SuperNeRF-GAN: A Universal 3D-Consistent Super-Resolution Framework for Efficient and Enhanced 3D-Aware Image Synthesis

Linzhi Huang; Peng Zheng; Rui Ma; Yi Chang; Yilin Wang; Yizhou Yu

REVIEW 4 major objections 6 minor 46 references

SuperNeRF-GAN: A Universal 3D-Consistent Super-Resolution Framework for Efficient and Enhanced 3D-Aware Image Synthesis

T0 review · 4 major / 6 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read A three-samples-per-ray depth map gives NeRF-GANs 3D-consistent super-resolution.

desk verdict Solid incremental 3D-consistent super-resolution for NeRF GANs, but the 'universal' claim overreaches given the three-sample-per-ray depth model. read the letter →

arxiv 2501.06770 v2 pith:DULOYAVI submitted 2025-01-12 cs.CV

Peng Zheng , Linzhi Huang , Yizhou Yu , Yi Chang , Yilin Wang , Rui Ma This is my paper · ORCID

classification cs.CV

keywords 3D-awareimagesynthesisneuralradiancefieldsgenerativeadversarialnetworks3D-consistentsuper-resolutiondepth-guidedrenderingmulti-depthmaptri-planerepresentationhigh-resolution

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

SuperNeRF-GAN sets out to establish that a pre-trained NeRF-based 3D-aware generator can be upgraded to produce high-resolution, view-consistent images without retraining the generator and without the 2D super-resolution step that breaks multi-view consistency. The method renders a low-resolution image plus depth and normal maps, learns a high-resolution NeRF from the low-resolution one, and then renders that high-resolution NeRF using only three depth layers per ray chosen from a boundary-correct multi-depth map. The paper reports that this drops sampling from 64 points per ray to 3, reaches 1024x1024 output on portraits, cat faces, and full bodies, and improves 3D consistency over the base generator while using less memory than a two-stage 3D-consistent alternative. A sympathetic reader would care because it promises a plug-in path from existing 3D GANs to high-resolution free-view rendering for virtual reality, video games, and avatar applications.

What carries the argument

The load-bearing object is the boundary-correct multi-depth map $\mathbf{D}_{\mathrm{HR}}$, a three-channel depth map that records depth values on both sides of object boundaries. It is built by eroding and dilating the low-resolution depth map, concatenating the two results with the original depth to form a three-channel aggregate, and super-resolving that aggregate with normal-map guidance: the normal map supplies local depth slopes $\Delta_x = N_x/N_z$ and $\Delta_y = N_y/N_z$, and a softmax weighting assigns each interpolated point to the most plausible side of a boundary. Rendering then follows $\mathrm{C}(\mathbf{r}) = \sum_{i=1}^{3} T_i(1-\exp(-\sigma_i\delta_i))c_i$, with $\delta_i$ the spacing between sorted depth values and $\sigma_i$, $c_i$ read from the high-resolution NeRF. This reduces the per-ray sampling count from 64 to 3, which is the mechanism that makes high-resolution, 3D-consistent rendering cheap enough to run within GAN training budgets.

What would settle it

Render a portrait at a viewpoint where some pixel's ray crosses at least four distinct surfaces, such as hair in front of an ear in front of a cheek with background behind it, and compare that pixel against a dense 64-sample render of the same latent code; a hole, ghost, or mismatched color at that pixel would show the three-channel depth map is not sufficient.

Watch

Extended reading notes

Core claim

The central claim is that a three-channel depth map, constructed from the low-resolution depth by dilation, erosion, and normal-guided upsampling, carries enough geometric information to replace dense ray sampling in high-resolution volume rendering. The depth-guided rendering equation sums color and density over exactly the three stored depth values per ray, interpolating $\sigma_i$ and $c_i$ from the high-resolution NeRF representation. As a result the method renders 1024x1024 images directly from the 3D representation instead of upsampling a low-resolution image in 2D, which is the source of the flicker and inconsistency the paper attributes to earlier methods. The reported trade-off is intentional: image quality remains slightly below 2D-super-resolution methods, but multi-view consistency, measured by reconstructing a neural surface from generated views and comparing rendered images, improves.

Load-bearing premise

The load-bearing premise is that every ray's visible content can be captured by exactly three depth layers; if a ray crosses more than three distinct surfaces, the depth-guided renderer has nowhere to store the missing layers and will drop content.

