Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

The Art of Deception: Color Visual Illusions and Diffusion Models

T0 review · 3 major / 5 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 Diffusion models reproduce human brightness and color illusions in their denoising trajectories and can generate new ones.

desk verdict Novel observation, but the headline metric is direction-blind; the paper needs a signed, direction-aware metric before the replication claim holds. read the letter →

arxiv 2412.10122 v1 pith:4XQBWZXV submitted 2024-12-13 cs.CV

Alex Gomez-Villa , Kai Wang , Alejandro C. Parraga , Bartlomiej Twardowski , Jesus Malo , Javier Vazquez-Corral , Joost van de Weijer This is my paper · ORCID
classification cs.CV
keywords visualillusionsdiffusionmodelsDDIMinversionbrightnessperceptioncolortext-to-imagegenerationpsychophysics
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

Diffusion models trained only on natural photographs appear to carry human-like low-level vision in their latent space. The paper tries to establish that the intermediate steps of DDIM inversion — the deterministic path that maps an image back toward noise — progressively shift physically identical patches in the same direction that humans perceive them, so the trajectory itself can be read as a perceptual prediction. The authors then turn this observation into a generation method: by adding a perceptual loss on target regions during text-to-image sampling, they plant brightness/color illusions into realistic generated images, and a psychophysical experiment confirms that human observers see the generated targets as different. If correct, this would mean diffusion priors are a usable model of human perception and a practical tool for creating perceptual effects.

What carries the argument

The load-bearing object is the DDIM inversion trajectory: the deterministic reverse mapping that sends an image $z_0$ to a sequence of latent states $z_t$ via the learned noise predictor. Instead of reading perception from the model's final output, the paper reads the decoded intermediate latents at small step counts, when gaussianization of the input is incomplete. The theoretical lens is that the forward diffusion attempts to Gaussianize images; illusory stimuli lie off the natural-image manifold, so the trajectory shows systematic 'corrections' that mirror human perceptual biases. A second mechanism is the generation loss $L = \gamma L_{\mathrm{VI}} + \beta L_{\mathrm{sim}}$, which pushes noisy latents so that target regions reach user-specified colors while staying harmonized with the background.

What would settle it

Compute a signed, direction-aware version of the paper's $\Delta I$ on a brightness-illusion dataset: instead of $|I_{\mathrm{out}} - I_{\mathrm{in}}|$, measure $I_{\mathrm{out}} - I_{\mathrm{in}}$ and count only cases where the sign matches the illusion's expected darkening. If the model's predicted shifts are actually direction-specific, this score should stay close to the absolute-value PAS; if it drops substantially, the claim that the trajectory mirrors human perception is not supported by the current metric.

Watch

Extended reading notes

Core claim

The central claim is that visual illusions are encoded in the intermediate representations of diffusion models. For an input image $z_0$, DDIM inversion produces a trajectory $z_1, \dots, z_T$; decoding the early states and measuring mean per-channel intensities inside illusion-relevant regions shows that regions humans judge as darker, lighter, or color-shifted are altered by the inversion in exactly that direction. The paper interprets this as the model mapping an out-of-distribution illusory stimulus toward the natural-image manifold, so the transformation is analogous to human adaptation to natural statistics. It supports the claim by reporting high Perception Accuracy Scores on a brightness-illusion dataset and on three color-illusion visual-question-answering datasets, and by showing that the effect is absent when only the autoencoder is used without the diffusion process. It then uses the same latent readout as a loss to generate new illusions and validates them with human observers.

Load-bearing premise

The quantitative argument depends on treating the mean absolute intensity change inside a target region as a faithful measure of a human-like illusion, which assumes that any change in the right direction is what the model produces and does not penalize changes in the opposite direction.

Editorial extensions

If this is right

  • A pretrained diffusion model can act as a vision model: perceptual judgments are obtained by DDIM inversion and read out from intermediate latents, without training or fine-tuning.
  • The same tool captures brightness and color illusions across laboratory stimuli, published illusion images, and natural photographs, without per-illusion parameter tuning.
  • Text-to-image generation can embed an illusion into a specified region and control its color, producing images that fool human observers in psychophysical testing.
  • Because the effect is tied to training on natural-image statistics, the latent shift can serve as a probe for what the model considers out-of-distribution.

