Pith. sign in

REVIEW 3 major objections 5 minor 88 references

Tuning-Free Latent Diffusion Models for Ultrahigh-Resolution Image Editing

T0 review · 3 major / 5 minor · reviewed 2026-07-08 · glm-5.2

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

Pith's one-line read Editing 8K images on a single GPU — no retraining required

desk verdict UltraDiffEdit combines existing patch-based, dilated, and upsample-guidance sampling into a coherent editing pipeline that works at 8K on a single 3090. The engineering is real and the code is public, but the U-IDS metric is almost certainly misconfigured, which undermines one of the key quality claims. read the letter →

arxiv 2607.06136 v1 pith:ZHJGTL7A submitted 2026-07-07 cs.CV cs.MM

Wanglong Lu , Lingming Su , Kaijie Shi , Minglun Gong , Xiaogang Jin , Hanli Zhao , Xianta Jiang This is my paper
classification cs.CVcs.MM
keywords editingimagelatenteditedresolutionsultradiffeditmodelsunedited
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper introduces UltraDiffEdit, a framework that extends pre-trained latent diffusion models (LDMs) to edit images at resolutions up to 8K using a single consumer-grade GPU with 24 GB of memory. The core problem is that LDMs are trained at fixed low resolutions (512 or 1024 pixels per side), and existing approaches either fail to preserve unedited regions, produce boundary artifacts, or run out of memory at high resolutions. UltraDiffEdit addresses this through three mechanisms: multi-patch encoding, which splits a high-resolution image into overlapping patches, encodes each separately, and reassembles them into a full latent feature map to preserve both edited and unedited detail; global-local consistency denoising, which uses a mask-guided cosine-decay fusion at every diffusion step to keep unedited regions stable while smoothly introducing edited content; and patch-based hybrid sampling, which combines local patch denoising, dilated global sampling, and a patch-based upsample guidance to capture detail at multiple spatial scales simultaneously. The pipeline operates in a coarse-to-fine multi-scale loop, using low-resolution edited results as structural references for the next higher scale. The authors demonstrate that this combination achieves leading PSNR, SSIM, FID, and LPIPS scores across three benchmark datasets compared to methods including DemoFusion, SDXL with super-resolution, and Flux, while keeping memory usage within 24 GB at resolutions where SDXL itself fails.

What carries the argument

Three components: (1) Multi-patch encoding — a shifted-crop function splits a high-resolution image into overlapping patches at the encoder's native resolution, encodes each patch separately, and reassembles them by averaging overlapping regions into a full-scale latent map. (2) Global-local consistency denoising — at each diffusion step t, a cosine-decay factor gamma_1 blends the diffused latent with the denoised latent, then a binary mask routes the blend so edited regions receive the fused signal while unedited regions retain their original latent values, preventing progressive corruption. (3) Patch-based hybrid sampling — the denoised latent is computed as a weighted sum of three sub-sam

What would settle it

If the cosine-decay fusion parameters beta_1=3 and beta_2=1 produce visible boundary artifacts or unedited-region distortion on a held-out dataset with different characteristics (e.g., aerial photography, medical scans, or non-natural images), the generality of the global-local consistency denoising claim would be undermined.

Watch

Extended reading notes

Core claim

The central discovery is that the three failure modes of high-resolution LDM editing — loss of unedited region detail, boundary artifacts between edited and unedited areas, and insufficient multi-scale receptive fields during denoising — can each be addressed by a specific, training-free intervention in latent space, and that these interventions compose into a pipeline that scales to 8K resolution on a single GPU. Multi-patch encoding preserves unedited detail by never downsampling the full image through the encoder at once. Global-local consistency denoising replaces DemoFusion's undifferentiated skip-residual fusion with mask-aware blending that decays over diffusion steps, keeping uneditd

Load-bearing premise

The fusion schedules that blend edited and unedited latent features at each diffusion step use hand-tuned decay parameters (beta_1=3 for mask-guided denoising, beta_2=1 for multi-scale sampling) with no systematic sensitivity analysis, so the entire quality advantage depends on these specific values being effective across all image types and editing scenarios.

Editorial extensions

If this is right

  • Any pre-trained LDM can be extended to ultra-high-resolution editing without fine-tuning, lowering the barrier for professional-grade image editing on consumer hardware.
  • The mask-aware denoising strategy could be adopted by existing high-resolution generation methods (e.g., DemoFusion, AccDiffusion) to support real-image editing workflows they currently cannot handle.
  • The multi-patch encoding approach generalizes to any encoder with fixed input size, suggesting applicability beyond diffusion models to other latent-space architectures.
  • The three benchmark datasets (DIV2KEdit, Syn2KEdit, UHRSDEdit) establish a standardized evaluation protocol for high-resolution image editing that future work can build on.

Reading between the lines

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

  • The cosine-decay fusion schedules (gamma_1 with beta_1=3, gamma_2 with beta_2=1) are the load-bearing hyperparameters — if they are overfit to the three benchmark datasets, the method may degrade on image types with very different texture statistics (e.g., medical imaging, satellite imagery, or non-photorealistic art).
  • The inference time grows roughly quadratically with resolution (confirmed by the authors' own Figure 13), meaning 8K editing takes approximately 16x longer than 2K — this limits practical use to offline workflows despite the memory feasibility.
  • The patch-based upsample guidance mechanism (Eq. 7) introduces an additional scale factor r=2 and a guidance weight w_t, adding a second level of patch decomposition inside an already patch-heavy pipeline; the interaction between these nested patch operations is not formally analyzed and could produce compounding boundary effects at extreme resolutions.
  • The method's reliance on the underlying LDM's quality (acknowledged by the authors in Figure 14) means improvements in base models transfer automatically, but so do their limitations — biases, artifacts, and prompt-misalignment issues propagate unchanged into high-resolution outputs.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. This paper introduces UltraDiffEdit, a tuning-free framework for extending pre-trained latent diffusion models (LDMs) to ultra-high-resolution (up to 8K) image editing. The method employs a multi-scale progressive editing strategy that iteratively blends high-resolution edited content with unedited areas in a coarse-to-fine manner. Key technical contributions include multi-patch encoding to preserve unedited visual details in the latent space, global-local consistency denoising to ensure smooth transitions at editing boundaries, and a patch-based hybrid sampling approach that integrates local, intermediate, and global features. The framework is evaluated on three newly introduced benchmark datasets (DIV2KEdit, Syn2KEdit, UHRSDEdit) against several baselines, demonstrating superior performance across metrics like PSNR, SSIM, FID, and LPIPS while operating within a single 24GB GPU memory budget.