Editorial extensions

If this is right

A user with a pre-trained tri-plane generator can add the module and obtain 1024x1024 renderings that stay stable when the camera moves, without retraining the base generator.
Depth-guided rendering lowers GPU memory for full-body 1024px synthesis from about 31GB to about 14GB in the reported comparison, making high-resolution 3D generation feasible on smaller hardware.
Because the high-resolution image is rendered from the NeRF rather than upsampled in 2D, latent-space interpolation remains 3D-consistent at every intermediate code.
The dilation-and-erosion aggregation plus normal guidance fixes boundary depth errors that the paper says prevent the prior two-stage method from working on portraits and cat faces.
The remaining bottleneck is the low-resolution dense sampling stage, so real-time rendering would need a replacement for that step, such as a coarse depth estimate or a non-NeRF representation.

Reading between the lines

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

Beyond the paper, the three-layer depth assumption suggests the method will start to fail on scenes with more than three transparent or layered surfaces per ray; adding more channels or an adaptive surface count is the natural extension.
Beyond the paper, the same boundary-correct depth map could be applied to 3D Gaussian splatting generators, where a depth-sorted set of splats would play the role of the three layers; the paper itself notes replacing NeRF with 3DGS is relatively unexplored.
Beyond the paper, the unsample loss that ties low- and high-resolution renderings may make the module useful for 3D GAN inversion and editing, since a single latent code now has to explain both resolutions consistently.

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.

Desk Editor's Note

read the letter

Short version: this is a useful, mostly honest extension of SH-HD that swaps naive neighbor depth aggregation for dilation/erosion plus normal-guided depth upsampling, and couples it with a shared-latent NeRF super-resolution module. It runs on EG3D across faces, cats, and full bodies, and the memory numbers are genuinely better (14 GB vs 31 GB for 1024²). That's a real contribution.

The ablation tables back up the two main components. Normal-guided depth SR gives a clear PSNR/SSIM lift in 3D-consistency, and dilation/erosion beats SH-HD's aggregation on portraits. I believe the components are new relative to the cited prior work, even though the paper is a direct follow-up to the authors' own SH-HD.

Soft spots, in rough order. The three-depth-layer assumption is the biggest one. Eq. 3 sums over exactly three surfaces per ray, and those three come from min/max erosion/dilation rather than any semantic reasoning. Rays crossing ears, glasses, hair, or clothing layers can in principle miss geometry. The stress-test note is right that the five-channel ablation in Table VII doesn't fix this because it still encodes min/max variations, not genuinely independent surfaces. The paper doesn't quantify how often such rays occur in FFHQ/AFHQ/DeepFashion. That said, the reported results show no obvious holes or ghosting on those datasets, and the limitation section is upfront about the quality tradeoff. So I'd treat it as a scope limitation, not a load-bearing flaw for the datasets evaluated.

Second, the evaluation has no error bars and some baselines are quoted from other papers. Table V's StyleNeRF/StyleSDF numbers come from GRAM-HD, not from a shared protocol. That's common in this literature but worth flagging for the referee.

Third, no code or data release. Given the authors' prior work, the components are plausibly reproducible, but the field would benefit from it.

Who's this for? Researchers building on EG3D or SH-HD who need 3D-consistent HR synthesis without the memory blowup. They'll get a workable recipe and a clear comparison. It doesn't change the fundamental tradeoff between 2D-SR quality and true 3D consistency, but it pushes the Pareto frontier.

My recommendation: send it to peer review. It's a solid subfield paper with an overclaimed title. A careful referee should ask for analysis of multi-layer depth failure modes and, ideally, code. Don't desk-reject it.

Referee Report

4 major / 6 minor

Summary. This paper presents SuperNeRF-GAN, a plug-in framework for converting a pre-trained NeRF-based 3D-aware generator into a 3D-consistent high-resolution generator. Given the generator's low-resolution tri-plane, the method volume-renders a 256x256 image together with depth and normal maps, upsamples the tri-plane with a StyleGAN2-based NeRF Super-Resolution module, and constructs a 1024x1024 three-channel 'boundary-correct multi-depth map' by dilation/erosion plus normal-guided depth interpolation. Rendering then samples only three points per ray at these depths. The method is evaluated on FFHQ portrait, AFHQ cat, and DeepFashion full-body generation, comparing FID/KID, NeuS-based PSNR/SSIM consistency, and training memory against EG3D, SH-HD, GRAM-HD, StyleNeRF, StyleSDF, EVA3D, VeRi3D, and GSM.

