Pith. sign in

REVIEW 4 major objections 5 minor 44 references

UVRM: A Scalable 3D Reconstruction Model from Unposed Videos

T0 review · 4 major / 5 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 UVRM trains a feed-forward 3D reconstruction model on monocular videos without any pose annotations, using a transformer to build pose-invariant latent features and a diffusion-based score distillation and pseudo-view augmentation…

desk verdict A genuinely new pose-free training pipeline for 3D reconstruction, worth a serious referee, but the central SDS property is asserted without proof and the experiments are still proof-of-concept scale. read the letter →

arxiv 2501.09347 v2 pith:VDOG3NVJ submitted 2025-01-16 cs.CV

Shiu-hong Kao , Xiao Li , Jinglu Wang , Yang Li , Chi-Keung Tang , Yu-Wing Tai , Yan Lu This is my paper · ORCID
classification cs.CV
keywords 3Dreconstructionpose-freetrainingmonocularvideotri-planerepresentationscoredistillationsamplinglargemodelneuralradiancefielddiffusion
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

The paper asks whether a 3D reconstruction model can be trained directly from monocular videos without any camera pose annotations, and its answer, with a system called UVRM, is yes. UVRM uses a transformer to compress an arbitrary number of video frames into a fixed set of pose-invariant latent tokens, which a style-based synthesizer decodes into a tri-plane neural radiance field. Training replaces pose-annotated render losses with score distillation sampling plus iteratively generated pseudo-views produced by a pretrained diffusion model, so no ground-truth camera poses are needed. Experiments on synthetic and real video collections show plausible object reconstructions, and the method is more robust than a per-object pose-free NeRF baseline.

What carries the argument

Three components carry the argument. The latent alignment encoder is a T5-style transformer with three learnable query tokens; it compresses an arbitrarily long sequence of VAE-encoded frames into a fixed three-token 3D latent, learning viewpoint invariance implicitly. The tri-plane synthesizer is a StyleGAN-style convolutional generator whose layers are modulated by the 3D latent through adaptive instance normalization, producing a three-axis-aligned feature grid that an MLP decodes to color and density for volumetric rendering. The training objective is the pose-free core: SDS gradients from a pretrained image-to-3D diffusion model with randomly sampled camera poses give weak supervision, while an iterative augmentation loop renders views from the current model, adds noise, lets the same diffusion model denoise them conditioned on a reference frame, and trains with MSE and LPIPS losses. The two signals are complementary, with SDS maintaining multi-view consistency and augmentation recovering high-frequency detail.

What would settle it

Train UVRM on a video whose true camera trajectory is deliberately far from the pose sampler in Eq. (8), such as all frames from a narrow frontal arc at fixed elevation, and check whether reconstruction collapses; or render a trained model's views, run a standard structure-from-motion pose estimator on them, and measure whether the estimated relative poses match the video's true relative poses up to a global affine transform.

Watch

Extended reading notes

Core claim

UVRM establishes that a feed-forward 3D reconstruction model can be trained and evaluated on monocular video without any pose information. The central claim is that a transformer, by attending over VAE-encoded frames, learns a pose-invariant latent representation of the object, and a style-based synthesizer turns this latent into a tri-plane field from which arbitrary novel views render. To supervise this without poses, the paper combines score distillation sampling, which matches the distribution of rendered views to the true view distribution up to a global affine transform, with an iterative analysis-by-synthesis step that uses the same diffusion model to turn rendered images into pseudo-views for pixel-wise losses. The result is a pose-free training pipeline that scales to multiple objects and produces stable 360-degree reconstructions where per-object pose-free NeRF methods fail.

Load-bearing premise

The load-bearing premise is that the score distillation loss, with its randomly sampled camera poses, still matches the true camera distribution closely enough to preserve relative geometry; the paper concedes this is only guaranteed when the sampled pose distribution equals the unknown ground-truth distribution, which is unlikely for real videos.

