pith. sign in

arxiv: 2604.18039 · v2 · submitted 2026-04-20 · 💻 cs.HC

HolmeSketcher: Generative 3D Sketch Mapping for Spatial Reconstruction in Crime Scene Investigation

Tianyi Xiao , Yizi Chen , Sidi Wu , Peter Kiefer , Yan Feng , Martin Raubal This is my paper

Pith reviewed 2026-05-10 04:20 UTC · model grok-4.3

classification 💻 cs.HC
keywords 3D sketch mappingcrime scene investigationextended realitygenerative modelingspatial accuracyuser studyHCIdeep learning
0
0 comments X

The pith

HolmeSketcher shows that generative 3D sketch mapping can improve spatial accuracy in crime scene reconstructions compared to 2D paper sketches.

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

The authors developed HolmeSketcher, a system with a 3D drawing interface and a deep learning backend that generates 3D objects and reconstructs scenes in extended reality for crime scene investigation. In a study with 15 participants, the system produced reconstructions with better spatial accuracy and interpretability than traditional paper-based 2D sketching. The trade-off was that users experienced higher task load and rated the system as less usable overall. Drawing on these results and interviews with three experts, the paper outlines three design implications for future 3D sketch mapping tools.

Core claim

By combining a front-end 3D drawing interface with a back-end deep learning pipeline, HolmeSketcher supports object generation and scene reconstruction in extended reality, leading to improved spatial accuracy and interpretability of crime scene models in user testing.

What carries the argument

The HolmeSketcher system, which integrates 3D sketching in extended reality with deep learning for generative object creation and scene assembly.

If this is right

  • Reconstructed crime scenes can capture three-dimensional spatial relationships more faithfully than flat drawings.
  • Investigators may communicate and analyze spatial evidence with greater clarity using XR-based tools.
  • Future CSI tools should address usability issues to make 3D sketching practical in field conditions.
  • Design guidelines can guide development of similar generative systems for other spatial documentation tasks.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Hybrid approaches mixing 2D sketches for quick notes with 3D for detailed reconstruction could minimize workload while retaining accuracy gains.
  • Improving the deep learning models to reduce artifacts might close the usability gap observed in the study.
  • Testing the system in real crime scenes rather than controlled studies would reveal practical barriers like time pressure and environmental factors.

Load-bearing premise

The deep learning pipeline reliably converts user-drawn 3D sketches into accurate three-dimensional object models without adding significant spatial errors.

What would settle it

Conducting a controlled experiment where multiple users reconstruct the same physical scene using both HolmeSketcher and paper methods, then measuring the deviation of the digital models from ground-truth measurements of object positions and orientations.

Figures

Figures reproduced from arXiv: 2604.18039 by Martin Raubal, Peter Kiefer, Sidi Wu, Tianyi Xiao, Yan Feng, Yizi Chen.

Figure 1
Figure 1. Figure 1: HolmeSketcher is a generative 3D sketch mapping tool for crime scene reconstruction that supports the externalization of spatial memory during investigative interviews. The workflow includes (a) 3D sketching, (b) an AI-based sketch-to-object pipeline, (c) material generation from voice input, (d) appearance adjustment, (e) object storage and retrieval for scene composition, and (f) object manipulation thro… view at source ↗
Figure 2
Figure 2. Figure 2: Example sketch maps for (a) scene documentation [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Interface of HolmeSketcher. Users sketch in the [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Sketch mapping workflow. (a) Users first align the workspace with the physical environment, then generate objects [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: The study procedure. Each participant completed both conditions with different stimuli. In C2, we disabled object [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: The design of the Stimuli, as well as examples of 2D and 3D sketch maps collected in the user study. The 2D sketch [PITH_FULL_IMAGE:figures/full_fig_p008_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Bar plots of user study results. Error bars indicate 95% confidence intervals. [PITH_FULL_IMAGE:figures/full_fig_p009_7.png] view at source ↗
read the original abstract

