Vector Field Synthesis with Sparse Streamlines Using Diffusion Model

Guoning Chen; Nguyen K. Phan; Ricardo Morales; Sebastian D. Espriella

arxiv: 2604.09838 · v1 · submitted 2026-04-10 · 💻 cs.CV

Vector Field Synthesis with Sparse Streamlines Using Diffusion Model

Nguyen K. Phan , Ricardo Morales , Sebastian D. Espriella , Guoning Chen This is my paper

Pith reviewed 2026-05-10 18:02 UTC · model grok-4.3

classification 💻 cs.CV

keywords vector field synthesisdiffusion modelssparse streamlinesphysical plausibilityconditional denoisingclassifier-free guidance2D vector fields

0 comments

The pith

A conditional diffusion model can reconstruct full 2D vector fields from sparse streamlines while preserving physical laws.

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

The paper presents a framework that takes a few coherent streamlines as input and generates a complete 2D vector field. It relies on a conditional denoising diffusion probabilistic model with classifier-free guidance to perform progressive reconstruction. The goal is to keep both geometric fidelity to the given lines and physical properties such as divergence-free behavior. A reader would care because many applications in fluid visualization and simulation start from limited observations yet need complete fields that do not violate known conservation rules. Experiments indicate the diffusion approach yields more flexible and consistent results than conventional optimization techniques.

Core claim

A conditional denoising diffusion probabilistic model with classifier-free guidance synthesizes plausible 2D vector fields from sparse, coherent streamline inputs by progressively denoising while preserving both geometric fidelity and physical constraints such as divergence-free or curl-free properties.

What carries the argument

Conditional denoising diffusion probabilistic model with classifier-free guidance, which performs step-by-step reconstruction from noise conditioned on the sparse streamlines.

If this is right

The generated fields adhere to physical laws while staying close to the sparse observations.
The method offers greater flexibility than traditional optimization-based vector field synthesis.
Physical consistency improves without requiring hand-crafted constraint terms or post-correction.
Progressive denoising allows control over the trade-off between fidelity and smoothness.

Where Pith is reading between the lines

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

The same conditioning strategy could be tested on 3D vector fields or time-dependent flows to check scalability.
Combining the diffusion output with downstream tasks such as particle tracing might reveal whether the preserved physics improves simulation stability.
If the model truly learns the constraints implicitly, retraining on datasets that deliberately violate physics could serve as a diagnostic for what the network actually captures.

Load-bearing premise

The diffusion model with classifier-free guidance will automatically preserve geometric fidelity to the input streamlines and physical constraints without any explicit enforcement or post-processing steps.

What would settle it

Generate fields from the same sparse inputs and measure that a large fraction of them exhibit non-zero divergence or curl values exceeding those of optimization baselines, or that the output streamlines deviate visibly from the supplied input lines.

Figures

Figures reproduced from arXiv: 2604.09838 by Guoning Chen, Nguyen K. Phan, Ricardo Morales, Sebastian D. Espriella.

**Figure 2.** Figure 2: Our vector field diffusion inference pipeline. The process [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗

**Figure 3.** Figure 3: Qualitative comparison across all three datasets: [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗

**Figure 4.** Figure 4: Vector field synthesis from hand-drawn streamlines. Each [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗

read the original abstract

We present a novel diffusion-based framework for synthesizing 2D vector fields from sparse, coherent inputs (i.e., streamlines) while maintaining physical plausibility. Our method employs a conditional denoising diffusion probabilistic model with classifier-free guidance, enabling progressive reconstruction that preserves both geometric and physical constraints. Experimental results demonstrate our method's ability to synthesize plausible vector fields that adhere to physical laws while maintaining fidelity to sparse input observations, outperforming traditional optimization-based approaches in terms of flexibility and physical consistency.

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.