Editorial extensions

If this is right

  • Pose-free training removes the need for structure-from-motion or pose annotations, making large-scale 2D video corpora usable for 3D reconstruction training.
  • Reconstruction time amortizes across objects: average per-object cost decreases as the number of objects in a collection grows, since a single forward pass handles each object.
  • The SDS plus pseudo-view augmentation recipe supplies a self-supervised training objective that does not depend on ground-truth poses, depth, or calibrated multi-view data.
  • Starting from one reference frame, the same pose-free training objective can propagate a 2D edit into the reconstructed 3D representation.

Reading between the lines

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

  • If the distribution-matching property of SDS holds for general video, the pipeline could in principle train on arbitrary internet video; the paper's experiments are confined to small curated collections, so large-scale scaling is a testable next step the authors do not perform.
  • The global affine ambiguity in the SDS match means reconstructed objects live in an arbitrary canonical coordinate frame, and applications needing metric scale would require an extra calibration signal.
  • The iterative augmentation schedule, moving from diffusion-guided coarse views to photometric fine views, resembles a curriculum that could transfer to other neural representations such as Gaussian splatting.
  • A direct diagnostic - running a standard structure-from-motion pose estimator on UVRM's rendered views and comparing relative poses to ground truth - would show whether the model learns genuine relative geometry or only plausible appearance.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper introduces UVRM, a 3D reconstruction model that takes an arbitrary number of monocular video frames without camera poses and outputs a triplane NeRF representation. To avoid pose annotations during training, the method combines score distillation sampling (SDS) conditioned on a reference frame with an iterative pseudo-view augmentation procedure in which a pretrained diffusion model is used to denoise the model's own renderings, which then serve as pseudo ground-truth views for a pixel-wise loss. The model is a transformer that compresses per-frame VAE latents into three learnable tokens, which condition a style-based triplane synthesizer. Experiments are reported on 20 objects from G-Objaverse Food (plus a 128-object scaling study) and on real CO3D Hydrant videos, comparing against Nope-NeRF and Zero123-XL.

Significance. If the central claim holds, UVRM would be a meaningful step toward scaling 3D reconstruction training to large unposed video corpora, removing the pose-estimation bottleneck that currently constrains LRM training. The architectural idea of pose-invariant latent alignment and the two-stage weak/self-supervision pipeline are interesting and potentially influential. However, the evidence in the paper is thin: the experiments use only a handful of synthetic objects plus one real-world category, the quantitative tables report single aggregate numbers without variance, no pose-supervised oracle baseline is included, and the key theoretical property of the SDS loss is asserted rather than proved. The method's reliance on the same diffusion prior for both SDS supervision and pseudo-view generation raises a circularity concern that the current experiments do not resolve.

