarking coding agents that build 3D scenes
14
  Project page for Code4Scene: the leaderboard, the paper's takeaways, and selected cases whose saved Unreal Engine scenes can
15
  be viewed in 3D. Each section has a Text-to-Scene tab (construction; selected by default, the 20-scene Research Track) and an
16
  Image-to-Scene tab (editing; the 75-case public evaluation). `scripts/t2s_only.py` removed Image-to-Scene on 2026-10-01;
17
- `scripts/i2s_tabs.py` brought it back as the second tab on 2026-10-04.
 
 
18
 
19
  Paper: [arXiv:2609.36777](https://arxiv.org/abs/2609.36777) · Code: [SimWorld-AI/Code4Scene](https://github.com/SimWorld-AI/Code4Scene)
20
 
@@ -42,7 +44,7 @@ Queue tests: `node --test tests/scene-assets.test.mjs`. Browser QA includes publ
42
 
43
  ## Geometry-first scene cards
44
 
45
- The homepage uses `data/progressive-scenes.json` to choose immutable packed derivatives for its 11 public scene/state variants. Geometry and meshopt bytes are unchanged; embedded images are moved into small packed batches. The original packed models remain available to the full case explorer. `assets/progressive-gltf.js` returns the geometry with neutral texture placeholders, then the visible card starts material work after its first rendered frame. Image updates preserve glTF samplers, color spaces, transforms and material parameters. At most two texture jobs run across the page; offscreen cards pause them, and a texture failure keeps the geometry visible with a separate retry control.
46
 
47
  Three.js r160, GLTFLoader, MeshoptDecoder, Sky and OrbitControls are served from `assets/vendor/three-r160/`, retaining the official MIT license. Both page import maps use this pinned same-origin version. Original sources: https://github.com/mrdoob/three.js/tree/r160 .
48
 
 
14
  Project page for Code4Scene: the leaderboard, the paper's takeaways, and selected cases whose saved Unreal Engine scenes can
15
  be viewed in 3D. Each section has a Text-to-Scene tab (construction; selected by default, the 20-scene Research Track) and an
16
  Image-to-Scene tab (editing; the 75-case public evaluation). `scripts/t2s_only.py` removed Image-to-Scene on 2026-10-01;
17
+ `scripts/i2s_tabs.py` brought it back as the second tab on 2026-10-04; `scripts/i2s_more_cases.py` adds four more
18
+ Image-to-Scene cases (six in all) and `scripts/harbor_card.py` features GPT-6 Astra's Industrial Harbor, which links to its
19
+ own viewer Space.
20
 
21
  Paper: [arXiv:2609.36777](https://arxiv.org/abs/2609.36777) · Code: [SimWorld-AI/Code4Scene](https://github.com/SimWorld-AI/Code4Scene)
22
 
 
44
 
45
  ## Geometry-first scene cards
46
 
47
+ The homepage uses `data/progressive-scenes.json` to choose immutable packed derivatives for 17 of its public scene/state variants (the suburban Image-to-Scene scenes are almost all geometry, so their cards load the packed model directly). Geometry and meshopt bytes are unchanged; embedded images are moved into small packed batches. The original packed models remain available to the full case explorer. `assets/progressive-gltf.js` returns the geometry with neutral texture placeholders, then the visible card starts material work after its first rendered frame. Image updates preserve glTF samplers, color spaces, transforms and material parameters. At most two texture jobs run across the page; offscreen cards pause them, and a texture failure keeps the geometry visible with a separate retry control.
48
 
49
  Three.js r160, GLTFLoader, MeshoptDecoder, Sky and OrbitControls are served from `assets/vendor/three-r160/`, retaining the official MIT license. Both page import maps use this pinned same-origin version. Original sources: https://github.com/mrdoob/three.js/tree/r160 .
50
 
