Pith. sign in

REVIEW 4 major objections 5 minor 71 references

Dense-Face: Personalized Face Generation Model via Dense Annotation Prediction

T0 review · 4 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 Dense-Face: a frozen Stable Diffusion that personalizes faces without losing text control

desk verdict A genuinely new frozen-SD face personalization architecture with best FID/CLIP-I in its comparison, but the dense-annotation branch—its most distinctive contribution—rests on unvalidated pseudo-labels, so the SOTA claim is conditional pending code, data, and label-quality evidence. read the letter →

arxiv 2412.18149 v1 pith:3KMPDB5W submitted 2024-12-24 cs.CV

Xiao Guo , Manh Tran , Jiaxin Cheng , Xiaoming Liu This is my paper · ORCID
classification cs.CV
keywords personalizedgenerationtext-to-imagediffusionfaceidentitypreservationdenseannotationpredictionposecontrollatentspaceblendingT2I-Dense-Facedataset
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

Dense-Face is a text-to-image personalization system that tries to generate a person's face in new poses and contexts from a single reference image, with no test-time fine-tuning. Its central claim is that a frozen Stable Diffusion model can keep its full text-editing ability while gaining identity preservation and pose control, provided the face-specific knowledge is injected through small plug-in modules rather than by fine-tuning the whole generator. The paper reports top or near-top performance on image-text alignment, identity preservation, and pose control against current personalized generation baselines, and it points to the dense-annotation objective as the source of face-domain knowledge. The work also contributes a dataset of roughly two million image-text pairs with dense face annotations to make this training possible.

What carries the argument

The load-bearing mechanism is the pose-controllable (PC) adapter, a set of trained residual weight matrices added to the query, key, and value projections of the cross-attention modules, which shifts the frozen Stable Diffusion toward the face domain when enabled. Around it sit a pose branch that encodes a sparse three-Euler-angle head-pose image, an annotation prediction module that converts concatenated internal UNet features into dense landmarks, a pseudo face mask, and a depth map, and a latent-space blending procedure that combines the text-editing mode's background with the face-generation mode's identity-preserving face. The training objective adds three auxiliary losses on those predicted annotations to the standard diffusion denoising loss.

What would settle it

Train Dense-Face twice with everything identical except that the dense annotation targets are replaced by random or shuffled labels; if FID and identity similarity stay at the same level, the dense prediction branch is not what carries the claimed gain. A complementary check is to compare the automatically generated pose, landmark, and depth labels against manual ground truth on a held-out face benchmark; large systematic errors would show the structural supervision is built on unverified pseudo-labels.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that identity fidelity and text controllability are not a trade-off in personalized face generation. Dense-Face keeps the pre-trained Stable Diffusion frozen and attaches a pose-controllable adapter that adds residual shifts to the query, key, and value projections of every cross-attention layer; a pose branch feeds yaw, pitch, and roll conditions; and an auxiliary module reads internal UNet features to predict dense face annotations. These components define two generation modes: one identical to the original Stable Diffusion, used to paint a scene from the text caption, and one conditioned on pose and identity, used to repaint the face region; the two outputs are merged by latent space blending. The paper argues that this design preserves the base model's text-editing ability while the dense-annotation prediction supplies the structural face knowledge needed for identity preservation, and it reports top or near-top scores on image fidelity, text alignment, identity similarity, and pose accuracy relative to strong personalized generation baselines.

Load-bearing premise

The load-bearing premise is that the automatically generated training annotations — head poses, 468 landmarks, pseudo masks, depth maps, and captions — are accurate enough to teach facial structure, yet the paper never evaluates their quality.

Editorial extensions

If this is right

  • Because the base Stable Diffusion weights stay frozen, the text-editing ability of the original model is retained by construction rather than recovered after fine-tuning damage.
  • Dense annotation prediction is doing measurable work: the paper's ablation shows removing it raises FID by about seven points, so structural supervision contributes to generation quality.
  • The two-mode design makes identity transfer compositional: the same text prompt can be paired with different reference identities and head poses, and the background comes from the text-only pass.
  • The learned identity embedding transfers to face swapping with competitive identity retrieval and pose accuracy, so the same representation supports both generation and editing.
  • The released two-million-pair dense-annotation dataset is itself a reusable asset for training and evaluating other face-generation methods.

Reading between the lines

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

  • Because the dense labels are produced by off-the-shelf estimators and their quality is never measured, the method's ceiling is set by those pseudo-labels; auditing and correcting them is an obvious next step the paper leaves open.
  • The recipe is not face-specific: a frozen text-to-image model plus a sparse-condition adapter plus auxiliary dense prediction could be transferred to hands, bodies, or animals whenever a cheap structural estimator exists.
  • If the reported text-alignment parity with the larger base model holds across more prompts, it would imply that full-model fine-tuning is unnecessary for domain personalization, potentially simplifying future pipelines.
  • A cleaner decomposition would be to run the face-generation mode alone, without the text-editing base image, and measure identity and pose accuracy directly; the paper reports the combined pipeline but not this isolated behavior.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper proposes Dense-Face, a personalized face-generation model built on a frozen Stable Diffusion 2.1 backbone. It introduces three trainable components: a pose branch (ControlNet-like, taking head pose Euler angles), a pose-controllable (PC) adapter that adds residual weights to the cross-attention layers, and a dense annotation prediction module that predicts landmarks, pseudo masks, and depth from UNet features. A new dataset, T2I-Dense-Face, is constructed from CASIA-WebFace and CelebA images restored by CodeFormer, with pseudo-labels from Hopenet (pose), Mediapipe (landmarks/depth), and multiple BLIP/OPT captioners filtered by alignment scores. At inference, a text-editing mode generates a base image, and a face-generation mode repaints the face region using latent-space blending. Experiments report FID, CLIP-T, CLIP-I, DINO, identity similarity (Arcface and Adaface), face diversity, and pose MAE, claiming state-of-the-art or competitive performance against SD variants, ControlNet, IP-Adapter, PhotoMaker, and InstantID, plus a face-swapping application.

