Pith. sign in

REVIEW 3 major objections 4 minor 77 references

NeuSEditor: From Multi-View Images to Text-Guided Neural Surface Edits

T0 review · 3 major / 4 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read NeuSEditor shows that text-guided editing of neural implicit surfaces preserves identity and geometry when the network separates background, source, and target renderers and adds a Phong-shaded posterior distillation loss.

desk verdict Solid incremental extension of PDS with a new architecture and a Phong-shaded distillation loss, but the ablation table undercuts the paper's headline claim for that loss. read the letter →

arxiv 2505.10827 v1 pith:NOVDXEAY submitted 2025-05-16 cs.CV

Nail Ibrahimli , Julian F. P. Kooij , Liangliang Nan This is my paper · ORCID
classification cs.CV
keywords text-guided3Deditingneuralimplicitsurfacessigneddistancefunctionsposteriordistillationsamplingscoreidentitypreservationmulti-viewreconstructionPhongshading
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

NeuSEditor claims that text-guided editing of a neural implicit 3D scene can preserve the original scene's identity if the network separates the scene into background, source foreground, and target foreground renderers, with the target initialized from and conditioned on the source. This makes the edit a learned delta layered on top of the original geometry instead of a fresh optimization that can overwrite it. The paper adds a geometry-aware distillation loss that aligns diffusion latents of the source image with latents of a Phong-shaded target render, pushing gradient signal into surface geometry rather than appearance alone. The claimed consequence is a simpler workflow: no iterative dataset updates and no source prompt, with better identity preservation, fewer floaters, and superior CLIP/LPIPS and user-study results on DTU, Blender, and IN2N data.

What carries the argument

The load-bearing mechanism is the three-renderer identity-preserving architecture together with the geometry-aware distillation loss. A background renderer uses an inverted-sphere density field; a source foreground renderer learns an SDF and the appearance of the original object; a target foreground renderer, initialized from and conditioned on the source, learns the edit. The Phong-aware loss term $\mathcal{L}_{\mathrm{PE}}$ computes stochastic latents (defined via the DDPM reverse-step mean, Eq. 9) for the source image and for a Phong-shaded version of the target surface, repeats the shading across three channels to fit the diffusion model input, and minimizes their squared difference; gradients are taken only through the renderer, omitting the diffusion U-Net Jacobian as in SDS. This is what carries the claimed geometric fidelity: the diffusion prior sees shading that varies with surface normal, so the gradient signal is tied to geometry rather than appearance alone.

What would settle it

Replace the Phong-shaded target render in $\mathcal{L}_{\mathrm{PE}}$ with a constant gray image of the same shape and rerun the ablation; if the multi-view LPIPS and floater counts stay the same, the geometric signal in the loss is not doing the claimed work. A complementary check is to render a set of shapes with Phong shading, push them through the diffusion encoder, and see whether latent-space distances between renderings correlate with surface-normal difference rather than with albedo difference.

Watch

Extended reading notes

Core claim

The central discovery is that identity preservation in text-guided implicit-surface editing is an architectural problem, not just a loss-tuning problem. The paper separates the scene into a background renderer (density field on an inverted sphere) and source and target foreground renderers (SDF-based, NeuS volume rendering), with the target conditioned on the source's geometry and feature outputs. On top of this, it defines a Phong-enhanced posterior distillation sampling objective, $\mathcal{L}_{\mathrm{PEPDS}} = \lambda_{\mathrm{PDS}}\mathcal{L}_{\mathrm{PDS}} + \lambda_{\mathrm{PE}}\mathcal{L}_{\mathrm{PE}}$, where $\mathcal{L}_{\mathrm{PDS}}$ aligns stochastic latents of source and target renderings and $\mathcal{L}_{\mathrm{PE}}$ aligns the source latent with the latent of a Phong-shaded target rendering computed from numerical SDF gradients. The paper argues that this combination keeps recognizable scene features, reduces floaters, and outperforms PDS and Instruct-NeRF2NeRF quantitatively and in a 41-participant user study across DTU, Blender, and IN2N datasets.

Load-bearing premise

The load-bearing assumption is that the diffusion model treats a Phong-shaded, three-channel copy of the target surface as a legitimate image, so the noise-recovery latent computed from it carries real geometric information that can steer the edit; if that input is out-of-distribution, the geometry-aware loss contributes no useful gradient and the architecture alone must explain the results.