cases.html CHANGED
@@ -87,7 +87,7 @@ table.cs td,table.cs th{white-space:nowrap}.panel code{font-size:.82em;backgroun
87
  <section class="alt" style="padding-top:34px;padding-bottom:40px"><div class="container">
88
  <div class="section-header" style="margin-bottom:1.6rem"><span class="section-label">Cases</span><h2 class="section-title">Every scene in 3D, with its scores</h2>
89
  <p class="section-desc">Each agent's result is the engine-native scene it saved, exported from Unreal Engine so you can turn it in the browser. Beside it, how the
90
- evaluator scored that scene. 6 cases, 41 scenes in 3D. Pick a case, then an agent.</p></div>
91
  <div class="casebar" id="casebar"></div>
92
  <div class="card" style="padding:1.4rem 1.5rem"><div class="case-head"><h3 id="ctitle"></h3><p><b id="ctrack"></b> · <span id="cnote"></span></p></div>
93
  <div class="case-input" id="cinput"></div>
@@ -99,7 +99,7 @@ evaluator scored that scene. 6 cases, 41 scenes in 3D. Pick a case, then an agen
99
  <footer><p>Code4Scene  ·  UC San Diego &amp; UC Berkeley  ·  <a href="https://arxiv.org/abs/2609.36777" rel="noopener" target="_blank">arXiv:2609.36777</a>  ·  <a href="https://github.com/SimWorld-AI/Code4Scene" rel="noopener" target="_blank">GitHub</a>  · 
100
  <a href="https://simworld.org/" rel="noopener" target="_blank">SimWorld Home</a></p>
101
  <p style="margin-top:.5rem;font-size:.75rem;color:rgba(255,255,255,.25)">Part of the SimWorld project.</p></footer>
102
- <script src="assets/paper.js?v=5427505a"></script><script src="data/site.js?v=20261004i2s"></script><script type="module">import * as THREE from "three";
103
  import { GLTFLoader } from "three/addons/loaders/GLTFLoader.js";
104
  import { OrbitControls } from "three/addons/controls/OrbitControls.js";
105
  import { MeshoptDecoder } from "three/addons/libs/meshopt_decoder.module.js";
 
87
  <section class="alt" style="padding-top:34px;padding-bottom:40px"><div class="container">
88
  <div class="section-header" style="margin-bottom:1.6rem"><span class="section-label">Cases</span><h2 class="section-title">Every scene in 3D, with its scores</h2>
89
  <p class="section-desc">Each agent's result is the engine-native scene it saved, exported from Unreal Engine so you can turn it in the browser. Beside it, how the
90
+ evaluator scored that scene. 10 cases, 58 scenes in 3D. Pick a case, then an agent.</p></div>
91
  <div class="casebar" id="casebar"></div>
92
  <div class="card" style="padding:1.4rem 1.5rem"><div class="case-head"><h3 id="ctitle"></h3><p><b id="ctrack"></b> · <span id="cnote"></span></p></div>
93
  <div class="case-input" id="cinput"></div>
 
99
  <footer><p>Code4Scene  ·  UC San Diego &amp; UC Berkeley  ·  <a href="https://arxiv.org/abs/2609.36777" rel="noopener" target="_blank">arXiv:2609.36777</a>  ·  <a href="https://github.com/SimWorld-AI/Code4Scene" rel="noopener" target="_blank">GitHub</a>  · 
100
  <a href="https://simworld.org/" rel="noopener" target="_blank">SimWorld Home</a></p>
101
  <p style="margin-top:.5rem;font-size:.75rem;color:rgba(255,255,255,.25)">Part of the SimWorld project.</p></footer>
102
+ <script src="assets/paper.js?v=5427505a"></script><script src="data/site.js?v=20261004i2smore"></script><script type="module">import * as THREE from "three";
103
  import { GLTFLoader } from "three/addons/loaders/GLTFLoader.js";
104
  import { OrbitControls } from "three/addons/controls/OrbitControls.js";
105
  import { MeshoptDecoder } from "three/addons/libs/meshopt_decoder.module.js";
data/progressive-scenes.json CHANGED
@@ -86,5 +86,53 @@
86
  "textureBytes": 2056820,
87
  "images": 232,
88
  "bundles": 8
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
89
  }
90
  }
 
