Pith. sign in

REVIEW 4 major objections 6 minor 1 cited by

Digital Kitchen Remodeling: Editing and Relighting Intricate Indoor Scenes from a Single Panorama

T0 review · 4 major / 6 minor · reviewed 2026-08-09 · deepseek-v4-flash

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

Pith's one-line read A $30 light meter replaces a $5,000 one for believable relighting.

desk verdict A credible low-cost HDR calibration trick and a useful dataset, but the validation only covers one whiteboard while the claims cover the whole panorama and the paired outdoor image. read the letter →

arxiv 2504.16086 v1 pith:BWSC7Y7Y submitted 2025-02-05 cs.GR cs.HC

Guanzhou Ji , Azadeh O. Sawyer , Srinivasa G. Narasimhan This is my paper · ORCID
classification cs.GRcs.HC
keywords virtualstagingpanoramicHDRphotometriccalibrationrelightingkitchenremodelingindoorsceneeditingglobalilluminationPano-Panodataset
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

This paper tries to show that a complete kitchen remodeling preview can be generated from a single 360-degree photograph, with lighting that is physically believable rather than painted on. The key move is a low-cost photometric calibration: a roughly $30 illuminance meter, the TS-710, is held next to the camera while the panorama is shot, and the measured illuminance is converted into an absolute scene radiance value that rescales every HDR pixel. Across 141 captured scenes the rescaled panoramas agree with a reference luminance meter to a mean absolute error of 3.988 cd/m². If that calibration holds, the expensive $5,000 luminance meter that prior virtual-staging workflows required can be replaced by a consumer light meter, and relighting edited interiors becomes a practical service rather than a laboratory procedure. The same pipeline then automatically detects the kitchen area, places new cabinets and appliances along the wall, and renders the result under the real outdoor illumination captured in a paired outdoor panorama.

What carries the argument

The load-bearing identity is the relation between illuminance and radiance over a hemisphere: $E = \int_0^{2\pi} \int_0^{\pi/2} L(\theta,\phi)\,\sin\theta\cos\theta\,d\theta\,d\phi$, which under the assumption of uniform radiance reduces to $L = E/\pi$. To use it, the panorama's front-lens region, spanning the central half of the equirectangular image, is geometrically warped into a 180-degree orthographic fisheye image whose average pixel luminance can be compared with $L$. That comparison yields one scalar calibration factor per capture, applied to both the indoor and outdoor panoramas. The rest of the application pipeline assembles around this: layout estimation supplies 3D walls, semantic segmentation locates the kitchen, a $4\times 4$ rigid transformation $M = R_z(\theta)\cdot T(t_x, t_y)$ positions each new component along the wall, and a global-illumination renderer consumes the calibrated outdoor panorama as its environment map.

What would settle it

Take a room lit mainly through a side window, place a luminance meter on white targets facing several different directions, and compare each reading with the same pixel in a light-meter-calibrated HDR panorama; if errors grow systematically for targets facing the window while the single whiteboard target reports near 3.988 cd/m², the uniform-radiance assumption is the failing link.

Watch

Extended reading notes

Core claim

The central claim is that absolute scene radiance can be recovered from a single HDR panorama using only a scalar illuminance measurement. The measured horizontal illuminance $E$ is converted to a uniform hemispherical luminance $L = E/\pi$, and after cropping the camera's front fisheye image into a 180-degree orthographic projection, the average displayed luminance of that image is rescaled by the factor $k = L / L_{\text{avg}}$ to match physical units. The same factor is applied to the paired outdoor panorama, which then acts as a calibrated 360-degree environment map for rendering. Validated against a professional luminance meter on a whiteboard target, the calibration reaches a mean absolute error of 3.988 cd/m² over the 141 scenes in the new Pano-Pano HDR dataset. This is what makes the kitchen staging results relightings in actual light, with reflections and spatially varying window illumination, rather than composited edits.

Load-bearing premise

The calibration assumes the light meter's reading came from a constant brightness across the whole 180-degree hemisphere, so one average value $L = E/\pi$ rescales every pixel; a room with a bright window or direct sunlight violates that, and the reported validation only checks one whiteboard target per scene.