Sketch mapping is widely used in crime scene investigation (CSI) to document, interpret, and communicate spatial information. However, it is typically performed on 2D media, which limits its ability to represent 3D spatial relationships. We present HolmeSketcher, a generative 3D sketch mapping system that combines a front-end 3D drawing interface with a back-end deep learning pipeline to support object generation and scene reconstruction in extended reality. In a within-subject user study (N = 15), HolmeSketcher improved the spatial accuracy and interpretability of reconstructed scenes, but with a clear trade-off of higher task load and lower usability compared with paper-based 2D sketch mapping. By integrating findings from the user study and expert interviews (N = 3), we further derive three design implications for next-generation 3D sketch mapping tools for CSI.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper presents HolmeSketcher, a system combining a front-end 3D drawing interface in extended reality with a back-end deep learning pipeline for generating 3D objects and reconstructing scenes from user sketches. It is positioned for crime scene investigation (CSI) to overcome limitations of traditional 2D sketch mapping. A within-subject user study (N=15) is reported to show gains in spatial accuracy and interpretability of reconstructions versus paper-based 2D methods, accompanied by higher task load and lower usability; expert interviews (N=3) yield three design implications for future 3D CSI tools.

Significance. If the user-study outcomes prove robust and the generative pipeline is shown to be faithful, the work would offer a concrete demonstration of XR+generative-AI integration for a high-stakes applied domain, potentially informing both HCI tool design and forensic documentation practices. The empirical focus on trade-offs (accuracy vs. workload) and the derivation of design implications are constructive contributions.

major comments (2)
  1. [Evaluation / user-study section] Evaluation / user-study section: The central claim that HolmeSketcher improves spatial accuracy rests on the N=15 within-subject comparison, yet the manuscript provides no quantitative validation (reconstruction error, IoU, positional deviation, or similar) of the back-end deep-learning pipeline's fidelity in converting sketches to 3D objects and layouts. Without such metrics, it is impossible to determine whether observed accuracy gains reflect genuine 3D spatial benefits or are constrained by model artifacts.
  2. [User-study description] User-study description: The abstract and results narrative state positive outcomes for spatial accuracy and interpretability but omit details on the precise accuracy measures employed, statistical tests performed, effect sizes, or controls for confounds (e.g., learning effects, interface familiarity). This absence prevents independent verification of the reported improvements.
minor comments (2)
  1. [System description] The manuscript would benefit from a dedicated subsection or table summarizing the deep-learning architecture, training data, and inference pipeline, including any hyper-parameters or loss functions used.
  2. [Results figures] Figure captions and axis labels in the results section should explicitly state the dependent variables and units used for spatial-accuracy and task-load metrics.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback, which highlights important opportunities to strengthen the evaluation of the generative pipeline and the transparency of the user study reporting. We will revise the manuscript accordingly to address these concerns while preserving the core contributions of the work.

read point-by-point responses

  1. Referee: [Evaluation / user-study section] Evaluation / user-study section: The central claim that HolmeSketcher improves spatial accuracy rests on the N=15 within-subject comparison, yet the manuscript provides no quantitative validation (reconstruction error, IoU, positional deviation, or similar) of the back-end deep-learning pipeline's fidelity in converting sketches to 3D objects and layouts. Without such metrics, it is impossible to determine whether observed accuracy gains reflect genuine 3D spatial benefits or are constrained by model artifacts.

    Authors: We agree that isolated quantitative validation of the back-end pipeline would help isolate its contribution from end-to-end system effects. The reported user study measured spatial accuracy via end-to-end scene reconstructions (e.g., object placement alignment with ground-truth layouts as rated and measured in the XR environment), but did not report separate fidelity metrics for the generative model. In the revised manuscript we will add a dedicated pipeline evaluation subsection reporting reconstruction error, IoU scores, and positional deviation on a held-out test set of sketches to clarify whether accuracy gains are limited by model artifacts. revision: yes

  2. Referee: [User-study description] User-study description: The abstract and results narrative state positive outcomes for spatial accuracy and interpretability but omit details on the precise accuracy measures employed, statistical tests performed, effect sizes, or controls for confounds (e.g., learning effects, interface familiarity). This absence prevents independent verification of the reported improvements.