Significance. The central idea is appealing and practically motivated: if a small number of depth layers can represent the relevant geometry, depth-guided rendering makes 1024x1024 synthesis much cheaper, and the framework is in principle agnostic to the underlying NeRF representation. The evaluation spans three datasets and several baselines, reports memory savings, and includes ablation studies for the depth aggregation and depth super-resolution steps. The authors are also honest in the Limitation section about the residual image-quality gap and the remaining dense-sampling cost at the low-resolution stage. If the depth-layer assumption and the evaluation protocol are tightened, the method could be a useful and general component for 3D-aware synthesis.

major comments (4)

[III-C2 and III-D1] The efficiency claim rests on the three-sample-per-ray rendering in Eq. (3), where the three depth values per pixel are the original, eroded, and dilated LR depths (Section III-D1). This construction encodes at most three surface crossings per ray. A ray that passes through more than three distinct surfaces, such as glasses, hair wisps, ears, or layered clothing, will silently miss the omitted layers, producing holes or ghosting exactly where 3D consistency matters. The paper targets exactly such scenes (FFHQ portraits, AFHQ cats, DeepFashion bodies) but provides no measurement of how often more than three depth discontinuities occur per ray in these datasets. The ablation in Table VII with a 5-channel map adds only a second min/max pair after repeated dilation/erosion, not independent surface depths, so it does not address this concern. The 'universal' claim in the abstract and Section I is therefore unsupported unless the authors either quantify the frequency of >3-layer rays on these datasets and show the failures are negligible, or redesign the map to capture all significant surface crossings.
[V-A, V.3, Tables IV-VI] The quantitative evidence for improved 3D consistency is not fully conclusive. The PSNR/SSIM in Table IV are computed by reconstructing a NeuS model from each method's synthesized images and comparing the reconstruction to the synthesized images; this conflates multi-view consistency with reconstruction fidelity and image sharpness. In addition, the baseline numbers for StyleNeRF, StyleSDF, GRAM-HD, EVA3D, VeRi3D, and GSM are taken from other papers (Tables IV, V, VI), which may use different yaw ranges, truncation, NeuS settings, or evaluation code. No error bars or multiple-seed runs are reported anywhere. Please rerun at least the main baselines under the same protocol, or provide confidence intervals from several seeds, so the claimed improvements in 3D consistency and quality are not an artifact of protocol differences.
[III-C1 and Section VI] The NeRF Super-Resolution module produces a new HR tri-plane representation, but the depth map used for depth-guided rendering is derived solely from the LR rendering. There is no mechanism that guarantees the HR tri-plane's density field has its surfaces exactly at the three depths stored in DHR. If the HR module shifts geometry or creates additional surfaces, the depth-guided sampling cannot recover them. The paper does not analyze the alignment between the LR depth map and the HR tri-plane, nor does it report a failure case. This is a second load-bearing assumption of the efficiency claim and should be explicitly tested, e.g., by comparing density peaks along sampled rays against the depth map.
[Table III and IV.2] The SH-HD baseline on FFHQ and AFHQ is not the original SH-HD: Section IV.2 states that the authors modified SH-HD by integrating its components into pre-trained EG3D models. The resulting 'SH-HD' may not reflect SH-HD's actual behavior on those datasets, so the conclusion that SH-HD fails on portraits and cats (Table III, Fig. 3) is not established by a faithful comparison. The authors should either run the original SH-HD code, if it can be adapted, or carefully justify that their modified version is equivalent up to the task-level adaptations.

minor comments (6)

[III-E2] The term 'unsample loss' is unclear; define it precisely and cite the exact loss formulation from SH-HD.
[III-D2, Eq. (8)] The placement of the square-root denominator in Eq. (8) is ambiguous; only the second term appears divided by the norm. Please add parentheses to clarify the intended expression.
[III-D2] The description of applying the normal-guided module 'twice' to go from 256 to 1024 is terse; clarify whether the module is applied recursively at 2x scales and whether the same weights are shared.
[Tables II and VII] Several table entries contain repeated digit groups (e.g., '4.654.65', '36.4436.44', '8.478.47'); please clean the table formatting.
[Fig. 3 caption] The caption says the results are 'synthesized by untrained SuperNeRF-GAN models'; clarify whether these are ablations without depth aggregation rather than untrained networks.
[Section VI heading] 'LIMIATION' is a typo and should read 'LIMITATION'.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the depth-guided rendering and NeRF super-resolution pipeline is defined by explicit equations and evaluated against external metrics, with self-citations limited to components and baselines.

full rationale

SuperNeRF-GAN's derivation chain is self-contained rather than circular. The HR image is rendered by Eq. 3, a discretized volume rendering over three depth values, and those depth values come from the boundary-correct multi-depth map DHR constructed from the LR depth map and normal map through explicit morphological operations (Eqs. 5-6), normal-guided upsampling (Eqs. 7-10), and sorting. The HR NeRF representation is produced by a separately trained StyleGAN2-based super-resolution module, not by fitting the evaluation metrics. The reported FID, KID, PSNR, and SSIM values are computed post hoc on generated images and NeuS reconstructions, and none of these targets appears in the training losses. The SH-HD self-citations supply the unsample loss, a baseline, and a pre-trained model for full-body synthesis, but the central boundary-correct depth construction is explicitly differentiated from SH-HD's neighbor-aware aggregation and is validated by ablations (Tables VII-VIII) rather than assumed from the citation. The paper's main limitation, that only three depth layers are sampled per ray, is a genuine robustness concern for scenes with more than three surface crossings, but it is not circular: the model actually renders from the constructed depth map instead of re-displaying its inputs. No equation or fitted parameter is equivalent to the claims by construction.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central claim rests on standard volume rendering, pre-trained GAN backbones, and the empirical assumption that a three-layer depth map captures all relevant geometry. The only hand-tuned values are the structuring element, the channel count, and the loss weight inherited from SH-HD. No constants are fitted to force the reported 3D-consistency numbers, and no new physical entities are introduced.