Desk Editor's Note private letter to a colleague

The paper applies conditional diffusion models to complete sparse streamlines into 2D vector fields, but the physical consistency claims rest on an unshown mechanism.

read the letter

The main thing here is a conditional denoising diffusion probabilistic model with classifier-free guidance that takes sparse coherent streamlines and produces a full 2D vector field. The setup treats the task as progressive generative reconstruction rather than direct optimization, which is a straightforward extension of existing diffusion techniques to this graphics and simulation setting. That framing is reasonable and could be useful for pipelines that already work with incomplete flow data. The abstract positions the approach as more flexible than traditional methods while still respecting geometric and physical constraints. What the work does well is identifying a practical niche where generative models might reduce the need for hand-tuned solvers. The idea of using the diffusion process itself to fill in missing information from streamlines makes sense on paper. The soft spots are in the execution details that are missing from the abstract. The central claim is that the outputs adhere to physical laws and show better physical consistency than baselines, yet there is no description of how that is achieved—no auxiliary loss term for divergence or curl, no mention of training exclusively on divergence-free data, and no post-processing projection step. Standard diffusion training minimizes a denoising objective on the observed distribution; it does not automatically embed PDE constraints. Without one of those mechanisms the generated fields can easily violate basic physical properties even if they look plausible. The abstract also asserts outperformance without any metrics, error bars, or baseline comparisons, so the strength of the results cannot be evaluated. This paper is for people in computer graphics or scientific visualization who need a data-driven way to complete sparse vector data. A reader already working with diffusion models in graphics might pick up the application, but anyone expecting rigorous physical guarantees will need the full experiments and loss formulation. I would send it to peer review so the authors can supply the missing quantitative evidence and clarify how the constraints are actually handled.

Referee Report

2 major / 1 minor

Summary. The manuscript proposes a conditional denoising diffusion probabilistic model (DDPM) with classifier-free guidance to synthesize 2D vector fields from sparse, coherent streamline inputs. It claims that the generated fields preserve geometric fidelity to the observations while adhering to physical laws (e.g., divergence-free or curl properties) and outperform traditional optimization-based methods in flexibility and physical consistency.

Significance. If the central claims were supported by quantitative evidence, the work would offer a flexible generative alternative to optimization for vector field reconstruction tasks in computer vision and scientific visualization. The application of modern conditional diffusion models to sparse physical data is a reasonable direction. However, the manuscript supplies no metrics, baselines, error bars, or enforcement details, so its significance cannot be assessed at present. No reproducible code, parameter-free derivations, or falsifiable predictions are presented.

major comments (2)

[Abstract] Abstract: The central claim that outputs 'adhere to physical laws' and exhibit 'physical consistency' is unsupported. Standard conditional DDPM training minimizes a denoising objective on the data distribution and does not embed hard PDE constraints; without (a) provably divergence-free training data, (b) an auxiliary loss term, or (c) post-hoc projection, generated fields can violate physics. No mechanism, loss formulation, or metric (e.g., mean ||∇·V||) is described.
[Experiments] No experimental section or results are supplied that report quantitative metrics, baseline comparisons, or verification of physical properties. The assertion of outperformance therefore cannot be evaluated and is load-bearing for the paper's contribution.

minor comments (1)

[Method] Notation for the vector field V and the conditioning on sparse streamlines should be defined explicitly in the method description.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments, which identify key areas where the manuscript requires clarification and additional content. We address each major point below and will revise the manuscript to strengthen the presentation of our contributions.

read point-by-point responses

Referee: [Abstract] Abstract: The central claim that outputs 'adhere to physical laws' and exhibit 'physical consistency' is unsupported. Standard conditional DDPM training minimizes a denoising objective on the data distribution and does not embed hard PDE constraints; without (a) provably divergence-free training data, (b) an auxiliary loss term, or (c) post-hoc projection, generated fields can violate physics. No mechanism, loss formulation, or metric (e.g., mean ||∇·V||) is described.

Authors: We agree that the abstract does not explicitly describe the mechanism ensuring adherence to physical laws. The training data consists of vector fields that satisfy the relevant physical properties (such as being divergence-free), so the learned distribution inherently favors physically plausible outputs. To address the concern, we will revise the abstract and main text to detail the data generation process and add quantitative verification using metrics such as mean divergence norm. revision: yes
Referee: [Experiments] No experimental section or results are supplied that report quantitative metrics, baseline comparisons, or verification of physical properties. The assertion of outperformance therefore cannot be evaluated and is load-bearing for the paper's contribution.