    Authors: We acknowledge that the current description of the user study is insufficiently detailed for full reproducibility. The manuscript reports the within-subject design with N=15 and overall outcomes, but we will expand the Methods and Results sections to explicitly define the spatial accuracy measures (e.g., 3D coordinate deviation and alignment scores), the statistical tests applied (paired t-tests or non-parametric equivalents), effect sizes, and controls such as counterbalancing of condition order and a pre-study interface familiarization phase to address learning effects and familiarity confounds. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical user study with no derivations or fitted predictions

full rationale

The paper presents HolmeSketcher as a system combining a 3D drawing interface with a deep learning back-end, then reports direct empirical results from a within-subject user study (N=15) and expert interviews (N=3). Claims of improved spatial accuracy and interpretability are grounded in measured study outcomes rather than any derivation chain. No equations, parameter fitting, self-referential predictions, or load-bearing self-citations appear; the design implications are synthesized from the collected data. The work is self-contained as a standard HCI system evaluation against external benchmarks (user performance metrics), with no reduction of results to inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claims rest on standard HCI assumptions about user study validity and the unstated performance of the deep learning component; no free parameters or invented entities are evident from the abstract.

axioms (1)
  • domain assumption The N=15 within-subject study participants provide a valid basis for comparing 2D and 3D sketching methods in CSI contexts
    Assumes participants represent target users and that order effects or learning do not bias results.

pith-pipeline@v0.9.0 · 5464 in / 1355 out tokens · 63720 ms · 2026-05-10T04:20:13.800203+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

98 extracted references · 98 canonical work pages

  1. [1]

    Meshal Albeedan, Hoshang Kolivand, Ramy Hammady, and Tanzila Saba. 2023. Seamless crime scene reconstruction in mixed reality for investigation training: a design and evaluation study.IEEE Transactions on Learning Technologies17 (2023), 856–873

    work page 2023

  2. [2]

    Flora Amato, Aniello Castiglione, Giovanni Cozzolino, and Fabio Narducci. 2020. A semantic-based methodology for digital forensics analysis.J. Parallel and Distrib. Comput.138 (2020), 172–177

    work page 2020

  3. [3]

    Oğuz Arslan, Artun Akdoğan, and Mustafa Doga Dogan. 2025. TinkerXR: In-Situ, Reality-Aware CAD and 3D Printing Interface for Novices. InProceedings of the ACM Symposium on Computational Fabrication. 1–19. doi:10.1145/3745778. 3766651

    work page doi:10.1145/3745778 2025

  4. [4]

    Bainbridge, Rebecca Chamberlain, Jeffrey Wammes, and Judith E

    Wilma A. Bainbridge, Rebecca Chamberlain, Jeffrey Wammes, and Judith E. Fan

    work page

  5. [5]

    doi:10.3758/s13421-024-01618-4

    Drawing as a Means to Characterize Memory and Cognition.Memory & Cognition53, 1 (2025), 1–5. doi:10.3758/s13421-024-01618-4

    work page doi:10.3758/s13421-024-01618-4 2025

  6. [6]

    Hmrishav Bandyopadhyay, Subhadeep Koley, Ayan Das, Ayan Kumar Bhunia, Aneeshan Sain, Pinaki Nath Chowdhury, Tao Xiang, and Yi-Zhe Song. 2024. Doodle your 3D: From abstract freehand sketches to precise 3D shapes. InCVPR

    work page 2024

  7. [7]

    Barlow, Richard P

    Claire M. Barlow, Richard P. Jolley, and Jenny L. Hallam. 2011. Drawings as Memory Aids: Optimising the Drawing Method to Facilitate Young Children’s Recall.Applied Cognitive Psychology25, 3 (2011), 480–487. doi:10.1002/acp.1716

    work page doi:10.1002/acp.1716 2011

  8. [8]