86
  "textureBytes": 2056820,
87
  "images": 232,
88
  "bundles": 8
89
+ },
90
+ "fef6016b97b176af9fef.bin": {
91
+ "geometry": "models/progressive-v1/geo-13f4cd9fc901d889707fc206.bin",
92
+ "originalBytes": 4864820,
93
+ "geometryBytes": 3641276,
94
+ "textureBytes": 1248976,
95
+ "images": 209,
96
+ "bundles": 7
97
+ },
98
+ "6ed621e057a53331469a.bin": {
99
+ "geometry": "models/progressive-v1/geo-d0f2f25f6a11dcb13b80fd71.bin",
100
+ "originalBytes": 4864728,
101
+ "geometryBytes": 3641184,
102
+ "textureBytes": 1248976,
103
+ "images": 209,
104
+ "bundles": 7
105
+ },
106
+ "843768ef1999c6f33178.bin": {
107
+ "geometry": "models/progressive-v1/geo-56b3730f534874fae19c22ef.bin",
108
+ "originalBytes": 4864728,
109
+ "geometryBytes": 3641180,
110
+ "textureBytes": 1248976,
111
+ "images": 209,
112
+ "bundles": 7
113
+ },
114
+ "d3e551d366b82c38cfe2.bin": {
115
+ "geometry": "models/progressive-v1/geo-7f4a328519e42f2fb2cd5bd8.bin",
116
+ "originalBytes": 4864836,
117
+ "geometryBytes": 3641288,
118
+ "textureBytes": 1248976,
119
+ "images": 209,
120
+ "bundles": 7
121
+ },
122
+ "ca03f1482ef5f2decbcb.bin": {
123
+ "geometry": "models/progressive-v1/geo-38a1193f3df64e3bf1e8f1b0.bin",
124
+ "originalBytes": 4864784,
125
+ "geometryBytes": 3641240,
126
+ "textureBytes": 1248976,
127
+ "images": 209,
128
+ "bundles": 7
129
+ },
130
+ "73ceabf2d178a47c7d70.bin": {
131
+ "geometry": "models/progressive-v1/geo-56b3730f534874fae19c22ef.bin",
132
+ "originalBytes": 4864728,
133
+ "geometryBytes": 3641180,
134
+ "textureBytes": 1248976,
135
+ "images": 209,
136
+ "bundles": 7
137
  }
138
  }
data/site.js CHANGED
The diff for this file is too large to render. See raw diff
 