major comments (4)
  1. [Eq. (12), Sec. 3.3] The self-supervision term in Eq. (12) is written as MSE(g_Θ(p'_i;V), I^A_i) + λ·LPIPS(g_Θ(p'_i;V), I^A_i), where I^A_i is defined in Eq. (9) as (g_Θ(p'_i;V), p'_i). Thus the two arguments to MSE and LPIPS are identical, making those terms identically zero and the described pseudo-view supervision vacuous. The target should presumably be the diffusion-augmented view (I^A_i)' from Eq. (11), not the original rendering. This is a load-bearing error in the training objective as written; as it stands, Eq. (12) does not implement the analysis-by-synthesis augmentation described in the text.
  2. [Sec. 3.3, Eq. (7)] The paper asserts that Eq. (7) matches the distribution of rendered views to the ground-truth view distribution up to a global affine camera transformation, without proof. The text immediately concedes that a perfect match is only theoretically guaranteed when the sampled pose distribution equals the unknown true distribution, which the fixed orbit sampler of Eq. (8) cannot ensure for real videos. Since the SDS term is conditioned only on a single reference image and uses a fixed orbit, it does not directly enforce agreement with the unseen frames of the input video. This property is the load-bearing justification for pose-free training; the manuscript needs either a formal argument or a concrete empirical diagnostic (e.g., comparing relative pose distributions of rendered views with those of the input video) to establish that the supervision is not merely making individual views plausible under the reference-conditioned diffusion prior.
  3. [Sec. 3.3, Sec. 4] The pseudoview augmentation (Eqs. 9–11) uses the same pretrained diffusion model D_φ that provides the SDS gradients in Eq. (7). Consequently, the pseudo ground-truth views are generated by the same prior that supervises the reconstruction, so the training objective may largely distill the diffusion prior into a consistent 3D shape rather than reconstructing the specific object in the video. Table 1 and Table A1 compare against Zero123-XL and Nope-NeRF, but they do not include a pose-supervised UVRM upper bound, nor do the real-data experiments report quantitative metrics on held-out CO3D frames. Without such comparisons, the results do not disentangle genuine reconstruction of the input video from prior distillation.
  4. [Sec. 4.2, Sec. 4.4, Table A1] The experiments are reported at a very small scale: Table 1 uses 20 objects only, the main real-data evaluation is qualitative on CO3D Hydrant, and Table A1 gives a Nope-NeRF success rate of 22% without defining the success criterion or clarifying how failures are incorporated into the reported PSNR/SSIM. Additionally, the Discussion in Sec. 3.3 states that UVRM trained with camera poses converges faster and achieves higher quality, but no experiment in the paper substantiates this claim. The authors should add a pose-supervised oracle baseline, report variance across objects, describe the success criterion, and provide quantitative real-video results.
minor comments (5)
  1. [Eq. (10)] The notation in Eq. (10) is circular: I^A_{i,t} appears on both sides. It should read I^A_{i,t} = √α_t I^A_i + √(1−α_t)ε, or similar, to denote the noisy version of the rendered image.
  2. [Sec. 3.1] There are several typos: 'loss fuction' in Sec. 3.1, 'we replacing' in Sec. 3.3, and 'Consequentially' in Sec. 3.3. Also 'V AE' in the Fig. 2 caption should be 'VAE'.
  3. [References] References [22] and [23] both cite Raffel et al. (T5) and appear to be the same work; one should be removed or the second should be a distinct citation.
  4. [Sec. 5, Appendix A.1] The Limitations paragraph states that the method has not yet been scaled to a large reconstruction model, yet the title and abstract emphasize 'Scalable.' The scaling claim is only demonstrated up to 128 objects, so the authors should temper the language or report results at a larger scale.
  5. [Table 2] The runtime comparison would benefit from reporting the number of training steps and the resolution/gpu type uniform across methods; currently the single-object UVRM time (990 min/obj) is slower than Nope-NeRF (380 min/obj), which is fine but could be explained more clearly.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the SDS and pseudo-view losses are anchored by an externally pretrained diffusion model and real reference frames; the unproved distribution-matching property is a correctness risk, not a circular reduction.

full rationale

UVRM's claimed derivation chain does not reduce to its inputs by construction. The pose-free training objective in Eq. (7) is a standard SDS term: gradients come from a fixed, externally pretrained diffusion model (Zero123-XL), conditioned on a real reference frame from the input video, and rendered at random poses from Eq. (8). The pseudo-view augmentation in Eqs. (9)-(12) generates targets by denoising the model's own renders with the same fixed diffusion model; this is self-referential self-distillation, but the diffusion weights are external and the reference frame and (through the transformer encoder) the other video frames are real inputs, so the prediction is not definitionally equal to its inputs. The paper does assert, without proof, in Sec. 3.3 that SDS preserves relative camera poses up to a global affine when the sampled pose distribution matches the unknown true distribution, and it concedes this is 'hardly the case when handling real-world videos'; that is a load-bearing correctness risk, not a circular reduction. No fitted parameter is renamed as a prediction, no uniqueness theorem is imported from the authors' prior work, and no known result is merely renamed. The quantitative comparisons are against external baselines (Zero123-XL, Nope-NeRF) using ground-truth poses only for evaluation. One textual ambiguity: Eq. (12) as printed compares renders to the model's own renders from Eq. (9) (no prime), which would be identically zero; the surrounding text and ablations indicate the augmented views (I^A_i)' are the intended targets, and even so the supervision is self-distillation through a fixed external diffusion prior, not an equivalence of output and input.