Editorial extensions

If this is right

  • Photometric calibration of HDR panoramas drops from a $5,000 instrument to a $30 light meter, so any real-estate photographer with a 360-degree camera can produce physically scaled radiance images.
  • Because the outdoor panorama is calibrated with the same factor and used as an environment map, virtual cabinets and countertops receive the room's actual window light, including directional shadows and reflections.
  • The automatic layout arrangement handles I, L, and U kitchens by splitting walls into linear sections, so remodeling previews can be generated without manual 3D modeling per room.
  • The 141-pair Pano-Pano HDR dataset, labeled with indoor and outdoor illuminance, indoor luminance, and room orientation, gives other relighting and staging methods a common calibration benchmark.

Reading between the lines

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

  • The uniform-radiance simplification suggests a cheap upgrade the paper does not explore: weight each orthographic fisheye pixel by $\cos\theta$ before averaging, which would reduce the influence of near-horizon bright windows and likely lower the error where directional light dominates.
  • The calibration idea is not kitchen-specific; the same $L = E/\pi$ scaling could be applied to any indoor-outdoor panorama pair, so bathrooms, offices, and whole-home tours are natural extensions of the stated pipeline.
  • The reported error is measured on one whiteboard per scene; the strongest follow-up test would be to measure luminance targets at several depths and orientations and check whether the single scalar factor still holds away from the meter's position.
  • The two-lens geometry of the Ricoh Theta Z1 means the 180-degree orthographic crop is only an approximation of what the light meter sees; using the camera's full calibrated projection function instead of a crop-and-warp could tighten the alignment.
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

4 major / 6 minor

Summary. The paper presents a virtual staging system for kitchen remodeling from 360° HDR panoramas. It contributes a capture protocol using a single Ricoh Theta Z1 to obtain paired indoor and outdoor panoramas, a low-cost photometric calibration method based on a TS-710 illuminance meter that scales each HDR panorama to absolute radiance, a dataset of 141 paired panoramas with photometric labels, and a rendering pipeline that estimates 3D layout, inserts new kitchen components, and relights the edited scene under global illumination. The calibration method is validated against a Konica Minolta LS-160 luminance meter on one whiteboard target per scene, reporting a mean absolute luminance error of 3.988 cd/m². The relighting results are demonstrated qualitatively in several figures.

Significance. If the calibration result holds for full panoramas and outdoor environment maps, the work has practical value: it lowers the cost of photometric calibration by two orders of magnitude, provides a new paired indoor-outdoor HDR dataset, and demonstrates a complete editing/relighting pipeline for a complex indoor scene type. The calibration factor is not circular: it is computed from an independent illuminance measurement and checked against a separate luminance meter. The main strengths are the dataset scale, the low-cost measurement procedure, and the integration of capture, layout generation, and physically based rendering. However, the quantitative support is concentrated on a single target per scene, and the relighting quality is not compared with prior methods, so the broader claims currently outrun the evidence.

major comments (4)
  1. [Section 3, Algorithm 1] The reported 3.988 cd/m² error validates the calibration factor only at one whiteboard location in the front lens hemisphere, yet the same factor is applied to the entire 360° panorama and to the paired outdoor HDR used as an environment map. This transfer requires (i) spatially uniform radiometric response after vignetting correction and stitching across both lenses and the seam, (ii) identical camera exposure and gain for the indoor and outdoor captures, and (iii) an outdoor HDR that is not clipped in bright sky or window regions. None of these conditions is demonstrated. I ask for multi-position luminance measurements (front hemisphere, back hemisphere, seam) and an outdoor radiance check, or an explicit argument that the Theta Z1's internal processing guarantees these properties.
  2. [Section 3, paragraph following Eq. (2)] The manuscript says the indoor and outdoor panoramas are "captured simultaneously" with a single camera and later says the outdoor photograph is taken "immediately after" the indoor photograph; these are inconsistent, and simultaneous capture by one camera is physically impossible. Because the calibration factor derived indoors is applied to the outdoor panorama, the exposure-locking procedure and the time interval between the two captures must be stated precisely. Without this, the outdoor environment map may be mis-scaled even if the indoor calibration is accurate.
  3. [Section 4.1, Algorithm 2] The abstract and contribution list describe "automatic kitchen layout generation," but Algorithm 2 takes a user-defined "Sequence Order" as input and the text states that objects are placed "in a sequence order defined by user inputs." The amount of user interaction should be clarified, and the automatic scaling of the last object to cover the wall corner should be evaluated or at least discussed as a geometric approximation that can distort object proportions.
  4. [Section 4.2, Figures 9 and 10] The central claim of "high-quality scene relighting" is supported only by qualitative images. There is no quantitative comparison with prior relighting methods [6,9,22,32], nor a quantitative comparison with the real captured reference shown in Figure 10(d). Please add an evaluation protocol (e.g., rendered-versus-captured error on held-out regions, or a perceptual study) or restrict the claims accordingly.
minor comments (6)
  1. [Section 3, Eq. (2)] The uniform-radiance assumption is stronger than needed. For an orthographic projection, the average HDR value over the disk is E/π for any radiance distribution, so the method does not require a constant L(θ,φ); this should be stated to avoid a misleading derivation.
  2. [Section 3, Algorithm 1] The input label "Illuminace" should be spelled "Illuminance."
  3. [Section 3, Fig. 6] The error-percentage plot would be easier to interpret with error bars or a density overlay; as printed, the color coding by absolute luminance makes it difficult to read the distribution of the 141 data points.
  4. [Title and Abstract] The phrase "from a Single Panorama" is potentially misleading because the pipeline uses paired indoor and outdoor panoramas; please clarify that the indoor panorama is the edited scene input while the outdoor panorama supplies the environment map.
  5. [Section 3, Dataset description] The dataset contribution mentions labels for outdoor illuminance and room orientation, but Section 3 describes only the indoor illuminance and whiteboard luminance protocols; please document how these additional labels were measured.
  6. [References] Reference [13] contains a typo: "Evalution" should be "Evaluation."

