Pith. sign in

REVIEW 3 major objections 6 minor 47 references

AccDiffusion v2: Towards More Accurate Higher-Resolution Diffusion Extrapolation

T0 review · 3 major / 6 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 AccDiffusion v2 claims that training-free patch-wise extrapolation can push SDXL to 2048–4096 pixel images with fewer repeated objects and less local distortion by per-patch prompts, canny-edge ControlNet conditioning, and interacting…

desk verdict Two of the three modules are credible; the third, as written, is a no-op, and the paper should not be accepted until that is fixed. read the letter →

arxiv 2412.02099 v2 pith:NMHUZPLS submitted 2024-12-03 cs.CV cs.AI

Zhihang Lin , Mingbao Lin , Wengyi Zhan , Rongrong Ji This is my paper · ORCID
classification cs.CVcs.AI
keywords high-resolutionimagegenerationdiffusionextrapolationtraining-freepatch-wisedenoisingpatch-content-awarepromptsControlNetdilatedsamplingobjectrepetitionsuppression
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

The paper tries to show that the two main failure modes of training-free higher-resolution diffusion extrapolation—repetitive objects and locally distorted structures—have identifiable causes and can be suppressed without any retraining. It claims that passing the same entire text prompt to every image patch is what produces repeated objects, while asking a patch for a global structure it cannot locally contain is what produces distortion. On that diagnosis it builds a patch-wise pipeline on SDXL that derives a dedicated prompt for each patch from cross-attention maps, feeds each patch canny-edge structure from the low-resolution draft through ControlNet, and makes global semantic information smoother by letting dilated samples interact. If correct, this gives a practical way to push a 1024-pixel pretrained diffusion model to 2048–4096 pixel images on a single consumer GPU, with quantitative results that match or beat eight training-free baselines.

What carries the argument

The load-bearing objects are three plug-in mechanisms operating inside latent diffusion patch-wise denoising. (1) The patch-content-aware prompt: for each word token, the cross-attention map from the low-resolution generation is thresholded at its mean, opened with morphological erosion and dilation, resized, and sampled into patches; a word is included in a patch's prompt only when its responsive mask covers enough of that patch (Eq. 12), so no patch is told to draw an object that is not there. (2) ControlNet-assisted local structure: the denoised low-resolution latent is decoded, upscaled, run through the Canny edge detector, and cropped into patch conditions that are fed, together with the patch prompt, into a ControlNet-conditioned denoiser (Eq. 13), so local geometry is anchored to the draft rather than inferred from incomplete text. (3) Dilated sampling with window interaction: the high-resolution latent is subsampled into dilation grids; before each denoising step a random bijection swaps the samples at each position inside a window, and after denoising the inverse bijection restores positions (Eqs. 16–17), letting different global-context samples inform one another. These three components are what carry the claim that repetition and distortion can be removed without changing the pretrained weights.

What would settle it

Generate a low-resolution draft of a single object, then run AccDiffusion v2 at 16x with a high-resolution prompt that explicitly names additional objects (e.g., draft shows one wolf, prompt says "three wolves running"). Because Eq. (12) drops tokens with no low-resolution attention and Eq. (13) anchors each patch to the draft's canny edges, the output should still show one wolf; an output that adds the new objects, or a quantitative comparison in which a detector counts repeated classes and finds no suppression advantage over DemoFusion, would undercut the claim.

Watch

Extended reading notes

Core claim

On the paper's own terms, AccDiffusion v2 establishes that accurate higher-resolution diffusion extrapolation can be decomposed into three corrections. First, repetitive generation is the adversarial outcome of one image-level prompt applied to every patch, so the method recovers a patch-content-aware prompt for each patch by thresholding the low-resolution cross-attention map (Eq. 12), keeping only the word tokens whose attended regions actually cover that patch. Second, local distortion persists even with per-patch prompts because a prompt such as "astronaut" urges a patch that contains only a hand to complete the whole object; the remedy is to condition each patch's denoising on the canny edges of the upscaled low-resolution image through ControlNet (Eq. 13), so local structure no longer has to be guessed from text. Third, global semantic information suppresses both artifacts, but conventional dilated sampling denoises its samples independently, producing inconsistent global guidance; AccDiffusion v2 permutes the dilated samples within each window by a position-wise bijection before each denoising step and restores positions afterward (Eqs. 16–17), yielding smoother global context. The paper reports that on SDXL this combination reaches the best FID, IS, and CLIP scores among the compared training-free extrapolation methods at 4x, 9x, and 16x resolutions, and that the visual artifacts of repetition and distortion are visibly reduced.

Load-bearing premise

The low-resolution generation is already the right content blueprint: Eq. (12) assumes its cross-attention maps tell which prompt words belong in each high-resolution patch, and Eq. (13) assumes its upscaled canny edges give the correct local structure, so if the ideal high-resolution image contains objects or composition absent from the low-resolution draft, the method is pulled back toward the draft and cannot deviate from it.