Significance. If the empirical claims hold, Dense-Face is a useful contribution: it preserves the text-editing ability of a frozen SD while adding identity and pose control through lightweight adapters, and it releases T2I-Dense-Face, a large-scale dataset with dense annotations that could benefit future work. The design choice of using two generation modes with latent blending is a reasonable way to decouple text controllability from face-domain specialization. The paper deserves credit for evaluating identity preservation with two different face-recognition embeddings (Arcface and Adaface) to mitigate overfitting to a single metric, for running ablations that isolate the PC-adapter, pose branch, and dense regularization, and for making the project page and dataset publicly available. However, the central SOTA claim is undercut by the absence of uncertainty quantification and by the lack of any validation of the automatically generated pseudo-labels, which are load-bearing for the dense-regularization branch.

major comments (4)
  1. [Sec. 3.4 / Sec. 3.3] The accuracy of the pseudo-labels in T2I-Dense-Face is never evaluated. The dense-annotation prediction branch (Sec. 3.3) is trained using Hopenet pose estimates, Mediapipe landmarks and depth, BLIP/OPT captions filtered by alignment scores, and CodeFormer-restored images, but the paper reports no quantitative measure of label quality (e.g., landmark NME, depth error, or pose MAE against a trusted reference). The ablation in Table 3a (last row) attributes a 7.0 FID improvement to 'dense regularization,' yet without evidence that the pseudo-labels are accurate, that improvement could reflect a generic regularization/denoising effect rather than genuine face-structure knowledge. The central claim that dense annotation prediction enables domain knowledge is therefore not uniquely supported. Please add quantitative evaluation of both the pseudo-label quality (including failure cases on low-resolution or restored images) and the accuracy of the predicted annotations, or compare against a subset with human-verified labels.
  2. [Table 2 / Table 3a] No error bars, confidence intervals, or significance tests are reported for any metric. Several claimed advantages in Table 2 are small (e.g., FID 95.70 vs. 97.53 for InstantID), and the identity similarity of Dense-Face is actually lower than InstantID on both Arcface (0.568 vs. 0.611) and Adaface (0.519 vs. 0.574). Without variance over seeds or over evaluation subjects/prompts, the statement that Dense-Face 'achieves state-of-the-art or competitive generation performance' is not statistically substantiated. Similarly, Table 3a compares models trained on 6,000 subjects with a single run; the reported differences (e.g., ID similarity 0.357 vs. 0.349) may be within run-to-run noise. Please report multiple seeds or a bootstrap/confidence interval over the 30 evaluation subjects, and use an appropriate significance test for the SOTA comparisons.
  3. [Eq. (8), Sec. 3.5] The hyperparameters λ1, λ2, and λ3 in Eq. (8) are never reported, and the caption-selection criterion (keeping the 'three highest scores' in Sec. 3.4) is not precisely specified (which alignment scorer? what threshold?). These values are necessary to reproduce the method and to interpret the dense-regularization ablation in Table 3a, since the relative weighting of LLD, LPM, and LDE determines how much the auxiliary losses influence training. Please provide the exact settings in Implementation Details, and ideally a small sensitivity study for λ1–λ3.
  4. [Sec. 4.2, Table 3(b)] The time-step analysis for dense annotation prediction is not quantitatively supported. The text states that fixed time steps (0, 200, 400, 600, 800) are all worse than randomly sampling t, but the figure (labeled 'Fig. 3b,' apparently Table 3(b)) only plots FID for fixed steps; no random-time-step baseline or numerical values are given. Since the training objective uses random time steps, this claim is load-bearing for the design choice and should be backed with a direct comparison (e.g., a horizontal line or error bar for the random-t policy) and the corresponding numbers.
minor comments (5)
  1. [Fig. 3 caption] The caption lists the PC-adapter weights as '(w′q, w′v, and w′v)'; the third term should be w′k to match Eq. (6).
  2. [Table 3a] The abbreviation 'ID bran.' is not defined; the text uses 'ID branch' only in the table. Please clarify what the ID branch refers to (presumably the MLP that maps cid to c′ in Eq. 4) and define it in Sec. 3.2.
  3. [Sec. 4.2] The estimator name is spelled inconsistently: 'Hopenet' in Sec. 3.4 and the references, but 'HopeNet' in the text after Table 3a. Please unify the spelling.
  4. [Fig. 1 and captions] The placeholder 'AS*' (used in figure captions and examples) is not defined in the main text; it appears to be a typo for 'S*'. Please define the subject placeholder notation consistently (e.g., 'S*' as in the supplementary).
  5. [Eq. (4)] The identity scale λ is set to 1e−2 without any ablation or sensitivity analysis. Since this parameter directly controls the strength of the identity text embedding, a brief ablation would help readers understand its role, though this is not required for the central claim.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the central personalization results are not forced by training inputs; only minor self-referential evaluation metrics appear and they do not drive the main claims.