Authors: The referee correctly notes that the submitted version lacks a dedicated experimental section with quantitative results. This omission was an oversight during submission. We will add a complete experimental section to the revised manuscript, including quantitative metrics, comparisons to optimization baselines, error bars, and explicit verification of physical properties to support the claims. revision: yes

Circularity Check

0 steps flagged

No circularity; purely data-driven method with no derivation chain

full rationale

The paper describes a conditional denoising diffusion probabilistic model trained on data to synthesize vector fields from sparse streamlines. No mathematical derivation, first-principles equations, or parameter-fitting steps are presented that could reduce predictions to inputs by construction. Claims of physical plausibility rest on experimental outcomes and the learned data distribution rather than any self-referential loop, self-citation load-bearing premise, or renamed ansatz. The approach is self-contained as a standard ML pipeline without the enumerated circularity patterns.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review yields no identifiable free parameters, axioms, or invented entities; the method is presented as a standard conditional diffusion pipeline without additional postulated quantities.

pith-pipeline@v0.9.0 · 5373 in / 1120 out tokens · 69359 ms · 2026-05-10T18:02:15.470129+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

Our method employs a conditional denoising diffusion probabilistic model with classifier-free guidance, enabling progressive reconstruction that preserves both geometric and physical constraints.
IndisputableMonolith/Foundation/AlphaCoordinateFixation.lean J_uniquely_calibrated_via_higher_derivative unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

We compute the loss only in the unknown regions: L=∥ε̂·(1−M)−ε·(1−M)∥²
IndisputableMonolith/Foundation/AlexanderDuality.lean alexander_duality_circle_linking unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

Physics error which combines normalized curl and divergence errors

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

31 extracted references · 31 canonical work pages

[1]

Berenjkoub, G

M. Berenjkoub, G. Chen, and T. Günther. V ortex boundary identifica- tion using convolutional neural network. In2020 IEEE Visualization Conference (VIS), pp. 261–265. IEEE, 2020. 3

work page 2020
[2]

G. Chen, V . Kwatra, L.-Y . Wei, C. D. Hansen, and E. Zhang. Design of 2d time-varying vector fields.IEEE Trans. Vis. Comput. Graph., 18(10):1717–1730, 2012. 1, 3

work page 2012
[3]

G. Chen, K. Mischaikow, R. S. Laramee, P. Pilarczyk, and E. Zhang. Vector Field Editing and Periodic Orbit Extraction Using Morse Decom- position.IEEE Transactions on Visualization and Computer Graphics, 13(4):769–785, Jul./Aug. 2007. 1, 3

work page 2007
[4]

de Goes, M

F. de Goes, M. Desbrun, and Y . Tong. Vector field processing on triangle meshes. InACM SIGGRAPH 2016 Courses, SIGGRAPH ’16, pp. 27:1–27:49. ACM, New York, NY , USA, 2016. doi: 10.1145/ 2897826.2927303 1

work page arXiv 2016
[5]

Fisher, P

M. Fisher, P. Schröder, M. Desbrun, and H. Hoppe. Design of tangent vector fields.ACM Transactions on Graphics, 26(3):56:1–56:9, 2007. 1

work page 2007
[6]

H. Fu, Y . Wei, C.-L. Tai, and L. Quan. Sketching hairstyles. In SBIM ’07: Proceedings of the 4th Eurographics workshop on Sketch- based interfaces and modeling, pp. 31–36. ACM, 2007. doi: 10.1145/ 1384429.1384439 1

work page arXiv 2007
[7]

P. Gu, J. Han, D. Z. Chen, and C. Wang. Reconstructing unsteady flow data from representative streamlines via diffusion and deep- learning-based denoising.IEEE Computer Graphics and Applications, 41(6):111–121, 2021. 1

work page 2021
[8]

J. Ho, A. Jain, and P. Abbeel. Denoising diffusion probabilistic models. Advances in neural information processing systems, 33:6840–6851,

work page
[9]

Y . A. Kapitanyuk, A. V . Proskurnikov, and M. Cao. A guiding vector- field algorithm for path-following control of nonholonomic mobile robots.IEEE Transactions on Control Systems Technology, 26(4):1372– 1385, 2017. 1

work page 2017
[10]

Y .-K. Lai, M. Jin, X. Xie, Y . He, J. Palacios, E. Zhang, S.-M. Hu, and X. Gu. Metric-driven rosy field design and remeshing.IEEE Transactions on Visualization and Computer Graphics, 16(1):95–108,

work page
[11]

doi: 10.1109/TVCG.2009.59 1

work page doi:10.1109/tvcg.2009.59 2009
[12]

S. R. Lindemann and S. M. LaValle. Smoothly blending vector fields for global robot navigation. InProceedings of the 44th IEEE Conference on Decision and Control, pp. 3553–3559. IEEE, 2005. 1

work page 2005
[13]