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

The method rests on several hand-chosen hyperparameters (k0, theta, beta, lambda, noise schedule) and on assumptions about the pretrained components. The most consequential assumptions are that the frozen VAE preserves enough information, that Zero123-XL provides a view-consistent prior, and that the random orbit camera distribution combined with SDS preserves relative camera poses. The latter two are asserted rather than derived, and the paper's own appendix observes view inconsistency in Zero123-XL. No new entities are introduced.

free parameters (5)
  • k0 (initial pseudo-view count) = 6, increasing by 5 per augmentation iteration
    Chosen by hand; ablation in Fig. A5 shows sensitivity and recommends 3-8, indicating no principled selection.
  • SDS camera sampling angle theta = pi/18 for SDS, 0 for augmentation
    Hand-chosen; the orbit camera distribution is central to the pose-free SDS property.
  • SDS loss weight beta = linearly increased from 1 to 25000
    Hand-chosen schedule balancing SDS and reconstruction losses; no sensitivity analysis.
  • LPIPS loss weight lambda = 1
    Hand-chosen; no ablation is reported for this value.
  • Denoising timestep schedule (t_max and floor 0.2) = t = max(1 - s_curr/s_total, 0.2) * t_max
    Manual schedule for pseudo-view noise level; not derived from a principle.
assumptions (5)
  • standard math Triplane NeRF with volumetric rendering can represent a 3D object and be optimized with image losses.
    Adopted from Chan et al. 2022 and prior NeRF works; standard in the field.
  • domain assumption The Stable Diffusion VAE encoder produces a latent space suitable for compressing video frames for 3D reconstruction.
    Used as a frozen image encoder in Sec. 3.2; no analysis of what information is lost.
  • domain assumption Pretrained Zero123-XL provides a view-consistent, object-consistent diffusion prior when conditioned on a single reference frame.
    Core to both the SDS loss and pseudo-view generation in Sec. 3.3; the paper itself notes view inconsistency at larger k in App. B.5.
  • ad hoc to paper The random orbit camera distribution in Eq. (8), combined with SDS, preserves relative camera poses between input views up to a global affine transformation.
    Asserted in Sec. 3.3 without proof; if false, the weak-supervision signal may not correspond to correct multi-view geometry.
  • ad hoc to paper Intermediate UVRM renderings during training can be interpreted as partially denoised samples, so applying the diffusion model to them produces valid pseudo-views.
    This analysis-by-synthesis argument in Sec. 3.3 is a heuristic that underlies Eqs. (9)-(12).

how reviews work

0 comments
Cite this review

Pith. "Pith review of UVRM: A Scalable 3D Reconstruction Model from Unposed Videos." pith.science (2026). https://pith.science/paper/VDOG3NVJ

@misc{pith2026250109347,
  author       = {Pith},
  title        = {Pith review of: UVRM: A Scalable 3D Reconstruction Model from Unposed Videos},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VDOG3NVJ}},
  note         = {Machine review of arXiv:2501.09347}
}
read the original abstract

Large Reconstruction Models (LRMs) have recently become a popular method for creating 3D foundational models. Training 3D reconstruction models with 2D visual data traditionally requires prior knowledge of camera poses for the training samples, a process that is both time-consuming and prone to errors. Consequently, 3D reconstruction training has been confined to either synthetic 3D datasets or small-scale datasets with annotated poses. In this study, we investigate the feasibility of 3D reconstruction using unposed video data of various objects. We introduce UVRM, a novel 3D reconstruction model capable of being trained and evaluated on monocular videos without requiring any information about the pose. UVRM uses a transformer network to implicitly aggregate video frames into a pose-invariant latent feature space, which is then decoded into a tri-plane 3D representation. To obviate the need for ground-truth pose annotations during training, UVRM employs a combination of the score distillation sampling (SDS) method and an analysis-by-synthesis approach, progressively synthesizing pseudo novel-views using a pre-trained diffusion model. We qualitatively and quantitatively evaluate UVRM's performance on the G-Objaverse and CO3D datasets without relying on pose information. Extensive experiments show that UVRM is capable of effectively and efficiently reconstructing a wide range of 3D objects from unposed videos.