    Michael J Bolliger, Ursula Buck, Michael J Thali, and Stephan A Bolliger. 2012. Reconstruction and 3D visualisation based on objective real 3D based documen- tation.Forensic science, medicine, and pathology8, 3 (2012), 208–217

    work page 2012

  9. [9]

    Neil Brewer, Sophie Harvey, and Carolyn Semmler. 2004. Improving Compre- hension of Jury Instructions with Audio-Visual Presentation.Applied Cognitive Psychology18, 6 (2004), 765–776. doi:10.1002/acp.1036

    work page doi:10.1002/acp.1036 2004

  10. [10]

    Williams

    Neil Brewer and Kipling D. Williams. 2007.Psychology and Law: An Empirical Perspective. Guilford Press

    work page 2007

  11. [11]

    John Brooke et al. 1996. SUS-A quick and dirty usability scale.Usability evaluation in industry189, 194 (1996), 4–7

    work page 1996

  12. [12]

    Tilt Brush. 2016. Tilt Brush. https://www.tiltbrush.com/

    work page 2016

  13. [13]

    2014.Investigative Interviewing

    Ray Bull. 2014.Investigative Interviewing. Springer Science & Business Media

    work page 2014

  14. [14]

    Carew and David Errickson

    Rachael M. Carew and David Errickson. 2019. Imaging in Forensic Science: Five Years On.Journal of Forensic Radiology and Imaging16 (2019), 24–33. doi:10.1016/j.jofri.2019.01.002

    work page doi:10.1016/j.jofri.2019.01.002 2019

  15. [15]

    Rachael M Carew, James French, and Ruth M Morgan. 2021. 3D forensic science: A new field integrating 3D imaging and 3D printing in crime reconstruction. Forensic Science International: Synergy3 (2021), 100205

    work page 2021

  16. [16]

    Tianrun Chen, Chaotao Ding, Shangzhan Zhang, Chunan Yu, Ying Zang, Zejian Li, Sida Peng, and Lingyun Sun. 2024. Rapid 3D model generation with intuitive 3D input. InCVPR

    work page 2024

  17. [17]

    Tianrun Chen, Chenglong Fu, Ying Zang, Lanyun Zhu, Jia Zhang, Papa Mao, and Lingyun Sun. 2023. Deep3dsketch+: Rapid 3D modeling from single free-hand sketches. InInternational Conference on Multimedia Modeling. Springer

    work page 2023

  18. [18]

    Tianrun Chen, Chenglong Fu, Lanyun Zhu, Papa Mao, Jia Zhang, Ying Zang, and Lingyun Sun. 2023. Deep3dsketch: 3D modeling from free-hand sketches with view-and structural-aware adversarial training.arXiv:2312.04435(2023)

    work page arXiv 2023

  19. [19]

    Yizi Chen, Sidi Wu, Tianyi Xiao, Nina Wiedemann, and Loic Landrieu. 2025. Order Matters: 3D Shape Generation from Sequential VR Sketches.arXiv preprint arXiv:2512.04761(2025)

    work page arXiv 2025

  20. [20]

    Ari Cover, Ricardo Deitoz Posser, Joao Pedro Assuncao Campos, and Rafael Rieder

    work page

  21. [21]

    InSymposium on Virtual and Augmented Reality

    Methodology of communication between a criminal database and a virtual reality environment for forensic study. InSymposium on Virtual and Augmented Reality. 215–222

    work page

  22. [22]

    Coral Dando, Fiona Gabbert, and Lorraine Hope. 2020. Supporting Older Eye- witnesses’ Episodic Memory: The Self-Administered Interview and Sketch Rein- statement of Context.Memory28, 6 (2020), 712–723. doi:10.1080/09658211.2020. 1757718

    work page doi:10.1080/09658211.2020 2020

  23. [23]

