Pith. sign in

REVIEW 3 major objections 6 minor 60 references

Controllable Shadow Generation with Single-Step Diffusion Models from Synthetic Data

T0 review · 3 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read A diffusion model trained only on synthetic Blender renders generates realistic, controllable shadows for cut-out object photos in one sampling step.

desk verdict A genuinely new task formulation and a well-ablated single-step rectified-flow result, but the real-image generalization claim overreaches because the training camera is fixed at eye level. read the letter →

arxiv 2412.11972 v1 pith:HO7W335Q submitted 2024-12-16 cs.CV

Onur Tasar , Clément Chadebec , Benjamin Aubin This is my paper · ORCID
classification cs.CV
keywords shadowgenerationdiffusionmodelsrectifiedflowsingle-stepinferencesyntheticdatacontrollableimagesynthesislatentcompositing
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 claims that a diffusion model can produce a realistic shadow map for a cut-out object image in a single sampling step, with direct control over the shadow's direction, softness, and intensity. To get there, the authors train the model exclusively on 257,612 synthetic renders of 9,872 3D meshes lit by a single square area light whose position and size are given as spherical coordinates. The central empirical claim is that a rectified-flow training objective makes one-step inference competitive with, and at one step better than, twenty-step epsilon-, v-, and sample-prediction diffusion training. If correct, this makes controllable shadow generation fast enough for real-time compositing and removes the need for 3D scene reconstruction at inference time.

What carries the argument

The load-bearing object is the rectified-flow objective, $\min_\psi \mathbb{E}\left[\|(x_1-x_0)-f_\psi(x_t,t)\|^2\right]$ with $x_t = x_1 t + (1-t)x_0$, which trains the denoiser to follow straight-line paths from noise to data and thereby makes a single Euler step sufficient at inference. It is paired with a conditioning mechanism that encodes the light-source parameters (polar angle $\theta$, azimuth $\phi$, area-light size $s$) as sinusoidal embeddings added to the timestep embedding, so that control signals enter the network without extra latent channels or external light maps.

What would settle it

Render or photograph a real object with known geometry and a known light position, run the model on the object cutout with matching ($\theta$, phi, s), and compare the predicted shadow map against a ray-traced ground truth; if soft IoU under matched real conditions falls far below the synthetic-test range (roughly 0.73-0.77), the synthetic-to-real generalization claim is not supported.

Watch

Extended reading notes

Core claim

The paper's central claim is that rectified flow, a diffusion training objective that forces the network to follow a straight interpolation path between noise and data, allows a latent diffusion model to predict high-quality shadow maps in a single sampling step. The model, built on an SDXL backbone with cross-attention removed, is conditioned on the object cutout, its binary mask, and scalar light parameters ($\theta$, $\phi$, $s$) encoded with sinusoidal embeddings injected into the timestep embedding; intensity is handled by scalar multiplication of the predicted map. On the authors' synthetic benchmark, the rectified-flow model reaches soft IoU around 0.73-0.77 in one step across three test tracks (softness, horizontal direction, vertical direction), outperforming $\varepsilon$-, $v$-, and sample-prediction models even at twenty steps. The authors further claim that the model, trained purely on synthetic Blender renders, generalizes to real photographs, and they release the synthetic dataset and a three-track public benchmark for future evaluation.

Load-bearing premise

The model is trained and quantitatively evaluated only on synthetic renders from the same Blender pipeline, so the claim that it generalizes to real-world photos rests on the assumption that those renders capture how real object photos cast shadows; the real-image evidence is qualitative only.

Editorial extensions

If this is right

  • One-step inference puts controllable shadow generation in real-time range for compositing, product photography, and video pipelines.
  • Because conditioning enters through scalar embeddings, shadow direction, softness, and intensity can be swept continuously without retraining or auxiliary control networks.
  • The approach is background-free: the same predicted shadow map can be blended into any target background.
  • Rectified flow, not only for generative synthesis, emerges as a strong default for dense-prediction diffusion tasks where output diversity is low.
  • The released three-track benchmark gives the community quantitative targets (IoU, RMSE, S-RMSE, ZNCC) for future shadow-generation methods.