Editorial extensions

If this is right

  • In the paper's settings, SDXL extrapolates to 2048×2048, 3072×3072, and 4096×4096 (4x, 9x, and 16x) in about 4, 15, and 35 minutes on a single consumer GPU, with the best FID/IS/CLIP scores among the compared training-free methods.
  • The recipe transfers to other latent diffusion backbones: the paper shows 4x generation with Stable Diffusion 1.5 (512² to 1024²) and Stable Diffusion 2.1 (768² to 1536²), with output quality tracking the backbone's quality.
  • Reducing the number of ControlNet steps trades local fidelity for speed; the ablation shows local artifacts reappear as ControlNet is used less often.
  • The three modules are complementary: removing any one of them reintroduces either small-object repetition, semantic inconsistency, or local distortion, so the full combination is what suppresses both artifact classes.
  • The paper's stated limits carry over: inference latency is high, quality depends on the base model, spurious text-like details can appear, and beyond about 8K (64x) detail degrades.

Reading between the lines

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

  • An implication the paper leaves implicit: AccDiffusion v2 inherits the composition of its low-resolution draft. Because both prompt selection and canny anchoring read from that draft, a user who wants new objects or a rearranged scene at high resolution should regenerate the draft; changing only the high-resolution prompt will not introduce content the draft lacks.
  • The mask machinery behind Eq. (12) could be reused outside extrapolation as a cheap, model-agnostic localizer: thresholding cross-attention maps at their mean and applying morphological opening yields per-word region masks that could drive spatially targeted prompt editing, region-aware inpainting, or attention debugging.
  • The window-interaction permutation in Eqs. (16)–(17) is plausibly a general cure for tiled or patch-wise generative inconsistency, not just for dilation grids: any process that denoises interleaved pieces independently may suffer the same non-smooth global semantics, and the bijection trick transfers directly.
  • Because the distortion remedy is coupled to the canny condition, a testable extension is applying AccDiffusion v2 to other structural priors (depth, pose, or an LLM-refined per-patch prompt) to see whether the gains come from structure conditioning per se or from canny edges specifically; the paper names this direction when it suggests vision-language models for prompt refinement.
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

3 major / 6 minor

Summary. The paper proposes AccDiffusion v2, a training-free patch-wise method for extrapolating pre-trained latent diffusion models (SDXL) to higher resolutions (up to 16x). It introduces three components: (i) patch-content-aware prompts derived from low-resolution cross-attention maps, (ii) ControlNet injection of canny edges from the upscaled low-resolution draft to reduce local distortion, and (iii) dilated sampling with a position-wise bijection (window interaction) intended to make global semantic information smoother. Experiments compare against eight training-free baselines at 4x, 9x, and 16x resolutions using FID, IS, and CLIP score, with qualitative comparisons and ablations.

Significance. If validated, the patch-content-aware prompt and ControlNet-structure contributions would be useful practical extensions for training-free high-resolution image generation on consumer GPUs. The code is released, the comparisons are external to eight baselines, and the paper includes ablations of each module. However, the third contribution as described in §4.3 is mathematically a no-op, and the quantitative evidence for state-of-the-art performance is thin without variance or significance testing. The main value lies in the prompt decoupling idea and the ControlNet local-structure prior, both of which are plausible and partially supported by the ablations.

major comments (3)
  1. [4.3] Equations (16) and (17) define a position-wise bijection f applied to the dilated samples before denoising and its inverse after denoising. Since each dilated sample is denoised independently by the same pretrained network and the paper introduces no cross-sample coupling, attention, or mixing between samples, the composition f^{-1}∘f is the identity and the operation has no effect on the latent variables. The claim that this 'window interaction' produces smoother global semantics (Fig. 7) is therefore unsupported as written; Fig. 7(a) versus 7(b) is a non-seed-matched qualitative pair with no quantitative metric. Please specify the actual coupling mechanism (e.g., mixing latents within the 2x2 windows after denoising) and provide matched-seed or metric-based evidence, or remove this contribution from the claims.
  2. [4.2] The ControlNet condition in Eq. (13) is built from canny edges of the upscaled low-resolution image I'. This makes the method's local-structure prior inherit any errors in the low-resolution draft. The paper claims the module 'suppresses local distortion,' but it does not test cases where the low-resolution draft already contains distortion or where the intended high-resolution composition contains new objects absent from the draft; Fig. 15 shows a failure case but no systematic test. Please add an experiment that perturbs or distorts the low-resolution draft and reports whether the ControlNet module corrects or propagates the error, or explicitly scope the claim to drafts that are structurally sound.
  3. [5.2] The main quantitative claim of state-of-the-art performance rests on Table 1, but at 4x the differences from the second-best method are small (e.g., FIDc 38.10 vs. 38.36, ISc 25.59 vs. 24.62), and no standard errors, confidence intervals, or multiple-run statistics are reported. Since the method is stochastic and the datasets are the authors' own selection from LAION-5B, the reported margins may not be significant. Please report mean and variance over multiple seeds or prompts and a significance test; if the margins are not robust, temper the abstract's SOTA claim.
minor comments (6)
  1. [Table 1] The header 'Resolusion' should be spelled 'Resolution'.
  2. [5.1 and 5.5.2] The hyperparameter c and the ControlNet step schedule are tuned on the authors' qualitative ablations (Figs. 13 and 14) and then frozen for the main benchmark; a quantitative sensitivity analysis across c values and step schedules would strengthen the claim that the chosen settings are not overfit to the displayed examples.
  3. [Fig. 12 caption] The caption uses '%' and '!' with unclear meaning; the caption text should explicitly say which symbol denotes removal and which denotes preservation.
  4. [Eq. (14)] The variable P2 is used as the number of dilated samples but is not explicitly defined; please state that P2 = hs * ws and clarify the indexing range.
  5. [References] Reference [11] lists the title as 'Democratising high-resolution image generation with no $$$'; the placeholder '$$$' should be corrected or expanded.
  6. [Abstract] The GitHub URL in the abstract contains a space ('AccDiffusion v2'); the hyperlink should use 'AccDiffusion_v2' to work correctly.