    Dando, R

    C. Dando, R. Wilcock, Becky Milne, and L. Henry. 2009. A Modified Cogni- tive Interview Procedure for Frontline Police Investigators.Applied Cognitive Psychology23, 5 (2009), 698–716. doi:10.1002/acp.1501

    work page doi:10.1002/acp.1501 2009

  24. [24]

    Haneen Deeb, Aldert Vrij, Sharon Leal, and Jennifer Burkhardt. 2021. The Effects of Sketching While Narrating on Information Elicitation and Deception Detection in Multiple Interviews.Acta Psychologica213 (2021), 103236. doi:10.1016/j.actpsy. 2020.103236

    work page doi:10.1016/j.actpsy 2021

  25. [25]

    Johanna Delanoy, Mathieu Aubry, Phillip Isola, Alexei A Efros, and Adrien Bousseau. 2018. 3D sketching using multi-view deep volumetric prediction. Proceedings of the ACM on Computer Graphics and Interactive Techniques(2018)

    work page 2018

  26. [26]

    Tobias Drey, Jan Gugenheimer, Julian Karlbauer, Maximilian Milo, and Enrico Rukzio. 2020. VRSketchIn: Exploring the Design Space of Pen and Tablet Interac- tion for 3D Sketching in Virtual Reality. InProceedings of the 2020 CHI Conference on Human Factors in Computing Systems (CHI ’20). 1–14. doi:10.1145/3313831. 3376628

    work page doi:10.1145/3313831 2020

  27. [27]

    2020.An introduction to crime scene investigation

    Aric W Dutelle. 2020.An introduction to crime scene investigation. Jones & Bartlett Learning

    work page 2020

  28. [28]

    Fan, Wilma A

    Judith E. Fan, Wilma A. Bainbridge, Rebecca Chamberlain, and Jeffrey D. Wammes

    work page

  29. [29]

    doi:10.1038/s44159-023-00212-w

    Drawing as a Versatile Cognitive Tool.Nature Reviews Psychology2, 9 (2023), 556–568. doi:10.1038/s44159-023-00212-w

    work page doi:10.1038/s44159-023-00212-w 2023

  30. [30]

    FARO. 2017. Tutorials for Crime Zone and Crash Zone. https: //knowledge.faro.com/Software/Legacy-Software/Legacy-CAD_Zone/Legacy- Crime_and_Crash_Zone/Tutorials_for_Crime_Zone_and_Crash_Zone

    work page 2017

  31. [31]

    Barry A. J. Fisher. 2003.Techniques of Crime Scene Investigation(7 ed.). CRC Press, Boca Raton. doi:10.1201/9781420058192

    work page doi:10.1201/9781420058192 2003

  32. [32]

    Gardner and Donna R

    Ross M. Gardner and Donna R. Krouskup. 2018. Crime Scene Sketching and Mapping. InPractical Crime Scene Processing and Investigation, Third Edition(3 ed.). CRC Press

    work page 2018

  33. [33]

    Maha Ghanem and Haidy M. Megahed. 2021. Crime Scene Processing: Docu- mentation and Evaluation. InCrime Scene Management within Forensic Science, Jaskaran Singh and Neeta Raj Sharma (Eds.). Springer Nature, Singapore, 15–35. doi:10.1007/978-981-16-4091-9_2

    work page doi:10.1007/978-981-16-4091-9_2 2021

  34. [34]

    Erik Mac Giolla, Pär Anders Granhag, and Zarah Vernham. 2017. Drawing-Based Deception Detection Techniques: A State-of-the-Art Review.Crime Psychology Review3, 1 (2017), 23–38. doi:10.1080/23744006.2017.1393986

    work page doi:10.1080/23744006.2017.1393986 2017

  35. [35]

    Songen Gu, Haoxuan Song, Binjie Liu, Qian Yu, Sanyi Zhang, Haiyong Jiang, Jin Huang, and Feng Tian. 2025. VRsketch2Gaussian: 3D VR Sketch Guided 3D Object Generation with Gaussian Splatting.arXiv:2503.12383(2025)

    work page arXiv 2025

  36. [36]