Editorial extensions

If this is right

  • Edits no longer require iterative dataset updates or a source text prompt; the identity is held in the source and background renderers while the target renderer learns only the change.
  • Foreground edits leave the background untouched because the background renderer is separate, and the paper also shows the separation works in reverse, letting background-only edits preserve the foreground.
  • The per-edit parameter cost is small and additive, about 14M parameters per edit on top of a 28M identity, so successive edits can be layered without retraining the identity.
  • Measured on 34 tasks across DTU, Blender, and IN2N data, identity preservation (LPIPS against a separately trained NeRF) and user preference improve over PDS and Instruct-NeRF2NeRF, with the largest gains on DTU and Blender.
  • Numerical SDF gradients used in the Phong shading produce cleaner meshes and fewer floaters than analytical gradients at similar CLIP alignment with the text prompt.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the architecture is the active ingredient in identity preservation, this three-way split should transfer to explicit representations such as Gaussian splatting: a source-conditioned target set of splats plus a normal-based shading loss could reproduce the identity gains without SDF rendering or marching cubes.
  • The Phong-latent alignment rests on the diffusion encoder accepting repeated-channel shading as a real image; a direct test would swap the Phong input for a depth or normal map rendered in the same three-channel format and see whether the floater reduction persists, revealing whether the signal is geometric or merely a second, differently textured view.
  • Because the head-to-head comparison gives PDS two prompts and NeuSEditor one, the reported gains may mix architectural benefit with prompt-count benefit; running the proposed architecture with the original two-prompt PDS loss would decompose the two.
  • The additive parameter budget suggests a continual-editing regime in which edits compose in order; one could check whether 'make it a church' followed by 'make it snow' yields the same result as a single 'snowy church' prompt without retraining the identity.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. NeuSEditor proposes a two-stage pipeline for text-guided editing of neural implicit surfaces reconstructed from multi-view images. In the first stage, a source SDF-based renderer learns the identity of the input scene, including a separately modeled background. In the second stage, a target renderer, initialized from and conditioned on the source renderer, is optimized with a posterior distillation loss (PDS) augmented by a geometry-aware loss (LPE) that aligns the stochastic latents of a Phong-shaded target rendering with the source latents. The paper reports quantitative metrics (CLIP, LPIPS), a 41-participant user study over 34 editing experiments on DTU, Blender, and IN2N data, and qualitative comparisons against IN2N, PDS-NeRF, and PDS-Splat, concluding that the method outperforms prior work in identity preservation and geometric consistency.

Significance. If the claims hold, the contribution is practically valuable: it offers single-prompt text-guided editing of neural implicit surfaces without iterative dataset updates, with an architecture that explicitly separates foreground identity, background, and target edits. The user study is substantial (41 participants, 34 experiments, randomized order), and the evaluation spans three benchmark datasets with several scenes each. The method also ships a useful set of ablations, including a comparison of analytical versus numerical SDF gradients. The central novelty, the Phong-enhanced distillation loss, is clearly stated and is potentially interesting to the 3D editing community, but its validation is incomplete, and one ablation result is internally inconsistent.

major comments (3)
  1. [§5.2, Table 1] The ablation table directly contradicts the prose claim that the Phong-aware loss improves fidelity of fine details. On single-view LPIPS, the full method ('Ours') is worse than 'Arch + PDS' for both backbones (Alex: 0.700 vs 0.685; VGG: 0.683 vs 0.676), while only the multi-view consistency metric improves. Since the central claim of the paper is that LPE improves rendering and geometric quality, this inconsistency is load-bearing and needs to be resolved, either by reporting per-scene results, by explaining why single-view perceptual similarity may degrade while the user study still favors the full method, or by softening the claim in §5.2.
  2. [§4.3, Eqs. (13)–(18)] The geometry-aware term LPE feeds a Phong-shaded grayscale image, repeated across three channels, into the Stable Diffusion encoder. This input is out-of-distribution for a model trained on natural images, and the paper provides no experiment isolating the effect of this channel conversion, no comparison against alternative shading representations (e.g., normal maps or albedo-only renderings), and no measurement that the resulting latents carry geometric information rather than systematic grayscale-pattern bias. If the latents are biased, the gradient in Eq. (18) could pull the optimization in a spurious direction. The authors should add a dedicated validation: for example, measure latent-space distances between Phong-shaded inputs and natural images, or ablate LPE with a different shading input to show the geometric signal comes from the shading itself.
  3. [§5.3.1, Table 2] The identity-preservation metric is the LPIPS distance between the edited rendering and a reference rendering of the unedited scene produced by a separately trained NeRF. This conflates the intended magnitude of the edit with catastrophic forgetting: a method that successfully makes a drastic edit will necessarily have a large LPIPS distance even if no scene content is forgotten. The claim that 'our method preserves the original scene significantly better than competing approaches' therefore needs a more controlled evaluation, such as masking the foreground/background or measuring identity-related features in regions that should remain unchanged.
minor comments (4)
  1. [§5.3.1] There is a typo: 'editting' should be 'editing'.
  2. [Table 2 and §5.3] The labels are inconsistent: Table 2 uses 'PDSGS', while the text and Tables 3–4 use 'PDSSplat' for the same method. Please unify the notation.
  3. [§5.3.2, Tables 3–4] The user study reports mean scores but no statistical significance tests, confidence intervals, or inter-rater variability. Given the strong claims, at least a pairwise significance test would be appropriate.
  4. [§10, Eq. (22)] The conditioning input to the target geometry network is written as MLPtgt(Gsrc(x), Htgt(x)); the notation is clear, but the figure and text would benefit from stating explicitly whether Gsrc(x) refers to the source SDF value, the concatenated feature vector, or both.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the method is an empirical pipeline validated by external metrics and external pretrained diffusion models.