free parameters (4)

Structuring element B for dilation and erosion
A square structuring element is used for depth aggregation in Section III-D1, but its size is not reported, so the boundary-correct depth map depends on a hand-chosen kernel.
Multi-depth channel count = 3 channels, with 5-channel ablation
The depth map uses three channels (DLR, Ddil, Dero) in the default setting. The ablation in Table VII shows that five channels from (D&E)^2 performs worse, indicating the channel count is a tuned design choice.
Unsample loss weight
The unsample loss from SH-HD is adopted to train the NeRF super-resolution module, but its weight is not specified in this paper, making it a hidden hyperparameter.
Truncation weight for high-resolution demo = 0.5 waveraged + 0.5 wrandom
In Section V-C3, the latent is set to a weighted average to balance quality and diversity for the high-resolution qualitative results, which is a hand-chosen parameter not central to the main evaluation.

assumptions (5)

standard math The NeRF volume rendering integral (Eq. 1) correctly describes how colors, normals, and depths are accumulated along rays.
Adopted from NeRF [1] without proof; the entire pipeline builds on this equation, including the depth and normal rendering.
domain assumption Pre-trained EG3D and SH-HD generators provide valid tri-plane NeRF representations, depth maps, and normal maps.
The framework requires a pre-trained generator capable of producing a NeRF representation and rendering normals and depth, as stated in Section III-B. Not all NeRF-based generators may provide accurate normals.
domain assumption Three depth layers per ray suffice for high-quality rendering (Eq. 3 sums over only three terms).
The boundary-correct multi-depth map has three channels; if a ray crosses more than three surfaces, content is missed. This is the load-bearing assumption behind the reduction from 64 to 3 samples.
domain assumption The normal map from the low-resolution representation gives accurate local depth gradients for normal-guided depth super-resolution (Eqs. 7-9).
The weight functions use Nx/Nz and Ny/Nz as depth differences; if normals are noisy, the super-resolved depth map will be wrong.
domain assumption NeuS reconstruction from 30 rendered images is a faithful proxy for 3D consistency.
The paper itself notes in Section V-A that quantitative improvements in 3D-consistency may appear subtle due to NeuS limitations, making this metric a fragile support for the consistency claim.

how reviews work

0 comments

Cite this review

Pith. "Pith review of SuperNeRF-GAN: A Universal 3D-Consistent Super-Resolution Framework for Efficient and Enhanced 3D-Aware Image Synthesis." pith.science (2026). https://pith.science/paper/DULOYAVI

@misc{pith2026250106770,
  author       = {Pith},
  title        = {Pith review of: SuperNeRF-GAN: A Universal 3D-Consistent Super-Resolution Framework for Efficient and Enhanced 3D-Aware Image Synthesis},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DULOYAVI}},
  note         = {Machine review of arXiv:2501.06770}
}

read the original abstract

Neural volume rendering techniques, such as NeRF, have revolutionized 3D-aware image synthesis by enabling the generation of images of a single scene or object from various camera poses. However, the high computational cost of NeRF presents challenges for synthesizing high-resolution (HR) images. Most existing methods address this issue by leveraging 2D super-resolution, which compromise 3D-consistency. Other methods propose radiance manifolds or two-stage generation to achieve 3D-consistent HR synthesis, yet they are limited to specific synthesis tasks, reducing their universality. To tackle these challenges, we propose SuperNeRF-GAN, a universal framework for 3D-consistent super-resolution. A key highlight of SuperNeRF-GAN is its seamless integration with NeRF-based 3D-aware image synthesis methods and it can simultaneously enhance the resolution of generated images while preserving 3D-consistency and reducing computational cost. Specifically, given a pre-trained generator capable of producing a NeRF representation such as tri-plane, we first perform volume rendering to obtain a low-resolution image with corresponding depth and normal map. Then, we employ a NeRF Super-Resolution module which learns a network to obtain a high-resolution NeRF. Next, we propose a novel Depth-Guided Rendering process which contains three simple yet effective steps, including the construction of a boundary-correct multi-depth map through depth aggregation, a normal-guided depth super-resolution and a depth-guided NeRF rendering. Experimental results demonstrate the superior efficiency, 3D-consistency, and quality of our approach. Additionally, ablation studies confirm the effectiveness of our proposed components.