    Benoit Guillard, Edoardo Remelli, Pierre Yvernay, and Pascal Fua. 2021. Sketch2mesh: Reconstructing and editing 3D shapes from sketches. InICCV

    work page 2021

  37. [37]

    David Harris, Mark Wilson, and Samuel Vine. 2020. Development and validation of a simulation workload measure: the simulation task load index (SIM-TLX). Virtual Reality24, 4 (2020), 557–566

    work page 2020

  38. [38]

    Hendrik Haupt. 2026. Enviro 3 - Sky and Weather. https://assetstore.unity.com/ packages/tools/particles-effects/enviro-3-sky-and-weather-236601

    work page 2026

  39. [39]

    Takeo Igarashi, Satoshi Matsuoka, and Hidehiko Tanaka. 2006. Teddy: a sketching interface for 3D freeform design. InACM SIGGRAPH 2006 Courses. 11–es

    work page 2006

  40. [40]

    Richard M Karp, Umesh V Vazirani, and Vijay V Vazirani. 1990. An optimal algorithm for on-line bipartite matching. InProceedings of the twenty-second annual ACM symposium on Theory of computing. 352–358

    work page 1990

  41. [41]

    Olga A Karpenko and John F Hughes. 2006. Smoothsketch: 3d free-form shapes from complex sketches. InACM SIGGRAPH 2006 Papers. 589–598

    work page 2006

  42. [42]

    Rubaiat Habib Kazi, Tovi Grossman, Hyunmin Cheong, Ali Hashemi, and George W Fitzmaurice. 2017. DreamSketch: Early Stage 3D Design Explorations with Sketching and Generative Design.. InUIST, Vol. 14. 401–414

    work page 2017

  43. [43]

    Florian Kern, Peter Kullmann, Elisabeth Ganal, Kristof Korwisi, René Stingl, Florian Niebling, and Marc Erich Latoschik. 2021. Off-The-Shelf Stylus: Using XR Devices for Handwriting and Sketching on Physically Aligned Virtual Surfaces. Conference’17, July 2017, Washington, DC, USA Xiao, et al. Frontiers in Virtual Reality2 (2021), 1–20

    work page 2021

  44. [44]

    Donghyun Kim, Subin Oh, and Taeshik Shon. 2023. Digital forensic approaches for metaverse ecosystems.Forensic Science International: Digital Investigation46 (2023), 301608

    work page 2023

  45. [45]

    Kevin Gonyop Kim, Jakub Krukar, Panagiotis Mavros, Jiayan Zhao, Peter Kiefer, Angela Schwering, Christoph Hölscher, and Martin Raubal. 2022. 3D Sketch Maps: Concept, Potential Benefits, and Challenges. In15th International Conference on Spatial Information Theory (COSIT 2022), Vol. 240. Schloss Dagstuhl, Leibniz- Zentrum für Informatik, 14

    work page 2022

  46. [46]

    Yongkwan Kim, Kyuhyoung Hong, and Junwon Yang. 2023. Feather: 3D sketch- book light as a feather. InACM SIGGRAPH 2023 Appy Hour. 1–2

    work page 2023

  47. [47]

    Juil Koo, Seungwoo Yoo, Minh Hieu Nguyen, and Minhyuk Sung. 2023. Salad: Part-level latent diffusion for 3d shape generation and manipulation. (2023), 14441–14451

    work page 2023

  48. [48]

    Bettina Laugwitz, Theo Held, and Martin Schrepp. 2008. Construction and evaluation of a user experience questionnaire. InHCI and Usability for Education and Work: 4th Symposium of the Workgroup Human-Computer Interaction and Usability Engineering of the Austrian Computer Society, USAB 2008, Graz, Austria, November 20-21, 2008. Proceedings 4. Springer, 63–76

    work page 2008

  49. [49]