Reading between the lines

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

  • One testable extension is to treat the signed latent drift along the trajectory as a continuous perceptual metric and compare it with human discrimination thresholds; the paper's own metric is unsigned, so it cannot yet separate darkening from brightening.
  • The same gaussianization argument could be tested on other low-level dimensions, such as contrast sensitivity, spatial frequency, or motion aftereffects, where human biases are also thought to reflect natural-image statistics.
  • If diffusion priors genuinely mirror human perception, then optimizing images to maximize latent drift could generate adversarial perceptual probes: stimuli that strongly fool humans but not models would reveal exactly where the learned prior diverges from human adaptation.
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 / 5 minor

Summary. The paper claims that DDIM inversion of diffusion models produces intermediate representations whose brightness/color shifts align with human visual illusion perception, and uses this to (1) predict illusions on BRI3L and VQA datasets and (2) generate new illusions with text-to-image diffusion models, the latter being validated by a psychophysical experiment. The quantitative replication claim rests on Eq. (8), a mean absolute intensity-change metric, and on manually masked VQA evaluations; the generation method optimizes a mean-intensity loss in the latent space.

Significance. If the quantitative replication claim were sound, this would be an interesting and nontrivial observation: diffusion models trained only on natural images might encode low-level perceptual biases, and their latent trajectories could serve as a computational model of brightness/color illusions. The paper has genuine strengths: the qualitative demonstrations in Fig. 2 and 4 are suggestive; the psychophysical experiment in Sec. 5.2 provides independent and credible evidence that the generated illusions fool humans; and no parameter is fitted to the benchmark labels. However, the main quantitative tool used to support the replication claim is direction-blind and degenerate, so the central claim is not currently established.

major comments (3)
  1. [Sec. 4.3, Eq. (8)] The metric Delta-I = mean(|I_out - I_in|) over the target mask is direction-blind: an output that brightens the target exactly as much as humans darken it yields the same Delta-I, so the PAS values in Table 1 cannot distinguish human-like darkening from brightening. More seriously, the identity output I_out = I_in gives Delta-I = 0 and satisfies Delta-I < tau * I_int for every positive tau, so a model that simply copies the input would score 100% PAS at all thresholds. The text's statement that lowering tau "requires the model to achieve greater darkness in the target regions" is inconsistent with the equation; lowering tau actually requires the output to stay closer to the input, and no change is always acceptable. Because BRI3L masks denote expected darkening regions, the reported DDIM PAS values (e.g., 84-100% at tau = 0.8-1.0) could be driven by near-identity outputs rather than by human-like perceptual shifts. The replication claim therefore needs a signed, direction-aware metric (and ideally a no-change baseline) before Table 1 can be interpreted.
  2. [Sec. 4.3, VQA-based evaluation] The evaluation on IllusionVQA, GVIL, and HallusionBench is underspecified: the paper states that the authors "manually create segmentation masks" and consider the answer aligned "if it matches the expected intensity" for the referenced regions, but it does not define how the expected intensity is derived, how ambiguous multiple-choice answers are scored, or how the masks are constructed and validated. Table 2 reports PAS values (e.g., DDIM 5 steps 90.40% on IllusionVQA) without specifying the question-answer matching procedure. Since these numbers are part of the central claim that diffusion models predict color illusions, the protocol must be detailed enough to be reproduced and checked for potential experimenter bias.
  3. [Sec. 5.2] The psychophysical experiment is a strong point, but the paper does not report statistical significance of the difference between the illusion images (64% mean "different") and control images (13% mean "different"). A confidence interval or hypothesis test over the 15 observers would strengthen the claim that the generation method reliably fools humans beyond chance.
minor comments (5)
  1. [Sec. 4.3, Eq. (8)] The notation is inconsistent: Eq. (8) uses |M| for the mask size while Eq. (3) uses M; please unify and define the symbols clearly.
  2. [Fig. 4] The caption says "Herman-grid" but the correct spelling is "Hermann grid"; also please ensure the reference [16] is formatted consistently.
  3. [Sec. 4.2] There is a typo "responsable" in Sec. 4.2; it should be "responsible".
  4. [References] References [42] and [43] appear to refer to the same BRI3L dataset (one as a 2023 citation and one as a 2024 arXiv preprint); please consolidate them.
  5. [Sec. 3.2] The intuitive notion of "Gaussianization" of the latent space is used informally; a brief formal definition or a reference to the iterative Gaussianization literature would help readers outside the specific field.