index.html CHANGED
@@ -45,7 +45,7 @@ svg.ch .dim{opacity:.42}svg.ch g.row:hover{opacity:1}svg.ch .hot{fill:var(--c4s-
45
  <a class="text-button" href="https://simworld-ai-industrial-harbor-viewer.static.hf.space/" rel="noopener" target="_blank">Explore the full scene in 3D ↗</a>
46
  <p class="feature-note">Streams about 1.2 GB of textures on first load.</p>
47
  </div>
48
- </article><article class="case-card"><div class="live-scene" data-agent="astra" data-case="nordic-harbour"><canvas aria-label="Interactive 3D scene: Nordic Harbour. Drag to orbit, scroll or use plus and minus to zoom." tabindex="0"></canvas><span class="scene-loading" role="status">Loading 3D scene…</span><span class="scene-hint">Drag to orbit · Scroll to zoom</span><button aria-label="Pause scene rotation" class="scene-motion">Ⅱ</button></div><div class="meta"><span class="eyebrow">Text-to-Scene</span><h4>Nordic Harbour</h4><p>GPT-6 Astra (max) · score 0.877</p><a class="text-button" href="cases.html#nordic-harbour">Compare all 14 agents ↗</a></div></article><article class="case-card"><div class="live-scene" data-agent="fable" data-case="medieval-big-farm-town"><canvas aria-label="Interactive 3D scene: Medieval Big Farm Town. Drag to orbit, scroll or use plus and minus to zoom." tabindex="0"></canvas><span class="scene-loading" role="status">Loading 3D scene…</span><span class="scene-hint">Drag to orbit · Scroll to zoom</span><button aria-label="Pause scene rotation" class="scene-motion">Ⅱ</button></div><div class="meta"><span class="eyebrow">Text-to-Scene</span><h4>Medieval Big Farm Town</h4><p>Claude Fable 5.1 (max) · score 0.830</p><a class="text-button" href="cases.html#medieval-big-farm-town">Compare all 14 agents ↗</a></div></article><article class="case-card"><div class="live-scene" data-agent="opus" data-case="egyptian-temple"><canvas aria-label="Interactive 3D scene: Egyptian Temple. Drag to orbit, scroll or use plus and minus to zoom." tabindex="0"></canvas><span class="scene-loading" role="status">Loading 3D scene…</span><span class="scene-hint">Drag to orbit · Scroll to zoom</span><button aria-label="Pause scene rotation" class="scene-motion">Ⅱ</button></div><div class="meta"><span class="eyebrow">Text-to-Scene</span><h4>Egyptian Temple</h4><p>Claude Opus 5 (max) · score 0.777</p><a class="text-button" href="cases.html#egyptian-temple">Compare all 14 agents ↗</a></div></article><article class="case-card"><div class="live-scene" data-agent="fable" data-case="bazaar"><canvas aria-label="Interactive 3D scene: Bazaar. Drag to orbit, scroll or use plus and minus to zoom." tabindex="0"></canvas><span class="scene-loading" role="status">Loading 3D scene…</span><span class="scene-hint">Drag to orbit · Scroll to zoom</span><button aria-label="Pause scene rotation" class="scene-motion">Ⅱ</button></div><div class="meta"><span class="eyebrow">Text-to-Scene</span><h4>Bazaar</h4><p>Claude Fable 5.1 (max) · score 0.930</p><a class="text-button" href="cases.html#bazaar">Compare all 14 agents ↗</a></div></article></div><div class="case-grid" data-view="i2s" style="display:none"><article class="case-card"><div class="live-scene" data-agent="astra" data-case="comp-31-old-industrial-pallet-bay-repair-v2"><canvas aria-label="Interactive 3D scene: Old Industrial Pallet Bay. Drag to orbit, scroll or use plus and minus to zoom." tabindex="0"></canvas><span class="scene-loading" role="status">Loading 3D scene…</span><span class="scene-hint">Drag to orbit · Scroll to zoom</span><button aria-label="Pause scene rotation" class="scene-motion">Ⅱ</button></div><div aria-label="Scene state" class="scene-versions"><button class="on" data-state="repair">Repair</button><button data-state="input">Corrupted input</button><button data-state="gt">Ground truth</button></div><div class="meta"><span class="eyebrow">Image-to-Scene</span><h4>Old Industrial Pallet Bay</h4><p>GPT-6 Astra (max) · score 1.000</p><a class="text-button" href="cases.html#comp-31-old-industrial-pallet-bay-repair-v2">Compare all 14 agents ↗</a></div></article><article class="case-card"><div class="live-scene" data-agent="fable" data-case="comp-61-new-york-mailbox-pair-repair-v2"><canvas aria-label="Interactive 3D scene: New York Mailbox Pair. Drag to orbit, scroll or use plus and minus to zoom." tabindex="0"></canvas><span class="scene-loading" role="status">Loading 3D scene…</span><span class="scene-hint">Drag to orbit · Scroll to zoom</span><button aria-label="Pause scene rotation" class="scene-motion">Ⅱ</button></div><div aria-label="Scene state" class="scene-versions"><button class="on" data-state="repair">Repair</button><button data-state="input">Corrupted input</button><button data-state="gt">Ground truth</button></div><div class="meta"><span class="eyebrow">Image-to-Scene</span><h4>New York Mailbox Pair</h4><p>Claude Fable 5.1 (max) · score 1.000</p><a class="text-button" href="cases.html#comp-61-new-york-mailbox-pair-repair-v2">Compare all 14 agents ↗</a></div></article></div>
49
  <div style="text-align:center;margin-top:2rem"><a class="btn btn-primary" href="cases.html"><svg aria-hidden="true" fill="none" stroke="currentColor" stroke-width="1.5" viewbox="0 0 16 16"><path d="M8 1.5l5.5 3v7L8 14.5l-5.5-3v-7z"></path><path d="M2.5 4.5L8 7.5l5.5-3M8 7.5v7"></path></svg> Open the case explorer</a></div>
50
  </div></section><section class="alt" id="takeaways"><div class="container">
51
  <div class="section-header centered"><span class="section-label">Findings</span><h2 class="section-title">Takeaways</h2>
@@ -70,4 +70,4 @@ svg.ch .dim{opacity:.42}svg.ch g.row:hover{opacity:1}svg.ch .hot{fill:var(--c4s-
70
  <a href="https://simworld.org/" rel="noopener" target="_blank">SimWorld Home</a></p>
71
  <p style="margin-top:.5rem;font-size:.75rem;color:rgba(255,255,255,.25)">Part of the SimWorld project.</p></footer>
72
  <script src="assets/paper.js?v=5427505a"></script>
73
- <script defer="" src="assets/leaderboard-bars.js?v=20261004i2s"></script><script src="data/site.js?v=20261004i2s"></script><script src="assets/pareto.js?v=20261004i2s" type="module"></script><script src="assets/scene-cards.js?v=20261002a" type="module"></script><script src="assets/task-tabs.js?v=20261001c" type="module"></script><script src="assets/leaderboard-controls.js?v=20261001c" type="module"></script><script src="assets/takeaway-bars.js?v=20261004kit" type="module"></script></body></html>
 