    Fisher, Aldert Vrij, Sharon Leal, and Samantha Mann

    Drew Leins, Ronald P. Fisher, Aldert Vrij, Sharon Leal, and Samantha Mann. 2011. Using Sketch Drawing to Induce Inconsistency in Liars.Legal and Criminological Psychology16, 2 (2011), 253–265. doi:10.1348/135532510X501775

    work page doi:10.1348/135532510x501775 2011

  50. [50]

    Vivian Liu. 2023. Beyond text-to-image: Multimodal prompts to explore gener- ative AI. InExtended Abstracts of the 2023 CHI Conference on Human Factors in Computing Systems. 1–6

    work page 2023

  51. [51]

    Zhaoliang Lun, Matheus Gadelha, Evangelos Kalogerakis, Subhransu Maji, and Rui Wang. 2017. 3D shape reconstruction from sketches via multi-view convolu- tional networks. In3DV. IEEE

    work page 2017

  52. [52]

    Ling Luo, Pinaki Nath Chowdhury, Tao Xiang, Yi-Zhe Song, and Yulia Gryadit- skaya. 2023. 3D VR sketch guided 3D shape prototyping and exploration. In ICCV

    work page 2023

  53. [53]

    Ling Luo, Yulia Gryaditskaya, Yongxin Yang, Tao Xiang, and Yi-Zhe Song. 2021. Fine-grained VR sketching: Dataset and insights. In3DV

    work page 2021

  54. [54]

    Zhongjin Luo, Jie Zhou, Heming Zhu, Dong Du, Xiaoguang Han, and Hongbo Fu. 2021. SimpModeling: Sketching implicit field to guide mesh modeling for 3d animalmorphic head design. InThe 34th annual ACM symposium on user interface software and technology. 854–863

    work page 2021

  55. [55]

    Kirk Luther, Brent Snook, Joseph Eastwood, and Ronald P Fisher. 2023. Sketching: the effect of a dual-modality technique on recall performance.Journal of Police and Criminal Psychology38, 2 (2023), 469–482

    work page 2023

  56. [56]

    Kirk Luther, Brent Snook, Joseph Eastwood, and Ronald P. Fisher. 2023. Sketching: The Effect of a Dual-Modality Technique on Recall Performance.Journal of Police and Criminal Psychology38, 2 (2023), 469–482. doi:10.1007/s11896-022-09544-4

    work page doi:10.1007/s11896-022-09544-4 2023

  57. [57]

    Wei Ma, Xiangyu Wang, Jun Wang, Xiaolei Xiang, and Junbo Sun. 2021. Gener- ative design in building information modelling (BIM): approaches and require- ments.Sensors21, 16 (2021), 5439

    work page 2021

  58. [58]

    Matej Vanco. 2023. Sculpting Pro. https://assetstore.unity.com/packages/tools/ modeling/sculpting-pro-201873

    work page 2023

  59. [59]

    Richard Mayne and Helen Green. 2020. Virtual reality for teaching and learning in crime scene investigation.Science & Justice60, 5 (2020), 466–472

    work page 2020

  60. [60]

    2023.Meta Quest Pro: Premium Mixed Reality

    Meta. 2023.Meta Quest Pro: Premium Mixed Reality. https://www.meta.com/ch/ en/quest/quest-pro/

    work page 2023

  61. [61]

    Meta. 2026. Meta XR Interaction SDK. https://assetstore.unity.com/packages/ tools/integration/meta-xr-interaction-sdk-265014

    work page 2026

  62. [62]

    Microsoft. 2025. Microsoft Visio: Diagramming & Flowcharts. https://www. microsoft.com/en-us/microsoft-365/visio/flowchart-software

    work page 2025

  63. [63]

    Paul Milgram and Fumio Kishino. 1994. A Taxonomy of Mixed Reality Visual Displays.IEICE Transactions on InformationE77-D, 12 (1994), 1321–1329

    work page 1994

  64. [64]