full rationale

NeuSEditor's central claims are empirical: an architecture with separate background, source, and target renderers, plus a distillation loss L_PEPDS = λ_PDS L_PDS + λ_PE L_PE that aligns stochastic latents from a frozen Stable Diffusion model. The source identity is learned in stage 1 from input images via photometric and eikonal losses and is retained as a separate renderer; the target renderer is initialized from and conditioned on the source. This is a design choice that plausibly explains identity preservation, not a fitted parameter later renamed as a prediction. The LPIPS identity-preservation evaluation uses an independently trained NeRF on the input images, so it is external to the proposed network. The LPE term feeds a repeated-grayscale Phong-shaded rendering into Stable Diffusion; whether this input is out-of-distribution is a correctness or robustness risk, not a circular reduction, because the Phong rendering is computed from target SDF gradients and compared with source latents through an external encoder. The ablation in Table 1 shows LPIPS(Alex) worsening when LPE is added (0.685 to 0.700) while multi-view consistency improves; this is an internal consistency concern, not evidence that any claim is definitionally forced. There is no load-bearing self-citation chain, no uniqueness theorem imported from the authors' prior work, no fitted input called a prediction, and no equation that reduces to its own inputs. The derivation is self-contained with respect to external benchmarks and baselines, so the appropriate finding is no significant circularity.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The method introduces no new physical or conceptual entities such as forces, particles, or dimensions. The free parameters and axioms listed above are the tunable choices and background assumptions the central claim depends on. The most fragile is the ad hoc Phong-shaded image assumption, which is the paper's own contribution and is not validated against the diffusion model's input distribution.

free parameters (4)
  • CFG guidance scale = 350 (100 for minimal edits)
    Chosen by hand per prompt. Fig. 4 shows it changes identity/edit tradeoff, and Tab. 7 footnote shows it was lowered for moustache/bowtie edits to encourage minimal changes. This tuned hyperparameter directly affects all results.
  • Phong loss weight lambda_PE = 0.2
    Set manually in Eq. (19), with lambda_PDS=1. The paper reports only one ablation value, suggesting qualitative tuning rather than a principled selection.
  • Identity learning iterations = 8K (DTU/Blender), versus 30K for baselines
    Hand-chosen to demonstrate identity preservation. The difference in training length between method and baselines is a confound in the comparison.
  • Numerical gradient step h = unspecified 'very small positive value'
    Used in Eqs. (21) and (23). The paper does not give the value, and the claim that numerical gradients outperform analytical gradients depends on this choice.
assumptions (4)
  • domain assumption Stochastic latent alignment between source and target images (PDS assumption) provides a valid editing gradient for 3D scenes.
    Inherited from PDS [25]. NeuSEditor's LPDS uses Eq. (10) directly, so if the source and target posterior latents do not share structure, the loss cannot preserve identity.
  • ad hoc to paper A Phong-shaded grayscale image repeated across three channels is a valid input to Stable Diffusion, and its stochastic latents encode geometric information.
    Section 4.3 states 'we convert the Phong shading to a three-channel image by simply repeating it across all three channels' without validation. The entire LPE term rests on this out-of-distribution input being meaningful to the diffusion model.
  • domain assumption Initializing the target renderer from the source renderer and conditioning it on source geometry makes edits additive and identity-preserving.
    Section 4.2 describes the initialization and conditioning. There is no proof that the learned residual edit stays bounded to the intended region, and source reconstruction errors propagate to the target.
  • domain assumption SDF-based rendering with progressive hash encoding (Neuralangelo) produces geometry accurate enough for downstream editing.
    Section 4.1 adopts Neuralangelo and NeuS. The method inherits their assumptions about photometric consistency and object-centric scenes.

how reviews work

0 comments
Cite this review

Pith. "Pith review of NeuSEditor: From Multi-View Images to Text-Guided Neural Surface Edits." pith.science (2026). https://pith.science/paper/NOVDXEAY

@misc{pith2026250510827,
  author       = {Pith},
  title        = {Pith review of: NeuSEditor: From Multi-View Images to Text-Guided Neural Surface Edits},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NOVDXEAY}},
  note         = {Machine review of arXiv:2505.10827}
}
read the original abstract

Implicit surface representations are valued for their compactness and continuity, but they pose significant challenges for editing. Despite recent advancements, existing methods often fail to preserve identity and maintain geometric consistency during editing. To address these challenges, we present NeuSEditor, a novel method for text-guided editing of neural implicit surfaces derived from multi-view images. NeuSEditor introduces an identity-preserving architecture that efficiently separates scenes into foreground and background, enabling precise modifications without altering the scene-specific elements. Our geometry-aware distillation loss significantly enhances rendering and geometric quality. Our method simplifies the editing workflow by eliminating the need for continuous dataset updates and source prompting. NeuSEditor outperforms recent state-of-the-art methods like PDS and InstructNeRF2NeRF, delivering superior quantitative and qualitative results. For more visual results, visit: neuseditor.github.io.

Figures

Figures reproduced from arXiv: 2505.10827 by the authors.