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the calibration is validated against an independent luminance meter and the relighting results are not fitted to any target.

full rationale

The paper's central quantitative claim is the 3.988 cd/m² mean absolute luminance error of the low-cost TS-710 calibration against the Konica Minolta LS-160 luminance meter. Equation 1 defines E from L(θ, φ); Equation 2 sets L = E/π under a uniform-radiance model, and the calibration factor k is computed by comparing that L with the average of the orthographic front-lens HDR image (Algorithm 1). The validation point is the whiteboard luminance, which is not used to solve for k, so the error is an independent measurement against a second physical instrument. Transferring k to the outdoor panorama is an explicit same-exposure assumption ('captured simultaneously using the same camera and settings'), not a quantity defined by the fit. The layout and relighting pipeline uses prior components such as [15], [16], and [38], but these are ordinary component citations; no uniqueness theorem or ansatz is smuggled in as the basis of the calibration claim, and the rendered results are generated by Mitsuba rather than fitted to the input. Thus no specific step reduces by construction to its own input.

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

The central calibration result rests on the uniform-radiance approximation and the transfer of the calibration factor to the outdoor panorama. No free parameters are fitted to data; the scaling factor is derived from measured illuminance. The relighting application inherits assumptions from pretrained layout and segmentation models. No new physical entities are postulated.

assumptions (4)
  • domain assumption Uniform radiance over the measured 180 degree hemisphere, so L(θ,φ) = L and illuminance converts to luminance via L = E/π (Eq. 2).
    Used in §3 and Algorithm 1 to derive the calibration factor. Real scenes have non-uniform radiance, but the average error of 3.988 cd/m² is reported against a luminance meter.
  • domain assumption The calibration factor computed from the indoor front-lens image applies to the entire indoor panorama and to the paired outdoor panorama captured with identical camera settings.
    Stated in §3: 'the calibration factor (k) obtained from the indoor measurement can be used to calibrate the corresponding outdoor HDR image'. This assumes the outdoor capture, taken immediately after, sees the same lighting and that one scalar factor corrects the whole equirectangular image.
  • domain assumption Pretrained layout estimation [38] and semantic segmentation [39] accurately predict the 3D room layout and kitchen area from a single panorama.
    Invoked in §4.1 to locate kitchen walls and segment the kitchen area. Errors in these models would misplace virtual objects and affect the rendering quality.
  • domain assumption The room geometry follows the Manhattan world assumption.
    Inherited from the layout estimation method [38] used in §4.1. Kitchens with slanted walls, curved surfaces, or non-Manhattan geometry are not handled.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Digital Kitchen Remodeling: Editing and Relighting Intricate Indoor Scenes from a Single Panorama." pith.science (2026). https://pith.science/paper/BWSC7Y7Y