full rationale

Walking the derivation chain, no load-bearing step reduces by construction to its inputs. The training objective (Eq. 8) is a denoising loss plus dense-annotation (landmark, mask, depth) losses; the headline evaluations (FID, CLIP-I, CLIP-T, DINO, Arcface/Adaface ID similarity, LPIPS) are not optimized in training, so the state-of-the-art/competitive claims are externally measurable. The identity condition uses an Arcface embedding (Sec. 3.2, Eq. 4), and identity preservation is also reported as Arcface cosine similarity, but the authors explicitly add Adaface "in case the proposed method becomes overfitting on the identity embedding from the Arcface" and use the same protocol for all baselines, so the metric is not forced by construction. The pose-accuracy ablation (Sec. 4.2) compares the condition head pose to a Hopenet estimate on the generated image; because the condition pose was itself produced by Hopenet (Sec. 3.4), this particular MAE is a Hopenet self-consistency check rather than independent pose ground truth, but it is an auxiliary ablation metric and does not by itself force the central generation claim. The main validity risk is the unvalidated T2I-Dense-Face pseudo-labels (Hopenet, Mediapipe, BLIP/OPT, CodeFormer), which is a data-quality/correctness concern, not circularity. There is no load-bearing self-citation chain; author self-citations are confined to broader-impact and unrelated prior work. Accordingly, no circular step is identified.

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

The method is built on standard diffusion assumptions and on the reliability of automatically generated pseudo-annotations. The free parameters are the three auxiliary loss weights lambda1, lambda2, lambda3 (not reported), the identity-text scale lambda=1e-2 (reported), and the per-image caption selection rule (top three). No new physical entities are introduced. The most consequential unstated dependency is the validity of Hopenet and Mediapipe outputs as ground truth for 2M training images.

free parameters (5)
  • lambda1 (landmark loss weight in Eq. 8)
    Introduced in Eq. 8 for the landmark heatmap loss L_LD, but its value is never reported.
  • lambda2 (pseudo-mask cross-entropy weight in Eq. 8)
    Introduced in Eq. 8 for the pseudo-mask loss L_PM, but its value is never reported.
  • lambda3 (depth loss weight in Eq. 8)
    Introduced in Eq. 8 for the depth loss L_DE, but its value is never reported.
  • lambda (identity text embedding scale) = 1e-2
    Set empirically in Eq. 4; controls how strongly the MLP-mapped identity embedding is added to the 'Face' text embedding.
  • Caption selection cutoff = top 3
    In Sec. 3.4, only the three captions with the highest text-image alignment scores are kept per image; this hand-chosen cutoff shapes the training distribution.
assumptions (5)
  • domain assumption Pre-trained Stable Diffusion cross-attention can be modified with additive residual weights without destroying text controllability.
    Eq. 6 defines q'=(wq+w'q)f, k'=(wk+w'k)c', v'=(wv+w'v)c'; the paper assumes the frozen base model's text-alignment ability survives this modification, and supports this with CLIP-T scores in Table 2.
  • domain assumption Arcface face embeddings lie in a manifold that an MLP can map into text-embedding space while preserving identity.
    Sec. 3.2 builds c' = lambda*MLP(cid) + cFace; the property of the faceNet space is assumed rather than analyzed.
  • domain assumption Hopenet, Mediapipe, and CodeFormer outputs are accurate enough to serve as ground truth for a 2M-pair dataset.
    Sec. 3.4 constructs T2I-Dense-Face from these automatic estimates without human verification or an accuracy audit.
  • domain assumption Latent-space blending of two diffusion branches produces artifact-free composites while preserving background.
    Sec. 3.6 relies on the blended latent diffusion literature [2,3,40,69] and shows only qualitative examples; no quantitative artifact metric is reported.
  • standard math Standard DDPM/Stable Diffusion noise and denoising assumptions apply as in Ho et al.
    Eqs. 1 and 3 adopt the standard latent diffusion objective; this is standard background and not specific to the paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Dense-Face: Personalized Face Generation Model via Dense Annotation Prediction." pith.science (2026). https://pith.science/paper/3KMPDB5W

@misc{pith2026241218149,
  author       = {Pith},
  title        = {Pith review of: Dense-Face: Personalized Face Generation Model via Dense Annotation Prediction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3KMPDB5W}},
  note         = {Machine review of arXiv:2412.18149}
}
read the original abstract

The text-to-image (T2I) personalization diffusion model can generate images of the novel concept based on the user input text caption. However, existing T2I personalized methods either require test-time fine-tuning or fail to generate images that align well with the given text caption. In this work, we propose a new T2I personalization diffusion model, Dense-Face, which can generate face images with a consistent identity as the given reference subject and align well with the text caption. Specifically, we introduce a pose-controllable adapter for the high-fidelity image generation while maintaining the text-based editing ability of the pre-trained stable diffusion (SD). Additionally, we use internal features of the SD UNet to predict dense face annotations, enabling the proposed method to gain domain knowledge in face generation. Empirically, our method achieves state-of-the-art or competitive generation performance in image-text alignment, identity preservation, and pose control.