Reading between the lines

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

  • The scalar-conditioning trick likely transfers to other physics parameters in image-conditioned diffusion (light intensity falloff, camera tilt, object scale), since it injects continuous control without architectural changes.
  • The near-deterministic nature of the image-plus-light to shadow-map mapping is probably why a single flow step suffices; tasks with more output variation (e.g., full relighting with reflections) may need more steps or distillation.
  • A quantitative real-world test is the natural next experiment: re-render scanned real objects under matched light parameters, or build a small real photo set with ray-traced reference shadows, to measure the synthetic-to-real gap.
  • Training data restricted to a single area light in a void means the model has never seen environment lighting or multiple lights; extending the dataset to environment maps would test whether controllability generalizes to everyday lighting.
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 / 6 minor

Summary. The paper proposes a controllable shadow generation pipeline for 2D object cutouts. It builds a large synthetic dataset of 257,612 renders from 9,872 artist-created meshes using Blender Cycles with a square area light, rendering for each light-parameter triple (θ, φ, s) a color image, a binary mask, and a gray-scale shadow map. The authors train an SDXL-based latent diffusion model with a rectified flow objective, conditioning on the object image and mask via latent concatenation and on the light parameters via sinusoidal embeddings added to the timestep embedding. They conduct ablations over prediction types (ε, v, sample, rectified flow), number of sampling steps, training iterations, and conditioning mechanisms (scalar vs. blob map), reporting IoU, RMSE, s-RMSE, and ZNCC on three test tracks with held-out meshes and 10 seeds. The main quantitative finding is that rectified flow with one sampling step outperforms the other prediction types with 20 steps. Qualitative results on real images demonstrate softness, direction, and intensity control, and the authors release the synthetic benchmark.

Significance. The quantitative core of the paper is solid: the multi-seed, multi-metric ablations with held-out meshes provide credible evidence that the rectified flow objective is particularly well-suited to single-step shadow prediction in this setting. The release of a public benchmark with three control tracks is a useful community resource. If the real-image generalization holds in broader conditions, the method could enable practical real-time shadow control for image compositing. However, the paper's second central claim—generalization to real-world images—is currently supported only by a qualitative gallery from a restricted camera viewpoint, and the experiments do not compare against prior shadow generation methods. These gaps limit the current evidence for the full set of claims.

major comments (3)
  1. [Sec. 4.4 and Abstract] The claim that the model "generalizes well to real-world images" is load-bearing but is supported only by qualitative examples. The training pipeline (Sec. 3.1) fixes the camera at the negative y-axis and moves it only along that axis, yielding eye-level views; the synthetic test tracks (Table 1) use the same camera setup. Real-world photographs frequently use non-eye-level camera angles, and the mapping from a 2D object cutout to a ground-plane shadow changes with camera pitch. The qualitative results in Table 3 and SM Sec. 8 are limited to apparently eye-level views, so the model's behavior under pitch variation is untested. I request either (a) an experiment that varies camera pitch on synthetic scenes (e.g., rendering the same object and light settings from multiple camera elevations and measuring shadow-map accuracy), (b) a quantitative real-image evaluation with known ground-truth shadows or a user study, or (c) a tempered claim that restricts generalization to the eye-level setting.
  2. [Sec. 4, Experiments] The paper does not compare the proposed method against any existing shadow generation approach. Related work discusses SSN [40], Pixel Height Maps [41], PixHt-Lab [42], ObjectDrop [51], and Desobav2 [23], but none of these are evaluated on the released benchmark or on an existing dataset. Without such a comparison, the contribution's practical benefit relative to the state of the art is not established. At minimum, I expect a comparison with SSN and one recent diffusion-based method (e.g., ObjectDrop or Desobav2) on the released test tracks, or an explanation of why a comparison is infeasible (e.g., conditioning differences). This is necessary to support the claim that the method produces "high-quality results" in the context of existing work.
  3. [Sec. 4.3, Fig. 11] The intensity conditioning ablation concludes that conditioning on S(θ, φ, s, I) is equivalent to post-hoc scaling by a scalar I, but the comparison is presented only as pairs of images. Because the paper uses quantitative metrics for other ablations, the same metrics (IoU, RMSE, ZNCC) should be reported for the two intensity-control variants; without them, the conclusion of equivalence is not established. This matters because the abstract and introduction list intensity control as one of the controllable shadow properties.
minor comments (6)
  1. [Sec. 3.1, Table 1] The sampling procedure for light parameters is under-specified. The intervals for θ, ϕ, and s imply up to 46 × 361 × 7 = 116,242 possible combinations, but 257,612 training images were rendered; clarify whether combinations are randomly sampled per iteration, and how the counts in Table 1 were derived.
  2. [Eq. (7)] The sinusoidal embedding formula for ω_i^d is non-standard; please double-check the indexing, since the term i·(i−1) suggests a quadratic schedule while the standard formulation uses 10000^{-2i/d}. If the formula is intentional, clarify; if it is a typo, correct it.
  3. [Sec. 3.2.3] The phrase "replicating it twice and concatenating them with the original" is ambiguous; it should state that the gray-scale channel is duplicated to form a 3-channel RGB input for the VAE.
  4. [Fig. 8 and Figs. 16-17] The x-axis for training iterations starts at 0 and then jumps to 10k; consider labeling all ticks consistently (0, 10k, 30k, ...) or removing the 0 tick to avoid implying a linear scale.
  5. [Sec. 4.1] The statement "We use 50 models to create a test set and consider the remaining 9,872 models for the trainings" is consistent with the total 9,922, but the test tracks use overlapping subsets (50, 15, 15 models); please state explicitly whether the 15-model tracks are disjoint from the 50-model track or drawn from the same held-out set.
  6. [Sec. 4.4] Please specify how the real foreground images were cut out (manual or automatic matting) and how the model inputs (object image and mask) were prepared for the real-image tests, since small cutout errors can affect shadow quality.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity found; the central result is an empirical comparison against an external physics-based renderer.

