Hugging Face's logo

Spaces:
dylanebert
/
3d-arena
Running

App Files Community
12
3d-arena
File size: 3,270 Bytes 
7ebede6
 
b1aa559
7ebede6
 
 
 
770810c
 
 
 
 
 
7ebede6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d366785
 
7ebede6
 
 
d366785
7ebede6
 
 
 
19ba03f
 
7e2471a
770810c
 
 
 
 
 
 
 
7e2471a
 
 
19ba03f
7e2471a
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
 
<div class="section" style="padding: 10px 0;">
    <p class="muted">
        The <span class="emphasis">3D Arena</span> leaderboard evaluates leading generative 3D models. It is inspired by
        <a class="emphasis" target="_blank" href="https://huggingface.co/spaces/TTS-AGI/TTS-Arena">TTS-Arena</a>
        and
        <a class="emphasis" target="_blank" href="https://chat.lmsys.org/">Chatbot Arena</a>.
    </p>
    <p class="muted">
        Read more about the platform and methodology in the official paper:
        <a class="emphasis" target="_blank" href="https://huggingface.co/papers/2506.18787"
            >3D Arena: An Open Platform for Generative 3D Evaluation</a
        >.
    </p>
</div>
<div class="section">
    <h3 class="emphasis">Motivation</h3>
    <p class="muted">
        Evaluating generated 3D quality quantitatively is challenging, and there is no standard practice for reporting
        critical factors like mesh topology in real-world applications.
    </p>
    <p class="muted">This leaderboard displays usable, final results side-by-side for the community to rank.</p>
</div>
<div class="section">
    <h3 class="emphasis">Criteria</h3>
    <p class="muted">
        Models must handle <span class="emphasis">image</span> (.png, .jpg) inputs and produce
        <span class="emphasis">mesh</span>
        (.obj, .glb) or <span class="emphasis">splat</span> (.splat, .ply) outputs. They should run end-to-end without human
        intervention, including UV unwrapping, texture mapping, and other post-processing.
    </p>
    <p class="muted">
        To preview how your results will look in the leaderboard, upload them to any <a
            class="emphasis"
            target="_blank"
            href="https://huggingface.co/spaces/gradio/model3D">gradio/model3d</a
        > viewer.
    </p>
</div>
<div class="section">
    <h3 class="emphasis">Contributing</h3>
    <p class="muted">
        All inputs and outputs are publicly available <a
            class="emphasis"
            target="_blank"
            href="https://huggingface.co/datasets/dylanebert/3d-arena">here</a
        >.
        Input image URLs are provided 
        <a class="emphasis" target="_blank" href="https://huggingface.co/datasets/dylanebert/3d-arena/raw/main/inputs.txt">here</a>.
    </p>
    <p class="muted">
        To contribute a model, run your model on all input images and create a pull request with your model's output.
        Refer to other examples in the dataset to understand the expected format.
    </p>
</div>
<div class="section">
    <h3 class="emphasis">Citing</h3>
    <div class="pre-wrapper">
        <pre>
            <code>
                @misc&lbrace;ebert20253darenaopenplatform,
                    title = &lbrace;3D Arena: An Open Platform for Generative 3D Evaluation&rbrace;, 
                    author = &lbrace;Dylan Ebert&rbrace;,
                    year = &lbrace;2025&rbrace;,
                    eprint = &lbrace;2506.18787&rbrace;,
                    archivePrefix = &lbrace;arXiv&rbrace;,
                    primaryClass = &lbrace;cs.CV&rbrace;,
                    url = &lbrace;https://arxiv.org/abs/2506.18787&rbrace;, 
                &rbrace;
            </code>
        </pre>
    </div>
</div>