45
  <a class="text-button" href="https://simworld-ai-industrial-harbor-viewer.static.hf.space/" rel="noopener" target="_blank">Explore the full scene in 3D ↗</a>
46
  <p class="feature-note">Streams about 1.2 GB of textures on first load.</p>
47
  </div>
48
+ </article><article class="case-card"><div class="live-scene" data-agent="astra" data-case="nordic-harbour"><canvas aria-label="Interactive 3D scene: Nordic Harbour. Drag to orbit, scroll or use plus and minus to zoom." tabindex="0"></canvas><span class="scene-loading" role="status">Loading 3D scene…</span><span class="scene-hint">Drag to orbit · Scroll to zoom</span><button aria-label="Pause scene rotation" class="scene-motion">Ⅱ</button></div><div class="meta"><span class="eyebrow">Text-to-Scene</span><h4>Nordic Harbour</h4><p>GPT-6 Astra (max) · score 0.877</p><a class="text-button" href="cases.html#nordic-harbour">Compare all 14 agents ↗</a></div></article><article class="case-card"><div class="live-scene" data-agent="fable" data-case="medieval-big-farm-town"><canvas aria-label="Interactive 3D scene: Medieval Big Farm Town. Drag to orbit, scroll or use plus and minus to zoom." tabindex="0"></canvas><span class="scene-loading" role="status">Loading 3D scene…</span><span class="scene-hint">Drag to orbit · Scroll to zoom</span><button aria-label="Pause scene rotation" class="scene-motion">Ⅱ</button></div><div class="meta"><span class="eyebrow">Text-to-Scene</span><h4>Medieval Big Farm Town</h4><p>Claude Fable 5.1 (max) · score 0.830</p><a class="text-button" href="cases.html#medieval-big-farm-town">Compare all 14 agents ↗</a></div></article><article class="case-card"><div class="live-scene" data-agent="opus" data-case="egyptian-temple"><canvas aria-label="Interactive 3D scene: Egyptian Temple. Drag to orbit, scroll or use plus and minus to zoom." tabindex="0"></canvas><span class="scene-loading" role="status">Loading 3D scene…</span><span class="scene-hint">Drag to orbit · Scroll to zoom</span><button aria-label="Pause scene rotation" class="scene-motion">Ⅱ</button></div><div class="meta"><span class="eyebrow">Text-to-Scene</span><h4>Egyptian Temple</h4><p>Claude Opus 5 (max) · score 0.777</p><a class="text-button" href="cases.html#egyptian-temple">Compare all 14 agents ↗</a></div></article><article class="case-card"><div class="live-scene" data-agent="fable" data-case="bazaar"><canvas aria-label="Interactive 3D scene: Bazaar. Drag to orbit, scroll or use plus and minus to zoom." tabindex="0"></canvas><span class="scene-loading" role="status">Loading 3D scene…</span><span class="scene-hint">Drag to orbit · Scroll to zoom</span><button aria-label="Pause scene rotation" class="scene-motion">Ⅱ</button></div><div class="meta"><span class="eyebrow">Text-to-Scene</span><h4>Bazaar</h4><p>Claude Fable 5.1 (max) · score 0.930</p><a class="text-button" href="cases.html#bazaar">Compare all 14 agents ↗</a></div></article></div><div class="case-grid" data-view="i2s" style="display:none"><article class="case-card"><div class="live-scene" data-agent="astra" data-case="comp-31-old-industrial-pallet-bay-repair-v2"><canvas aria-label="Interactive 3D scene: Old Industrial Pallet Bay. Drag to orbit, scroll or use plus and minus to zoom." tabindex="0"></canvas><span class="scene-loading" role="status">Loading 3D scene…</span><span class="scene-hint">Drag to orbit · Scroll to zoom</span><button aria-label="Pause scene rotation" class="scene-motion">Ⅱ</button></div><div aria-label="Scene state" class="scene-versions"><button class="on" data-state="repair">Repair</button><button data-state="input">Corrupted input</button><button data-state="gt">Ground truth</button></div><div class="meta"><span class="eyebrow">Image-to-Scene</span><h4>Old Industrial Pallet Bay</h4><p>GPT-6 Astra (max) · score 1.000</p><a class="text-button" href="cases.html#comp-31-old-industrial-pallet-bay-repair-v2">Compare all 14 agents ↗</a></div></article><article class="case-card"><div class="live-scene" data-agent="fable" data-case="comp-61-new-york-mailbox-pair-repair-v2"><canvas aria-label="Interactive 3D scene: New York Mailbox Pair. Drag to orbit, scroll or use plus and minus to zoom." tabindex="0"></canvas><span class="scene-loading" role="status">Loading 3D scene…</span><span class="scene-hint">Drag to orbit · Scroll to zoom</span><button aria-label="Pause scene rotation" class="scene-motion">Ⅱ</button></div><div aria-label="Scene state" class="scene-versions"><button class="on" data-state="repair">Repair</button><button data-state="input">Corrupted input</button><button data-state="gt">Ground