full rationale

The paper's derivation chain is not circular. The ground-truth shadow maps are produced by Blender's Cycles ray-tracing renderer from artist-created 3D meshes, which is an external physical simulation independent of the trained model. The model is trained to predict those rendered shadow maps conditioned on object image, mask, and light parameters, so the target is not defined in terms of the model's output. The single-step advantage of rectified flow is established by ablations comparing four prediction objectives at multiple sampling-step counts; this is an empirical finding, not an identity forced by the loss definitions. The published evaluation tracks use held-out 3D meshes rendered by the same external pipeline, and the stated real-image generalization claim is supported only qualitatively, which is a limitation in evidence strength rather than a circular reduction. The only self-citation, Flash Diffusion [1], appears in a related-work list of few-step sampling attempts and is not used in the method, loss, or uniqueness argument, so it is not load-bearing. No fitted parameter is renamed as a prediction, no uniqueness theorem is imported from the authors' prior work, and no known result is repackaged under new coordinates. The intensity-control ablation is an empirical check that post-hoc scaling approximates explicit intensity conditioning; it is not a claim that the two are identical by construction. Overall, the central claims stand on independently rendered data and controlled experiments, so no circular step is present.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

The paper contributes a trained predictor, not a closed-form derivation. Its numerical results depend on hand-set scene constants and domain assumptions about synthetic-to-real transfer, rather than on fitted free parameters in the physics sense.

free parameters (2)
  • sphere radius r = 8 (scene units)
    Set by hand in Sec. 3.1 and Fig. 4. The light distance determines how a fixed angular change translates into shadow displacement, so the controllability mapping is tied to this constant.
  • training light parameter ranges = theta 0-45 deg, phi 0-360 deg, s 2-8
    Chosen in Table 1. The model and benchmark only operate inside these ranges; behavior outside them is not demonstrated.
assumptions (5)
  • domain assumption Blender cycles renders with a single square area light produce physically correct shadow ground truths suitable for training.
    The entire training signal is synthetic (Sec. 3.1); real-world validity is only checked qualitatively in Sec. 4.4.
  • domain assumption A 2D object image plus binary mask is sufficient to infer the 3D geometry relevant for shadow prediction.
    The model conditions only on the object image and mask (Sec. 3.2.3), so hidden surfaces or self-occlusions must be recovered from appearance alone.
  • domain assumption The frozen SDXL VAE, trained on natural images, compresses replicated grayscale shadow maps without losing information critical for one-step generation.
    The target shadow is transformed into a 3-channel image and encoded with the VAE (Sec. 3.2.3); no analysis of VAE reconstruction error for shadow maps is provided.
  • domain assumption A single Euler step of the rectified flow ODE is an accurate approximation for this task.
    Flow matching theory does not guarantee one-step exactness; the paper relies on empirical evidence in Sec. 4.2.1.
  • domain assumption Real-world lighting can be approximated by one controllable area light with direction and softness parameters.
    The training set uses exactly one light (Sec. 3.1); environment lighting and multiple lights are not modeled.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Controllable Shadow Generation with Single-Step Diffusion Models from Synthetic Data." pith.science (2026). https://pith.science/paper/HO7W335Q