Significance. The paper addresses a highly relevant practical problem: applying pre-trained LDMs to ultra-high-resolution image editing without requiring fine-tuning or excessive GPU memory. The approach of decomposing the editing process into multi-patch encoding, boundary-aware denoising, and hybrid sampling is well-motivated and technically sound. The authors provide clear pseudo-code (Algorithm 1) and mathematical formulations for each component. The release of source code and three new benchmark datasets adds significant value to the reproducibility and future research in the community. The demonstration of 8K image editing on a single RTX 3090 is a notable engineering achievement.

major comments (3)
  1. Table I: The U-IDS metric returns exactly 0.00% for all seven baselines on both DIV2KEdit and Syn2KEdit, while UltraDiffEdit scores 4.58% and 5.55%. In Table II, U-IDS is again 0.00% for all eight compared generation methods. Getting exactly 0.00% across multiple diverse methods and datasets strongly suggests either a threshold set so high that only the method's own outputs pass, or an implementation issue in how baseline images are fed to the discriminator. If U-IDS is systematically broken for baselines, it is uninformative and should be either corrected or removed. The authors should verify the metric implementation or clarify the thresholding mechanism.
  2. Section IV.B and Table I: The evaluation framework may conflate region preservation with editing quality. PSNR and SSIM are computed against the original unedited image, inherently rewarding methods that preserve unmasked areas. Since UltraDiffEdit's specific design goal is multi-patch encoding + global-local consistency denoising to preserve unedited areas, high PSNR/SSIM scores are expected by construction and do not necessarily indicate better edits in the masked regions. The crop-based metrics (FID crop, LPIPS crop) partially address this but still include unedited patches. The authors should clarify this distinction in the text and ideally report metrics computed strictly on the masked/edited regions to isolate editing quality from background preservation.
  3. Section III.C, Eq. (5) and Section III.D, Eq. (6): The hybrid sampling weights use cosine decay factors γ₁ and γ₂ with scaling factors β₁=3 and β₂=1, and the upsample guidance uses w_t = 0.2×I(t>500). These are hand-tuned hyperparameters set without systematic justification or sensitivity analysis. The entire quality advantage over DemoFusion's skip-residual approach hinges on these specific fusion schedules being universally effective. The authors should provide a sensitivity analysis for β₁, β₂, and w_t across diverse image types to demonstrate that the method is not overfit to the three benchmark datasets.
minor comments (5)
  1. Section IV.A: The benchmark datasets (DIV2KEdit, Syn2KEdit, UHRSDEdit) are self-created. While the authors describe their construction in the supplementary document, the evaluation on self-created datasets introduces a mild circularity concern. The authors should consider evaluating on at least one externally established high-resolution editing benchmark if one exists, or explicitly acknowledge this limitation.
  2. Fig. 2: The notation in the figure (e.g., subscripts and superscripts for latent variables like z_t^s) is somewhat difficult to parse due to formatting. Improving the clarity of the labels in this figure would aid reader comprehension.
  3. Section IV.C, Fig. 10: The time cost differences for various strides (e.g., 10198.4s vs. 9561.6s) are reported, but the resolution at which these times were measured is not explicitly stated in the figure caption. Please specify the image resolution used for this timing experiment.
  4. Table I: Several baselines report 'N/A' for UHRSDEdit due to out-of-memory (OOM) errors. It would be helpful to specify the GPU memory limit (e.g., 24GB or 80GB) at which these baselines encountered OOM, to provide context for the feasibility comparison.
  5. Section IV.D, Table V: The runtime and memory comparison is conducted on an NVIDIA A800 (80GB), whereas the main experiments and the core claim of single-GPU feasibility are demonstrated on an RTX 3090 (24GB). Please clarify why the A800 was used for this specific table and ensure the memory usage reported for UltraDiffEdit (11,014 MB) is consistent with the 24GB GPU claim.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the thorough and constructive review. The referee raises three major points concerning (1) the U-IDS metric returning 0.00% for all baselines, (2) the conflation of region preservation with editing quality in PSNR/SSIM, and (3) the lack of sensitivity analysis for hand-tuned hyperparameters. We address each point below and describe the revisions we will make.

read point-by-point responses

  1. Referee: Table I: The U-IDS metric returns exactly 0.00% for all seven baselines on both DIV2KEdit and Syn2KEdit, while UltraDiffEdit scores 4.58% and 5.55%. In Table II, U-IDS is again 0.00% for all eight compared generation methods. Getting exactly 0.00% across multiple diverse methods and datasets strongly suggests either a threshold set so high that only the method's own outputs pass, or an implementation issue in how baseline images are fed to the discriminator. If U-IDS is systematically broken for baselines, it is uninformative and should be either corrected or removed. The authors should verify the metric implementation or clarify the thresholding mechanism.

    Authors: The referee is correct that uniform 0.00% across all baselines is suspicious and warrants investigation. We have re-examined our U-IDS implementation, which follows the protocol from the original Co-Modulated GAN paper (Zhao et al., ICLR 2021). The U-IDS metric uses a discriminator trained on real images to detect whether edited images are perceptually distinguishable from unedited ones. The 0.00% values arise because the baseline methods (SDXL+bicubic, DemoFusion, etc.) produce outputs that the discriminator consistently flags as non-real, particularly at 2K resolution where upscaling artifacts are pronounced. However, we acknowledge that this explanation is insufficient and that the metric may not be discriminative enough to serve as a useful comparison axis in its current form. We agree with the referee's recommendation. In the revision, we will either (a) correct the implementation if a bug is found upon further verification, or (b) remove U-IDS from the main tables if the metric cannot reliably distinguish among methods. We will also add a clarifying note about the metric's limitations in our setting. revision: yes

  2. Referee: Section IV.B and Table I: The evaluation framework may conflate region preservation with editing quality. PSNR and SSIM are computed against the original unedited image, inherently rewarding methods that preserve unmasked areas. Since UltraDiffEdit's specific design goal is multi-patch encoding + global-local consistency denoising to preserve unedited areas, high PSNR/SSIM scores are expected by construction and do not necessarily indicate better edits in the masked regions. The crop-based metrics (FID crop, LPIPS crop) partially address this but still include unedited patches. The authors should clarify this distinction in the text and ideally report metrics computed strictly on the masked/edited regions to isolate editing quality from background preservation.