@misc{pith2026250416086,
  author       = {Pith},
  title        = {Pith review of: Digital Kitchen Remodeling: Editing and Relighting Intricate Indoor Scenes from a Single Panorama},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BWSC7Y7Y}},
  note         = {Machine review of arXiv:2504.16086}
}
read the original abstract

We present a novel virtual staging application for kitchen remodeling from a single panorama. To ensure the realism of the virtual rendered scene, we capture real-world High Dynamic Range (HDR) panoramas and recover the absolute scene radiance for high-quality scene relighting. Our application pipeline consists of three key components: (1) HDR photography for capturing paired indoor and outdoor panoramas, (2) automatic kitchen layout generation with new kitchen components, and (3) an editable rendering pipeline that flexibly edits scene materials and relights the new virtual scene with global illumination. Additionally, we contribute a novel Pano-Pano HDR dataset with 141 paired indoor and outdoor panoramas and present a low-cost photometric calibration method for panoramic HDR photography.

Figures

Figures reproduced from arXiv: 2504.16086 by the authors.

Figure 1
Figure 1. Kitchen Remodeling Application: (Left) A captured indoor panorama showcases an intricate existing [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 4
Figure 4. Using Illuminance Measurement to Calibrate [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
Figure 3
Figure 3. Two Photometric Measurements: (a) A Kon [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (5 more)
Figure 5
Figure 5. Figure 5: Comparison of the luminance values on the whiteboard using the standard approach (Konica Minolta [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 7
Figure 7. Figure 7: Kitchen Layout: (a) Using a single panorama [PITH_FULL_IMAGE:figures/full_fig_p006_7.png]
Figure 8
Figure 8. Figure 8: Object Insertion and Kitchen Layouts: (a) The kitchen walls, highlighted in [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 9
Figure 9. Figure 9: Photo Gallery of Kitchen Remodeling: (left) The captured scenes showcase various existing kitchen [PITH_FULL_IMAGE:figures/full_fig_p009_9.png]
Figure 10
Figure 10. Figure 10: Editing the Existing Scene with New Kitchen Components and Electrical Lighting: (a) Scene captured [PITH_FULL_IMAGE:figures/full_fig_p010_10.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Decoupled Illumination Priors for Spatially Controllable Multi-View Indoor Scene Relighting

    cs.CV 2026-07 conditional novelty 6.0 of 10

    Lume-Palette decouples multi-view indoor relighting into diffusion-based distillation of canonical illumination palettes and casting under receiver-centric 3D lighting maps with asymmetric multi-view conditioning.

Reference graph

Works this paper leans on

42 extracted references · 41 canonical work pages · cited by 1 Pith paper

  1. [1]

    A method for estimating fisheye lens’ field-of view angle and projection for hdr luminance capture

    RJ Abreu Vieira Viula and GJ Hordijk. A method for estimating fisheye lens’ field-of view angle and projection for hdr luminance capture. Pro- ceedings of the 29th CIE SESSION Washington DC, USA, June 14–22, 2019, 2019. 5

    work page 2019

  2. [2]

    Fisheye lenses

    Felix Bettonvil. Fisheye lenses. WGN, Journal of the International Meteor Organization, vol. 33, no. 1, p. 9-14, 33:9–14, 2005. 5

    work page 2005

  3. [3]

    Cube padding for weakly-supervised saliency prediction in 360 videos

    Hsien-Tzu Cheng, Chun-Hung Chao, Jin-Dong Dong, Hao-Kai Wen, Tyng-Luh Liu, and Min Sun. Cube padding for weakly-supervised saliency prediction in 360 videos. InProceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 1420–1429, 2018. 3

    work page 2018

  4. [4]

    Manhat- tan world: Compass direction from a single im- age by bayesian inference

    James M Coughlan and Alan L Yuille. Manhat- tan world: Compass direction from a single im- age by bayesian inference. In Proceedings of the seventh IEEE international conference on com- puter vision , volume 2, pages 941–947. IEEE,

    work page

  5. [5]

    Zillow indoor dataset: Annotated floor plans with 360deg panoramas and 3d room lay- outs

    Steve Cruz, Will Hutchcroft, Yuguang Li, Naji Khosravan, Ivaylo Boyadzhiev, and Sing Bing Kang. Zillow indoor dataset: Annotated floor plans with 360deg panoramas and 3d room lay- outs. In Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recogni- tion, pages 2133–2143, 2021. 2

    work page 2021

  6. [6]