Figure 1
Figure 1. NeuSEditor is a novel method for text-guided neural surface editing from multi-view calibrated images. It generates an edited [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Our network architecture retains information about the identity (learned source) of the original scene, including background and [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Overview of our two stage scene editing pipeline. In the first stage (source learning) the network captures the scene’s identity, [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (9 more)
Figure 4
Figure 4. Figure 4: Influence of hyperparameter guidance scale. Left: two sample input images from the DTU dataset (scan105 scene) and the [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: The left column displays an input view from the DTU [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Comparison between IN2N, PDSNeRF, PDSSplat, and our method (in this order) across two frames. Edit prompts for ours and IN2N, and the target prompts for the PDS models are also provided below each row. 8 [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: Our network architecture retains information about the identity of the original scene, including background and foreground [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]
Figure 8
Figure 8. Figure 8: NeuSEditor integrates three dedicated renderers: a background renderer utilizing density fields, source and target foreground [PITH_FULL_IMAGE:figures/full_fig_p013_8.png]
Figure 9
Figure 9. Figure 9: Background editing results. Top row shows the full model render, while the bottom row displays the edited background under [PITH_FULL_IMAGE:figures/full_fig_p014_9.png]
Figure 10
Figure 10. Figure 10: This figure shows textured mesh reconstruction results under varying text prompts. The top-left quadrant depicts results without [PITH_FULL_IMAGE:figures/full_fig_p015_10.png]
Figure 11
Figure 11. Figure 11: Source and target geometry networks. 10. SDF gradient computation We have observed that, during editing, the numerical com￾putation of the SDF gradient (via finite differences) results in cleaner and smoother geometry compared to using the analytical gradient via torc…
Figure 12
Figure 12. Figure 12: Experiments conducted on the IN2N dataset (bear scene) demonstrate both analytical and numerical solutions for SDF gradient [PITH_FULL_IMAGE:figures/full_fig_p016_12.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

77 extracted references · 50 canonical work pages

  1. [1]

    Large-scale data for multiple-view stereopsis

    Henrik Aanæs, Rasmus Ramsbøl Jensen, George V ogiatzis, Engin Tola, and Anders Bjorholm Dahl. Large-scale data for multiple-view stereopsis. IJCV, 2016. 2, 6

    work page 2016

  2. [2]

    Cross-image attention for zero- shot appearance transfer

    Yuval Alaluf, Daniel Garibi, Or Patashnik, Hadar Averbuch- Elor, and Daniel Cohen-Or. Cross-image attention for zero- shot appearance transfer. In ACM SIGGRAPH, 2024. 1

    work page 2024

  3. [3]

    Tim Brooks, Aleksander Holynski, and Alexei A. Efros. In- structPix2Pix: Learning to follow image editing instructions. In CVPR, 2023. 1, 2

    work page 2023

  4. [4]

    Tensorf: Tensorial radiance fields

    Anpei Chen, Zexiang Xu, Andreas Geiger, Jingyi Yu, and Hao Su. Tensorf: Tensorial radiance fields. In ECCV, 2022. 2

    2022

  5. [5]

    Fan- tasia3D: Disentangling geometry and appearance for high- quality text-to-3D content creation

    Rui Chen, Yongwei Chen, Ningxin Jiao, and Kui Jia. Fan- tasia3D: Disentangling geometry and appearance for high- quality text-to-3D content creation. In ICCV, 2023. 1

    work page 2023

  6. [6]

    Gensdf: Two- stage learning of generalizable signed distance functions

    Gene Chou, Ilya Chugunov, and Felix Heide. Gensdf: Two- stage learning of generalizable signed distance functions. In NeurIPS, 2022. 1

    work page 2022

  7. [7]

    Diffusion-sdf: Conditional generative modeling of signed distance func- tions

    Gene Chou, Yuval Bahat, and Felix Heide. Diffusion-sdf: Conditional generative modeling of signed distance func- tions. 2023

    work page 2023

  8. [8]

    Obja- verse: A universe of annotated 3d objects

    Matt Deitke, Dustin Schwenk, Jordi Salvador, Luca Weihs, Oscar Michel, Eli VanderBilt, Ludwig Schmidt, Kiana Ehsani, Aniruddha Kembhavi, and Ali Farhadi. Obja- verse: A universe of annotated 3d objects. arXiv preprint arXiv:2212.08051, 2022

    arXiv 2022

Show all 77 references
  1. [9]

    Objaverse-xl: A universe of 10m+ 3d objects

    Matt Deitke, Ruoshi Liu, Matthew Wallingford, Huong Ngo, Oscar Michel, Aditya Kusupati, Alan Fan, Chris- tian Laforte, Vikram V oleti, Samir Yitzhak Gadre, Eli VanderBilt, Aniruddha Kembhavi, Carl V ondrick, Georgia Gkioxari, Kiana Ehsani, Ludwig Schmidt, and Ali Farhadi. Obja...