Figures

Figures reproduced from arXiv: 2501.09347 by the authors.

Figure 1
Figure 1. Different from previous methods, which either focus [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. UVRM architecture. We propose UVRM, a transformer-based reconstruction model for pose-free monocular video inputs. It first encodes each input view into latent space with a VAE encoder [14]. Next, it adopts a T5-based transformer encoder [22] to extract a pose-invariant feature by implicitly aligning the image latent sequence. The extracted feature are then used modulate a style-based [12] synthesizer to output a tr… view at source ↗
Figure 3
Figure 3. Illustration of our pose-free training framework, where [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Iterative augmentation pipeline. We iteratively alter￾nate between (left, green) weakly supervise training of the UVRM model with score distillation sampling (SDS) on the current set of reference view frames, and (right, orange) generating new set of novel pseudo-views…
Figure 6
Figure 6. Figure 6: Comparison of pose-free training on single object. We compare UVRM with Nope-NeRF [2], the state of the art for NeRF reconstruction with unknown camera poses. In general, UVRM supports more stable reconstruction for 360-degree and sparse views, while Nope-NeRF is shown…
Figure 7
Figure 7. Figure 7: Ablation study. Comparison between the full UVRM, the UVRM without certain components, and Zero123-XL [17]. Specifi￾cally, Zero123-XL synthesizes high-frequency image but lacks view consistency. 4.2. Ablation Study Pose-free Alignment. We first validate the capability …
Figure 8
Figure 8. Figure 8: Results of UVRM with pose-free training on video collections from the G-Objeverse dataset. The first three rows contains subjects that are also in previous experiments ( [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 9
Figure 9. Figure 9: Multi-object results on real-world data, i.e. CO3D Hydrant [24]. We show that UVRM can not only be trained on multiple objects, but it can also tackle complex objects with non￾symmetry and varying shadow. but this time using our proposed pose-free training pipeline in …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

44 extracted references · 16 canonical work pages

  1. [1]

    Score distillation sampling with learned manifold cor- rective

    Thiemo Alldieck, Nikos Kolotouros, and Cristian Sminchis- escu. Score distillation sampling with learned manifold cor- rective. arXiv preprint arXiv:2401.05293, 2024. 3

    arXiv 2024

  2. [2]

    Nope-nerf: Optimising neu- ral radiance field with no pose prior

    Wenjing Bian, Zirui Wang, Kejie Li, Jia-Wang Bian, and Victor Adrian Prisacariu. Nope-nerf: Optimising neu- ral radiance field with no pose prior. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 4160–4169, 2023. 2, 6, 7, 8, 11

    work page 2023

  3. [3]

    Efficient geometry-aware 3d generative adversarial networks

    Eric R Chan, Connor Z Lin, Matthew A Chan, Koki Nagano, Boxiao Pan, Shalini De Mello, Orazio Gallo, Leonidas J Guibas, Jonathan Tremblay, Sameh Khamis, et al. Efficient geometry-aware 3d generative adversarial networks. In Pro- ceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 16123–16133, 2022. 2, 3, 4

    work page 2022

  4. [4]

    pixelsplat: 3d gaussian splats from image pairs for scalable generalizable 3d reconstruction

    David Charatan, Sizhe Li, Andrea Tagliasacchi, and Vincent Sitzmann. pixelsplat: 3d gaussian splats from image pairs for scalable generalizable 3d reconstruction. In arXiv, 2023. 2

    work page 2023

  5. [5]

    Explicit correspondence matching for generalizable neural radiance fields

    Yuedong Chen, Haofei Xu, Qianyi Wu, Chuanxia Zheng, Tat-Jen Cham, and Jianfei Cai. Explicit correspondence matching for generalizable neural radiance fields. arXiv preprint arXiv:2304.12294, 2023

    arXiv 2023

  6. [6]