Figures

Figures reproduced from arXiv: 2501.06770 by the authors.

**Figure 1.** Effectiveness of our proposed SuperNeRF-GAN. The images in the first row are synthesized by existing pre-trained [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗

**Figure 2.** Pipeline of the proposed SuperNeRF-GAN framework. Given a random noise [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

**Figure 3.** The left two figures demonstrate the effectiveness of [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Figures from the paper (6 more)

**Figure 4.** Figure 4: Effectiveness of Normal-Guided Depth Super [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]

**Figure 5.** Figure 5: Qualitative comparison among 3D-aware image synthesis methods. The results of other methods are taken from their [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]

**Figure 6.** Figure 6: Qualitative comparison of geometry quality. EG3D and our method generate results with more detailed geometry than [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

**Figure 7.** Figure 7: Interpolation in the latent space produces smooth transitions, with each intermediate result maintaining 3D-consistency. [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]

**Figure 8.** Figure 8: 3D-consistent image synthesis. The images synthesized [PITH_FULL_IMAGE:figures/full_fig_p009_8.png]

**Figure 9.** Figure 9: High-resolution (1024 × 1024) portrait synthesis. Our method produces high-quality portraits with rich details. REFERENCES [1] B. Mildenhall, P. P. Srinivasan, M. Tancik, J. T. Barron, R. Ramamoorthi, and R. Ng, “Nerf: Representing scenes as neural radiance fields for …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

46 extracted references · 30 canonical work pages

[1]

Nerf: Representing scenes as neural radiance fields for view synthesis,

B. Mildenhall, P. P. Srinivasan, M. Tancik, J. T. Barron, R. Ramamoorthi, and R. Ng, “Nerf: Representing scenes as neural radiance fields for view synthesis,” Communications of the ACM , vol. 65, no. 1, pp. 99–106, 2021

2021
[2]

Mip-nerf: A multiscale representation for anti- aliasing neural radiance fields,

J. T. Barron, B. Mildenhall, M. Tancik, P. Hedman, R. Martin-Brualla, and P. P. Srinivasan, “Mip-nerf: A multiscale representation for anti- aliasing neural radiance fields,” in Proceedings of the IEEE/CVF inter- national conference on computer vision , 2021, pp. 5855–5864

work page 2021
[3]

Mip-nerf 360: Unbounded anti-aliased neural radiance fields,

J. T. Barron, B. Mildenhall, D. Verbin, P. P. Srinivasan, and P. Hedman, “Mip-nerf 360: Unbounded anti-aliased neural radiance fields,” in Pro- ceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 5470–5479

work page 2022
[4]

Tensorf: Tensorial radiance fields,

A. Chen, Z. Xu, A. Geiger, J. Yu, and H. Su, “Tensorf: Tensorial radiance fields,” in European conference on computer vision . Springer, 2022, pp. 333–350

2022
[5]

3D-SSGAN: Lifting 2D Semantics for 3D-Aware Compositional Portrait Synthesis

R. Liu, P. Zheng, Y . Wang, and R. Ma, “3d-ssgan: Lifting 2d se- mantics for 3d-aware compositional portrait synthesis,” arXiv preprint arXiv:2401.03764, 2024

work page Pith review arXiv 2024
[6]

Semantic 3D-aware Portrait Synthesis and Manipulation Based on Compositional Neural Radiance Field

T. Ma, B. Li, Q. He, J. Dong, and T. Tan, “Semantic 3d-aware portrait synthesis and manipulation based on compositional neural radiance field,” arXiv preprint arXiv:2302.01579 , 2023

work page Pith review arXiv 2023
[7]

AttriHuman-3D: Editable 3D Human Avatar Generation with Attribute Decomposition and Indexing

F. Yang, T. Chen, X. He, Z. Cai, L. Yang, S. Wu, and G. Lin, “Attrihuman-3d: Editable 3d human avatar generation with attribute decomposition and indexing,” arXiv preprint arXiv:2312.02209 , 2023

work page Pith review arXiv 2023
[8]

3d-aware conditional image synthesis,

K. Deng, G. Yang, D. Ramanan, and J.-Y . Zhu, “3d-aware conditional image synthesis,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2023, pp. 4434–4445

work page 2023

Show all 46 references

[9]

Rodin: A generative model for sculpting 3d digital avatars using diffusion,

T. Wang, B. Zhang, T. Zhang, S. Gu, J. Bao, T. Baltrusaitis, J. Shen, D. Chen, F. Wen, Q. Chen et al. , “Rodin: A generative model for sculpting 3d digital avatars using diffusion,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2023, pp....