    Authors: This is a fair and important point. We agree that PSNR and SSIM, computed over the full image, inherently reward preservation of unedited regions, and that UltraDiffEdit's design is specifically optimized for this. We will add an explicit discussion in Section IV.B clarifying that PSNR and SSIM primarily reflect region preservation fidelity (which is a core design goal of our method), while perceptual metrics (FID, LPIPS, CLIP-S) better reflect editing quality. More importantly, we will add masked-region-only metrics in the revision: we will compute PSNR, SSIM, and LPIPS strictly on the masked/edited regions for all methods on DIV2KEdit and Syn2KEdit, so that editing quality can be assessed independently of background preservation. This will require re-running evaluations with masked-region cropping, which we will complete for the revision. revision: yes

  3. Referee: Section III.C, Eq. (5) and Section III.D, Eq. (6): The hybrid sampling weights use cosine decay factors gamma_1 and gamma_2 with scaling factors beta_1=3 and beta_2=1, and the upsample guidance uses w_t = 0.2*I(t>500). These are hand-tuned hyperparameters set without systematic justification or sensitivity analysis. The entire quality advantage over DemoFusion's skip-residual approach hinges on these specific fusion schedules being universally effective. The authors should provide a sensitivity analysis for beta_1, beta_2, and w_t across diverse image types to demonstrate that the method is not overfit to the three benchmark datasets.

    Authors: We agree that a sensitivity analysis for these hyperparameters would strengthen the paper and address the concern about overfitting. We will add a sensitivity study in the revision that varies beta_1 in {1, 2, 3, 4, 5}, beta_2 in {0.5, 1, 2, 3}, and w_t in {0.1, 0.2, 0.3, 0.4} across all three benchmark datasets, reporting FID and LPIPS for each configuration. We note that the cosine decay schedule itself is motivated by the diffusion process dynamics (early steps establish global structure, later steps refine detail), so the functional form is principled rather than arbitrary. The scaling factors control the relative emphasis and were selected based on qualitative pilot experiments, but we acknowledge this was not documented. We will also add a brief discussion of why these schedules are expected to generalize: the decay follows the noise schedule of the diffusion model, which is dataset-independent. That said, we cannot fully rule out that alternative settings might perform better on other image distributions, and we will state this limitation honestly. revision: yes

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found

full rationale

The paper introduces UltraDiffEdit, a tuning-free framework for ultra-high-resolution image editing. The methodological derivation (Eqs. 1-7) is self-contained: it builds on externally published techniques (MultiDiffusion [18], DemoFusion [4], upsample guidance [57]) and uses pre-trained models (SDXL, ControlNet) as black-box inputs. The core technical contributions—multi-patch encoding (Eq. 4), global-local consistency denoising (Eq. 5), and patch-based hybrid sampling (Eq. 6)—are defined by their own equations with explicitly stated hyperparameters (β₁=3, β₂=1, r=2). No equation reduces to its own input by construction, and no 'prediction' is a renamed fit. The benchmark datasets (DIV2KEdit, Syn2KEdit, UHRSDEdit) are self-created but derive from external sources (DIV2K [20], UHRSD [21]) and are evaluated with standard metrics (PSNR, SSIM, FID, LPIPS, CLIP-S, U-IDS). The self-citations present (e.g., [22], [25], [84]) are for related prior work by the authors but are not load-bearing for the central derivation. The U-IDS=0.00% anomaly for baselines is a correctness/evaluation concern, not a circularity issue. No circularity detected.

Assumptions & free parameters 5 free parameters · 3 assumptions · 0 invented entities

The paper introduces no new entities (particles, forces, dimensions, etc.). It introduces algorithmic modules (multi-patch encoding, global-local consistency denoising, patch-based hybrid sampling) but these are engineering constructs, not postulated physical or mathematical objects. The free parameters are all hand-set hyperparameters without systematic optimization. The axioms are standard domain assumptions from the diffusion model literature plus one ad-hoc choice of decay schedule.

free parameters (5)
  • β₁ (denoising decay scaling factor) = 3
    Controls the cosine decay schedule for global-local consistency denoising fusion strength. Set by hand; no sensitivity analysis provided.
  • β₂ (sampling decay scaling factor) = 1
    Controls the cosine decay schedule for hybrid sampling weights. Set by hand; no sensitivity analysis provided.
  • r (upsample guidance scale factor) = 2
    Determines the kernel enlargement factor for upsample guidance sampling. Set by hand.
  • w_t (guidance scale weight) = 0.2 × I(t>500)
    Binary-onset guiding scale for upsample guidance. Set by hand without justification for the threshold or magnitude.
  • G (phase set) = {1, S}
    Controls the number of progressive editing stages. Authors test alternatives (Fig. 11) but default to two-stage.
assumptions (3)
  • domain assumption Pre-trained LDM encoders can be applied to sub-patches of a larger image and the resulting latents merged by averaging without introducing systematic artifacts.
    Section III.B, Eq. 4: multi-patch encoding assumes that averaging overlapping encoded sub-latent feature maps produces a valid large-scale latent. This is standard in MultiDiffusion but unproven for editing tasks where boundary consistency matters.
  • ad hoc to paper A cosine decay schedule is the appropriate temporal weighting for both denoising fusion and hybrid sampling.
    Eqs. 5-6: the choice of cosine decay (vs. linear, exponential, or learned schedules) is not justified; it is asserted.
  • domain assumption DDIM scheduler with 50 steps and guidance scale 7.5 is sufficient for all denoising paths.
    Section IV.A: fixed across all experiments without ablation on scheduler choice or step count.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Tuning-Free Latent Diffusion Models for Ultrahigh-Resolution Image Editing." pith.science (2026). https://pith.science/paper/ZHJGTL7A

@misc{pith2026260706136,
  author       = {Pith},
  title        = {Pith review of: Tuning-Free Latent Diffusion Models for Ultrahigh-Resolution Image Editing},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZHJGTL7A}},
  note         = {Machine review of arXiv:2607.06136}
}
read the original abstract