J. Page. Super-resolution of turbulence with dynamics in the loss. Journal of Fluid Mechanics, 1002:R3, 2025. 1, 3

work page 2025
[14]

Palacios, C

J. Palacios, C. Ma, W. Chen, L.-Y . Wei, and E. Zhang. Tensor field design in volumes. InSIGGRAPH ASIA 2016 Technical Briefs, pp. 1–4. 2016. 1

work page 2016
[15]

Palacios and E

J. Palacios and E. Zhang. Rotational symmetry field design on surfaces. ACM Transactions on Graphics (SIGGRAPH 07), 26(3):56:1–56:10,

work page
[16]

Raissi, P

M. Raissi, P. Perdikaris, and G. E. Karniadakis. Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations. Journal of Computational Physics, 378:686–707, 2019. 1

work page 2019
[17]

N. Ray, B. Vallet, W.-C. Li, and B. Levy. N-symmetry direction field design.ACM Transactions on Graphics, 27(2):10:1–10:13, 2008. 1

work page 2008
[18]

D. Shu, Z. Li, and A. B. Farimani. A physics-informed diffusion model for high-fidelity flow field reconstruction.Journal of Computational Physics, 478:111972, 2023. 1, 2, 3

work page 2023
[19]

D. Shu, W. Zhen, Z. Li, and A. B. Farimani. Inpainting computational fluid dynamics with deep learning.arXiv preprint arXiv:2402.17185,

work page arXiv
[20]

J. Stam. Stable fluids. InProceedings of the 26th annual conference on Computer graphics and interactive techniques, SIGGRAPH ’99, pp. 121–128. ACM Press/Addison-Wesley Publishing Co., 1999. doi: 10. 1145/311535.311548 1

work page arXiv 1999
[21]

J. Stam. Flows on surfaces of arbitrary topology.ACM Transactions on Graphics (SIGGRAPH 03), 22(3):724–731, July 2003. doi: 10. 1145/882262.882338 1

work page arXiv 2003
[22]

G. Turk. Texture synthesis on surfaces. InProceedings of the 28th annual conference on Computer graphics and interactive techniques, SIGGRAPH ’01, pp. 347–354, 2001. 1, 3

work page 2001
[23]

van Wijk

J. van Wijk. Image based flow visualization for curved surfaces. In Proceedings IEEE Visualization ’03, pp. 123–130. IEEE Computer Society, 2003. 1

work page 2003
[24]

Vaxman, M

A. Vaxman, M. Campen, O. Diamanti, D. Bommes, K. Hildebrandt, M. Ben-Chen, and D. Panozzo. Directional field synthesis, design, and processing. InSIGGRAPH ASIA 2016 Courses, SA ’16, pp. 15:1–15:30. ACM, New York, NY , USA, 2016. doi: 10.1145/2988458.2988478 1

work page doi:10.1145/2988458.2988478 2016
[25]

von Funck, H

W. von Funck, H. Theisel, and H.-P. Seidel. Vector field based shape deformations.ACM Transactions on Graphics, 25(3):1118–1125, 2006. doi: 10.1145/1141911.1142002 1

work page doi:10.1145/1141911.1142002 2006
[26]

M. Wang, J. Tao, J. Ma, Y . Shen, and C. Wang. Flowvisual: A visu- alization app for teaching and understanding 3d flow field concepts. Electronic Imaging, 28:1–10, 2016. 1

work page 2016
[27]

L. Y . Wei and M. Levoy. Texture synthesis over arbitrary manifold surfaces. InProceedings of the 28th annual conference on Computer graphics and interactive techniques, SIGGRAPH ’01, pp. 355–360,

work page
[28]

Topological Sep- aration of Vortices

A. Zafar, Z. Poorshayegh, D. Yang, and G. Chen. Topological separa- tion of vortices. In2024 IEEE Visualization and Visual Analytics (VIS), pp. 311–315, 2024. doi: 10.1109/VIS55277.2024.00070 1

work page doi:10.1109/vis55277.2024.00070 2024
[29]

Synthesizing realistic avatars for enhanced virtual communication.IEEE Transactions on Visualization and Computer Graphics, 29(5):1802–1815, 2023

A. Zafar, D. Yang, and G. Chen. Extract and characterize hairpin vortices in turbulent flows.IEEE Transactions on Visualization and Computer Graphics, 30(1):716–726, 2024. doi: 10.1109/TVCG.2023. 3326603 1

work page doi:10.1109/tvcg.2023 2024
[30]