    2023 arXiv

  2. [10]

    Hyperdiffusion: Generating implicit neural fields with weight-space diffusion

    Ziya Erkoc ¸, Fangchang Ma, Qi Shan, Matthias Nießner, and Angela Dai. Hyperdiffusion: Generating implicit neural fields with weight-space diffusion. In ICCV, 2023. 1

    2023

  3. [11]

    Text-to-audio generation using instruc- tion tuned llm and latent diffusion model

    Deepanway Ghosal, Navonil Majumder, Ambuj Mehrish, and Soujanya Poria. Text-to-audio generation using instruc- tion tuned llm and latent diffusion model. arXiv preprint arXiv:2304.13731, 2023. 1

    2023 arXiv

  4. [12]

    Implicit geometric regularization for learning shapes

    Amos Gropp, Lior Yariv, Niv Haim, Matan Atzmon, and Yaron Lipman. Implicit geometric regularization for learning shapes. 2020. 4

    2020

  5. [13]

    Instant neural surface reconstruction, 2022

    Yuan-Chen Guo. Instant neural surface reconstruction, 2022. https://github.com/bennyguo/instant-nsr-pl. 1

    2022

  6. [14]

    threestudio: A unified framework for 3d content generation

    Yuan-Chen Guo, Ying-Tian Liu, Ruizhi Shao, Christian Laforte, Vikram V oleti, Guan Luo, Chia-Hao Chen, Zi- Xin Zou, Chen Wang, Yan-Pei Cao, and Song-Hai Zhang. threestudio: A unified framework for 3d content generation. https://github.com/threestudio- project/ threestudio, 2023. 1

    2023

  7. [15]

    Instruct-NeRF2NeRF: Editing 3D scenes with instructions

    Ayaan Haque, Matthew Tancik, Alexei Efros, Aleksander Holynski, and Angjoo Kanazawa. Instruct-NeRF2NeRF: Editing 3D scenes with instructions. In ICCV, 2023. 1, 2, 6, 7

    2023

  8. [16]

    Delta de- noising score

    Amir Hertz, Kfir Aberman, and Daniel Cohen-Or. Delta de- noising score. In ICCV, 2023. 2, 3, 4

    2023

  9. [17]

    Classifier-free diffusion guidance

    Jonathan Ho and Tim Salimans. Classifier-free diffusion guidance. In NeurIPS 2021 Workshop on Deep Generative Models and Downstream Applications, 2021. 1, 5

    2021

  10. [18]

    Denoising diffu- sion probabilistic models

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffu- sion probabilistic models. NeurIPS, 2020. 1, 3, 4

    2020

  11. [19]

    LoRA: Low-rank adaptation of large language models

    Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen- Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. LoRA: Low-rank adaptation of large language models. In ICLR, 2022. 2, 4

    2022

  12. [20]

    Make-an-audio: Text-to-audio genera- tion with prompt-enhanced diffusion models

    Rongjie Huang, Jiawei Huang, Dongchao Yang, Yi Ren, Luping Liu, Mingze Li, Zhenhui Ye, Jinglin Liu, Xiang Yin, and Zhou Zhao. Make-an-audio: Text-to-audio genera- tion with prompt-enhanced diffusion models. arXiv preprint arXiv:2301.12661, 2023. 1

    2023 arXiv

  13. [21]

    An edit friendly DDPM noise space: Inversion and manipulations

    Inbar Huberman-Spiegelglas, Vladimir Kulikov, and Tomer Michaeli. An edit friendly DDPM noise space: Inversion and manipulations. In CVPR, 2024. 2, 3

    2024

  14. [22]

    Shap-e: Generat- ing conditional 3d implicit functions

    Heewoo Jun and Alex Nichol. Shap-e: Generat- ing conditional 3d implicit functions. arXiv preprint arXiv:2305.02463, 2023. 1

    2023 arXiv

  15. [23]

    Spad: Spatially aware multi-view diffusers

    Yash Kant, Aliaksandr Siarohin, Ziyi Wu, Michael Vasilkovsky, Guocheng Qian, Jian Ren, Riza Alp Guler, Bernard Ghanem, Sergey Tulyakov, and Igor Gilitschenski. Spad: Spatially aware multi-view diffusers. In CVPR, 2024. 1, 7

    2024

  16. [24]

    3d gaussian splatting for real-time radiance field rendering

    Bernhard Kerbl, Georgios Kopanas, Thomas Leimk ¨uhler, and George Drettakis. 3d gaussian splatting for real-time radiance field rendering. ACM TOG, 2023. 1, 2

    2023

  17. [25]

    Posterior distil- lation sampling

    Juil Koo, Chanho Park, and Minhyuk Sung. Posterior distil- lation sampling. In CVPR, 2024. 1, 2, 3, 4, 5, 7

    2024

  18. [26]

    Blip: Bootstrapping language-image pre-training for unified vision-language understanding and generation

    Junnan Li, Dongxu Li, Caiming Xiong, and Steven Hoi. Blip: Bootstrapping language-image pre-training for unified vision-language understanding and generation. 2022. 1

    2022

  19. [27]

    Nerfacc: Efficient sampling accelerates nerfs

    Ruilong Li, Hang Gao, Matthew Tancik, and Angjoo Kanazawa. Nerfacc: Efficient sampling accelerates nerfs. arXiv preprint arXiv:2305.04966, 2023. 1

    2023 arXiv

  20. [28]

    Neuralangelo: High-fidelity neural surface reconstruction

    Zhaoshuo Li, Thomas M ¨uller, Alex Evans, Russell H Tay- lor, Mathias Unberath, Ming-Yu Liu, and Chen-Hsuan Lin. Neuralangelo: High-fidelity neural surface reconstruction. In CVPR, 2023. 1, 3, 4

    2023

  21. [29]

    Magic3D: High-resolution text-to-3D content creation

    Chen-Hsuan Lin, Jun Gao, Luming Tang, Towaki Takikawa, Xiaohui Zeng, Xun Huang, Karsten Kreis, Sanja Fidler, Ming-Yu Liu, and Tsung-Yi Lin. Magic3D: High-resolution text-to-3D content creation. In CVPR, 2023. 1