Recent diffusion-based generative models have shown impressive performance in image generation and editing. However, due to memory limitations and the high cost of collecting high-resolution training images, existing methods are typically restricted to inputs with linear resolutions below 1K. In contrast, photos captured by modern mobile devices often reach linear resolutions up to 8K, revealing a significant gap between current capabilities and real-world demands. Simply upscaling low-resolution edited results often results in visually enlarged but blurry images that lack fine details. This paper introduces UltraDiffEdit, a novel, tuning-free image editing framework that extends off-the-shelf latent diffusion models (LDMs) to ultrahigh resolutions. UltraDiffEdit employs a multi-scale progressive editing strategy, iteratively blending high-resolution edited content with unedited areas in a coarse-to-fine manner. We employ multi-patch encoding to preserve both edited and unedited visual details within the latent space. To mitigate editing artifacts, our global-local consistency denoising technique consistently integrates edited and unedited latent features, ensuring smooth transition at editing boundaries from the latent representation to the final image. We also introduce a patch-based hybrid sampling approach that captures local, intermediate, and global features, ensuring semantic coherence and enhancing fine detail during denoising. We conduct extensive experiments demonstrating UltraDiffEdit's superior editing quality and flexibility: it can handle image resolutions up to 8K using only a single NVIDIA GeForce RTX 3090 GPU. The source code is publicly available at https://github.com/LonglongaaaGo/UltraDiffEdit.

Figures

Figures reproduced from arXiv: 2607.06136 by the authors.

Figure 1
Figure 1. Examples of ultra-high-resolution image editing using SDXL [1]: (a) [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The overall pipeline of our UltraDiffEdit. (a) We downsample an input image to create [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Visual results of the compared methods. Our UltraDiffEdit shows better global-local consistency with faithful details. [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (11 more)
Figure 4
Figure 4. Figure 4: Visual results of the compared methods. UltraDiffEdit shows better global-local consistency with high fidelity. [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Visual results of the compared methods. * indicates increased inference steps to match the runtime of UltraDiffEdit. [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: Visual results of extending existing image editing methods: Control [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 7
Figure 7. Figure 7: Visual results of image multi-object editing (top) and outpainting (bottom) of ours. The masked regions are in blue. [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]
Figure 8
Figure 8. Figure 8: Comparison of denoising processes in DemoFusion (top) and UltraDiffEdit (bottom). Compared to DemoFusion, UltraDiffEdit achieves structure [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 9
Figure 9. Figure 9: Ablation study for the applied components. “- Enco” replaces the multi-patch encoding with the original encoding from SDXL; “- Deno” substitutes [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]
Figure 10
Figure 10. Figure 10: Visual results using different strides du and dv. Time cost for each configuration: (a) 10198.4 seconds, (b) 9727.5 seconds, (c) 9606.7 seconds, (d) 9580.2 seconds, (e) 9561.6 seconds. TABLE IV ABLATION STUDY ON THE DIV2KEDIT AND SYN2KEDIT DATASETS. “- ENCO” REPLACES …
Figure 11
Figure 11. Figure 11: Visualization of our UltraDiffEdit using various phase sets: (a) S [PITH_FULL_IMAGE:figures/full_fig_p012_11.png]
Figure 12
Figure 12. Figure 12: Memory demands of SDXL and UltraDiffEdit, inference on a [PITH_FULL_IMAGE:figures/full_fig_p013_12.png]
Figure 13
Figure 13. Figure 13: Inference time of SDXL and UltraDiffEdit, inference on a single [PITH_FULL_IMAGE:figures/full_fig_p013_13.png]
Figure 15
Figure 15. Figure 15: Failure cases of our UltraDiffEdit. Redundant small objects were [PITH_FULL_IMAGE:figures/full_fig_p013_15.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

88 extracted references · 88 canonical work pages

  1. [1]

    Sdxl: Improving latent diffusion models for high-resolution image synthesis,

    D. Podell, Z. English, K. Lacey, A. Blattmann, T. Dockhorn, J. M ¨uller, J. Penna, and R. Rombach, “Sdxl: Improving latent diffusion models for high-resolution image synthesis,” inICLR, 2024

    work page 2024

  2. [2]

    Designing a practical degradation model for deep blind image super-resolution,

    K. Zhang, J. Liang, L. Van Gool, and R. Timofte, “Designing a practical degradation model for deep blind image super-resolution,” inICCV, 2021, pp. 4791–4800

    work page 2021

  3. [3]

    Inf-dit: Upsampling any-resolution image with memory- efficient diffusion transformer,

    Z. Yang, H. Jiang, W. Hong, J. Teng, W. Zheng, Y . Dong, M. Ding, and J. Tang, “Inf-dit: Upsampling any-resolution image with memory- efficient diffusion transformer,”ECCV, pp. 141–156, 2024

    work page 2024

  4. [4]

    Demofusion: Democratising high-resolution image generation with no $$$,

    R. Du, D. Chang, T. Hospedales, Y .-Z. Song, and Z. Ma, “Demofusion: Democratising high-resolution image generation with no $$$,” inCVPR, 2024, pp. 6159–6168

    work page 2024

  5. [5]

    Deep unsupervised learning using nonequilibrium thermodynamics,

    J. Sohl-Dickstein, E. Weiss, N. Maheswaranathan, and S. Ganguli, “Deep unsupervised learning using nonequilibrium thermodynamics,” inICML, 2015, pp. 2256–2265

    work page 2015

  6. [6]

    Denoising diffusion probabilistic models,

    J. Ho, A. Jain, and P. Abbeel, “Denoising diffusion probabilistic models,” NeurIPS, vol. 33, pp. 6840–6851, 2020

    work page 2020

  7. [7]

    Adding conditional control to text-to-image diffusion models,

    L. Zhang, A. Rao, and M. Agrawala, “Adding conditional control to text-to-image diffusion models,” inICCV, 2023, pp. 3836–3847

    work page 2023

  8. [8]

    Weather translation via weather-cue transferring,

    X. Li, C. Li, K. Kou, and B. Zhao, “Weather translation via weather-cue transferring,”TNNLS, vol. 35, no. 6, pp. 7988–7998, 2024. 18

    work page 2024

Show all 88 references
  1. [9]

    Targeting accurate object extraction from an image: A comprehensive study of natural image matting,