Figures

Figures reproduced from arXiv: 2412.18149 by the authors.

Figure 1
Figure 1. Performance comparison of different personalized methods. Abstract. The text-to-image (T2I) personalization diffusion model can generate images of the novel concept based on the user input text caption. However, existing T2I personalized methods either require test-time fine￾tuning or fail to generate images that align well with the given text cap￾tion. In this work, we propose a new T2I personalization diffusion mo… view at source ↗
Figure 2
Figure 2. (a) Our proposed Dense-Face introduces additional components, including a pose branch and PC-adapter, on the top of the pre-trained SD. These two components enable Dense-Face to have two generation modes: text-editing mode and face-generation mode. These two modes are jointly used via the latent space blending (Sec. 3.6) for the personalized generation. For example, given one of reference subject images, the text-ed… view at source ↗
Figure 3
Figure 3. We propose Dense-Face for personalized image generation, which introduces additional components, such as a pose-controllable (PC) adapter, pose branch (i.e., ϵpose) and annotation prediction module (i.e., ϵdense) on the top of the pre-trained T2I-SD. The input includes captions, head pose and reference image (i.e., Ipose and Iid). The output includes generated faces (Itar.) and dense face annotations (e.g., face dep… view at source ↗
Figures from the paper (10 more)
Figure 4
Figure 4. Figure 4: Given the reference image and target head pose (not shown in Figure), the proposed method generates the id-preserved image, depth map, pseudo mask and dense landmark. Such annotation predictions follow the insight that human faces possess inherent structural characteri…
Figure 5
Figure 5. Figure 5: We use various image caption methods via the third-party implementation [60] to generate captions. These captions have different text-image alignment scores. We only include captions with three highest scores in training samples. Landmark and Pseudo Face Mask According…
Figure 7
Figure 7. Figure 7: Algorithm Block [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]
Figure 9
Figure 9. Figure 9: Dense-Face can place subjects in diverse contexts with changed attributes, such as hair color and clothes. Model FID ↓ ID Similarity ↑ Pose Accuracy ↓ Full 100.3 0.357 5.83 w/o PC-ada 109.2 0.151 5.74 w/o (PC-ada + ID bran.) 113.3 0.093 5.71 w/o pose bran. 115.4 0.321 …
Figure 1
Figure 1. Figure 1: Additional comparisons among different personalized generation methods. Our proposed Dense-Face generates images with a consistent identity with the reference image, which can even be an old photo [PITH_FULL_IMAGE:figures/full_fig_p017_1.png]
Figure 2
Figure 2. Figure 2: Dense-Face can place subjects in diverse contexts with changed attributes, such as hair color and clothes [PITH_FULL_IMAGE:figures/full_fig_p018_2.png]
Figure 3
Figure 3. Figure 3: Two samples from the proposed T2I-Dense-Face. Reference image Generated image Depth image Pseudo mask Dense landmarks [PITH_FULL_IMAGE:figures/full_fig_p019_3.png]
Figure 4
Figure 4. Figure 4: Additional visualizations on the dense annotation prediction. The proposed Dense-Face can generate high-fidelity identity-preserved images and corresponding an￾notations (e.g., depth image, pseudo mask, and landmarks). Generated images can be at large pose-views [PITH…
Figure 5
Figure 5. Figure 5: Additional visualizations on different subject stylizations. CVL Computer Vision Lab Source Target Faceshifter HiFi-Face Dense￾Face Source Target Faceshifter HiFi-Face Dense￾Face [PITH_FULL_IMAGE:figures/full_fig_p020_5.png]
Figure 6
Figure 6. Figure 6: Additional Face Swapping results from the proposed method. Dense-Face achieves a comparable identity preservation to that of the previous work [PITH_FULL_IMAGE:figures/full_fig_p020_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

71 extracted references · 54 canonical work pages

  1. [1]

    JAMA network open (2020) 4

    AbdAlmageed, W., Mirzaalian, H., Guo, X., Randolph, L.M., Tanawat- tanacharoen, V.K., Geffner, M.E., Ross, H.M., Kim, M.S.: Assessment of facial morphologic features in patients with congenital adrenal hyperplasia using deep learning. JAMA network open (2020) 4

    work page 2020

  2. [2]

    ACM Transac- tions on Graphics (TOG)42(4), 1–11 (2023) 3, 9, 10

    Avrahami, O., Fried, O., Lischinski, D.: Blended latent diffusion. ACM Transac- tions on Graphics (TOG)42(4), 1–11 (2023) 3, 9, 10

    work page 2023

  3. [3]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Avrahami, O., Lischinski, D., Fried, O.: Blended diffusion for text-driven editing of natural images. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 18208–18218 (2022) 3, 9, 10

    work page 2022

  4. [4]