    Mvsplat: Efficient 3d gaussian splatting from sparse multi-view images

    Yuedong Chen, Haofei Xu, Chuanxia Zheng, Bohan Zhuang, Marc Pollefeys, Andreas Geiger, Tat-Jen Cham, and Jianfei Cai. Mvsplat: Efficient 3d gaussian splatting from sparse multi-view images. arXiv preprint arXiv:2403.14627, 2024. 2

    arXiv 2024

  7. [7]

    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. 1, 6

    arXiv 2022

  8. [8]

    Instantsplat: Unbounded sparse-view pose-free gaus- sian splatting in 40 seconds, 2024

    Zhiwen Fan, Wenyan Cong, Kairun Wen, Kevin Wang, Jian Zhang, Xinghao Ding, Danfei Xu, Boris Ivanovic, Marco Pavone, Georgios Pavlakos, Zhangyang Wang, and Yue Wang. Instantsplat: Unbounded sparse-view pose-free gaus- sian splatting in 40 seconds, 2024. 2

    2024

Show all 44 references
  1. [9]

    Efros, and Xiaolong Wang

    Yang Fu, Sifei Liu, Amey Kulkarni, Jan Kautz, Alexei A. Efros, and Xiaolong Wang. Colmap-free 3d gaussian splat- ting. arXiv preprint arXiv:2312.07504, 2023. 2

    2023 arXiv

  2. [10]

    Get3d: A generative model of high quality 3d tex- tured shapes learned from images

    Jun Gao, Tianchang Shen, Zian Wang, Wenzheng Chen, Kangxue Yin, Daiqing Li, Or Litany, Zan Gojcic, and Sanja Fidler. Get3d: A generative model of high quality 3d tex- tured shapes learned from images. Advances In Neural In- formation Processing Systems, 35:31841–31854, 2022. 3

    2022

  3. [11]

    Lrm: Large reconstruction model for single image to 3d

    Yicong Hong, Kai Zhang, Jiuxiang Gu, Sai Bi, Yang Zhou, Difan Liu, Feng Liu, Kalyan Sunkavalli, Trung Bui, and Hao Tan. Lrm: Large reconstruction model for single image to 3d. arXiv preprint arXiv:2311.04400, 2023. 2

    2023 arXiv

  4. [12]

    A style-based generator architecture for generative adversarial networks

    Tero Karras, Samuli Laine, and Timo Aila. A style-based generator architecture for generative adversarial networks. In Proceedings of the IEEE/CVF conference on computer vi- sion and pattern recognition, pages 4401–4410, 2019. 3, 4, 6

    2019

  5. [13]

    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 Trans. Graph., 42(4):139–1,

  6. [14]

    Auto-encoding variational bayes

    Diederik P Kingma. Auto-encoding variational bayes. arXiv preprint arXiv:1312.6114, 2013. 3

    2013 arXiv

  7. [15]

    Adam: A method for stochastic opti- mization

    Diederik P Kingma. Adam: A method for stochastic opti- mization. arXiv preprint arXiv:1412.6980, 2014. 6

    2014 arXiv

  8. [16]

    M-lrm: Multi-view large reconstruction model

    Mengfei Li, Xiaoxiao Long, Yixun Liang, Weiyu Li, Yuan Liu, Peng Li, Xiaowei Chi, Xingqun Qi, Wei Xue, Wenhan Luo, et al. M-lrm: Multi-view large reconstruction model. arXiv preprint arXiv:2406.07648, 2024. 2

    2024 arXiv

  9. [17]

    Zero-1-to-3: Zero-shot one image to 3d object, 2023

    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, 2023. 4, 6, 7, 8, 12

    2023

  10. [18]

    Progressively optimized local radiance fields for robust view synthesis

    Andreas Meuleman, Yu-Lun Liu, Chen Gao, Jia-Bin Huang, Changil Kim, Min H Kim, and Johannes Kopf. Progressively optimized local radiance fields for robust view synthesis. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition, pages 16539–16548, 2023. 2

    2023

  11. [19]

    Nerf: Representing scenes as neural radiance fields for view syn- thesis

    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. Communications of the ACM, 65(1):99–106, 2021. 1, 2, 3