@misc{pith2026241211972,
  author       = {Pith},
  title        = {Pith review of: Controllable Shadow Generation with Single-Step Diffusion Models from Synthetic Data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HO7W335Q}},
  note         = {Machine review of arXiv:2412.11972}
}
read the original abstract

Realistic shadow generation is a critical component for high-quality image compositing and visual effects, yet existing methods suffer from certain limitations: Physics-based approaches require a 3D scene geometry, which is often unavailable, while learning-based techniques struggle with control and visual artifacts. We introduce a novel method for fast, controllable, and background-free shadow generation for 2D object images. We create a large synthetic dataset using a 3D rendering engine to train a diffusion model for controllable shadow generation, generating shadow maps for diverse light source parameters. Through extensive ablation studies, we find that rectified flow objective achieves high-quality results with just a single sampling step enabling real-time applications. Furthermore, our experiments demonstrate that the model generalizes well to real-world images. To facilitate further research in evaluating quality and controllability in shadow generation, we release a new public benchmark containing a diverse set of object images and shadow maps in various settings. The project page is available at https://gojasper.github.io/controllable-shadow-generation-project/

Figures

Figures reproduced from arXiv: 2412.11972 by the authors.

Figure 1
Figure 1. Our single-step model enables the generation of realistic shadows with precise control over their direction, softness, and intensity. [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Example renders with unprocessed (first two) and pro [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 4
Figure 4. Spherical coordinate system. θ, ϕ, and r represent the polar angle, azimuthal angle, and the radius. s corresponds to the size of the area light. We place the camera at negative y-axis. (a) Image (b) Mask (c) Shadow Map [PITH_FULL_IMAGE:figures/full_fig_p003_4.png] view at source ↗
Figures from the paper (16 more)
Figure 5
Figure 5. Figure 5: An example image from our dataset and its annotations. [PITH_FULL_IMAGE:figures/full_fig_p003_5.png]
Figure 6
Figure 6. Figure 6: Our controllable shadow generation pipeline. We first remove the background of the input image, providing us with a binary [PITH_FULL_IMAGE:figures/full_fig_p005_6.png]
Figure 7
Figure 7. Figure 7: Example renders from each test track. Track 1: Softness control. Tracks 2-3: Horizontal and vertical shadow direction control. [PITH_FULL_IMAGE:figures/full_fig_p006_7.png]
Figure 8
Figure 8. Figure 8: IoU vs number of sampling (first plot) and training iter [PITH_FULL_IMAGE:figures/full_fig_p006_8.png]
Figure 11
Figure 11. Figure 11: Intensity Control. The first two and the last two images [PITH_FULL_IMAGE:figures/full_fig_p007_11.png]
Figure 10
Figure 10. Figure 10: Comparison between our timestep and blob condition [PITH_FULL_IMAGE:figures/full_fig_p007_10.png]
Figure 12
Figure 12. Figure 12: Number of 3D meshes for each category in our synthetic dataset. [PITH_FULL_IMAGE:figures/full_fig_p012_12.png]
Figure 13
Figure 13. Figure 13: Renders for one 3D mesh from the horizontal shadow direction control track. [PITH_FULL_IMAGE:figures/full_fig_p013_13.png]
Figure 14
Figure 14. Figure 14: Renders for two 3D meshes from the vertical shadow direction control track. [PITH_FULL_IMAGE:figures/full_fig_p013_14.png]
Figure 15
Figure 15. Figure 15: Renders for two 3D meshes from the softness control track. [PITH_FULL_IMAGE:figures/full_fig_p014_15.png]
Figure 17
Figure 17. Figure 17: Plots comparing methods trained for various iterations, [PITH_FULL_IMAGE:figures/full_fig_p014_17.png]
Figure 18
Figure 18. Figure 18: Softness control. The fixed light parameters are [PITH_FULL_IMAGE:figures/full_fig_p015_18.png]
Figure 19
Figure 19. Figure 19: Softness control. The fixed light parameters are [PITH_FULL_IMAGE:figures/full_fig_p015_19.png]
Figure 20
Figure 20. Figure 20: Horizontal shadow direction control. The fixed light parameters are [PITH_FULL_IMAGE:figures/full_fig_p016_20.png]
Figure 21
Figure 21. Figure 21: Vertical shadow direction control. The fixed light parameters are [PITH_FULL_IMAGE:figures/full_fig_p016_21.png]
Figure 22
Figure 22. Figure 22: Shadow direction and softness control by changing the value for each of [PITH_FULL_IMAGE:figures/full_fig_p017_22.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