    2023

  22. [30]

    Zero-1-to-3: Zero-shot one image to 3d object

    Ruoshi Liu, Rundi Wu, Basile Van Hoorick, Pavel Tok- makov, Sergey Zakharov, and Carl V ondrick. Zero-1-to-3: Zero-shot one image to 3d object. In ICCV, 2023. 1

    2023

  23. [31]

    Neural vol- umes: learning dynamic renderable volumes from images

    Stephen Lombardi, Tomas Simon, Jason Saragih, Gabriel Schwartz, Andreas Lehrmann, and Yaser Sheikh. Neural vol- umes: learning dynamic renderable volumes from images. ACM TOG, 2019. 2

    2019

  24. [32]

    SDEdit: Guided 9 image synthesis and editing with stochastic differential equa- tions

    Chenlin Meng, Yutong He, Yang Song, Jiaming Song, Jia- jun Wu, Jun-Yan Zhu, and Stefano Ermon. SDEdit: Guided 9 image synthesis and editing with stochastic differential equa- tions. In ICLR, 2022. 1

    2022

  25. [33]

    Latent-nerf for shape-guided generation of 3D shapes and textures

    Gal Metzer, Elad Richardson, Or Patashnik, Raja Giryes, and Daniel Cohen-Or. Latent-nerf for shape-guided generation of 3D shapes and textures. In CVPR, 2023. 1

    2023

  26. [34]

    Srinivasan, Matthew Tancik, Jonathan T

    Ben Mildenhall, Pratul P. Srinivasan, Matthew Tancik, Jonathan T. Barron, Ravi Ramamoorthi, and Ren Ng. NeRF: Representing scenes as neural radiance fields for view syn- thesis. In ECCV, 2020. 1, 2, 3, 4, 6

    2020

  27. [35]

    Instant neural graphics primitives with a multires- olution hash encoding

    Thomas M ¨uller, Alex Evans, Christoph Schied, and Alexan- der Keller. Instant neural graphics primitives with a multires- olution hash encoding. ACM TOG, 2022. 1, 2, 3, 5

    2022

  28. [36]

    Contrastive denoising score for text-guided latent diffusion image editing

    Hyelin Nam, Gihyun Kwon, Geon Yeong Park, and Jong Chul Ye. Contrastive denoising score for text-guided latent diffusion image editing. In CVPR, 2024. 4

    2024

  29. [37]

    Point-e: A system for generat- ing 3d point clouds from complex prompts

    Alex Nichol, Heewoo Jun, Prafulla Dhariwal, Pamela Mishkin, and Mark Chen. Point-e: A system for generat- ing 3d point clouds from complex prompts. arXiv preprint arXiv:2212.08751, 2022. 1

    2022 arXiv

  30. [38]

    Differentiable volumetric rendering: Learn- ing implicit 3d representations without 3d supervision

    Michael Niemeyer, Lars Mescheder, Michael Oechsle, and Andreas Geiger. Differentiable volumetric rendering: Learn- ing implicit 3d representations without 3d supervision. In CVPR, 2020. 1, 2

    2020

  31. [39]

    Floaters no more: Radiance field gradient scaling for improved near-camera training

    Julien Philip and Valentin Deschaintre. Floaters no more: Radiance field gradient scaling for improved near-camera training. Eurographics Symposium on Rendering, 2023. 1, 5

    2023

  32. [40]

    Barron, and Ben Milden- hall

    Ben Poole, Ajay Jain, Jonathan T. Barron, and Ben Milden- hall. Dreamfusion: Text-to-3D using 2D diffusion. In ICLR,

  33. [41]

    Learn- ing transferable visual models from natural language super- vision

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learn- ing transferable visual models from natural language super- vision. 2021. 1, 5

    2021

  34. [42]

    Dreambooth3D: Subject-driven text-to-3D generation

    Amit Raj, Srinivas Kaza, Ben Poole, Michael Niemeyer, Nataniel Ruiz, Ben Mildenhall, Shiran Zada, Kfir Aberman, Michael Rubinstein, Jonathan Barron, et al. Dreambooth3D: Subject-driven text-to-3D generation. In ICCV, 2023. 1

    2023

  35. [43]

    Hierarchical text-conditional image gen- eration with clip latents

    Aditya Ramesh, Prafulla Dhariwal, Alex Nichol, Casey Chu, and Mark Chen. Hierarchical text-conditional image gen- eration with clip latents. arXiv preprint arXiv:2204.06125,

    arXiv

  36. [44]

    High-resolution image syn- thesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj¨orn Ommer. High-resolution image syn- thesis with latent diffusion models. In CVPR, 2022. 3, 4

    2022

  37. [45]

    Dreambooth: Fine tuning text-to-image diffusion models for subject-driven generation

    Nataniel Ruiz, Yuanzhen Li, Varun Jampani, Yael Pritch, Michael Rubinstein, and Kfir Aberman. Dreambooth: Fine tuning text-to-image diffusion models for subject-driven generation. In CVPR, 2023. 1

    2023

  38. [46]