    2021

  12. [20]

    Gta: A geometry-aware attention mechanism for multi-view transformers

    Takeru Miyato, Bernhard Jaeger, Max Welling, and Andreas Geiger. Gta: A geometry-aware attention mechanism for multi-view transformers. arXiv preprint arXiv:2310.10375,

    arXiv

  13. [21]

    Dreamfusion: Text-to-3d using 2d diffusion

    Ben Poole, Ajay Jain, Jonathan T Barron, and Ben Milden- hall. Dreamfusion: Text-to-3d using 2d diffusion. arXiv preprint arXiv:2209.14988, 2022. 3

    2022 arXiv

  14. [22]

    Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, and Peter J. Liu. Exploring the limits of transfer learning with a unified text-to-text transformer. Journal of Machine Learn- ing Research, 21(140):1–67, 2020. 3

    2020

  15. [23]

    Exploring the limits of transfer learning with a unified text-to-text transformer

    Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, and Peter J Liu. Exploring the limits of transfer learning with a unified text-to-text transformer. Journal of machine learning research, 21(140):1–67, 2020. 6

    2020

  16. [24]

    Com- mon objects in 3d: Large-scale learning and evaluation of real-life 3d category reconstruction

    Jeremy Reizenstein, Roman Shapovalov, Philipp Henzler, Luca Sbordone, Patrick Labatut, and David Novotny. Com- mon objects in 3d: Large-scale learning and evaluation of real-life 3d category reconstruction. In International Con- ference on Computer Vision, 2021. 2, 6, 8

    2021

  17. [25]

    High-resolution image syn- thesis with latent diffusion models, 2021

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj¨orn Ommer. High-resolution image syn- thesis with latent diffusion models, 2021. 4

    2021

  18. [26]

    High-resolution image synthesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj ¨orn Ommer. High-resolution image synthesis with latent diffusion models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 10684–10695, 2022. 6

    2022

  19. [27]

    Structure-from-motion revisited

    Johannes Lutz Sch ¨onberger and Jan-Michael Frahm. Structure-from-motion revisited. In Conference on Com- puter Vision and Pattern Recognition (CVPR), 2016. 5 9

    2016

  20. [28]

    Structure- from-motion revisited

    Johannes L Schonberger and Jan-Michael Frahm. Structure- from-motion revisited. In Proceedings of the IEEE con- ference on computer vision and pattern recognition , pages 4104–4113, 2016. 2

    2016

  21. [29]

    Correspondence-guided sfm-free 3d gaussian splatting for nvs

    Wei Sun, Xiaosong Zhang, Fang Wan, Yanzhao Zhou, Yuan Li, Qixiang Ye, and Jianbin Jiao. Correspondence-guided sfm-free 3d gaussian splatting for nvs. arXiv preprint arXiv:2408.08723, 2024. 2

    2024 arXiv

  22. [30]

    Dreamgaussian: Generative gaussian splatting for effi- cient 3d content creation

    Jiaxiang Tang, Jiawei Ren, Hang Zhou, Ziwei Liu, and Gang Zeng. Dreamgaussian: Generative gaussian splatting for effi- cient 3d content creation. arXiv preprint arXiv:2309.16653,

    arXiv

  23. [31]

    Attention is all you need

    A Vaswani. Attention is all you need. Advances in Neural Information Processing Systems, 2017. 4

    2017

  24. [32]

    NeuS: Learning neural im- plicit surfaces by volume rendering for multi-view recon- struction

    Peng Wang, Lingjie Liu, Yuan Liu, Christian Theobalt, Taku Komura, and Wenping Wang. NeuS: Learning neural im- plicit surfaces by volume rendering for multi-view recon- struction. In NeurIPS, 2021. 2

    2021

  25. [33]

    Pf-lrm: Pose-free large reconstruction model for joint pose and shape prediction

    Peng Wang, Hao Tan, Sai Bi, Yinghao Xu, Fujun Luan, Kalyan Sunkavalli, Wenping Wang, Zexiang Xu, and Kai Zhang. Pf-lrm: Pose-free large reconstruction model for joint pose and shape prediction. arXiv preprint arXiv:2311.12024, 2023. 2