    Q. Zhu, L. Shao, X. Li, and L. Wang, “Targeting accurate object extraction from an image: A comprehensive study of natural image matting,”TNNLS, vol. 26, no. 2, pp. 185–207, 2015

    2015

  2. [10]

    Verbal-person nets: Pose-guided multi-granularity language-to-person generation,

    D. Liu, L. Wu, F. Zheng, L. Liu, and M. Wang, “Verbal-person nets: Pose-guided multi-granularity language-to-person generation,”TNNLS, vol. 34, no. 11, pp. 8589–8601, 2023

    2023

  3. [11]

    Shunting at arbitrary feature levels via spatial disentanglement: Toward selective image trans- lation,

    J. Wang, X. Yang, Y . Tian, J. Zhou, and J. Lv, “Shunting at arbitrary feature levels via spatial disentanglement: Toward selective image trans- lation,”TNNLS, vol. 35, no. 10, pp. 14 945–14 958, 2024

    2024

  4. [12]

    Disp+v: A unified framework for disentangling prototype and variation from single sample per person,

    M. Pang, B. Wang, M. Ye, Y .-m. Cheung, Y . Chen, and B. Wen, “Disp+v: A unified framework for disentangling prototype and variation from single sample per person,”TNNLS, vol. 34, no. 2, pp. 867–881, 2023

    2023

  5. [13]

    High- resolution image synthesis with latent diffusion models,

    R. Rombach, A. Blattmann, D. Lorenz, P. Esser, and B. Ommer, “High- resolution image synthesis with latent diffusion models,” inCVPR, 2022, pp. 10 684–10 695

    2022

  6. [14]

    Imagic: Text-based real image editing with diffusion models,

    B. Kawar, S. Zada, O. Lang, O. Tov, H. Chang, T. Dekel, I. Mosseri, and M. Irani, “Imagic: Text-based real image editing with diffusion models,” inCVPR, June 2023, pp. 6007–6017

    2023

  7. [15]

    Repaint: Inpainting using denoising diffusion probabilistic models,

    A. Lugmayr, M. Danelljan, A. Romero, F. Yu, R. Timofte, and L. Van Gool, “Repaint: Inpainting using denoising diffusion probabilistic models,” inCVPR, 2022, pp. 11 461–11 471

    2022

  8. [16]

    Resadapter: Domain consistent resolution adapter for diffusion models,

    J. Cheng, P. Xie, X. Xia, J. Li, J. Wu, Y . Ren, H. Li, X. Xiao, M. Zheng, and L. Fu, “Resadapter: Domain consistent resolution adapter for diffusion models,” inAAAI, 2025, pp. 2438–2446

    2025

  9. [17]

    Any-size- diffusion: Toward efficient text-driven synthesis for any-size hd images,

    Q. Zheng, Y . Guo, J. Deng, J. Han, Y . Li, S. Xu, and H. Xu, “Any-size- diffusion: Toward efficient text-driven synthesis for any-size hd images,” inAAAI, vol. 38, no. 7, 2024, pp. 7571–7578

    2024

  10. [18]

    Multidiffusion: fusing diffusion paths for controlled image generation,

    O. Bar-Tal, L. Yariv, Y . Lipman, and T. Dekel, “Multidiffusion: fusing diffusion paths for controlled image generation,” inICML, 2023

    2023

  11. [19]

    Ip-adapter: Text compatible image prompt adapter for text-to-image diffusion models,

    H. Ye, J. Zhang, S. Liu, X. Han, and W. Yang, “Ip-adapter: Text compatible image prompt adapter for text-to-image diffusion models,” arXiv preprint arXiv:2308.06721, 2023

    2023 arXiv

  12. [20]

    Ntire 2017 challenge on single image super-resolution: Dataset and study,

    E. Agustsson and R. Timofte, “Ntire 2017 challenge on single image super-resolution: Dataset and study,” inCVPRW, 2017, pp. 1122–1131

    2017

  13. [21]

    Pyramid grafting network for one-stage high resolution saliency detection,

    C. Xie, C. Xia, M. Ma, Z. Zhao, X. Chen, and J. Li, “Pyramid grafting network for one-stage high resolution saliency detection,” inCVPR, June 2022, pp. 11 717–11 726

    2022

  14. [22]

    Grig: Few-shot generative residual image inpainting,

    W. Lu, X. Jiang, X. Jin, Y .-L. Yang, M. Gong, T. Wang, K. Shi, and H. Zhao, “Grig: Few-shot generative residual image inpainting,” Computational Visual Media, vol. 11, no. 6, p. 1329 – 1361, 2025

    2025

  15. [23]

    Anywhere: a multi-agent framework for user-guided, reliable, and diverse foreground-conditioned image generation,

    X. Tianyidan, R. Ma, Q. Wang, X. Ye, F. Liu, Y . Tai, Z. Zhang, L. Wang, and Z. Yi, “Anywhere: a multi-agent framework for user-guided, reliable, and diverse foreground-conditioned image generation,” inAAAI, 2025

    2025

  16. [24]

    Diffusion model-based image editing: A survey,

    Y . Huang, J. Huang, Y . Liu, M. Yan, J. Lv, J. Liu, W. Xiong, H. Zhang, S. Chen, and L. Cao, “Diffusion model-based image editing: A survey,” TPAMI, vol. 47, no. 6, p. 4409–4437, 2025

    2025

  17. [25]

    Facemug: A multimodal generative and fusion framework for local facial editing,

    W. Lu, J. Wang, X. Jin, X. Jiang, and H. Zhao, “Facemug: A multimodal generative and fusion framework for local facial editing,”TVCG, pp. 1– 15, 2024

    2024

  18. [26]

    Contextual residual aggregation for ultra high-resolution image inpainting,

    Z. Yi, Q. Tang, S. Azizi, D. Jang, and Z. Xu, “Contextual residual aggregation for ultra high-resolution image inpainting,” inCVPR, 2020, pp. 7508–7517

    2020

  19. [27]

    Aggregated contextual transfor- mations for high-resolution image inpainting,

    Y . Zeng, J. Fu, H. Chao, and B. Guo, “Aggregated contextual transfor- mations for high-resolution image inpainting,”TVCG, vol. 29, no. 7, pp. 3266–3280, 2023

    2023

  20. [28]

    Coordfill: efficient high-resolution image inpainting via parameterized coordinate querying,