    Stable diffusion v1.5

    RunwayML. Stable diffusion v1.5. https : / / huggingface . co / stable - diffusion - v1 - 5 / stable-diffusion-v1-5 , 2022. 4

    2022

  39. [47]

    Structure-from-motion revisited

    Johannes Lutz Sch ¨onberger and Jan-Michael Frahm. Structure-from-motion revisited. In CVPR, 2016. 1, 6

    2016

  40. [48]

    Laion-400m: Open dataset of clip-filtered 400 million image-text pairs

    Christoph Schuhmann, Richard Vencu, Romain Beaumont, Robert Kaczmarczyk, Clayton Mullis, Aarush Katta, Theo Coombes, Jenia Jitsev, and Aran Komatsuzaki. Laion-400m: Open dataset of clip-filtered 400 million image-text pairs. arXiv preprint arXiv:2111.02114, 2021. 1

    2021 arXiv

  41. [49]

    Laion-5b: An open large-scale dataset for training next generation image-text models

    Christoph Schuhmann, Romain Beaumont, Richard Vencu, Cade Gordon, Ross Wightman, Mehdi Cherti, Theo Coombes, Aarush Katta, Clayton Mullis, Mitchell Worts- man, et al. Laion-5b: An open large-scale dataset for training next generation image-text models. In NeurIPS, 2022. 1

    2022

  42. [50]

    MVDream: Multi-view diffusion for 3D generation

    Yichun Shi, Peng Wang, Jianglong Ye, Mai Long, Kejie Li, and Xiao Yang. MVDream: Multi-view diffusion for 3D generation. arXiv preprint arXiv:2308.16512, 2023. 1, 7

    2023 arXiv

  43. [51]

    Diffusion-based signed distance fields for 3d shape gener- ation

    Jaehyeok Shim, Changwoo Kang, and Kyungdon Joo. Diffusion-based signed distance fields for 3d shape gener- ation. In CVPR, 2023. 1

    2023

  44. [52]

    Denois- ing diffusion implicit models

    Jiaming Song, Chenlin Meng, and Stefano Ermon. Denois- ing diffusion implicit models. In ICLR, 2021. 1

    2021

  45. [53]

    Generative modeling by esti- mating gradients of the data distribution

    Yang Song and Stefano Ermon. Generative modeling by esti- mating gradients of the data distribution. In NeurIPS, 2019

    2019

  46. [54]

    Score-based generative modeling through stochastic differential equa- tions

    Yang Song, Jascha Sohl-Dickstein, Diederik P Kingma, Ab- hishek Kumar, Stefano Ermon, and Ben Poole. Score-based generative modeling through stochastic differential equa- tions. In ICLR, 2021. 1

    2021

  47. [55]

    Direct voxel grid optimization: Super-fast convergence for radiance fields reconstruction

    Cheng Sun, Min Sun, and Hwann-Tzong Chen. Direct voxel grid optimization: Super-fast convergence for radiance fields reconstruction. In CVPR, 2022. 2

    2022

  48. [56]

    Nerfstudio: A modular framework for neural radiance field development

    Matthew Tancik, Ethan Weber, Evonne Ng, Ruilong Li, Brent Yi, Justin Kerr, Terrance Wang, Alexander Kristof- fersen, Jake Austin, Kamyar Salahi, Abhik Ahuja, David McAllister, and Angjoo Kanazawa. Nerfstudio: A modular framework for neural radiance field development. In ACM SI...

    2023

  49. [57]

    Instruct-gs2gs: Editing 3d gaussian splats with instructions, 2024

    Cyrus Vachha and Ayaan Haque. Instruct-gs2gs: Editing 3d gaussian splats with instructions, 2024. 2

    2024

  50. [58]

    Neus: Learning neural implicit surfaces by volume rendering for multi-view reconstruction

    Peng Wang, Lingjie Liu, Yuan Liu, Christian Theobalt, Taku Komura, and Wenping Wang. Neus: Learning neural implicit surfaces by volume rendering for multi-view reconstruction. In NeurIPS, 2021. 2, 3, 4, 5, 1

    2021

  51. [59]

    Prolificdreamer: High-fidelity and diverse text-to-3d generation with variational score distilla- tion

    Zhengyi Wang, Cheng Lu, Yikai Wang, Fan Bao, Chongxuan Li, Hang Su, and Jun Zhu. Prolificdreamer: High-fidelity and diverse text-to-3d generation with variational score distilla- tion. In NeurIPS, 2023. 1

    2023

  52. [60]

    A latent space of stochastic diffusion models for zero-shot image editing and guidance

    Chen Henry Wu and Fernando De la Torre. A latent space of stochastic diffusion models for zero-shot image editing and guidance. In ICCV, 2023. 2, 3

    2023

  53. [61]

    Point-nerf: Point- based neural radiance fields

    Qiangeng Xu, Zexiang Xu, Julien Philip, Sai Bi, Zhixin Shu, Kalyan Sunkavalli, and Ulrich Neumann. Point-nerf: Point- based neural radiance fields. In CVPR, 2022. 2

    2022

  54. [62]