60 extracted references · 56 canonical work pages

  1. [40]

    Ssn: Soft shadow network for image compositing

    Yichen Sheng, Jianming Zhang, and Bedrich Benes. Ssn: Soft shadow network for image compositing. In CVPR, pages 4380–4390, 2021. 2, 5, 7

    work page 2021

  2. [41]

    Controllable shadow generation using pixel height maps

    Yichen Sheng, Yifan Liu, Jianming Zhang, Wei Yin, A Cengiz Oztireli, He Zhang, Zhe Lin, Eli Shechtman, and Bedrich Benes. Controllable shadow generation using pixel height maps. In ECCV, pages 240–256. Springer, 2022. 2

    work page 2022

  3. [42]

    Pixht-lab: Pixel height based light effect generation for image compositing

    Yichen Sheng, Jianming Zhang, Julien Philip, Yan- nick Hold-Geoffroy, Xin Sun, He Zhang, Lu Ling, and Bedrich Benes. Pixht-lab: Pixel height based light effect generation for image compositing. In CVPR, pages 16643–16653, 2023. 2

    work page 2023

  4. [51]

    Object- drop: Bootstrapping counterfactuals for photorealistic object removal and insertion, 2024

    Daniel Winter, Matan Cohen, Shlomi Fruchter, Yael Pritch, Alex Rav-Acha, and Yedid Hoshen. Object- drop: Bootstrapping counterfactuals for photorealistic object removal and insertion, 2024. 2

    work page 2024

  5. [23]

    Shadow generation for com- posite image using diffusion model

    Qingyang Liu, Junqi You, Jianting Wang, Xinhao Tao, Bo Zhang, and Li Niu. Shadow generation for com- posite image using diffusion model. In CVPR, pages 8121–8130, 2024. 2

    work page 2024

  6. [1]

    Flash diffusion: Accelerating any conditional diffusion model for few steps image gen- eration, 2024

    Clement Chadebec, Onur Tasar, Eyal Benaroche, and Benjamin Aubin. Flash diffusion: Accelerating any conditional diffusion model for few steps image gen- eration, 2024. 2

    work page 2024

  7. [2]

    Diffusion models in vi- sion: A survey

    Florinel-Alin Croitoru, Vlad Hondru, Radu Tudor Ionescu, and Mubarak Shah. Diffusion models in vi- sion: A survey. PAMI, 45(9):10850–10869, 2023. 4

    work page 2023

  8. [3]

    Nerdi: Single-view nerf syn- thesis with language-guided diffusion as general image priors

    Congyue Deng, Chiyu Jiang, Charles R Qi, Xinchen Yan, Yin Zhou, Leonidas Guibas, Dragomir Anguelov, et al. Nerdi: Single-view nerf syn- thesis with language-guided diffusion as general image priors. In CVPR, pages 20637–20647, 2023. 2

    work page 2023

Show all 60 references
  1. [4]

    Blenderproc, 2019

    Maximilian Denninger, Martin Sundermeyer, Do- minik Winkelbauer, Youssef Zidan, Dmitry Olefir, Mohamad Elbadrawy, Ahsan Lodhi, and Harinandan Katam. Blenderproc, 2019. 2

    2019

  2. [5]

    Scaling rectified flow transformers for high-resolution image synthesis

    Patrick Esser, Sumith Kulal, Andreas Blattmann, Rahim Entezari, Jonas M¨uller, Harry Saini, Yam Levi, Dominik Lorenz, Axel Sauer, Frederic Boesel, et al. Scaling rectified flow transformers for high-resolution image synthesis. In ICML, 2024. 2

    2024

  3. [6]

    Generative adversarial nets

    Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. Generative adversarial nets. NIPS, 27, 2014. 2

    2014

  4. [7]

    Modeling and anima- tion using blender: blender 2.80: the rise of Eevee

    Ezra Thess Mendoza Guevarra. Modeling and anima- tion using blender: blender 2.80: the rise of Eevee . Apress, 2019. 5

    2019

  5. [8]

    Lotus: Diffusion-based visual foun- dation model for high-quality dense prediction, 2024