Circularity Check

1 steps flagged · score 6.0 of 10

Eq. 8 defines 'aligned with human perception' as small absolute change, so identity output scores perfect PAS; the quantitative replication claim reduces to a trivial closeness criterion.

  1. self definitional [Section 4.3, Eq. (8) and alignment condition; Table 1]
    "ΔI = 1/|M | ∑_{i∈r} |Iout(r) − Iin(r)| ... We consider the model’s response to be aligned with human perception when ΔI < τ∗ Iint ... Since BRI3L aims to detect regions that become darker, the metric becomes more harder as the value of τ decreases, requiring the model to achieve greater darkness in the target regions."

    By Eq. (8), alignment is defined as a small absolute mean deviation between output and input inside the mask. The trivial output Iout = Iin gives ΔI = 0, which satisfies ΔI < τ·Iint for every τ > 0. Therefore a model that copies the input is scored as 'aligned with human perception' by construction, and the PAS in Table 1 does not require the darkening direction annotated in BRI3L. The accompanying sentence that lowering τ 'requires greater darkness' contradicts the equation: lowering τ only requires staying closer to the input, and no change is always acceptable. Hence the reported DDIM PAS values (84.17–100%) are not evidence of human-like illusion prediction; the quantitative evaluation reduces to a closeness-to-input criterion rather than to perceptual direction.

full rationale

The replication claim is the paper's quantitative backbone. In Section 4.3, alignment is defined by ΔI < τ·Iint where ΔI is mean absolute intensity change. Because Iout = Iin gives ΔI = 0, the criterion is satisfied by doing nothing; the metric is direction-blind and cannot distinguish human-like darkening from brightening or from no change. The text's assertion that decreasing τ requires greater darkness is inconsistent with Eq. (8). Thus the high PAS scores in Table 1 are, at least in part, a consequence of the metric's definition rather than of diffusion-specific illusion behavior. This is a genuine self-definitional reduction of the central quantitative prediction. I do not find additional circularity: the generation loss (Eqs. 4–7) is not fitted to human labels and is validated by an external psychophysical experiment (64% vs 13% 'different' judgments), which breaks the loop for the generation claim. The qualitative demonstrations in Figs. 2 and 4 are independent of the metric. Self-citations to prior work by Malo and coauthors (e.g., manifold equalization, iterative Gaussianization) are used as explanatory context, not as the load-bearing evidence for the observation, and no uniqueness theorem or fitted parameter is imported. Overall, partial circularity: the quantitative replication result reduces by construction, while the generation contribution has independent support.

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

No new physical entities are postulated. The central claim depends on two domain assumptions, latent mean intensity as a perception proxy and defective gaussianization as the cause of alignment, plus hand-set generation hyperparameters. No free parameters are fitted to benchmark labels.

free parameters (5)
  • gamma = 0.5
    Weight of the perceptual loss L_VI in Eq. 4; chosen by hand and varied in a supplementary ablation.
  • beta = 1.0
    Weight of the similarity loss L_sim in Eq. 4; chosen by hand.
  • N = 5
    Number of optimization iterations applied at each diffusion step during generation; chosen by hand.
  • guidance_scale = 10
    Classifier-free guidance scale used in generation; chosen to make images follow the prompt.
  • DDIM_inversion_steps = 5 and 10
    Inversion step counts used for replication; no sensitivity analysis beyond these two values is reported.
assumptions (2)
  • domain assumption Out-of-distribution stimuli are not properly gaussianized by diffusion models, and defective gaussianization is linked to human illusory perception.
    Section 3.2 motivates the DDIM inversion observation by analogy to manifold equalization and cites prior work on Gaussianization flows; the link is not directly measured in the paper.
  • domain assumption Mean intensity of a decoded latent target region is a valid measurement of perceived intensity of that region.
    Equation 3 defines mint and the PAS scores in Section 4.3 rely on it; the paper does not calibrate this proxy against human perceptual magnitudes.