    2023 arXiv

  26. [34]

    Nerf–: Neural radiance fields without known camera parameters

    Zirui Wang, Shangzhe Wu, Weidi Xie, Min Chen, and Victor Adrian Prisacariu. Nerf–: Neural radiance fields without known camera parameters. arXiv preprint arXiv:2102.07064, 2021. 2

    2021 arXiv

  27. [35]

    latentsplat: Autoencoding variational gaussians for fast generalizable 3d reconstruction

    Christopher Wewer, Kevin Raj, Eddy Ilg, Bernt Schiele, and Jan Eric Lenssen. latentsplat: Autoencoding variational gaussians for fast generalizable 3d reconstruction. In arXiv,

  28. [36]

    Synsin: End-to-end view synthesis from a sin- gle image

    Olivia Wiles, Georgia Gkioxari, Richard Szeliski, and Justin Johnson. Synsin: End-to-end view synthesis from a sin- gle image. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 7467–7477,

  29. [37]

    Sinerf: Sinusoidal neural radiance fields for joint pose estimation and scene reconstruction

    Yitong Xia, Hao Tang, Radu Timofte, and Luc Van Gool. Sinerf: Sinusoidal neural radiance fields for joint pose estimation and scene reconstruction. arXiv preprint arXiv:2210.04553, 2022. 2

    2022 arXiv

  30. [38]

    Murf: Multi-baseline radiance fields

    Haofei Xu, Anpei Chen, Yuedong Chen, Christos Sakaridis, Yulun Zhang, Marc Pollefeys, Andreas Geiger, and Fisher Yu. Murf: Multi-baseline radiance fields. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 20041–20050, 2024. 2

    2024

  31. [39]

    Barron, Alberto Rodriguez, Phillip Isola, and Tsung-Yi Lin

    Lin Yen-Chen, Pete Florence, Jonathan T. Barron, Alberto Rodriguez, Phillip Isola, and Tsung-Yi Lin. inerf: Inverting neural radiance fields for pose estimation. In2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 1323–1330, 2021. 2

    2021

  32. [40]

    pixelnerf: Neural radiance fields from one or few images

    Alex Yu, Vickie Ye, Matthew Tancik, and Angjoo Kanazawa. pixelnerf: Neural radiance fields from one or few images. In Proceedings of the IEEE/CVF conference on computer vi- sion and pattern recognition, pages 4578–4587, 2021. 2

    2021

  33. [41]

    Gs-lrm: Large recon- struction model for 3d gaussian splatting

    Kai Zhang, Sai Bi, Hao Tan, Yuanbo Xiangli, Nanxuan Zhao, Kalyan Sunkavalli, and Zexiang Xu. Gs-lrm: Large recon- struction model for 3d gaussian splatting. arXiv preprint arXiv:2404.19702, 2024. 2

    2024 arXiv

  34. [42]

    The unreasonable effectiveness of deep features as a perceptual metric

    Richard Zhang, Phillip Isola, Alexei A Efros, Eli Shecht- man, and Oliver Wang. The unreasonable effectiveness of deep features as a perceptual metric. In Proceedings of the IEEE conference on computer vision and pattern recogni- tion, pages 586–595, 2018. 5

    2018

  35. [43]

    Stereo magnification: Learning view synthesis using multiplane images

    Tinghui Zhou, Richard Tucker, John Flynn, Graham Fyffe, and Noah Snavely. Stereo magnification: Learning view synthesis using multiplane images. arXiv preprint arXiv:1805.09817, 2018. 2

    2018 arXiv

  36. [44]

    a cupcake with whipped cream

    Qi Zuo, Xiaodong Gu, Yuan Dong, Zhengyi Zhao, Weihao Yuan, Lingteng Qiu, Liefeng Bo, and Zilong Dong. High- fidelity 3d textured shapes generation by sparse encoding and adversarial decoding. In European Conference on Computer Vision, 2024. 2, 6 10 Appendix This document conta...

    2024

Pith tools

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