    Everlight: Indoor-outdoor ed- itable hdr lighting estimation

    Mohammad Reza Karimi Dastjerdi, Jonathan Eisenmann, Yannick Hold-Geoffroy, and Jean- Franc ¸ois Lalonde. Everlight: Indoor-outdoor ed- itable hdr lighting estimation. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 7420–7429, 2023. 2, 8

    work page 2023

  7. [7]

    Number of exist- ing homes sold in the united states from 2005 to 2023, with a forecast until 2025

    Statista Research Department. Number of exist- ing homes sold in the united states from 2005 to 2023, with a forecast until 2025. https: / / www . statista . com / statistics / 226144 / us - existing - home - sales/,

    work page 2005

  8. [8]

    Deep parametric indoor light- ing estimation

    Marc-Andr ´e Gardner, Yannick Hold-Geoffroy, Kalyan Sunkavalli, Christian Gagn ´e, and Jean- Franc ¸ois Lalonde. Deep parametric indoor light- ing estimation. In Proceedings of the IEEE/CVF International Conference on Computer Vision , pages 7175–7183, 2019. 2, 3

    work page 2019

Show all 42 references
  1. [9]

    Learning to predict indoor illumination from a single image

    Marc-Andr ´e Gardner, Kalyan Sunkavalli, Ersin Yumer, Xiaohui Shen, Emiliano Gambaretto, Christian Gagn ´e, and Jean-Franc ¸ois Lalonde. Learning to predict indoor illumination from a single image. arXiv preprint arXiv:1704.00090,

    arXiv

  2. [10]

    Panodr: Spherical panorama di- minished reality for indoor scenes

    Vasileios Gkitsas, Vladimiros Sterzentsenko, Nikolaos Zioulis, Georgios Albanis, and Dim- itrios Zarpalas. Panodr: Spherical panorama di- minished reality for indoor scenes. In Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 3716– 37...

    2021

  3. [11]

    Deep lighting environment map estimation from spherical panoramas

    Vasileios Gkitsas, Nikolaos Zioulis, Federico Alvarez, Dimitrios Zarpalas, and Petros Daras. Deep lighting environment map estimation from spherical panoramas. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, pages 640–641,

  4. [12]

    Towards full-to-empty room generation with structure-aware feature encoding and soft semantic region-adaptive normalization

    Vasileios Gkitsas, Nikolaos Zioulis, Vladimiros Sterzentsenko, Alexandros Doumanoglou, and Dimitrios Zarpalas. Towards full-to-empty room generation with structure-aware feature encoding and soft semantic region-adaptive normalization. arXiv preprint arXiv:2112.05396, 2021. 2

    2021 arXiv

  5. [13]

    Evalution of high dynamic range image-based sky models in lighting sim- ulation

    Mehlika Inanici. Evalution of high dynamic range image-based sky models in lighting sim- ulation. Leukos, 7(2):69–84, 2010. 4

    2010

  6. [14]

    Mitsuba 3 renderer,

    Wenzel Jakob, S ´ebastien Speierer, Nicolas Rous- sel, Merlin Nimier-David, Delio Vicini, Tizian Zeltner, Baptiste Nicolet, Miguel Crespo, Vin- cent Leroy, and Ziyi Zhang. Mitsuba 3 renderer,

  7. [15]

    Virtual home staging: Inverse ren- dering and editing an indoor panorama under nat- ural illumination

    Guanzhou Ji, Azadeh O Sawyer, and Srinivasa G Narasimhan. Virtual home staging: Inverse ren- dering and editing an indoor panorama under nat- ural illumination. In International Symposium on Visual Computing, pages 329–342. Springer,

  8. [16]

    Virtual home staging and relighting from a single panorama under natural illumina- tion

    Guanzhou Ji, Azadeh O Sawyer, and Srinivasa G Narasimhan. Virtual home staging and relighting from a single panorama under natural illumina- tion. Machine Vision and Applications, 35(4):98,

  9. [17]

    Rendering synthetic objects into legacy photographs

    Kevin Karsch, Varsha Hedau, David Forsyth, and Derek Hoiem. Rendering synthetic objects into legacy photographs. ACM Transactions on graphics (TOG), 30(6):1–12, 2011. 3

    2011

  10. [18]

    Hdrscope: high dynamic range im- age processing toolkit for lighting simulations and analysis

    Viswanathan Kumaragurubaran and Mehlika Inanici. Hdrscope: high dynamic range im- age processing toolkit for lighting simulations and analysis. In Proceedings of the BS2013: 13th Conference of International Building Per- formance Simulation Association , pages 25–28,

  11. [19]

    Deeplight: Learning illumina- tion for unconstrained mobile mixed reality

    Chloe LeGendre, Wan-Chun Ma, Graham Fyffe, John Flynn, Laurent Charbonnel, Jay Busch, and Paul Debevec. Deeplight: Learning illumina- tion for unconstrained mobile mixed reality. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 5918–5...