how reviews work

0 comments
Cite this review

Pith. "Pith review of The Art of Deception: Color Visual Illusions and Diffusion Models." pith.science (2026). https://pith.science/paper/4XQBWZXV

@misc{pith2026241210122,
  author       = {Pith},
  title        = {Pith review of: The Art of Deception: Color Visual Illusions and Diffusion Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4XQBWZXV}},
  note         = {Machine review of arXiv:2412.10122}
}
read the original abstract

Visual illusions in humans arise when interpreting out-of-distribution stimuli: if the observer is adapted to certain statistics, perception of outliers deviates from reality. Recent studies have shown that artificial neural networks (ANNs) can also be deceived by visual illusions. This revelation raises profound questions about the nature of visual information. Why are two independent systems, both human brains and ANNs, susceptible to the same illusions? Should any ANN be capable of perceiving visual illusions? Are these perceptions a feature or a flaw? In this work, we study how visual illusions are encoded in diffusion models. Remarkably, we show that they present human-like brightness/color shifts in their latent space. We use this fact to demonstrate that diffusion models can predict visual illusions. Furthermore, we also show how to generate new unseen visual illusions in realistic images using text-to-image diffusion models. We validate this ability through psychophysical experiments that show how our model-generated illusions also fool humans.

Figures

Figures reproduced from arXiv: 2412.10122 by the authors.

Figure 1
Figure 1. Observation: Denoising Diffusion Implicit Models (DDIM) have human-like visual illusions. Application: text-to-image generated visual illusions. Examples (a) and (b) show how the responses to physically equal patches shift differently in the path to the latent space. In contrast, (c) shows how physically different stimuli get more similar along the path, achieving color constancy. The observation of this human-like … view at source ↗
Figure 2
Figure 2. DDIM inversion of the Brightness Contrast illusion [5] using Stable Diffusion. Top row: Image-space visualization (decoded latents) showing (left) Original illusion with two identical gray squares (marked in red) against black and white backgrounds, and inversion results using 3, 10, and 20 steps. Bottom row: Histograms of the corresponding latent representations. The model gradually reproduces the perceptual differ… view at source ↗
Figure 3
Figure 3. Overview of our visual illusion generation pipeline. The process modifies noisy latent representations, zt, through a custom loss function, L, that guides the generation toward percep￾tually ambiguous outputs. 3.3.1. Visual perception inside generation pipeline. Since text-to-image (T2I) diffusion models often struggle to perfectly align with textual prompts, a variety of aug￾mentation methods have been developed to… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Qualitative replication of visual illusions (please enlarge display): For each item, the original image is on the left and its DDIM inversion on the right. The observation done in [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Qualitative results of visual illusion generation (viewing at larger scale recommended). The first row presents prior work by (a) Gomez-Villa et al. [14], (b) Hirsch et al. [17], and (c) Roy et al. [43]. The second and third rows display results from our method (text p…

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. Denoising Models Develop Human-Like Perceptual Illusion Representations Across Architectures

    cs.CV 2026-07 conditional novelty 6.0 of 10

    Denoising diffusion models encode visual illusions in internal layers, yet these representations do not influence the generated image.

Reference graph

Works this paper leans on

57 extracted references · 46 canonical work pages · cited by 1 Pith paper

  1. [1]

    Contrast sensitivity function in deep networks

    Arash Akbarinia, Yaniv Morgenstern, and Karl R Gegenfurt- ner. Contrast sensitivity function in deep networks. Neural Networks, 164:228–244, 2023. 2

    work page 2023

  2. [2]

    DeepFloyd IF: a novel state- of-the-art open-source text-to-image model with a high de- gree of photorealism and language understanding

    DeepFloyd Lab at StabilityAI. DeepFloyd IF: a novel state- of-the-art open-source text-to-image model with a high de- gree of photorealism and language understanding. https: //www.deepfloyd.ai/deepfloyd- if , 2023. Re- trieved on 2023-11-08. 5

    work page 2023

  3. [3]