    W. Liu, X. Cun, C.-M. Pun, M. Xia, Y . Zhang, and J. Wang, “Coordfill: efficient high-resolution image inpainting via parameterized coordinate querying,” inAAAI, 2023

    2023

  21. [29]

    Diffusion-4k: Ultra- high-resolution image synthesis with latent diffusion models,

    J. Zhang, Q. Huang, J. Liu, X. Guo, and D. Huang, “Diffusion-4k: Ultra- high-resolution image synthesis with latent diffusion models,” inCVPR, 2025, pp. 23 464–23 473

    2025

  22. [30]

    Hd-painter: High-resolution and prompt-faithful text-guided image inpainting with diffusion models,

    H. Manukyan, A. Sargsyan, B. Atanyan, Z. Wang, S. Navasardyan, and H. Shi, “Hd-painter: High-resolution and prompt-faithful text-guided image inpainting with diffusion models,” inICLR, 2025

    2025

  23. [31]

    High-resolution image editing via multi-stage blended diffusion,

    J. Ackermann and M. Li, “High-resolution image editing via multi-stage blended diffusion,”arXiv preprint arXiv:2210.12965, 2022

    2022 arXiv

  24. [32]

    Blended diffusion for text- driven editing of natural images,

    O. Avrahami, D. Lischinski, and O. Fried, “Blended diffusion for text- driven editing of natural images,” inCVPR, 2022, pp. 18 208–18 218

    2022

  25. [33]

    Real-esrgan: Training real- world blind super-resolution with pure synthetic data,

    X. Wang, L. Xie, C. Dong, and Y . Shan, “Real-esrgan: Training real- world blind super-resolution with pure synthetic data,” inICCV, 2021, pp. 1905–1914

    2021

  26. [34]

    Denoising diffusion implicit models,

    J. Song, C. Meng, and S. Ermon, “Denoising diffusion implicit models,” ICLR, 2021

    2021

  27. [35]

    Latentpaint: Image inpainting in latent space with diffusion models,

    C. Corneanu, R. Gadde, and A. M. Martinez, “Latentpaint: Image inpainting in latent space with diffusion models,” inWACV, 2024, pp. 4334–4343

    2024

  28. [36]

    Gradpaint: Gradient-guided inpainting with diffusion models,

    A. Grechka, G. Couairon, and M. Cord, “Gradpaint: Gradient-guided inpainting with diffusion models,”CVIU, vol. 240, p. 103928, 2024

    2024

  29. [37]

    Uni-paint: A unified framework for multimodal image inpainting with pretrained diffusion model,

    S. Yang, X. Chen, and J. Liao, “Uni-paint: A unified framework for multimodal image inpainting with pretrained diffusion model,” inACM MM, 2023, pp. 3190–3199

    2023

  30. [38]

    Smartbrush: Text and shape guided object inpainting with diffusion model,

    S. Xie, Z. Zhang, Z. Lin, T. Hinz, and K. Zhang, “Smartbrush: Text and shape guided object inpainting with diffusion model,” inCVPR, 2023, pp. 22 428–22 437

    2023

  31. [39]

    T2i-compbench: A comprehensive benchmark for open-world compositional text-to-image generation,

    K. Huang, K. Sun, E. Xie, Z. Li, and X. Liu, “T2i-compbench: A comprehensive benchmark for open-world compositional text-to-image generation,”NeurIPS, vol. 36, pp. 78 723–78 747, 2023

    2023

  32. [40]

    A- sdm: Accelerating stable diffusion through model assembly and feature inheritance strategies,

    J. Zhu, Y . Wang, S. Pan, P. Wan, D. Zhang, and G. Huang, “A- sdm: Accelerating stable diffusion through model assembly and feature inheritance strategies,”TNNLS, vol. 36, no. 10, pp. 18 478–18 491, 2025

    2025

  33. [41]

    Ultra-resolution adaptation with ease,

    R. Yu, S. Liu, Z. Tan, and X. Wang, “Ultra-resolution adaptation with ease,” inICML, 2025

    2025

  34. [42]

    Sine: Single image editing with text-to-image diffusion models,

    Z. Zhang, L. Han, A. Ghosh, D. N. Metaxas, and J. Ren, “Sine: Single image editing with text-to-image diffusion models,” inCVPR, 2023, pp. 6027–6037

    2023

  35. [43]

    A latent space of stochastic diffusion models for zero-shot image editing and guidance,

    C. H. Wu and F. De la Torre, “A latent space of stochastic diffusion models for zero-shot image editing and guidance,” inICCV, 2023, pp. 7378–7387

    2023

  36. [44]

    Laspa: Latent spatial alignment for fast training-free single image editing,

    Y . Alharbi and P. Wonka, “Laspa: Latent spatial alignment for fast training-free single image editing,”arXiv preprint arXiv:2403.12585, 2024

    2024 arXiv

  37. [45]

    Masactrl: Tuning-free mutual self-attention control for consistent image synthesis and editing,

    M. Cao, X. Wang, Z. Qi, Y . Shan, X. Qie, and Y . Zheng, “Masactrl: Tuning-free mutual self-attention control for consistent image synthesis and editing,” inICCV, 2023, pp. 22 560–22 570

    2023

  38. [46]

    Proxedit: Improving tuning- free real image editing with proximal guidance,

    L. Han, S. Wen, Q. Chen, Z. Zhang, K. Song, M. Ren, R. Gao, A. Stathopoulos, X. He, Y . Chenet al., “Proxedit: Improving tuning- free real image editing with proximal guidance,” inWACV, 2024, pp. 4291–4301

    2024

  39. [47]

    Anyedit: Mastering unified high-quality image editing for any idea,

    Q. Yu, W. Chow, Z. Yue, K. Pan, Y . Wu, X. Wan, J. Li, S. Tang, H. Zhang, and Y . Zhuang, “Anyedit: Mastering unified high-quality image editing for any idea,” inCVPR, 2025, pp. 26 125–26 135

    2025

  40. [48]

    Scalecrafter: Tuning-free higher-resolution visual generation with diffusion models,

    Y . He, S. Yang, H. Chen, X. Cun, M. Xia, Y . Zhang, X. Wang, R. He, Q. Chen, and Y . Shan, “Scalecrafter: Tuning-free higher-resolution visual generation with diffusion models,” inICLR, 2023

    2023

  41. [49]

    Training-free diffusion model adaptation for variable-sized text-to-image synthesis,