    Jing He, Haodong Li, Wei Yin, Yixun Liang, Leheng Li, Kaiqiang Zhou, Hongbo Liu, Bingbing Liu, and Ying-Cong Chen. Lotus: Diffusion-based visual foun- dation model for high-quality dense prediction, 2024. 2

    2024

  6. [9]

    Distill- ing the knowledge in a neural network

    Geoffrey Hinton, Oriol Vinyals, and Jeff Dean. Distill- ing the knowledge in a neural network. arXiv preprint arXiv:1503.02531, 2015. 2

    2015 arXiv

  7. [10]

    Denoising diffusion probabilistic models

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffusion probabilistic models. NIPS, 33:6840–6851,

  8. [11]

    Shadow gen- eration for composite image in real-world scenes

    Yan Hong, Li Niu, and Jianfu Zhang. Shadow gen- eration for composite image in real-world scenes. In AAAI, pages 914–922, 2022. 2

    2022

  9. [12]

    Diffusion for natural image matting, 2023

    Yihan Hu, Yiheng Lin, Wei Wang, Yao Zhao, Yunchao Wei, and Humphrey Shi. Diffusion for natural image matting, 2023. 2

    2023

  10. [13]

    Blender cycles: lighting and rendering cookbook

    Bernardo Iraci. Blender cycles: lighting and rendering cookbook. Packt Publishing Ltd, 2013. 5

    2013

  11. [14]

    Repurposing diffusion-based image generators for monocular depth estimation

    Bingxin Ke, Anton Obukhov, Shengyu Huang, Nando Metzger, Rodrigo Caye Daudt, and Konrad Schindler. Repurposing diffusion-based image generators for monocular depth estimation. In CVPR, 2024. 2

    2024

  12. [15]

    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. TOG, 42(4), 2023. 2

    2023

  13. [16]

    Variational diffusion models

    Diederik Kingma, Tim Salimans, Ben Poole, and Jonathan Ho. Variational diffusion models. NIPS, 34: 21696–21707, 2021. 4

    2021

  14. [17]

    Auto-encoding variational bayes,

    Diederik P Kingma. Auto-encoding variational bayes,

  15. [18]

    Fast template matching

    John P Lewis et al. Fast template matching. In Vision interface, pages 15–19. Quebec City, QC, Canada,

  16. [19]

    Playground v2.5: Three insights towards enhancing aesthetic quality in text-to-image generation, 2024

    Daiqing Li, Aleks Kamko, Ehsan Akhgari, Ali Sa- bet, Linmiao Xu, and Suhail Doshi. Playground v2.5: Three insights towards enhancing aesthetic quality in text-to-image generation, 2024. 2

    2024

  17. [20]

    Controlnet++: Improving conditional controls with efficient consistency feedback

    Ming Li, Taojiannan Yang, Huafeng Kuang, Jie Wu, Zhaoning Wang, Xuefeng Xiao, and Chen Chen. Controlnet++: Improving conditional controls with efficient consistency feedback. In ECCV, pages 129–

  18. [21]

    Flow matching for generative modeling

    Yaron Lipman, Ricky TQ Chen, Heli Ben-Hamu, Maximilian Nickel, and Matthew Le. Flow matching for generative modeling. In ICLR, 2023. 2, 4

    2023

  19. [22]

    One- 2-3-45: Any single image to 3d mesh in 45 seconds without per-shape optimization

    Minghua Liu, Chao Xu, Haian Jin, Linghao Chen, Mukund Varma T, Zexiang Xu, and Hao Su. One- 2-3-45: Any single image to 3d mesh in 45 seconds without per-shape optimization. NIPS, 36, 2024. 2

    2024

  20. [24]

    Shadows shed light on 3d objects, 2022

    Ruoshi Liu, Sachit Menon, Chengzhi Mao, Dennis Park, Simon Stent, and Carl V ondrick. Shadows shed light on 3d objects, 2022. 2

    2022

  21. [25]

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

    Ruoshi Liu, Rundi Wu, Basile Van Hoorick, Pavel Tokmakov, Sergey Zakharov, and Carl V ondrick. Zero-1-to-3: Zero-shot one image to 3d object. In CVPR, pages 9298–9309, 2023. 2

    2023

  22. [26]

    Flow straight and fast: Learning to generate and transfer data with rectified flow

    Xingchao Liu, Chengyue Gong, and Qiang Liu. Flow straight and fast: Learning to generate and transfer data with rectified flow. In ICLR, 2023. 4

