Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
97 tokens/sec
GPT-4o
53 tokens/sec
Gemini 2.5 Pro Pro
43 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
47 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

SRGS: Super-Resolution 3D Gaussian Splatting (2404.10318v2)

Published 16 Apr 2024 in cs.CV

Abstract: Recently, 3D Gaussian Splatting (3DGS) has gained popularity as a novel explicit 3D representation. This approach relies on the representation power of Gaussian primitives to provide a high-quality rendering. However, primitives optimized at low resolution inevitably exhibit sparsity and texture deficiency, posing a challenge for achieving high-resolution novel view synthesis (HRNVS). To address this problem, we propose Super-Resolution 3D Gaussian Splatting (SRGS) to perform the optimization in a high-resolution (HR) space. The sub-pixel constraint is introduced for the increased viewpoints in HR space, exploiting the sub-pixel cross-view information of the multiple low-resolution (LR) views. The gradient accumulated from more viewpoints will facilitate the densification of primitives. Furthermore, a pre-trained 2D super-resolution model is integrated with the sub-pixel constraint, enabling these dense primitives to learn faithful texture features. In general, our method focuses on densification and texture learning to effectively enhance the representation ability of primitives. Experimentally, our method achieves high rendering quality on HRNVS only with LR inputs, outperforming state-of-the-art methods on challenging datasets such as Mip-NeRF 360 and Tanks & Temples. Related codes will be released upon acceptance.

User Edit Pencil Streamline Icon: https://streamlinehq.com
Authors (10)
  1. Xiang Feng (16 papers)
  2. Yongbo He (3 papers)
  3. Yubo Wang (53 papers)
  4. Yan Yang (119 papers)
  5. Zhenzhong Kuang (5 papers)
  6. Yu Jun (1 paper)
  7. Jianping Fan (51 papers)
  8. Jiajun Ding (6 papers)
  9. Wen Li (107 papers)
  10. Yifei Chen (58 papers)
Citations (1)

Summary

  • The paper introduces a unified LaTeX template that consolidates multiple ACM formatting styles to simplify manuscript preparation.
  • It supports various documentation types and submission stages, ensuring uniformity for dual-anonymous reviews, camera-ready articles, and more.
  • The template enhances accessibility and facilitates metadata extraction, effectively integrating with the ACM Digital Library.

Exploring the ACM's New Unified LaTeX Template

Introduction to the Unified LaTeX Template

The Association for Computing Machinery (ACM) introduced a unified LaTeX document class, known as acmart, to streamline the publication process across its various conferences and journals. This development aimed to provide a consistent formatting style that accommodates the diverse needs of ACM publications while incorporating features for accessibility and metadata extraction crucial for the ACM Digital Library's future initiatives. This template consolidates the unique features of numerous ACM and Special Interest Group (SIG)-specific LaTeX templates into a single framework, thereby simplifying the document preparation process for authors.

Key Features of the Template

The acmart document class supports a wide range of documentation types, from dual-anonymous initial submission papers to camera-ready articles for journals and SIGCHI extended abstracts. One of the template's strengths lies in its versatility; authors can prepare documents for various stages of publication by selecting the appropriate template style and parameters. This flexibility is further enhanced by the inclusion of parameters such as anonymous and review for dual-anonymous conference submissions, authorversion for post-peer-review submissions, and screen for producing color hyperlinks.

Impact on Publication Accessibility and Metadata Extraction

The unified template is designed with an emphasis on accessibility. By adopting a consistent style that is compatible with the ACM Digital Library's metadata extraction tools, the acmart class facilitates the creation of accessible content for a broad audience, including individuals with disabilities. This approach not only enhances the user experience by ensuring content is easily navigable but also aids in the seamless integration of articles into the ACM Digital Library.

Practical Implications for Authors and Researchers

For authors, migrating to the acmart template simplifies the preparation of manuscripts regardless of the ACM publication outlet. The template's adaptability means researchers can focus more on the content of their papers rather than on formatting nuances. However, it's important to note that modifications to the template, such as adjusting margins or overriding the default typeface, are discouraged and may lead to the submission being returned for corrections. This ensures a uniform appearance across ACM publications but requires authors to strictly adhere to the template's guidelines.

Theoretical Implications and Future Developments

The introduction of the acmart LaTeX template represents a significant step toward standardizing academic publishing formats within the computing research community. Theoretically, this uniformity could pave the way for more streamlined publishing protocols across other academic disciplines. Looking ahead, the ACM may continue to refine the template based on user feedback and the evolving needs of the digital library, potentially introducing new features that further reduce the burden of manuscript preparation on authors.

Conclusion

The ACM's unified LaTeX template, acmart, marks a milestone in academic publishing within computing research. By providing a consistent format that meets the varied requirements of ACM's conferences and journals, the template significantly eases the publication process for authors. Its focus on accessibility and compatibility with the ACM Digital Library's metadata extraction efforts further underscores ACM's commitment to creating an inclusive and efficient digital repository for computing research. As the template evolves, it will undoubtedly continue to influence the standards of academic publishing within the computing community and beyond.