    In: International Conference on Learning Representations (2021) 3

    Baranchuk, D., Voynov, A., Rubachev, I., Khrulkov, V., Babenko, A.: Label- efficient semantic segmentation with diffusion models. In: International Conference on Learning Representations (2021) 3

    work page 2021

  5. [5]

    In: ICCV (2017) 9

    Bulat, A., Tzimiropoulos, G.: How far are we from solving the 2d & 3d face align- ment problem?(and a dataset of 230,000 3d facial landmarks). In: ICCV (2017) 9

    work page 2017

  6. [6]

    In: Proceedings of the IEEE/CVF international conference on computer vision

    Caron, M., Touvron, H., Misra, I., Jégou, H., Mairal, J., Bojanowski, P., Joulin, A.: Emerging properties in self-supervised vision transformers. In: Proceedings of the IEEE/CVF international conference on computer vision. pp. 9650–9660 (2021) 10

    work page 2021

  7. [7]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Chan, E.R., Lin, C.Z., Chan, M.A., Nagano, K., Pan, B., De Mello, S., Gallo, O., Guibas, L.J., Tremblay, J., Khamis, S., et al.: Efficient geometry-aware 3d generative adversarial networks. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 16123–16133 (2022) 2

    work page 2022

  8. [8]

    In: Proceedings of the 28th ACM International Conference on Multimedia

    Chen, R., Chen, X., Ni, B., Ge, Y.: Simswap: An efficient framework for high fidelity face swapping. In: Proceedings of the 28th ACM International Conference on Multimedia. pp. 2003–2011 (2020) 14

    work page 2020

Show all 71 references
  1. [9]

    arXiv preprint arXiv:2302.08908 (2023) 4, 6

    Cheng, J., Liang, X., Shi, X., He, T., Xiao, T., Li, M.: Layoutdiffuse: Adapt- ing foundational diffusion models for layout-to-image generation. arXiv preprint arXiv:2302.08908 (2023) 4, 6

    2023 arXiv

  2. [10]

    arXiv preprint arXiv:2311.00213 (2023) 4

    Cheng, J., Xiao, T., He, T.: Consistent video-to-video transfer using synthetic dataset. arXiv preprint arXiv:2311.00213 (2023) 4

    2023 arXiv

  3. [11]

    arXiv preprint arXiv:2409.04847 (2024) 4

    Cheng, J., Zhao, Z., He, T., Xiao, T., Zhou, Y., Zhang, Z.: Rethinking the training and evaluation of rich-context layout-to-image generation. arXiv preprint arXiv:2409.04847 (2024) 4

    2024 arXiv

  4. [12]

    In: IEEE Conf

    Choi, Y., Uh, Y., Yoo, J., Ha, J.W.: Stargan v2: Diverse image synthesis for mul- tiple domains. In: IEEE Conf. Comput. Vis. Pattern Recog. (2020) 2

    2020

  5. [13]

    arXiv preprint arXiv:2210.11416 (2022) 9

    Chung, H.W., Hou, L., Longpre, S., Zoph, B., Tay, Y., Fedus, W., Li, Y., Wang, X., Dehghani, M., Brahma, S., et al.: Scaling instruction-finetuned language models. arXiv preprint arXiv:2210.11416 (2022) 9

    2022 arXiv

  6. [14]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Deng, J., Guo, J., Xue, N., Zafeiriou, S.: Arcface: Additive angular margin loss for deep face recognition. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 4690–4699 (2019) 6, 11

    2019

  7. [15]

    In: Proceedings of the European conference on computer vision (ECCV)

    Feng, Y., Wu, F., Shao, X., Wang, Y., Zhou, X.: Joint 3d face reconstruction and dense alignment with position map regression network. In: Proceedings of the European conference on computer vision (ECCV). pp. 534–551 (2018) 3

    2018

  8. [16]

    In: The Eleventh International Conference on Learning Representations (2022) 2, 4, 10 8 X

    Gal, R., Alaluf, Y., Atzmon, Y., Patashnik, O., Bermano, A.H., Chechik, G., Cohen-or, D.: An image is worth one word: Personalizing text-to-image generation using textual inversion. In: The Eleventh International Conference on Learning Representations (2022) 2, 4, 10 8 X. Guo et al

    2022

  9. [17]

    ACM Transactions on Graphics (TOG)42(4), 1–13 (2023) 2, 4

    Gal, R., Arar, M., Atzmon, Y., Bermano, A.H., Chechik, G., Cohen-Or, D.: Encoder-based domain tuning for fast personalization of text-to-image models. ACM Transactions on Graphics (TOG)42(4), 1–13 (2023) 2, 4

    2023

  10. [18]

    ACM Transactions on Graphics (TOG)42(4), 1–13 (2023) 6

    Gal, R., Arar, M., Atzmon, Y., Bermano, A.H., Chechik, G., Cohen-Or, D.: Encoder-based domain tuning for fast personalization of text-to-image models. ACM Transactions on Graphics (TOG)42(4), 1–13 (2023) 6

    2023

  11. [19]

    Gu, J., Wang, Y., Zhao, N., Fu, T.J., Xiong, W., Liu, Q., Zhang, Z., Zhang, H., Zhang, J., Jung, H., Wang, X.E.: Photoswap: Personalized subject swapping in images (2023) 3, 4, 6, 11

    2023

  12. [20]