2023
[10]

Gram-hd: 3d-consistent image generation at high resolution with generative radiance manifolds,

J. Xiang, J. Yang, Y . Deng, and X. Tong, “Gram-hd: 3d-consistent image generation at high resolution with generative radiance manifolds,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 2195–2205

2023
[11]

Semantichuman-hd: High- resolution semantic disentangled 3d human generation,

P. Zheng, T. Liu, Z. Yi, and R. Ma, “Semantichuman-hd: High- resolution semantic disentangled 3d human generation,” arXiv preprint arXiv:2403.10166, 2024

2024 arXiv
[12]

Generative adversarial nets,

I. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, and Y . Bengio, “Generative adversarial nets,” Advances in neural information processing systems , vol. 27, 2014

2014
[13]

A style-based generator architecture for generative adversarial networks,

T. Karras, S. Laine, and T. Aila, “A style-based generator architecture for generative adversarial networks,” in Proceedings of the IEEE/CVF JOURNAL OF LATEX CLASS FILES, XXX 11 conference on computer vision and pattern recognition, 2019, pp. 4401– 4410

2019
[14]

Analyzing and improving the image quality of stylegan,

T. Karras, S. Laine, M. Aittala, J. Hellsten, J. Lehtinen, and T. Aila, “Analyzing and improving the image quality of stylegan,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 8110–8119

2020
[15]

Semanticstylegan: Learning compositional generative priors for controllable image synthesis and editing,

Y . Shi, X. Yang, Y . Wan, and X. Shen, “Semanticstylegan: Learning compositional generative priors for controllable image synthesis and editing,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2022, pp. 11 254–11 264

2022
[16]

Denoising diffusion probabilistic models,

J. Ho, A. Jain, and P. Abbeel, “Denoising diffusion probabilistic models,” Advances in neural information processing systems , vol. 33, pp. 6840– 6851, 2020

2020
[17]

Instructpix2pix: Learning to follow image editing instructions,

T. Brooks, A. Holynski, and A. A. Efros, “Instructpix2pix: Learning to follow image editing instructions,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2023, pp. 18 392–18 402

2023
[18]

High- resolution image synthesis with latent diffusion models,

R. Rombach, A. Blattmann, D. Lorenz, P. Esser, and B. Ommer, “High- resolution image synthesis with latent diffusion models,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 10 684–10 695

2022
[19]

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

N. Ruiz, Y . Li, V . Jampani, Y . Pritch, M. Rubinstein, and K. Aberman, “Dreambooth: Fine tuning text-to-image diffusion models for subject- driven generation,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2023, pp. 22 500–22 510

2023
[20]

Stylerig: Rigging stylegan for 3d control over portrait images,

A. Tewari, M. Elgharib, G. Bharaj, F. Bernard, H.-P. Seidel, P. P ´erez, M. Zollhofer, and C. Theobalt, “Stylerig: Rigging stylegan for 3d control over portrait images,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2020, pp. 6142–6151

2020
[21]

Style and pose control for image synthesis of humans from a single monocular view,

K. Sarkar, V . Golyanik, L. Liu, and C. Theobalt, “Style and pose control for image synthesis of humans from a single monocular view,” arXiv preprint arXiv:2102.11263, 2021

2021 arXiv
[22]

Styleflow: Attribute- conditioned exploration of stylegan-generated images using conditional continuous normalizing flows,

R. Abdal, P. Zhu, N. J. Mitra, and P. Wonka, “Styleflow: Attribute- conditioned exploration of stylegan-generated images using conditional continuous normalizing flows,” ACM Transactions on Graphics (ToG) , vol. 40, no. 3, pp. 1–21, 2021

2021
[23]

Keep it smpl: Automatic estimation of 3d human pose and shape from a single image,

F. Bogo, A. Kanazawa, C. Lassner, P. Gehler, J. Romero, and M. J. Black, “Keep it smpl: Automatic estimation of 3d human pose and shape from a single image,” in Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, P...