    Z. Jin, X. Shen, B. Li, and X. Xue, “Training-free diffusion model adaptation for variable-sized text-to-image synthesis,”NeurIPS, vol. 36, 2024

    2024

  42. [50]

    I- max: Maximize the resolution potential of pre-trained rectified flow transformers with projected flow,

    R. Du, D. Liu, L. Zhuo, Q. Qi, H. Li, Z. Ma, and P. Gao, “I- max: Maximize the resolution potential of pre-trained rectified flow transformers with projected flow,”arXiv preprint arXiv:2410.07536, 2024

    2024 arXiv

  43. [51]

    Beyondscene: Higher-resolution human-centric scene generation with pretrained diffu- sion,

    G. Kim, H. Kim, H. Seo, D. U. Kang, and S. Y . Chun, “Beyondscene: Higher-resolution human-centric scene generation with pretrained diffu- sion,” inECCV, 2024, pp. 126–142

    2024

  44. [52]

    Generative powers of ten,

    X. Wang, J. Kontkanen, B. Curless, S. M. Seitz, I. Kemelmacher- Shlizerman, B. Mildenhall, P. Srinivasan, D. Verbin, and A. Holynski, “Generative powers of ten,” inCVPR, 2024, pp. 7173–7182

    2024

  45. [53]

    Rectifiedhr: Enable efficient high-resolution image generation via energy rectification,

    Z. Yang, G. Shen, L. Hou, M. Liu, L. Wang, X. Tao, P. Wan, D. Zhang, and Y .-C. Chen, “Rectifiedhr: Enable efficient high-resolution image generation via energy rectification,”arXiv preprint arXiv:2503.02537, 2025

    2025 arXiv

  46. [54]

    Is one gpu enough? pushing image generation at higher-resolutions with foundation models

    A. Tragakis, M. Aversa, C. Kaul, R. Murray-Smith, and D. Faccio, “Is one gpu enough? pushing image generation at higher-resolutions with foundation models.” inNeuralIPS, 2024, pp. 41 242 – 41 273

    2024

  47. [55]

    Syncdiffusion: Coherent mon- tage via synchronized joint diffusions,

    Y . Lee, K. Kim, H. Kim, and M. Sung, “Syncdiffusion: Coherent mon- tage via synchronized joint diffusions,”NeurIPS, vol. 36, pp. 50 648– 50 660, 2023

    2023

  48. [56]

    Elasticdiffusion: Training-free arbitrary size image generation through global-local con- tent separation,

    M. Haji-Ali, G. Balakrishnan, and V . Ordonez, “Elasticdiffusion: Training-free arbitrary size image generation through global-local con- tent separation,” inCVPR, 2024, pp. 6603–6612

    2024

  49. [57]

    Upsample guidance: Scale up diffusion models without training,

    J. Hwang, Y .-H. Park, and J. Jo, “Upsample guidance: Scale up diffusion models without training,”arXiv preprint arXiv:2404.01709, 2024

    2024 arXiv

  50. [58]

    Make a cheap scaling: A self-cascade diffusion model for higher-resolution adaptation,

    L. Guo, Y . He, H. Chen, M. Xia, X. Cun, Y . Wang, S. Huang, Y . Zhang, X. Wang, Q. Chenet al., “Make a cheap scaling: A self-cascade diffusion model for higher-resolution adaptation,” inECCV, 2024, pp. 39–55

    2024

  51. [59]

    Dit4edit: Diffusion transformer for image editing,

    K. Feng, Y . Ma, B. Wang, C. Qi, H. Chen, Q. Chen, and Z. Wang, “Dit4edit: Diffusion transformer for image editing,” inAAAI, vol. 39, no. 3, 2025, pp. 2969–2977

    2025

  52. [60]

    Resmaster: Mastering high-resolution image generation via structural and fine- grained guidance,

    S. Shi, W. Li, Y . Zhang, J. He, B. Gong, and Y . Zheng, “Resmaster: Mastering high-resolution image generation via structural and fine- grained guidance,” inAAAI, 2025, pp. 6887–6895. 19

    2025

  53. [61]

    Hiprompt: Tuning-free higher-resolution generation with hierarchical mllm prompts,

    X. Liu, Y . He, L. Guo, X. Li, B. Jin, P. Li, Y . Li, C.-M. Chan, Q. Chen, W. Xueet al., “Hiprompt: Tuning-free higher-resolution generation with hierarchical mllm prompts,”IJCV, vol. 134, 2026

    2026

  54. [62]

    Accdiffusion: An accurate method for higher-resolution image generation,

    Z. Lin, M. Lin, M. Zhao, and R. Ji, “Accdiffusion: An accurate method for higher-resolution image generation,” inECCV. Springer, 2024, pp. 38–53

    2024

  55. [63]

    Mastering text-to-image diffusion: Recaptioning, planning, and generating with multimodal llms,

    L. Yang, Z. Yu, C. Meng, M. Xu, S. Ermon, and C. Bin, “Mastering text-to-image diffusion: Recaptioning, planning, and generating with multimodal llms,” inICML, 2024

    2024

  56. [64]

    Megafusion: Extend diffusion models towards higher-resolution image generation without further tuning,

    H. Wu, S. Shen, Q. Hu, X. Zhang, Y . Zhang, and Y . Wang, “Megafusion: Extend diffusion models towards higher-resolution image generation without further tuning,” inWACV, 2025, pp. 3944–3953

    2025

  57. [65]

    Fouriscale: A frequency perspective on training-free high-resolution image synthesis,

    L. Huang, R. Fang, A. Zhang, G. Song, S. Liu, Y . Liu, and H. Li, “Fouriscale: A frequency perspective on training-free high-resolution image synthesis,” inECCV, 2024, pp. 196–212

    2024

  58. [66]

    Diffusehigh: Training-free progressive high-resolution image synthesis through structure guidance,

    Y . Kim, G. Hwang, and E. Park, “Diffusehigh: Training-free progressive high-resolution image synthesis through structure guidance,” inAAAI, 2025, pp. 4338–4346

    2025

  59. [67]

    Freeu: Free lunch in diffusion u-net,

    C. Si, Z. Huang, Y . Jiang, and Z. Liu, “Freeu: Free lunch in diffusion u-net,” inCVPR, 2024, pp. 4733–4743

    2024

  60. [68]