    Noise masking of white’s illusion exposes the weakness of current spatial filtering models of lightness perception

    Torsten Betz, Robert Shapley, Felix A Wichmann, and Mar- ianne Maertens. Noise masking of white’s illusion exposes the weakness of current spatial filtering models of lightness perception. Journal of vision, 15(14):1–1, 2015. 5

    work page 2015

  4. [4]

    A multiscale spa- tial filtering account of the white effect, simultaneous bright- ness contrast and grating induction

    Barbara Blakeslee and Mark E McCourt. A multiscale spa- tial filtering account of the white effect, simultaneous bright- ness contrast and grating induction. Vision research, 39(26): 4361–4377, 1999. 6, 7

    work page 1999

  5. [5]

    E. Brucke. uber erganzungs und contrasfarben. Wiener Sitzungsber, 51, 1865. 3, 4

    work page

  6. [6]

    Diffusion illusions: Hiding images in plain sight

    Ryan Burgert, Xiang Li, Abe Leite, Kanchana Ranasinghe, and Michael S Ryoo. Diffusion illusions: Hiding images in plain sight. arXiv preprint arXiv:2312.03817, 2023. 2, 3, 4

    arXiv 2023

  7. [7]

    Attend-and-excite: Attention-based se- mantic guidance for text-to-image diffusion models

    Hila Chefer, Yuval Alaluf, Yael Vinker, Lior Wolf, and Daniel Cohen-Or. Attend-and-excite: Attention-based se- mantic guidance for text-to-image diffusion models. ACM Transactions on Graphics (TOG), 2023. 5

    work page 2023

  8. [8]

    What are lightness illusions and why do we see them? PLoS computational biology , 3 (9):e180, 2007

    David Corney and R Beau Lotto. What are lightness illusions and why do we see them? PLoS computational biology , 3 (9):e180, 2007. 2

    work page 2007

Show all 57 references
  1. [9]

    The many colours of ‘the dress’

    Karl R Gegenfurtner, Marina Bloj, and Matteo Toscani. The many colours of ‘the dress’. Current Biology, 25(13):R543– R544, 2015. 2

    2015

  2. [10]

    Visual ana- grams: Generating multi-view optical illusions with diffu- sion models

    Daniel Geng, Inbum Park, and Andrew Owens. Visual ana- grams: Generating multi-view optical illusions with diffu- sion models. Proceedings of the IEEE Conference on Com- puter Vision and Pattern Recognition, 2024. 3, 4

    2024

  3. [11]

    Factor- ized diffusion: Perceptual illusions by noise decomposition

    Daniel Geng, Inbum Park, and Andrew Owens. Factor- ized diffusion: Perceptual illusions by noise decomposition. arXiv preprint arXiv:2404.11615, 2024. 3, 4

    2024 arXiv

  4. [12]

    Convolutional neural networks can be deceived by visual illusions

    Alex Gomez-Villa, Adrian Martin, Javier Vazquez-Corral, and Marcelo Bertalm ´ıo. Convolutional neural networks can be deceived by visual illusions. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 12309–12317, 2019. 2, 7

    2019

  5. [13]

    Color illusions also de- ceive cnns for low-level vision tasks: Analysis and implica- tions

    Alex Gomez-Villa, Adrian Mart ´ın, Javier Vazquez-Corral, Marcelo Bertalm´ıo, and Jes´us Malo. Color illusions also de- ceive cnns for low-level vision tasks: Analysis and implica- tions. Vision Research, 176:156–174, 2020. 2, 6, 7

    2020

  6. [14]

    Gomez-Villa, A

    A. Gomez-Villa, A. Mart ´ın, J. Vazquez-Corral, M. Bertalm´ıo, and J. Malo. On the synthesis of visual illusions using deep generative models. Journal of Vision, 22(8):2–2,

  7. [15]

    Hallusionbench: an advanced diagnos- tic suite for entangled language hallucination and visual il- lusion in large vision-language models

    Tianrui Guan, Fuxiao Liu, Xiyang Wu, Ruiqi Xian, Zongxia Li, Xiaoyu Liu, Xijun Wang, Lichang Chen, Furong Huang, Yaser Yacoob, et al. Hallusionbench: an advanced diagnos- tic suite for entangled language hallucination and visual il- lusion in large vision-language models. In P...