    2023

  23. [27]

    Instaflow: One step is enough for high-quality diffusion-based text-to-image generation

    Xingchao Liu, Xiwen Zhang, Jianzhu Ma, Jian Peng, et al. Instaflow: One step is enough for high-quality diffusion-based text-to-image generation. In ICLR,

  24. [28]

    Decoupled weight decay regularization

    Ilya Loshchilov and Frank Hutter. Decoupled weight decay regularization. In ICLR, 2019. 6

    2019

  25. [29]

    Latent consistency models: Synthesizing high-resolution images with few-step inference, 2023

    Simian Luo, Yiqin Tan, Longbo Huang, Jian Li, and Hang Zhao. Latent consistency models: Synthesizing high-resolution images with few-step inference, 2023. 2

    2023

  26. [30]

    On distillation of guided diffusion models

    Chenlin Meng, Robin Rombach, Ruiqi Gao, Diederik Kingma, Stefano Ermon, Jonathan Ho, and Tim Sali- mans. On distillation of guided diffusion models. In CVPR, pages 14297–14306, 2023. 2

    2023

  27. [31]

    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. InCVPR, pages 12663–12673, 2023. 2

    2023

  28. [32]

    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 synthesis. In ECCV, 2020. 2

    2020

  29. [33]

    T2i- adapter: Learning adapters to dig out more control- lable ability for text-to-image diffusion models

    Chong Mou, Xintao Wang, Liangbin Xie, Yanze Wu, Jian Zhang, Zhongang Qi, and Ying Shan. T2i- adapter: Learning adapters to dig out more control- lable ability for text-to-image diffusion models. In AAAI, pages 4296–4304, 2024. 2

    2024

  30. [34]

    Sdxl: Improving latent diffusion models for high-resolution image synthesis, 2023

    Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas M ¨uller, Joe Penna, and Robin Rombach. Sdxl: Improving latent diffusion models for high-resolution image synthesis, 2023. 2, 4, 5

    2023

  31. [35]

    High- resolution image synthesis with latent diffusion mod- els

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj ¨orn Ommer. High- resolution image synthesis with latent diffusion mod- els. In CVPR, pages 10684–10695, 2022. 2, 4

    2022

  32. [36]

    U-net: Convolutional networks for biomedical image segmentation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U-net: Convolutional networks for biomedical image segmentation. In MICCAI, pages 234–241. Springer,

  33. [37]

    Progressive distilla- tion for fast sampling of diffusion models

    Tim Salimans and Jonathan Ho. Progressive distilla- tion for fast sampling of diffusion models. In ICLR,

  34. [38]

    Adversarial diffusion distillation

    Axel Sauer, Dominik Lorenz, Andreas Blattmann, and Robin Rombach. Adversarial diffusion distillation. In ECCV, pages 87–103. Springer, 2025. 2

    2025

  35. [39]

    Saurabh Saxena, Abhishek Kar, Mohammad Norouzi, and David J. Fleet. Monocular depth estimation using diffusion models, 2023. 2

    2023

  36. [43]

    Realistic ray trac- ing

    Peter Shirley and R Keith Morley. Realistic ray trac- ing. AK Peters, Ltd., 2008. 1, 5

    2008

  37. [44]

    Deep unsupervised learning using nonequilibrium thermodynamics

    Jascha Sohl-Dickstein, Eric Weiss, Niru Mah- eswaranathan, and Surya Ganguli. Deep unsupervised learning using nonequilibrium thermodynamics. In ICML, pages 2256–2265. PMLR, 2015. 2, 4

    2015

  38. [45]

    Score-based generative modeling through stochastic differential equations

    Yang Song, Jascha Sohl-Dickstein, Diederik P Kingma, Abhishek Kumar, Stefano Ermon, and Ben Poole. Score-based generative modeling through stochastic differential equations. In ICLR, 2020. 2, 4

    2020

  39. [46]

    Objectstitch: Object compositing with diffu- sion model

    Yizhi Song, Zhifei Zhang, Zhe Lin, Scott Cohen, Brian Price, Jianming Zhang, Soo Ye Kim, and Daniel Aliaga. Objectstitch: Object compositing with diffu- sion model. In CVPR, pages 18310–18319, 2023. 2

    2023

  40. [47]