    In: Proceeding of Thirty-eighth Conference on Neural Information Processing Systems

    Guo, X., Asnani, V., Liu, S., Liu, X.: Tracing hyperparameter dependencies for model parsing via learnable graph pooling network. In: Proceeding of Thirty-eighth Conference on Neural Information Processing Systems. Vancouver, Canada (De- cember 2024) 14

    2024

  13. [21]

    In: AAAI (2019) 4

    Guo, X., Choi, J.: Human motion prediction via learning local structure represen- tations and temporal dependencies. In: AAAI (2019) 4

    2019

  14. [22]

    In: International Journal of Computer Vision (December 2024) 14

    Guo, X., Liu, X., Masi, I., Liu, X.: Language-guided hierarchical fine-grained image forgery detection and localization. In: International Journal of Computer Vision (December 2024) 14

    2024

  15. [23]

    In: In Proceeding of IEEE Computer Vision and Pattern Recognition (2023) 14

    Guo, X., Liu, X., Ren, Z., Grosz, S., Masi, I., Liu, X.: Hierarchical fine-grained image forgery detection and localization. In: In Proceeding of IEEE Computer Vision and Pattern Recognition (2023) 14

    2023

  16. [24]

    In: ECCV (2022) 4

    Guo, X., Liu, Y., Jain, A., Liu, X.: Multi-domain learning for updating face anti- spoofing models. In: ECCV (2022) 4

    2022

  17. [25]

    arXiv preprint arXiv:2401.00551 (2023) 4

    Han, Y., Zhang, J., Zhu, J., Li, X., Ge, Y., Li, W., Wang, C., Liu, Y., Liu, X., Tai, Y.: A generalist facex via learning unified facial representation. arXiv preprint arXiv:2401.00551 (2023) 4

    2023 arXiv

  18. [26]

    Advances in Neural Information Processing Systems33, 6840–6851 (2020) 5, 14

    Ho, J., Jain, A., Abbeel, P.: Denoising diffusion probabilistic models. Advances in Neural Information Processing Systems33, 6840–6851 (2020) 5, 14

    2020

  19. [27]

    In: Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition

    Huang, Z., Chan, K.C., Jiang, Y., Liu, Z.: Collaborative diffusion for multi-modal face generation and editing. In: Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition. pp. 6080–6090 (2023) 4, 6

    2023

  20. [28]

    Ju, X., Zeng, A., Zhao, C., Wang, J., Zhang, L., Xu, Q.: HumanSD: A native skeleton-guided diffusion model for human image generation (2023) 3, 4

    2023

  21. [29]

    In: CVPR

    Karras, T., Laine, S., Aila, T.: A style-based generator architecture for generative adversarial networks. In: CVPR. pp. 4401–4410 (2019) 2, 11

    2019

  22. [30]

    In: CVPR (2020) 2

    Karras, T., Laine, S., Aittala, M., Hellsten, J., Lehtinen, J., Aila, T.: Analyzing and improving the image quality of stylegan. In: CVPR (2020) 2

    2020

  23. [31]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Kim, M., Jain, A.K., Liu, X.: Adaface: Quality adaptive margin for face recogni- tion. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 18750–18759 (2022) 11

    2022

  24. [32]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Kim, Y., Lee, J., Kim, J.H., Ha, J.W., Zhu, J.Y.: Dense text-to-image genera- tion with attention modulation. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 7701–7711 (2023) 3

    2023

  25. [33]

    In: Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition

    Kumari, N., Zhang, B., Zhang, R., Shechtman, E., Zhu, J.Y.: Multi-concept cus- tomization of text-to-image diffusion. In: Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition. pp. 1931–1941 (2023) 2, 4, 11

    2023

  26. [34]

    arXiv preprint arXiv:2301.12597 (2023) 9 Title Suppressed Due to Excessive Length 9

    Li, J., Li, D., Savarese, S., Hoi, S.: Blip-2: Bootstrapping language-image pre- training with frozen image encoders and large language models. arXiv preprint arXiv:2301.12597 (2023) 9 Title Suppressed Due to Excessive Length 9

    2023 arXiv

  27. [35]

    In: International Con- ference on Machine Learning

    Li, J., Li, D., Xiong, C., Hoi, S.: Blip: Bootstrapping language-image pre-training for unified vision-language understanding and generation. In: International Con- ference on Machine Learning. pp. 12888–12900. PMLR (2022) 9

    2022

  28. [36]

    CVPR (2020) 2, 14

    Li, L., Bao, J., Yang, H., Chen, D., Wen, F.: Faceshifter: Towards high fidelity and occlusion aware face swapping. CVPR (2020) 2, 14

    2020

  29. [37]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Li,Y., Liu, H.,Wu,Q., Mu, F., Yang, J.,Gao, J.,Li,C., Lee, Y.J.:Gligen: Open-set grounded text-to-image generation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 22511–22521 (2023) 2, 4

    2023

  30. [38]

    In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2024) 1, 4, 11

    Li, Z., Cao, M., Wang, X., Qi, Z., Cheng, M.M., Shan, Y.: Photomaker: Cus- tomizing realistic human photos via stacked id embedding. In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2024) 1, 4, 11

    2024

  31. [39]