Zhang, J

E. Zhang, J. Hays, and G. Turk. Interactive tensor field design and visualization on surfaces.IEEE transactions on visualization and computer graphics, 13(1):94–107, 2006. 1

work page 2006
[31]

Zhang, K

E. Zhang, K. Mischaikow, and G. Turk. Vector field design on surfaces. ACM Transactions on Graphics, 25(4):1294–1326, 2006. 1, 3

work page 2006

[1] [1]

Berenjkoub, G

M. Berenjkoub, G. Chen, and T. Günther. V ortex boundary identifica- tion using convolutional neural network. In2020 IEEE Visualization Conference (VIS), pp. 261–265. IEEE, 2020. 3

work page 2020

[2] [2]

G. Chen, V . Kwatra, L.-Y . Wei, C. D. Hansen, and E. Zhang. Design of 2d time-varying vector fields.IEEE Trans. Vis. Comput. Graph., 18(10):1717–1730, 2012. 1, 3

work page 2012

[3] [3]

G. Chen, K. Mischaikow, R. S. Laramee, P. Pilarczyk, and E. Zhang. Vector Field Editing and Periodic Orbit Extraction Using Morse Decom- position.IEEE Transactions on Visualization and Computer Graphics, 13(4):769–785, Jul./Aug. 2007. 1, 3

work page 2007

[4] [4]

de Goes, M

F. de Goes, M. Desbrun, and Y . Tong. Vector field processing on triangle meshes. InACM SIGGRAPH 2016 Courses, SIGGRAPH ’16, pp. 27:1–27:49. ACM, New York, NY , USA, 2016. doi: 10.1145/ 2897826.2927303 1

work page arXiv 2016

[5] [5]

Fisher, P

M. Fisher, P. Schröder, M. Desbrun, and H. Hoppe. Design of tangent vector fields.ACM Transactions on Graphics, 26(3):56:1–56:9, 2007. 1

work page 2007

[6] [6]

H. Fu, Y . Wei, C.-L. Tai, and L. Quan. Sketching hairstyles. In SBIM ’07: Proceedings of the 4th Eurographics workshop on Sketch- based interfaces and modeling, pp. 31–36. ACM, 2007. doi: 10.1145/ 1384429.1384439 1

work page arXiv 2007

[7] [7]

P. Gu, J. Han, D. Z. Chen, and C. Wang. Reconstructing unsteady flow data from representative streamlines via diffusion and deep- learning-based denoising.IEEE Computer Graphics and Applications, 41(6):111–121, 2021. 1

work page 2021

[8] [8]

J. Ho, A. Jain, and P. Abbeel. Denoising diffusion probabilistic models. Advances in neural information processing systems, 33:6840–6851,

work page

[9] [9]

Y . A. Kapitanyuk, A. V . Proskurnikov, and M. Cao. A guiding vector- field algorithm for path-following control of nonholonomic mobile robots.IEEE Transactions on Control Systems Technology, 26(4):1372– 1385, 2017. 1

work page 2017

[10] [10]

Y .-K. Lai, M. Jin, X. Xie, Y . He, J. Palacios, E. Zhang, S.-M. Hu, and X. Gu. Metric-driven rosy field design and remeshing.IEEE Transactions on Visualization and Computer Graphics, 16(1):95–108,

work page

[11] [11]

doi: 10.1109/TVCG.2009.59 1

work page doi:10.1109/tvcg.2009.59 2009

[12] [12]

S. R. Lindemann and S. M. LaValle. Smoothly blending vector fields for global robot navigation. InProceedings of the 44th IEEE Conference on Decision and Control, pp. 3553–3559. IEEE, 2005. 1

work page 2005

[13] [13]

J. Page. Super-resolution of turbulence with dynamics in the loss. Journal of Fluid Mechanics, 1002:R3, 2025. 1, 3

work page 2025

[14] [14]

Palacios, C

J. Palacios, C. Ma, W. Chen, L.-Y . Wei, and E. Zhang. Tensor field design in volumes. InSIGGRAPH ASIA 2016 Technical Briefs, pp. 1–4. 2016. 1

work page 2016

[15] [15]

Palacios and E

J. Palacios and E. Zhang. Rotational symmetry field design on surfaces. ACM Transactions on Graphics (SIGGRAPH 07), 26(3):56:1–56:10,

work page

[16] [16]