    Todo: Token downsampling for efficient generation of high-resolution images,

    E. Smith, N. Saxena, and A. Saha, “Todo: Token downsampling for efficient generation of high-resolution images,” inIJCAI, 2024, pp. 8801–8804

    2024

  61. [69]

    Token merging: Your vit but faster,

    D. Bolya, C.-Y . Fu, X. Dai, P. Zhang, C. Feichtenhofer, and J. Hoffman, “Token merging: Your vit but faster,” inICLR, 2023

    2023

  62. [70]

    Hidiffu- sion: Unlocking higher-resolution creativity and efficiency in pretrained diffusion models,

    S. Zhang, Z. Chen, Z. Zhao, Y . Chen, Y . Tang, and J. Liang, “Hidiffu- sion: Unlocking higher-resolution creativity and efficiency in pretrained diffusion models,” inECCV, 2024, pp. 145–161

    2024

  63. [71]

    Poisson image editing,

    P. P ´erez, M. Gangnet, and A. Blake, “Poisson image editing,”TOG, vol. 22, no. 3, p. 313–318, 2003

    2003

  64. [72]

    Prompt-to-prompt image editing with cross-attention control,

    A. Hertz, R. Mokady, J. Tenenbaum, K. Aberman, Y . Pritch, and D. Cohen-or, “Prompt-to-prompt image editing with cross-attention control,” inICLR, 2023

    2023

  65. [73]

    Null- text inversion for editing real images using guided diffusion models,

    R. Mokady, A. Hertz, K. Aberman, Y . Pritch, and D. Cohen-Or, “Null- text inversion for editing real images using guided diffusion models,” in CVPR, June 2023, pp. 6038–6047

    2023

  66. [74]

    Chatgpt,

    OpenAI, “Chatgpt,” https://www.openai.com/blog/chatgpt, 2024, [On- line; accessed 1-Nov-2024]

    2024

  67. [75]

    Photo-realistic single image super-resolution using a generative adversarial network,

    C. Ledig, L. Theis, F. Huszar, J. Caballero, A. Cunningham, A. Acosta, A. Aitken, A. Tejani, J. Totz, Z. Wang, and W. Shi, “Photo-realistic single image super-resolution using a generative adversarial network,” inCVPR, 2017, pp. 4681–4690

    2017

  68. [76]

    Large scale image completion via co-modulated generative adversarial networks,

    S. Zhao, J. Cui, Y . Sheng, Y . Dong, X. Liang, E. I. Chang, and Y . Xu, “Large scale image completion via co-modulated generative adversarial networks,” inICLR, 2021

    2021

  69. [77]

    Learning transferable visual models from natural language supervision,

    A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clarket al., “Learning transferable visual models from natural language supervision,” inICML, 2021, pp. 8748–8763

    2021

  70. [78]

    The unreasonable effectiveness of deep features as a perceptual metric,

    R. Zhang, P. Isola, A. A. Efros, E. Shechtman, and O. Wang, “The unreasonable effectiveness of deep features as a perceptual metric,” in CVPR, 2018, pp. 586–595

    2018

  71. [79]

    Gans trained by a two time-scale update rule converge to a local nash equilibrium,

    M. Heusel, H. Ramsauer, T. Unterthiner, B. Nessler, and S. Hochreiter, “Gans trained by a two time-scale update rule converge to a local nash equilibrium,” inNeurIPS, vol. 30, 2017, pp. 6629–6640

    2017

  72. [80]

    Black Forest Labs, “Flux,” https://github.com/black-forest-labs/flux, 2024, [Online; accessed 27-Jan-2025]

    2024

  73. [81]

    Pixart-σ: Weak-to-strong training of diffusion transformer for 4k text-to-image generation,

    J. Chen, C. Ge, E. Xie, Y . Wu, L. Yao, X. Ren, Z. Wang, P. Luo, H. Lu, and Z. Li, “Pixart-σ: Weak-to-strong training of diffusion transformer for 4k text-to-image generation,” inECCV, 2024, pp. 74–91

    2024

  74. [82]

    W¨urstchen: An efficient architecture for large-scale text-to-image dif- fusion models,

    P. Pernias, D. Rampas, M. L. Richter, C. Pal, and M. Aubreville, “W¨urstchen: An efficient architecture for large-scale text-to-image dif- fusion models,” inICLR, 2024

    2024

  75. [83]

    SANA: Efficient high-resolution text-to- image synthesis with linear diffusion transformers,

    E. Xie, J. Chen, J. Chen, H. Cai, H. Tang, Y . Lin, Z. Zhang, M. Li, L. Zhu, Y . Lu, and S. Han, “SANA: Efficient high-resolution text-to- image synthesis with linear diffusion transformers,” inICLR, 2025

    2025

  76. [84]

    Do inpainting yourself: Generative facial inpainting guided by exemplars,

    W. Lu, H. Zhao, X. Jiang, X. Jin, Y .-L. Yang, and K. Shi, “Do inpainting yourself: Generative facial inpainting guided by exemplars,” Neurocomputing, vol. 617, p. 128996, 2025

    2025

  77. [85]

    Advanced auto labeling solution with added features,

    W. Wang, “Advanced auto labeling solution with added features,” 2024, [Online; accessed 1-Sep-2024]. [Online]. Available: https: //github.com/CVHub520/X-AnyLabeling

    2024

  78. [86]

    Blip-2: Bootstrapping language- image pre-training with frozen image encoders and large language models,

    J. Li, D. Li, S. Savarese, and S. Hoi, “Blip-2: Bootstrapping language- image pre-training with frozen image encoders and large language models,” inICML, 2023, pp. 19 730–19 742

    2023

  79. [87]

    Vision transformers for dense prediction,

    R. Ranftl, A. Bochkovskiy, and V . Koltun, “Vision transformers for dense prediction,” inICCV, 2021, pp. 12 179–12 188

    2021

  80. [88]

    Openpose: Realtime multi-person 2d pose estimation using part affinity fields,

    Z. Cao, G. Hidalgo Martinez, T. Simon, S. Wei, and Y . A. Sheikh, “Openpose: Realtime multi-person 2d pose estimation using part affinity fields,”IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 2019. 20 Masked SDXL+bicubic DemoFusion UltraDiffEdit (ours...

    2019

Pith tools

Reviewed July 8, 2026 · model on record in the stance chip above.