    In: Proceedings of the IEEE/CVF Interna- tional Conference on Computer Vision

    Li, Z., Zhou, Q., Zhang, X., Zhang, Y., Wang, Y., Xie, W.: Open-vocabulary object segmentation with diffusion models. In: Proceedings of the IEEE/CVF Interna- tional Conference on Computer Vision. pp. 7667–7676 (2023) 3

    2023

  32. [40]

    arXiv preprint arXiv:2401.01207 (2024) 3, 9, 10

    Liu, R., Ma, B., Zhang, W., Hu, Z., Fan, C., Lv, T., Ding, Y., Cheng, X.: To- wards a simultaneous and granular identity-expression control in personalized face generation. arXiv preprint arXiv:2401.01207 (2024) 3, 9, 10

    2024 arXiv

  33. [41]

    arXiv preprint arXiv:2310.08579 (2023) 3, 4

    Liu, X., Ren, J., Siarohin, A., Skorokhodov, I., Li, Y., Lin, D., Liu, X., Liu, Z., Tulyakov, S.: Hyperhuman: Hyper-realistic human generation with latent struc- tural diffusion. arXiv preprint arXiv:2310.08579 (2023) 3, 4

    2023 arXiv

  34. [42]

    Retrieved August15(2018), 11 (2018) 8

    Liu, Z., Luo, P., Wang, X., Tang, X.: Large-scale celebfaces attributes (celeba) dataset. Retrieved August15(2018), 11 (2018) 8

    2018

  35. [43]

    arXiv preprint arXiv:2409.10958 (2024) 14

    Pan, Y., Liu, X., Luo, S., Xin, Y., Guo, X., Liu, X., Min, X., Zhai, G.: Towards effective user attribution for latent diffusion models via watermark-informed blend- ing. arXiv preprint arXiv:2409.10958 (2024) 14

    2024 arXiv

  36. [44]

    arXiv preprint arXiv:2312.06354 (2023) 3

    Peng, X., Zhu, J., Jiang, B., Tai, Y., Luo, D., Zhang, J., Lin, W., Jin, T., Wang, C., Ji, R.: Portraitbooth: A versatile portrait model for fast identity-preserved personalization. arXiv preprint arXiv:2312.06354 (2023) 3

    2023 arXiv

  37. [45]

    In: International conference on machine learning

    Radford, A., Kim, J.W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., et al.: Learning transferable visual models from natural language supervision. In: International conference on machine learning. pp. 8748–8763. PMLR (2021) 11

    2021

  38. [46]

    In: CVPR (2022) 11

    Rombach, R., Blattmann, A., Lorenz, D., Esser, P., Ommer, B.: High-resolution image synthesis with latent diffusion models. In: CVPR (2022) 11

    2022

  39. [47]

    In: MICCAI

    Ronneberger, O., Fischer, P., Brox, T.: U-net: Convolutional networks for biomed- ical image segmentation. In: MICCAI. pp. 234–241 (2015) 5

    2015

  40. [48]

    In: ICCV (2019) 14

    Rössler, A., Cozzolino, D., Verdoliva, L., Riess, C., Thies, J., Nießner, M.: Face- forensics++: Learning to detect manipulated facial images. In: ICCV (2019) 14

    2019

  41. [49]

    In: The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops (June 2018) 8, 12

    Ruiz, N., Chong, E., Rehg, J.M.: Fine-grained head pose estimation without key- points. In: The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops (June 2018) 8, 12

    2018

  42. [50]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Ruiz, N., Li, Y., Jampani, V., Pritch, Y., Rubinstein, M., Aberman, K.: Dream- booth: Fine tuning text-to-image diffusion models for subject-driven generation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 22500–22510 (2023) 2, 4, 11

    2023

  43. [51]

    Advances in Neural Information Processing Systems35, 25278–25294 (2022) 2 10 X

    Schuhmann, C., Beaumont, R., Vencu, R., Gordon, C., Wightman, R., Cherti, M., Coombes, T., Katta, A., Mullis, C., Wortsman, M., et al.: Laion-5b: An open large- scale dataset for training next generation image-text models. Advances in Neural Information Processing Systems35, 2...

    2022

  44. [52]

    In: NeurIPS (2024) 14

    Song, X., Guo, X., Zhang, J., Li, Q., Bai, L., Liu, X., Zhai, G., Liu, X.: On learn- ing multi-modal forgery representation for diffusion generated video detection. In: NeurIPS (2024) 14

    2024

  45. [53]

    In: Proceedings of the Euro- pean conference on computer vision (ECCV)

    Tang, Z., Peng, X., Geng, S., Wu, L., Zhang, S., Metaxas, D.: Quantized densely connected u-nets for efficient landmark localization. In: Proceedings of the Euro- pean conference on computer vision (ECCV). pp. 339–354 (2018) 9

    2018

  46. [54]

    arXiv preprint arXiv:2308.12469 (2023) 3

    Tian, J., Aggarwal, L., Colaco, A., Kira, Z., Gonzalez-Franco, M.: Diffuse, attend, and segment: Unsupervised zero-shot segmentation using stable diffusion. arXiv preprint arXiv:2308.12469 (2023) 3