    Diffsound: Discrete diffusion model for text-to-sound generation

    Dongchao Yang, Jianwei Yu, Helin Wang, Wen Wang, Chao Weng, Yuexian Zou, and Dong Yu. Diffsound: Discrete diffusion model for text-to-sound generation. IEEE/ACM Transactions on Audio, Speech, and Language Processing ,

  55. [63]

    Multiview neu- ral surface reconstruction by disentangling geometry and ap- pearance

    Lior Yariv, Yoni Kasten, Dror Moran, Meirav Galun, Matan Atzmon, Basri Ronen, and Yaron Lipman. Multiview neu- ral surface reconstruction by disentangling geometry and ap- pearance. In NeurIPS, 2020. 1, 2

    2020

  56. [64]

    V ol- ume rendering of neural implicit surfaces

    Lior Yariv, Jiatao Gu, Yoni Kasten, and Yaron Lipman. V ol- ume rendering of neural implicit surfaces. InNeurIPS, 2021. 2, 3, 4 10

    2021

  57. [65]

    gsplat: An open-source library for Gaussian splatting

    Vickie Ye, Ruilong Li, Justin Kerr, Matias Turkulainen, Brent Yi, Zhuoyang Pan, Otto Seiskari, Jianbo Ye, Jeffrey Hu, Matthew Tancik, and Angjoo Kanazawa. gsplat: An open-source library for Gaussian splatting. arXiv preprint arXiv:2409.06765, 2024. 2, 7

    2024 arXiv

  58. [66]

    PlenOctrees for real-time rendering of neural radiance fields

    Alex Yu, Ruilong Li, Matthew Tancik, Hao Li, Ren Ng, and Angjoo Kanazawa. PlenOctrees for real-time rendering of neural radiance fields. In ICCV, 2021. 2

    2021

  59. [67]

    Sdfstudio: A unified framework for surface reconstruction, 2022

    Zehao Yu, Anpei Chen, Bozidar Antic, Songyou Peng, Apra- tim Bhattacharyya, Michael Niemeyer, Siyu Tang, Torsten Sattler, and Andreas Geiger. Sdfstudio: A unified framework for surface reconstruction, 2022. 1, 3

    2022

  60. [68]

    Nerf++: Analyzing and improving neural radiance fields

    Kai Zhang, Gernot Riegler, Noah Snavely, and Vladlen Koltun. Nerf++: Analyzing and improving neural radiance fields. arXiv preprint arXiv:2010.07492, 2020. 3, 1

    2010 arXiv

  61. [69]

    Adding conditional control to text-to-image diffusion models, 2023

    Lvmin Zhang, Anyi Rao, and Maneesh Agrawala. Adding conditional control to text-to-image diffusion models, 2023. 2, 4

    2023

  62. [70]

    The unreasonable effectiveness of deep features as a perceptual metric

    Richard Zhang, Phillip Isola, Alexei A Efros, Eli Shechtman, and Oliver Wang. The unreasonable effectiveness of deep features as a perceptual metric. In CVPR, 2018. 5

    2018

  63. [71]

    HiFA: High-fidelity text-to- 3D with advanced diffusion guidance

    Joseph Zhu and Peiye Zhuang. HiFA: High-fidelity text-to- 3D with advanced diffusion guidance. In ICLR, 2024. 1

    2024

  64. [72]

    Dreameditor: Text-driven 3d scene editing with neural fields

    Jingyu Zhuang, Chen Wang, Liang Lin, Lingjie Liu, and Guanbin Li. Dreameditor: Text-driven 3d scene editing with neural fields. In SIGGRAPH Asia, 2023. 1 11 NeuSEditor: From Multi-View Images to Text-Guided Neural Surface Edits Supplementary Material Input imageBackground (𝐼𝑏𝑔...

    2023

  65. [73]

    Network Architecture Details As shown in Fig. 7, our network architecture simultane- ously captures three critical components: (1) the back- ground scene context, (2) the source input’s foreground el- ements, and (3) the target edited scene composition. The architecture compri...

  66. [74]

    additive learning

    Background editing As discussed in Sec. 7, the background is modeled using ra- diance fields, with a separate set of hash grids as positional encodings. This background model utilizes its own geom- etry and renderer subnetworks. Our method also supports background editing. Sim...

  67. [75]

    make it a church’’). However, like most generative tech- 3 𝑦!

    A voiding mode collapse Mode collapse is a common issue in generative AI pipelines (e.g. GANs, GPTs, text-to-3Ds) where the generator learns to produce a limited set of (or similar) outputs, ignoring the full diversity of the target data distribution. Attentive read- ers may h...

  68. [76]

    turn the bear into a polar bear

    SDF gradient computation We have observed that, during editing, the numerical com- putation of the SDF gradient (via finite differences) results in cleaner and smoother geometry compared to using the analytical gradient via torch.autograd.grad. Fig. 11 shows the geometry netwo...

  69. [77]

    up” vector to be normal of the dominant ground plane, with the scene center defined as the closest “intersection

    Dataset and benchmark details Algorithm 1 Generate DTU Spherical Camera Poses 1: function DTUS PHERIC POSES (cams, n steps) 2: center← [0, 0, 0] 3: cam center← mean(cams) 4: eigvecs← eigenvectors(camsT× cams) 5: up← eigvecs[:, 1] 6: rot dir← cross(up, cam center) 7: max angle←...

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.