2016
[24]

3d mor- phable face models—past, present, and future,

B. Egger, W. A. Smith, A. Tewari, S. Wuhrer, M. Zollhoefer, T. Beeler, F. Bernard, T. Bolkart, A. Kortylewski, S. Romdhani et al. , “3d mor- phable face models—past, present, and future,” ACM Transactions on Graphics (ToG), vol. 39, no. 5, pp. 1–38, 2020

2020
[25]

Graf: Generative radiance fields for 3d-aware image synthesis,

K. Schwarz, Y . Liao, M. Niemeyer, and A. Geiger, “Graf: Generative radiance fields for 3d-aware image synthesis,” Advances in Neural Information Processing Systems , vol. 33, pp. 20 154–20 166, 2020

2020
[26]

Stylesdf: High-resolution 3d-consistent image and geometry generation,

R. Or-El, X. Luo, M. Shan, E. Shechtman, J. J. Park, and I. Kemelmacher-Shlizerman, “Stylesdf: High-resolution 3d-consistent image and geometry generation,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2022, pp. 13 503–13 513

2022
[27]

Stylenerf: A style-based 3d-aware generator for high-resolution image synthesis,

J. Gu, L. Liu, P. Wang, and C. Theobalt, “Stylenerf: A style-based 3d-aware generator for high-resolution image synthesis,” arXiv preprint arXiv:2110.08985, 2021

2021 arXiv
[28]

3d-aware image synthe- sis via learning structural and textural representations,

Y . Xu, S. Peng, C. Yang, Y . Shen, and B. Zhou, “3d-aware image synthe- sis via learning structural and textural representations,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2022, pp. 18 430–18 439

2022
[29]

Efficient geometry-aware 3d generative adversarial networks,

E. R. Chan, C. Z. Lin, M. A. Chan, K. Nagano, B. Pan, S. De Mello, O. Gallo, L. J. Guibas, J. Tremblay, S. Khamis et al. , “Efficient geometry-aware 3d generative adversarial networks,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2022...

2022
[30]

Ag3d: Learning to generate 3d avatars from 2d image collections,

Z. Dong, X. Chen, J. Yang, M. J. Black, O. Hilliges, and A. Geiger, “Ag3d: Learning to generate 3d avatars from 2d image collections,” in Proceedings of the IEEE/CVF international conference on computer vision, 2023, pp. 14 916–14 927

2023
[31]

Xagen: 3d expressive human avatars generation,

Z. Xu, J. Zhang, J. H. Liew, J. Feng, and M. Z. Shou, “Xagen: 3d expressive human avatars generation,” Advances in Neural Information Processing Systems, vol. 36, 2024

2024
[32]

Urban- giraffe: Representing urban scenes as compositional generative neural feature fields,

Y . Yang, Y . Yang, H. Guo, R. Xiong, Y . Wang, and Y . Liao, “Urban- giraffe: Representing urban scenes as compositional generative neural feature fields,” in Proceedings of the IEEE/CVF International Confer- ence on Computer Vision , 2023, pp. 9199–9210

2023
[33]

Diffusiongan3d: Boosting text-guided 3d generation and domain adaptation by combining 3d gans and diffusion priors,

B. Lei, K. Yu, M. Feng, M. Cui, and X. Xie, “Diffusiongan3d: Boosting text-guided 3d generation and domain adaptation by combining 3d gans and diffusion priors,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp. 10 487–10 497

2024
[34]

Panohead: Geometry-aware 3d full-head synthesis in 360deg,

S. An, H. Xu, Y . Shi, G. Song, U. Y . Ogras, and L. Luo, “Panohead: Geometry-aware 3d full-head synthesis in 360deg,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2023, pp. 20 950–20 959

2023
[35]

Ide-3d: Interactive disentangled editing for high-resolution 3d-aware portrait synthesis,

J. Sun, X. Wang, Y . Shi, L. Wang, J. Wang, and Y . Liu, “Ide-3d: Interactive disentangled editing for high-resolution 3d-aware portrait synthesis,” ACM Transactions on Graphics (ToG) , vol. 41, no. 6, pp. 1–10, 2022

2022
[36]

Nerffaceediting: Disentangled face editing in neural radiance fields,

K. Jiang, S.-Y . Chen, F.-L. Liu, H. Fu, and L. Gao, “Nerffaceediting: Disentangled face editing in neural radiance fields,” in SIGGRAPH Asia 2022 Conference Papers, 2022, pp. 1–9

2022
[37]

Gram: Generative radi- ance manifolds for 3d-aware image generation,

Y . Deng, J. Yang, J. Xiang, and X. Tong, “Gram: Generative radi- ance manifolds for 3d-aware image generation,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 10 673–10 683

2022
[38]

Eva3d: Composi- tional 3d human generation from 2d image collections,

F. Hong, Z. Chen, Y . Lan, L. Pan, and Z. Liu, “Eva3d: Composi- tional 3d human generation from 2d image collections,” arXiv preprint arXiv:2210.04888, 2022

2022 arXiv
[39]

Gaussian shell maps for efficient 3d human generation,

R. Abdal, W. Yifan, Z. Shi, Y . Xu, R. Po, Z. Kuang, Q. Chen, D.-Y . Yeung, and G. Wetzstein, “Gaussian shell maps for efficient 3d human generation,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp. 9441–9451

2024
[40]