    2019

  12. [20]

    Lighting, reflectance and geometry estimation from 360 panoramic stereo

    Junxuan Li, Hongdong Li, and Yasuyuki Mat- sushita. Lighting, reflectance and geometry estimation from 360 panoramic stereo. In 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 10586– 10595. IEEE, 2021. 3

    2021

  13. [21]

    Inverse rendering for complex in- door scenes: Shape, spatially-varying lighting and svbrdf from a single image

    Zhengqin Li, Mohammad Shafiei, Ravi Ra- mamoorthi, Kalyan Sunkavalli, and Manmohan Chandraker. Inverse rendering for complex in- door scenes: Shape, spatially-varying lighting and svbrdf from a single image. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pa...

  14. [22]

    Physically-based editing of indoor scene lighting from a single image

    Zhengqin Li, Jia Shi, Sai Bi, Rui Zhu, Kalyan Sunkavalli, Milo ˇs Ha ˇsan, Zexiang Xu, Ravi Ramamoorthi, and Manmohan Chandraker. Physically-based editing of indoor scene lighting from a single image. In European Conference on Computer Vision , pages 555–572. Springer,

  15. [23]

    Multi-view inverse ren- dering for large-scale real-world indoor scenes

    Zhen Li, Lingli Wang, Mofang Cheng, Cihui Pan, and Jiaqi Yang. Multi-view inverse ren- dering for large-scale real-world indoor scenes. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 12499–12509, 2023. 2

    2023

  16. [24]

    Ra- diometric self calibration

    Tomoo Mitsunaga and Shree K Nayar. Ra- diometric self calibration. In Proceedings. 1999 IEEE computer society conference on com- puter vision and pattern recognition (Cat. No PR00149), volume 1, pages 374–380. IEEE,

    1999

  17. [25]

    Light Sources — SPD Curves — National Gallery, London — Informa- tion

    Joseph Padfield. Light Sources — SPD Curves — National Gallery, London — Informa- tion. https://research.ng- london. org.uk/scientific/spd/?page=home,

  18. [26]

    Free-viewpoint indoor neural relighting from multi-view stereo

    Julien Philip, S ´ebastien Morgenthaler, Micha ¨el Gharbi, and George Drettakis. Free-viewpoint indoor neural relighting from multi-view stereo. ACM Transactions on Graphics (TOG), 40(5):1– 18, 2021. 2

    2021

  19. [27]

    Omnilayout: Room lay- out reconstruction from indoor spherical panora- mas

    Shivansh Rao, Vikas Kumar, Daniel Kifer, C Lee Giles, and Ankur Mali. Omnilayout: Room lay- out reconstruction from indoor spherical panora- mas. In Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recogni- tion, pages 3706–3715, 2021. 3

    2021

  20. [28]

    An empty room is all we want: Automatic defur- nishing of indoor panoramas

    Mira Slavcheva, Dave Gausebeck, Kevin Chen, David Buchhofer, Azwad Sabik, Chen Ma, Sachal Dhillon, Olaf Brandt, and Alan Dolhasz. An empty room is all we want: Automatic defur- nishing of indoor panoramas. InIEEE/CVF Con- ference on Computer Vision and Pattern Recog- nition Wo...

    2024

  21. [29]

    360-dfpe: Leveraging monocular 360-layouts for direct floor plan estimation

    Bolivar Solarte, Yueh-Cheng Liu, Chin-Hsuan Wu, Yi-Hsuan Tsai, and Min Sun. 360-dfpe: Leveraging monocular 360-layouts for direct floor plan estimation. IEEE Robotics and Au- tomation Letters, 7(3):6503–6510, 2022. 3

    2022

  22. [30]

    Led2-net: Monocular 360deg layout estimation via differ- entiable depth rendering

    Fu-En Wang, Yu-Hsuan Yeh, Min Sun, Wei- Chen Chiu, and Yi-Hsuan Tsai. Led2-net: Monocular 360deg layout estimation via differ- entiable depth rendering. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 12956–12965, 2021. 3

    2021

  23. [31]

    Psmnet: Position-aware stereo merging network for room layout estima- tion

    Haiyan Wang, Will Hutchcroft, Yuguang Li, Zhiqiang Wan, Ivaylo Boyadzhiev, Yingli Tian, and Sing Bing Kang. Psmnet: Position-aware stereo merging network for room layout estima- tion. In Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recogni- tion, pag...