    2024

  8. [16]

    Eine erscheinung simultanen contrastes

    Ludimar Hermann. Eine erscheinung simultanen contrastes. Archiv f ¨ur die gesamte Physiologie des Menschen und der Tiere, 3(1):13–15, 1870. 7

  9. [17]

    Color visual illusions: A statistics-based computational model

    Elad Hirsch and Ayellet Tal. Color visual illusions: A statistics-based computational model. Advances in neural information processing systems, 33:9447–9458, 2020. 2, 8

    2020

  10. [18]

    Classifier-free diffusion guidance

    Jonathan Ho and Tim Salimans. Classifier-free diffusion guidance. NeurIPS 2021 Workshop on Deep Generative Models and Downstream Applications, 2022. 3

    2021

  11. [19]

    Do neural networks show gestalt phenomena? an exploration of the law of closure

    Been Kim, Emily Reif, Martin Wattenberg, and Samy Bengio. Do neural networks show gestalt phenomena? an exploration of the law of closure. arXiv preprint arXiv:1903.01069, 2(8), 2019. 2

    1903 arXiv

  12. [20]

    Lightness, brightness and trans- parency: A quarter century of new ideas, captivating demon- strations and unrelenting controversy

    Frederick AA Kingdom. Lightness, brightness and trans- parency: A quarter century of new ideas, captivating demon- strations and unrelenting controversy. Vision Research, 51 (7):652–673, 2011. 3

    2011

  13. [21]

    Barutan-seijin

    Akiyoshi Kitaoka. Barutan-seijin. Akiyoshi’s illusion Web- site, 2003. Accessed: 2024-11-13. 7

    2003

  14. [22]

    Akiyoshi Kitaoka. Robot. Akiyoshi’s illusion Website, 2005. Accessed: 2024-11-13. 7

    2005

  15. [23]

    Akiyoshi Kitaoka. Shiosai. Akiyoshi’s illusion Website,

  16. [24]

    A ma- chine learning model perceiving brightness optical illusions: Quantitative evaluation with psychophysical data

    Yuki Kubota, Atsushi Hiyama, and Masahiko Inami. A ma- chine learning model perceiving brightness optical illusions: Quantitative evaluation with psychophysical data. In Pro- ceedings of the Augmented Humans International Confer- ence 2021, pages 174–182, 2021. 2

    2021

  17. [25]

    Diffusion-based image translation using disentangled style and content representa- tion

    Gihyun Kwon and Jong Chul Ye. Diffusion-based image translation using disentangled style and content representa- tion. In The Eleventh International Conference on Learning Representations, 2023. 3

    2023

  18. [26]

    Laparra and J

    V . Laparra and J. Malo. Visual aftereffects and sensory non- linearities from a single statistical framework. Front. Human Neurosci., 9, 2015. 2

    2015

  19. [27]

    Laparra, G

    V . Laparra, G. Camps, and J. Malo. Iterative gaussianization: From ICA to random rotations. IEEE Trans. Neur. Nets, 22 (4):537–549, 2011. 2

    2011

  20. [28]

    Laparra, S

    V . Laparra, S. Jim´enez, G. Camps, and J. Malo. Nonlineari- ties and adaptation of color vision from sequential principal curves analysis. Neural Comp., 24(10):2751–2788, 2012. 2

    2012

  21. [29]

    Laparra, JE

    V . Laparra, JE. Johnson, G. Camps, R. Santos, and J. Malo. Estimating information theoretic measures via multidimen- sional gaussianization. IEEE Trans. Patt. Anal. Mach. Intell.,

  22. [30]

    Contrast sensitivity functions in autoencoders

    Qiang Li, Alex Gomez-Villa, Marcelo Bertalm ´ıo, and Jes ´us Malo. Contrast sensitivity functions in autoencoders. Jour- nal of Vision, 22(6):8–8, 2022. 2, 6

    2022

  23. [31]