    Miller and Peter Massey

    Marilyn T. Miller and Peter Massey. 2018.The Crime Scene: A Visual Guide. Academic Press

    work page 2018

  65. [65]

    Moodkie. 2026. Easy Save 3 Documentation - The Complete Save and Load Solution for Unity. https://docs.moodkie.com/product/easy-save-3/

    work page 2026

  66. [66]

    OpenAI. 2025. Introducing 4o Image Generation. https://openai.com/index/ introducing-4o-image-generation/

    work page 2025

  67. [67]

    OpenAI. 2026. Whisper. https://github.com/openai/whisper

    work page 2026

  68. [68]

    Jamie K Pringle, Ian G Stimpson, Adam J Jeffery, Kristopher D Wisniewski, Timothy Grossey, Luke Hobson, Vivienne Heaton, Vladimir Zholobenko, and Steven L Rogers. 2022. Extended reality (XR) virtual practical and educational eGaming to provide effective immersive environments for learning and teaching in forensic science.Science & Justice62, 6 (2022), 696–707

    work page 2022

  69. [69]

    Dimitar Rangelov, Sierd Waanders, Kars Waanders, Maurice van Keulen, and Radoslav Miltchev. 2024. Impact of camera settings on 3D Reconstruction quality: Insights from NeRF and Gaussian Splatting.Sensors24, 23 (2024), 7594

    work page 2024

  70. [70]

    Hamid Rezatofighi, Nathan Tsoi, JunYoung Gwak, Amir Sadeghian, Ian Reid, and Silvio Savarese. 2019. Generalized intersection over union: A metric and a loss for bounding box regression. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 658–666

    work page 2019

  71. [71]

    Vincenzo Rinaldi, Lucina Hackman, and Niamh NicDaeid. 2022. Virtual reality as a collaborative tool for digitalised crime scene examination. InInternational conference on extended reality. 154–161

    work page 2022

  72. [72]

    Alan Saalfeld. 1999. Topologically Consistent Line Simplification with the Douglas-Peucker Algorithm.Cartography and Geographic Information Science26, 1 (1999), 7–18

    work page 1999

  73. [73]

    Chipofya, and Sahib Jan

    Angela Schwering, Jakub Krukar, Charu Manivannan, M. Chipofya, and Sahib Jan. 2022. Generalized, Inaccurate, Incomplete: How to Comprehensively Ana- lyze Sketch Maps beyond Their Metric Correctness. InProceedings of the 15th International Conference on Spatial Information Theory (COSIT 2022). Dagstuhl

    work page 2022

  74. [74]

    Gravity Sketch. 2024. Gravity Sketch. https://www.gravitysketch.com/

    work page 2024

  75. [75]

    George Stiny and James Gips. 1971. Shape grammars and the generative specifi- cation of painting and sculpture.. InIFIP congress (2), Vol. 2. Citeseer, 125–135

    work page 1971

  76. [76]

    Conway, and Randy Pausch

    Richard Stoakley, Matthew J. Conway, and Randy Pausch. 1995. Virtual Reality on a WIM: Interactive Worlds in Miniature. InProceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI ’95). 265–272. doi:10.1145/223904. 223938

    work page doi:10.1145/223904 1995

  77. [77]

    Trancite. 2025. Easy Street Draw. https://www.trancite.com/easystreetdraw

    work page 2025

  78. [78]

    Barbara Tversky. 1993. Cognitive maps, cognitive collages, and spatial mental models. InEuropean conference on spatial information theory. Springer, 14–24

    work page 1993

  79. [79]

    2024.Unity Real-Time Development Platform

    Unity. 2024.Unity Real-Time Development Platform. https://unity.com/

    work page 2024

  80. [80]

    Vandenberg and Allan R

    Steven G. Vandenberg and Allan R. Kuse. 1978. Mental Rotations, a Group Test of Three-Dimensional Spatial Visualization.Perceptual and Motor Skills47, 2 (1978), 599–604

    work page 1978

Showing first 80 references.