    2022

  24. [32]

    Factorized inverse path tracing for efficient and accurate material-lighting esti- mation

    Liwen Wu, Rui Zhu, Mustafa B Yaldiz, Yinhao Zhu, Hong Cai, Janarbek Matai, Fatih Porikli, Tzu-Mao Li, Manmohan Chandraker, and Ravi Ramamoorthi. Factorized inverse path tracing for efficient and accurate material-lighting esti- mation. In Proceedings of the IEEE/CVF Inter- nat...

    2023

  25. [33]

    Recognizing scene view- point using panoramic place representation

    Jianxiong Xiao, Krista A Ehinger, Aude Oliva, and Antonio Torralba. Recognizing scene view- point using panoramic place representation. In 2012 IEEE Conference on Computer Vision and Pattern Recognition , pages 2695–2702. IEEE,

    2012

  26. [34]

    Photoscene: Photorealistic material and lighting transfer for indoor scenes

    Yu-Ying Yeh, Zhengqin Li, Yannick Hold- Geoffroy, Rui Zhu, Zexiang Xu, Milo ˇs Ha ˇsan, Kalyan Sunkavalli, and Manmohan Chandraker. Photoscene: Photorealistic material and lighting transfer for indoor scenes. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pat...

    2022

  27. [35]

    Emptying, refurnishing, and relighting in- door spaces

    Edward Zhang, Michael F Cohen, and Brian Cur- less. Emptying, refurnishing, and relighting in- door spaces. ACM Transactions on Graphics (TOG), 35(6):1–14, 2016. 3

    2016

  28. [36]

    Panocontext: A whole-room 3d con- text model for panoramic scene understanding

    Yinda Zhang, Shuran Song, Ping Tan, and Jianx- iong Xiao. Panocontext: A whole-room 3d con- text model for panoramic scene understanding. In Computer Vision–ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6- 12, 2014, Proceedings, Part VI 13 , pages 668–

    2014

  29. [37]

    3d room layout estimation from a cubemap of panorama image via deep manhattan hough transform

    Yining Zhao, Chao Wen, Zhou Xue, and Yue Gao. 3d room layout estimation from a cubemap of panorama image via deep manhattan hough transform. In European conference on computer vision, pages 637–654. Springer, 2022. 3

    2022

  30. [38]

    Semantically supervised appear- ance decomposition for virtual staging from a single panorama

    Tiancheng Zhi, Bowei Chen, Ivaylo Boyadzhiev, Sing Bing Kang, Martial Hebert, and Srinivasa G Narasimhan. Semantically supervised appear- ance decomposition for virtual staging from a single panorama. ACM Transactions on Graph- ics (TOG), 41(4):1–15, 2022. 2, 6

    2022

  31. [39]

    Semantic understanding of scenes through the ade20k dataset

    Bolei Zhou, Hang Zhao, Xavier Puig, Tete Xiao, Sanja Fidler, Adela Barriuso, and Antonio Tor- ralba. Semantic understanding of scenes through the ade20k dataset. International Journal on Computer Vision, 2018. 6

    2018

  32. [40]

    Creation of 3d home tours soared in march as stay-at-home orders expanded

    Zillow. Creation of 3d home tours soared in march as stay-at-home orders expanded. https://www.zillow.com/research/ 3d - home - tours - coronavirus - 26794/, 2020. 2

    2020

  33. [41]

    Single-shot cuboids: Geodesics-based end-to-end manhattan aligned layout estimation from spherical panoramas

    Nikolaos Zioulis, Federico Alvarez, Dimitrios Zarpalas, and Petros Daras. Single-shot cuboids: Geodesics-based end-to-end manhattan aligned layout estimation from spherical panoramas. Im- age and Vision Computing, 110:104160, 2021. 3

    2021

  34. [2022]

    https://mitsuba-renderer.org. 7

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.