Raissi, P

M. Raissi, P. Perdikaris, and G. E. Karniadakis. Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations. Journal of Computational Physics, 378:686–707, 2019. 1

work page 2019

[17] [17]

N. Ray, B. Vallet, W.-C. Li, and B. Levy. N-symmetry direction field design.ACM Transactions on Graphics, 27(2):10:1–10:13, 2008. 1

work page 2008

[18] [18]

D. Shu, Z. Li, and A. B. Farimani. A physics-informed diffusion model for high-fidelity flow field reconstruction.Journal of Computational Physics, 478:111972, 2023. 1, 2, 3

work page 2023

[19] [19]

D. Shu, W. Zhen, Z. Li, and A. B. Farimani. Inpainting computational fluid dynamics with deep learning.arXiv preprint arXiv:2402.17185,

work page arXiv

[20] [20]

J. Stam. Stable fluids. InProceedings of the 26th annual conference on Computer graphics and interactive techniques, SIGGRAPH ’99, pp. 121–128. ACM Press/Addison-Wesley Publishing Co., 1999. doi: 10. 1145/311535.311548 1

work page arXiv 1999

[21] [21]

J. Stam. Flows on surfaces of arbitrary topology.ACM Transactions on Graphics (SIGGRAPH 03), 22(3):724–731, July 2003. doi: 10. 1145/882262.882338 1

work page arXiv 2003

[22] [22]

G. Turk. Texture synthesis on surfaces. InProceedings of the 28th annual conference on Computer graphics and interactive techniques, SIGGRAPH ’01, pp. 347–354, 2001. 1, 3

work page 2001

[23] [23]

van Wijk

J. van Wijk. Image based flow visualization for curved surfaces. In Proceedings IEEE Visualization ’03, pp. 123–130. IEEE Computer Society, 2003. 1

work page 2003

[24] [24]

Vaxman, M

A. Vaxman, M. Campen, O. Diamanti, D. Bommes, K. Hildebrandt, M. Ben-Chen, and D. Panozzo. Directional field synthesis, design, and processing. InSIGGRAPH ASIA 2016 Courses, SA ’16, pp. 15:1–15:30. ACM, New York, NY , USA, 2016. doi: 10.1145/2988458.2988478 1

work page doi:10.1145/2988458.2988478 2016

[25] [25]

von Funck, H

W. von Funck, H. Theisel, and H.-P. Seidel. Vector field based shape deformations.ACM Transactions on Graphics, 25(3):1118–1125, 2006. doi: 10.1145/1141911.1142002 1

work page doi:10.1145/1141911.1142002 2006

[26] [26]

M. Wang, J. Tao, J. Ma, Y . Shen, and C. Wang. Flowvisual: A visu- alization app for teaching and understanding 3d flow field concepts. Electronic Imaging, 28:1–10, 2016. 1

work page 2016

[27] [27]

L. Y . Wei and M. Levoy. Texture synthesis over arbitrary manifold surfaces. InProceedings of the 28th annual conference on Computer graphics and interactive techniques, SIGGRAPH ’01, pp. 355–360,

work page

[28] [28]

Topological Sep- aration of Vortices

A. Zafar, Z. Poorshayegh, D. Yang, and G. Chen. Topological separa- tion of vortices. In2024 IEEE Visualization and Visual Analytics (VIS), pp. 311–315, 2024. doi: 10.1109/VIS55277.2024.00070 1

work page doi:10.1109/vis55277.2024.00070 2024

[29] [29]

Synthesizing realistic avatars for enhanced virtual communication.IEEE Transactions on Visualization and Computer Graphics, 29(5):1802–1815, 2023

A. Zafar, D. Yang, and G. Chen. Extract and characterize hairpin vortices in turbulent flows.IEEE Transactions on Visualization and Computer Graphics, 30(1):716–726, 2024. doi: 10.1109/TVCG.2023. 3326603 1

work page doi:10.1109/tvcg.2023 2024

[30] [30]

Zhang, J

E. Zhang, J. Hays, and G. Turk. Interactive tensor field design and visualization on surfaces.IEEE transactions on visualization and computer graphics, 13(1):94–107, 2006. 1

work page 2006

[31] [31]

Zhang, K

E. Zhang, K. Mischaikow, and G. Turk. Vector field design on surfaces. ACM Transactions on Graphics, 25(4):1294–1326, 2006. 1, 3

work page 2006