Veri3d: Generative vertex-based radiance fields for 3d controllable human image synthesis,

X. Chen, J. Huang, Y . Bin, L. Yu, and Y . Liao, “Veri3d: Generative vertex-based radiance fields for 3d controllable human image synthesis,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 8986–8997

2023
[41]

Bilateral normal integration,

X. Cao, H. Santo, B. Shi, F. Okura, and Y . Matsushita, “Bilateral normal integration,” in European Conference on Computer Vision . Springer, 2022, pp. 552–567

2022
[42]

Stargan v2: Diverse image synthesis for multiple domains,

Y . Choi, Y . Uh, J. Yoo, and J.-W. Ha, “Stargan v2: Diverse image synthesis for multiple domains,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 8188– 8197

2020
[43]

Deepfashion: Power- ing robust clothes recognition and retrieval with rich annotations,

Z. Liu, P. Luo, S. Qiu, X. Wang, and X. Tang, “Deepfashion: Power- ing robust clothes recognition and retrieval with rich annotations,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 1096–1104

2016
[44]

Gans trained by a two time-scale update rule converge to a local nash equilibrium,

M. Heusel, H. Ramsauer, T. Unterthiner, B. Nessler, and S. Hochreiter, “Gans trained by a two time-scale update rule converge to a local nash equilibrium,” Advances in neural information processing systems , vol. 30, 2017

2017
[45]

Demysti- fying mmd gans,

M. Bi ´nkowski, D. J. Sutherland, M. Arbel, and A. Gretton, “Demysti- fying mmd gans,” arXiv preprint arXiv:1801.01401 , 2018

2018 arXiv
[46]

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

P. Wang, L. Liu, Y . Liu, C. Theobalt, T. Komura, and W. Wang, “Neus: Learning neural implicit surfaces by volume rendering for multi-view reconstruction,” arXiv preprint arXiv:2106.10689 , 2021

2021 arXiv

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.

[1] [1]

Nerf: Representing scenes as neural radiance fields for view synthesis,

B. Mildenhall, P. P. Srinivasan, M. Tancik, J. T. Barron, R. Ramamoorthi, and R. Ng, “Nerf: Representing scenes as neural radiance fields for view synthesis,” Communications of the ACM , vol. 65, no. 1, pp. 99–106, 2021

2021

[2] [2]

Mip-nerf: A multiscale representation for anti- aliasing neural radiance fields,

J. T. Barron, B. Mildenhall, M. Tancik, P. Hedman, R. Martin-Brualla, and P. P. Srinivasan, “Mip-nerf: A multiscale representation for anti- aliasing neural radiance fields,” in Proceedings of the IEEE/CVF inter- national conference on computer vision , 2021, pp. 5855–5864

work page 2021

[3] [3]

Mip-nerf 360: Unbounded anti-aliased neural radiance fields,

J. T. Barron, B. Mildenhall, D. Verbin, P. P. Srinivasan, and P. Hedman, “Mip-nerf 360: Unbounded anti-aliased neural radiance fields,” in Pro- ceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 5470–5479

work page 2022

[4] [4]

Tensorf: Tensorial radiance fields,

A. Chen, Z. Xu, A. Geiger, J. Yu, and H. Su, “Tensorf: Tensorial radiance fields,” in European conference on computer vision . Springer, 2022, pp. 333–350

2022

[5] [5]

3D-SSGAN: Lifting 2D Semantics for 3D-Aware Compositional Portrait Synthesis

R. Liu, P. Zheng, Y . Wang, and R. Ma, “3d-ssgan: Lifting 2d se- mantics for 3d-aware compositional portrait synthesis,” arXiv preprint arXiv:2401.03764, 2024

work page Pith review arXiv 2024

[6] [6]

Semantic 3D-aware Portrait Synthesis and Manipulation Based on Compositional Neural Radiance Field

T. Ma, B. Li, Q. He, J. Dong, and T. Tan, “Semantic 3d-aware portrait synthesis and manipulation based on compositional neural radiance field,” arXiv preprint arXiv:2302.01579 , 2023

work page Pith review arXiv 2023

[7] [7]

AttriHuman-3D: Editable 3D Human Avatar Generation with Attribute Decomposition and Indexing

F. Yang, T. Chen, X. He, Z. Cai, L. Yang, S. Wu, and G. Lin, “Attrihuman-3d: Editable 3d human avatar generation with attribute decomposition and indexing,” arXiv preprint arXiv:2312.02209 , 2023

work page Pith review arXiv 2023

[8] [8]

3d-aware conditional image synthesis,

K. Deng, G. Yang, D. Ramanan, and J.-Y . Zhu, “3d-aware conditional image synthesis,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2023, pp. 4434–4445

work page 2023