truth</button></div><div class="meta"><span class="eyebrow">Image-to-Scene</span><h4>New York Mailbox Pair</h4><p>Claude Fable 5.1 (max) · score 1.000</p><a class="text-button" href="cases.html#comp-61-new-york-mailbox-pair-repair-v2">Compare all 14 agents ↗</a></div></article><article class="case-card"><div class="live-scene" data-agent="opus" data-case="comp-72-suburban-umbrella-pair-repair-v2"><canvas aria-label="Interactive 3D scene: Suburban Umbrella Pair. Drag to orbit, scroll or use plus and minus to zoom." tabindex="0"></canvas><span class="scene-loading" role="status">Loading 3D scene…</span><span class="scene-hint">Drag to orbit · Scroll to zoom</span><button aria-label="Pause scene rotation" class="scene-motion">Ⅱ</button></div><div aria-label="Scene state" class="scene-versions"><button class="on" data-state="repair">Repair</button><button data-state="input">Corrupted input</button><button data-state="gt">Ground truth</button></div><div class="meta"><span class="eyebrow">Image-to-Scene</span><h4>Suburban Umbrella Pair</h4><p>Claude Opus 5 (max) · score 1.000</p><a class="text-button" href="cases.html#comp-72-suburban-umbrella-pair-repair-v2">Compare all 14 agents ↗</a></div></article><article class="case-card"><div class="live-scene" data-agent="astra" data-case="comp-17-rosies-service-bin-reset-v3"><canvas aria-label="Interactive 3D scene: Rosie's Service Bin. Drag to orbit, scroll or use plus and minus to zoom." tabindex="0"></canvas><span class="scene-loading" role="status">Loading 3D scene…</span><span class="scene-hint">Drag to orbit · Scroll to zoom</span><button aria-label="Pause scene rotation" class="scene-motion">Ⅱ</button></div><div aria-label="Scene state" class="scene-versions"><button class="on" data-state="repair">Repair</button><button data-state="input">Corrupted input</button><button data-state="gt">Ground truth</button></div><div class="meta"><span class="eyebrow">Image-to-Scene</span><h4>Rosie's Service Bin</h4><p>GPT-6 Astra (max) · score 1.000</p><a class="text-button" href="cases.html#comp-17-rosies-service-bin-reset-v3">Compare all 14 agents ↗</a></div></article><article class="case-card"><div class="live-scene" data-agent="astra" data-case="comp-18-rosies-roadside-billboard-reset-v3"><canvas aria-label="Interactive 3D scene: Rosie's Roadside Billboard. Drag to orbit, scroll or use plus and minus to zoom." tabindex="0"></canvas><span class="scene-loading" role="status">Loading 3D scene…</span><span class="scene-hint">Drag to orbit · Scroll to zoom</span><button aria-label="Pause scene rotation" class="scene-motion">Ⅱ</button></div><div aria-label="Scene state" class="scene-versions"><button class="on" data-state="repair">Repair</button><button data-state="input">Corrupted input</button><button data-state="gt">Ground truth</button></div><div class="meta"><span class="eyebrow">Image-to-Scene</span><h4>Rosie's Roadside Billboard</h4><p>GPT-6 Astra (max) · score 1.000</p><a class="text-button" href="cases.html#comp-18-rosies-roadside-billboard-reset-v3">Compare all 14 agents ↗</a></div></article><article class="case-card"><div class="live-scene" data-agent="fable" data-case="comp-51-suburban-driveway-mailbox-recovery-v2"><canvas aria-label="Interactive 3D scene: Suburban Driveway Mailbox. Drag to orbit, scroll or use plus and minus to zoom." tabindex="0"></canvas><span class="scene-loading" role="status">Loading 3D scene…</span><span class="scene-hint">Drag to orbit · Scroll to zoom</span><button aria-label="Pause scene rotation" class="scene-motion">Ⅱ</button></div><div aria-label="Scene state" class="scene-versions"><button class="on" data-state="repair">Repair</button><button data-state="input">Corrupted input</button><button data-state="gt">Ground truth</button></div><div class="meta"><span class="eyebrow">Image-to-Scene</span><h4>Suburban Driveway Mailbox</h4><p>Claude Fable 5.1 (max) · score 1.000</p><a class="text-button" href="cases.html#comp-51-suburban-driveway-mailbox-recovery-v2">Compare all 14 agents ↗</a></div></article></div>
49
  <div style="text-align:center;margin-top:2rem"><a class="btn btn-primary" href="cases.html"><svg aria-hidden="true" fill="none" stroke="currentColor" stroke-width="1.5" viewbox="0 0 16 16"><path d="M8 1.5l5.5 3v7L8 14.5l-5.5-3v-7z"></path><path d="M2.5 4.5L8 7.5l5.5-3M8 7.5v7"></path></svg> Open the case explorer</a></div>
50
  </div></section><section class="alt" id="takeaways"><div class="container">
51
  <div class="section-header centered"><span class="section-label">Findings</span><h2 class="section-title">Takeaways</h2>
 