    Vi- sual illusion cognition dataset construction and recognition performance by deep neural networks

    Tingting Li, Fanyu Wang, Ying Zhou, and Zhenping Xie. Vi- sual illusion cognition dataset construction and recognition performance by deep neural networks. In 2022 IEEE 8th In- ternational Conference on Cloud Computing and Intelligent Systems (CCIS), pages 90–94. IEEE, 2022. 2

    2022

  24. [32]

    A comparative biology approach to dnn modeling of vision: A focus on differences, not similarities

    Ben Lonnqvist, Alban Bornet, Adrien Doerig, and Michael H Herzog. A comparative biology approach to dnn modeling of vision: A focus on differences, not similarities. Journal of vision, 21(10):17–17, 2021. 2

    2021

  25. [33]

    Grating induction: A new explanation for stationary phantom gratings

    Mark McCourt. Grating induction: A new explanation for stationary phantom gratings. Vision Research, 34(12):1609– 1617, 1994. 7

    1994

  26. [34]

    SDEdit: Guided image synthesis and editing with stochastic differential equa- tions

    Chenlin Meng, Yutong He, Yang Song, Jiaming Song, Jia- jun Wu, Jun-Yan Zhu, and Stefano Ermon. SDEdit: Guided image synthesis and editing with stochastic differential equa- tions. In International Conference on Learning Representa- tions, 2022. 3

    2022

  27. [35]

    Does enhanced shape bias improve neural net- work robustness to common corruptions? arXiv preprint arXiv:2104.09789, 2021

    Chaithanya Kumar Mummadi, Ranjitha Subramaniam, Robin Hutmacher, Julien Vitay, V olker Fischer, and Jan Hen- drik Metzen. Does enhanced shape bias improve neural net- work robustness to common corruptions? arXiv preprint arXiv:2104.09789, 2021. 2

    2021 arXiv

  28. [36]

    Is clip fooled by optical illusions? 2023

    Jerry Ngo, Swami Sankaranarayanan, and Phillip Isola. Is clip fooled by optical illusions? 2023. 2, 4

    2023

  29. [37]

    The confetti illusion

    David Novick and Akiyoshi Kitaoka. The confetti illusion. Journal of Illusion, 2, 2021. 7

    2021

  30. [38]

    To- ward a unified chromatic induction model.Journal of Vision, 10(12):5–5, 2010

    Xavier Otazu, C Alejandro Parraga, and Maria Vanrell. To- ward a unified chromatic induction model.Journal of Vision, 10(12):5–5, 2010. 6, 7

    2010

  31. [39]

    Linguistic bind- ing in diffusion models: Enhancing attribute correspondence through attention map alignment

    Royi Rassin, Eran Hirsch, Daniel Glickman, Shauli Rav- fogel, Yoav Goldberg, and Gal Chechik. Linguistic bind- ing in diffusion models: Enhancing attribute correspondence through attention map alignment. Advances in Neural Infor- mation Processing Systems, 36, 2023. 5

    2023

  32. [40]

    Varia- tional inference with normalizing flows

    Danilo Jimenez Rezende and Shakir Mohamed. Varia- tional inference with normalizing flows. ICML ArXiv , abs/1505.05770, 2015. 2

    2015 arXiv

  33. [41]

    High-resolution image synthesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj ¨orn Ommer. High-resolution image synthesis with latent diffusion models. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10684–10695, 2022. 3

    2022

  34. [42]

    Bri3l: A brightness illusion image dataset for identification and localization of regions of illusory perception

    Aniket Roy, Anirban Roy, Soma Mitra, and Kuntal Ghosh. Bri3l: A brightness illusion image dataset for identification and localization of regions of illusory perception. 2023. 5

    2023

  35. [43]

    Bri3l: A brightness illusion image dataset for identification and localization of regions of illusory perception

    Aniket Roy, Anirban Roy, Soma Mitra, and Kuntal Ghosh. Bri3l: A brightness illusion image dataset for identification and localization of regions of illusory perception. arXiv preprint arXiv:2402.04541, 2024. 3, 8

    2024 arXiv

  36. [44]