    2023 arXiv

  47. [55]

    arXiv preprint arXiv:2401.15975 (2024) 2

    Wang, Q., Jia, X., Li, X., Li, T., Ma, L., Zhuge, Y., Lu, H.: Stableidentity: Inserting anybody into anywhere at first sight. arXiv preprint arXiv:2401.15975 (2024) 2

    2024 arXiv

  48. [56]

    arXiv preprint arXiv:2401.07519 (2024) 1, 4, 6, 11

    Wang, Q., Bai, X., Wang, H., Qin, Z., Chen, A.: Instantid: Zero-shot identity- preserving generation in seconds. arXiv preprint arXiv:2401.07519 (2024) 1, 4, 6, 11

    2024 arXiv

  49. [57]

    In: The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2021) 8

    Wang, X., Li, Y., Zhang, H., Shan, Y.: Towards real-world blind face restoration with generative facial prior. In: The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2021) 8

    2021

  50. [58]

    arXiv preprint arXiv:2106.09965 (2021) 14

    Wang, Y., Chen, X., Zhu, J., Chu, W., Tai, Y., Wang, C., Li, J., Wu, Y., Huang, F., Ji, R.: Hififace: 3d shape and semantic prior guided high fidelity face swapping. arXiv preprint arXiv:2106.09965 (2021) 14

    2021 arXiv

  51. [59]

    arXiv preprint arXiv:2302.13848 (2023) 4

    Wei, Y., Zhang, Y., Ji, Z., Bai, J., Zhang, L., Zuo, W.: Elite: Encoding visual concepts into textual embeddings for customized text-to-image generation. arXiv preprint arXiv:2302.13848 (2023) 4

    2023 arXiv

  52. [60]

    arXiv preprint arXiv:1910.03771 (2019) 9

    Wolf, T., Debut, L., Sanh, V., Chaumond, J., Delangue, C., Moi, A., Cistac, P., Rault, T., Louf, R., Funtowicz, M., et al.: Huggingface’s transformers: State-of- the-art natural language processing. arXiv preprint arXiv:1910.03771 (2019) 9

    2019 arXiv

  53. [61]

    In: European Conference on Computer Vision

    Wood, E., Baltrušaitis, T., Hewitt, C., Johnson, M., Shen, J., Milosavljević, N., Wilde, D., Garbin, S., Sharp, T., Stojiljković, I., et al.: 3d face reconstruction with dense landmarks. In: European Conference on Computer Vision. pp. 160–

  54. [62]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Xu, J., Liu, S., Vahdat, A., Byeon, W., Wang, X., De Mello, S.: Open-vocabulary panoptic segmentation with text-to-image diffusion models. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 2955– 2966 (2023) 3

    2023

  55. [63]

    Foundations and Trends in Privacy and Security (January 2024) 14

    Yao, Y., Guo, X., Asnani, V., Gong, Y., Liu, J., Lin, X., Liu, X., Liu, S.: Reverse engineering of deceptions on machine- and human-centric attacks. Foundations and Trends in Privacy and Security (January 2024) 14

    2024

  56. [64]

    Ye, H., Zhang, J., Liu, S., Han, X., Yang, W.: Ip-adapter: Text compatible image prompt adapter for text-to-image diffusion models (2023) 1, 11

    2023

  57. [65]

    arXiv preprint arXiv:1411.7923 (2014) 8

    Yi, D., Lei, Z., Liao, S., Li, S.Z.: Learning face representation from scratch. arXiv preprint arXiv:1411.7923 (2014) 8

    2014 arXiv

  58. [66]

    Advances in Neural Information Processing Systems36 (2024) 3

    Zhang, J., Herrmann, C., Hur, J., Polania Cabrera, L., Jampani, V., Sun, D., Yang, M.H.: A tale of two features: Stable diffusion complements dino for zero-shot semantic correspondence. Advances in Neural Information Processing Systems36 (2024) 3

    2024

  59. [67]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Zhang, L., Rao, A., Agrawala, M.: Adding conditional control to text-to-image diffusion models. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 3836–3847 (2023) 2, 4, 6, 11

    2023

  60. [68]

    arXiv preprint arXiv:2205.01068 (2022) 9 Title Suppressed Due to Excessive Length 11

    Zhang, S., Roller, S., Goyal, N., Artetxe, M., Chen, M., Chen, S., Dewan, C., Diab, M., Li, X., Lin, X.V., et al.: Opt: Open pre-trained transformer language models. arXiv preprint arXiv:2205.01068 (2022) 9 Title Suppressed Due to Excessive Length 11

    2022 arXiv

  61. [69]

    arXiv preprint arXiv:2306.07596 (2023) 3, 9, 10

    Zhang, X., Guo, J., Yoo, P., Matsuo, Y., Iwasawa, Y.: Paste, inpaint and harmonize via denoising: Subject-driven image editing with pre-trained diffusion model. arXiv preprint arXiv:2306.07596 (2023) 3, 9, 10

    2023 arXiv

  62. [70]

    In: European Conference on Computer Vision (2025) 14

    Zhang, Y., Colman, B., Guo, X., Shahriyari, A., Bharaj, G.: Common sense reason- ing for deepfake detection. In: European Conference on Computer Vision (2025) 14

    2025

  63. [177]

    Springer (2022) 3, 9

    2022

Pith tools

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