70
  <a href="https://simworld.org/" rel="noopener" target="_blank">SimWorld Home</a></p>
71
  <p style="margin-top:.5rem;font-size:.75rem;color:rgba(255,255,255,.25)">Part of the SimWorld project.</p></footer>
72
  <script src="assets/paper.js?v=5427505a"></script>
73
+ <script defer="" src="assets/leaderboard-bars.js?v=20261004i2s"></script><script src="data/site.js?v=20261004i2smore"></script><script src="assets/pareto.js?v=20261004i2s" type="module"></script><script src="assets/scene-cards.js?v=20261002a" type="module"></script><script src="assets/task-tabs.js?v=20261001c" type="module"></script><script src="assets/leaderboard-controls.js?v=20261001c" type="module"></script><script src="assets/takeaway-bars.js?v=20261004kit" type="module"></script></body></html>
media/i2s/comp-17-rosies-service-bin-reset-v3/astra.jpg ADDED
media/i2s/comp-17-rosies-service-bin-reset-v3/deepseek.jpg ADDED
media/i2s/comp-17-rosies-service-bin-reset-v3/deepseek_full.jpg ADDED

Git LFS Details

  • SHA256: 6489c63201ac0f57da4c61eb698fd2ff375b00c84d2b3124fa8abc16c7d788df
  • Pointer size: 131 Bytes
  • Size of remote file: 120 kB
media/i2s/comp-17-rosies-service-bin-reset-v3/fable.jpg ADDED
media/i2s/comp-17-rosies-service-bin-reset-v3/gemini.jpg ADDED
media/i2s/comp-17-rosies-service-bin-reset-v3/gemma-off.jpg ADDED

Git LFS Details

  • SHA256: ab293c1acffa3cb6a33b5bdc88e42374cff0c3ab8c483b32fe9a2c63894e17a7
  • Pointer size: 131 Bytes
  • Size of remote file: 107 kB
media/i2s/comp-17-rosies-service-bin-reset-v3/gemma-on.jpg ADDED
media/i2s/comp-17-rosies-service-bin-reset-v3/glm.jpg ADDED
media/i2s/comp-17-rosies-service-bin-reset-v3/grok.jpg ADDED
media/i2s/comp-17-rosies-service-bin-reset-v3/gt.jpg ADDED

Git LFS Details

  • SHA256: ba1961232dd5aec0fbe844dd31de214b704d7fd3b3abd24692869e0c3518e01e
  • Pointer size: 131 Bytes
  • Size of remote file: 121 kB
media/i2s/comp-17-rosies-service-bin-reset-v3/inkling.jpg ADDED
media/i2s/comp-17-rosies-service-bin-reset-v3/muse.jpg ADDED
media/i2s/comp-17-rosies-service-bin-reset-v3/opus.jpg ADDED
media/i2s/comp-17-rosies-service-bin-reset-v3/qwen-off.jpg ADDED
media/i2s/comp-17-rosies-service-bin-reset-v3/qwen-on.jpg ADDED
media/i2s/comp-17-rosies-service-bin-reset-v3/qwen-on_full.jpg ADDED