    Illusionvqa: A challenging optical illu- sion dataset for vision language models

    Haz Sameen Shahgir, Khondker Salman Sayeed, Abhik Bhattacharjee, Wasi Uddin Ahmad, Yue Dong, and Ri- fat Shahriyar. Illusionvqa: A challenging optical illu- sion dataset for vision language models. arXiv preprint arXiv:2403.15952, 2024. 2, 5

    2024 arXiv

  37. [45]

    Shapiro and Dejan Todorovic

    Arthur G. Shapiro and Dejan Todorovic. The Oxford Com- pendium of Visual Illusions, chapter 22. Oxford University Press, 2017. Figure 3.2. 7

    2017

  38. [46]

    Denois- ing diffusion implicit models

    Jiaming Song, Chenlin Meng, and Stefano Ermon. Denois- ing diffusion implicit models. In International Conference on Learning Representations, 2021. 3

    2021

  39. [47]

    Imagenet-trained deep neu- ral networks exhibit illusion-like response to the scintillating grid

    Eric D Sun and Ron Dekel. Imagenet-trained deep neu- ral networks exhibit illusion-like response to the scintillating grid. Journal of Vision, 21(11):15–15, 2021. 2

    2021

  40. [48]

    Investi- gating color illusions from the perspective of computational color constancy

    Oguzhan Ulucan, Diclehan Ulucan, and Marc Ebner. Investi- gating color illusions from the perspective of computational color constancy. arXiv preprint arXiv:2312.13114, 2023. 2

    2023 arXiv

  41. [49]

    Interpolating between im- ages with diffusion models

    Clinton Wang and Polina Golland. Interpolating between im- ages with diffusion models. 2023. 3

    2023

  42. [50]

    Tokencompose: Grounding diffusion with token-level supervision

    Zirui Wang, Zhizhou Sha, Zheng Ding, Yilin Wang, and Zhuowen Tu. Tokencompose: Grounding diffusion with token-level supervision. Proceedings of the IEEE Confer- ence on Computer Vision and Pattern Recognition, 2024. 5

    2024

  43. [51]

    Wang, Evan Montoya, David Munechika, Haoyang Yang, Benjamin Hoover, and Duen Horng Chau

    Zijie J. Wang, Evan Montoya, David Munechika, Haoyang Yang, Benjamin Hoover, and Duen Horng Chau. Diffu- sionDB: A large-scale prompt gallery dataset for text-to- image generative models. arXiv:2210.14896 [cs], 2022. 7

    2022 arXiv

  44. [52]

    Illusory motion repro- duced by deep neural networks trained for prediction

    Eiji Watanabe, Akiyoshi Kitaoka, Kiwako Sakamoto, Masaki Yasugi, and Kenta Tanaka. Illusory motion repro- duced by deep neural networks trained for prediction. Fron- tiers in psychology, 9:340023, 2018. 2

    2018

  45. [53]

    Optical illusions images dataset

    Robert Max Williams and Roman V Yampolskiy. Optical illusions images dataset. arXiv preprint arXiv:1810.00415, 2, 2018. 2, 4

    2018 arXiv

  46. [54]

    Muller-lyer illusion is replicated by higher layer of pre-trained deep neu- ral network for object recognition

    Hongtao Zhang, Zhen Li, and Shinichi Yoshida. Muller-lyer illusion is replicated by higher layer of pre-trained deep neu- ral network for object recognition. 2

  47. [55]

    Yichi Zhang, Jiayi Pan, Yuchen Zhou, Rui Pan, and Joyce Chai. Grounding visual illusions in language: Do vision- language models perceive illusions like humans? In Pro- ceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 5718–5728, 2023. 2, 4, 5

    2023

  48. [56]

    Enhancing semantic fidelity in text- to-image synthesis: Attention regulation in diffusion models

    Yang Zhang, Teoh Tze Tzun, Lim Wei Hern, Tiviatis Sim, and Kenji Kawaguchi. Enhancing semantic fidelity in text- to-image synthesis: Attention regulation in diffusion models. arXiv preprint arXiv:2403.06381, 2024. 5

    2024 arXiv

  49. [2008]

    Accessed: 2024-11-13. 7

    2024

Pith tools

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