    Rigid-body dynamics with friction and impact

    David E Stewart. Rigid-body dynamics with friction and impact. SIAM review, 42(1):3–39, 2000. 3

    2000

  41. [48]

    Single image portrait relighting.TOG, 38(4):1–12,

    Tiancheng Sun, Jonathan T Barron, Yun-Ta Tsai, Zex- iang Xu, Xueming Yu, Graham Fyffe, Christoph Rhe- mann, Jay Busch, Paul Debevec, and Ravi Ramamoor- thi. Single image portrait relighting.TOG, 38(4):1–12,

  42. [49]

    The art and style of product pho- tography

    J Dennis Thomas. The art and style of product pho- tography. John Wiley & Sons, 2013. 1

    2013

  43. [50]

    Matting by generation

    Zhixiang Wang, Baiang Li, Jian Wang, Yu-Lun Liu, Jinwei Gu, Yung-Yu Chuang, and Shin’ichi Satoh. Matting by generation. In SIGGRAPH, 2024. 2

    2024

  44. [52]

    Neurallift-360: Lifting an in-the-wild 2d photo to a 3d object with 360deg views

    Dejia Xu, Yifan Jiang, Peihao Wang, Zhiwen Fan, Yi Wang, and Zhangyang Wang. Neurallift-360: Lifting an in-the-wild 2d photo to a 3d object with 360deg views. In CVPR, pages 4479–4489, 2023. 2

    2023

  45. [53]

    Ufogen: You forward once large scale text-to- image generation via diffusion gans

    Yanwu Xu, Yang Zhao, Zhisheng Xiao, and Tingbo Hou. Ufogen: You forward once large scale text-to- image generation via diffusion gans. arXiv preprint arXiv:2311.09257, 2023. 2

    2023 arXiv

  46. [54]

    Diffusion models: A comprehen- sive survey of methods and applications

    Ling Yang, Zhilong Zhang, Yang Song, Shenda Hong, Runsheng Xu, Yue Zhao, Wentao Zhang, Bin Cui, and Ming-Hsuan Yang. Diffusion models: A comprehen- sive survey of methods and applications. ACM Com- puting Surveys, 56(4):1–39, 2023. 4

    2023

  47. [55]

    Stablenormal: Reducing diffu- sion variance for stable and sharp normal.TOG, 2024

    Chongjie Ye, Lingteng Qiu, Xiaodong Gu, Qi Zuo, Yushuang Wu, Zilong Dong, Liefeng Bo, Yuliang Xiu, and Xiaoguang Han. Stablenormal: Reducing diffu- sion variance for stable and sharp normal.TOG, 2024. 2

    2024

  48. [56]

    Adding conditional control to text-to-image diffusion models

    Lvmin Zhang, Anyi Rao, and Maneesh Agrawala. Adding conditional control to text-to-image diffusion models. In CVPR, pages 3836–3847, 2023. 2

    2023

  49. [57]

    Shad- owgan: Shadow synthesis for virtual objects with con- ditional adversarial networks

    Shuyang Zhang, Runze Liang, and Miao Wang. Shad- owgan: Shadow synthesis for virtual objects with con- ditional adversarial networks. CVM, 5:105–115, 2019. 2 Controllable Shadow Generation with Single-Step Diffusion Models from Synthetic Data Supplementary Material Figure 12. ...

    2019

  50. [58]

    4.1, we gathered 9,922 3D meshes created by professional artists publicly available on blenderkit 2 with a free of use, representing a diverse array of real-world object categories

    Synthetic Dataset As mentioned in Sec. 4.1, we gathered 9,922 3D meshes created by professional artists publicly available on blenderkit 2 with a free of use, representing a diverse array of real-world object categories. Fig. 12 displays the cate- gory names and number of 3D m...

  51. [59]

    8 compares models trained with various prediction types across multiple sampling steps and training iterations, using only IoU as the evaluation metric

    Quantitative Analysis Fig. 8 compares models trained with various prediction types across multiple sampling steps and training iterations, using only IoU as the evaluation metric. The figure presents the average values computed from all images over all our 3 tracks across 10 s...

  52. [60]

    Some 3D meshes, such as those in the seating set category (see Fig

    Qualitative Analysis on Real Images For a comprehensive qualitative analysis, we gathered a di- verse collection of object images from Unsplash 3 and Pex- els 4. Some 3D meshes, such as those in the seating set category (see Fig. 12), consist of multiple objects. Conse- quentl...

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.