Git LFS Details

  • SHA256: ffab5f556dd0fc03e915973b4c0ca0a8e388edcda69a21d048b4b2d52e6ab6e3
  • Pointer size: 131 Bytes
  • Size of remote file: 121 kB
media/i2s/comp-17-rosies-service-bin-reset-v3/ref-1.jpg ADDED
media/i2s/comp-17-rosies-service-bin-reset-v3/sol.jpg ADDED
media/i2s/comp-18-rosies-roadside-billboard-reset-v3/astra.jpg ADDED
media/i2s/comp-18-rosies-roadside-billboard-reset-v3/deepseek.jpg ADDED
media/i2s/comp-18-rosies-roadside-billboard-reset-v3/fable.jpg ADDED
media/i2s/comp-18-rosies-roadside-billboard-reset-v3/gemini.jpg ADDED
media/i2s/comp-18-rosies-roadside-billboard-reset-v3/gemini_full.jpg ADDED

Git LFS Details

  • SHA256: d4a3e4f578bc9ea1865cb24b82241e24544fc92b7b7c828226ee50c182f9618d
  • Pointer size: 131 Bytes
  • Size of remote file: 174 kB
media/i2s/comp-18-rosies-roadside-billboard-reset-v3/gemma-off.jpg ADDED
media/i2s/comp-18-rosies-roadside-billboard-reset-v3/gemma-on.jpg ADDED
media/i2s/comp-18-rosies-roadside-billboard-reset-v3/glm.jpg ADDED
media/i2s/comp-18-rosies-roadside-billboard-reset-v3/grok.jpg ADDED
media/i2s/comp-18-rosies-roadside-billboard-reset-v3/gt.jpg ADDED

Git LFS Details

  • SHA256: f930a1f3e404abc654034e456a01814d3053a94a761094d5f029b7992972f048
  • Pointer size: 131 Bytes
  • Size of remote file: 126 kB
media/i2s/comp-18-rosies-roadside-billboard-reset-v3/inkling.jpg ADDED
media/i2s/comp-18-rosies-roadside-billboard-reset-v3/inkling_full.jpg ADDED

Git LFS Details

  • SHA256: 942130788e8eb398f08400d4993f5cc52ca5413cb3a4d418c91de6649047f1bd
  • Pointer size: 131 Bytes
  • Size of remote file: 106 kB
media/i2s/comp-18-rosies-roadside-billboard-reset-v3/muse.jpg ADDED
media/i2s/comp-18-rosies-roadside-billboard-reset-v3/opus.jpg ADDED
media/i2s/comp-18-rosies-roadside-billboard-reset-v3/qwen-off.jpg ADDED
media/i2s/comp-18-rosies-roadside-billboard-reset-v3/qwen-on.jpg ADDED
media/i2s/comp-18-rosies-roadside-billboard-reset-v3/ref-1.jpg ADDED
media/i2s/comp-18-rosies-roadside-billboard-reset-v3/sol.jpg ADDED
media/i2s/comp-51-suburban-driveway-mailbox-recovery-v2/astra.jpg ADDED
media/i2s/comp-51-suburban-driveway-mailbox-recovery-v2/deepseek.jpg ADDED
media/i2s/comp-51-suburban-driveway-mailbox-recovery-v2/fable.jpg ADDED
media/i2s/comp-51-suburban-driveway-mailbox-recovery-v2/gemini.jpg ADDED
media/i2s/comp-51-suburban-driveway-mailbox-recovery-v2/gemini_full.jpg ADDED

Git LFS Details

  • SHA256: 3c689122ab458bad0b239356548cb45bc470854ffd715782eedb46da394e1af0
  • Pointer size: 131 Bytes
  • Size of remote file: 285 kB
media/i2s/comp-51-suburban-driveway-mailbox-recovery-v2/gemma-off.jpg ADDED
media/i2s/comp-51-suburban-driveway-mailbox-recovery-v2/gemma-on.jpg ADDED
media/i2s/comp-51-suburban-driveway-mailbox-recovery-v2/glm.jpg ADDED
media/i2s/comp-51-suburban-driveway-mailbox-recovery-v2/grok.jpg ADDED