Circularity Check

1 steps flagged · score 6.0 of 10

Dilated sampling 'interaction' is a no-op: Eq. (17) inverts Eq. (16) after independent denoising, so the claimed global-semantic improvement is identity by construction.

  1. self definitional [Section 4.3, Eqs. (16)-(17)]
    "we enable window interaction among different samples prior to each denoising process through a bijective function: Dt_{k,h,w} = Dt_{f^{h,w}_t(k),h,w}, f^{h,w}_t : {1,2,...,P2} -> {1,2,...,P2}. We then perform standard denoising progress on {Dk_t} to obtain {Dk_{t-1}}. Before applying Eq. (15) to {Dk_{t-1}}, we recover the position by using the inverse mapping (f^{h,w}_t)^{-1} ...: Dt-1_{k,h,w} = Dt-1_{(f^{h,w}_t)^{-1}(k),h,w}."

    The permutation in Eq. (16) is applied before denoising and inverted by Eq. (17) after denoising. Since each dilated sample is denoised independently by the same pretrained model—no term couples the samples, no attention across samples is introduced, and no window mixing is defined—the composition (f)^{-1} after denoising exactly cancels f before denoising. If the conditioning is identical across samples (as for global-prompt dilated sampling), the whole operation is the identity map on the latent variables; if per-sample prompts are used, the operation merely shuffles which prompt conditions which sample, which is not 'information exchange'.

full rationale

The paper's headline comparisons (Table 1, Fig. 9) are against eight external baselines and use public SDXL and ControlNet checkpoints, so the main SOTA claim is not circular. The threshold c (Eq. 12) and ControlNet-step schedule (Fig. 14) are selected from the authors' own ablations and then frozen for benchmarks; that is ordinary hyperparameter selection, not fitted-input-renamed-as-prediction, because the reported FID/IS/CLIP numbers are not optimized on those hyperparameters. The patch-content-aware prompt construction (Eq. 9-12) and the canny-structure injection (Eq. 13) are algorithmic pipelines validated externally, not defined in terms of the target output. The single concrete circular step is in Sec. 4.3: the window-interaction permutation is undone before recombination, and since denoising is per-sample and independent, the composition is exactly the identity (or, with per-sample conditions, a condition shuffle). Consequently, the claimed causal role of 'dilated sampling with window interaction' in producing Fig. 7(b) and in suppressing repetition/distortion is unsupported by the paper's own equations. The remaining two modules still give AccDiffusion v2 independent content against external baselines, so the paper is partially, not wholly, circular.

Assumptions & free parameters 5 free parameters · 5 assumptions · 1 invented entities

The honest count of what the paper adds: a threshold c with a visible tradeoff, a per-word attention threshold, a ControlNet step schedule, an unspecified morphological kernel size, and an inherited mixing schedule from DemoFusion. The load-bearing assumptions are that the low-resolution draft is a faithful blueprint for both content allocation (patch prompts) and structure (canny edges). The permutation entity has no independent evidence and is internally inconsistent with per-sample denoising; it is listed here because the paper presents it as a contribution.

free parameters (5)
  • patch-prompt inclusion threshold c (Eq. 12) = 0.3
    Chosen from the visual ablation in Sec. 5.5.2 (Fig. 13) and frozen for the quantitative comparison; it directly controls the repetition-versus-detail tradeoff and is the main dial of the patch-prompt module.
  • attention mask binarization threshold (Eq. 9) = per-word mean of the attention map
    The per-word mean threshold is justified by Table 2's differing attention ranges, but it is a hand-set design choice affecting which words enter every patch prompt.
  • number of ControlNet steps = 50 of 50 denoising steps in the main results
    Ablated in Sec. 5.5.3 (Fig. 14); the main 16x timing of 35 minutes corresponds to applying ControlNet at every step, so the reported results use the most expensive schedule.
  • morphological opening kernel size (Eq. 11) = unspecified
    The paper states only that erosion and dilation are applied to clean small connected areas; the kernel size is not reported.
  • global-semantic mixing weight eta (Eq. 15) = cosine schedule from 1 to 0
    Inherited unchanged from DemoFusion [11] and not re-tuned, but it controls how much dilated global information is mixed into each patch.
assumptions (5)
  • domain assumption The cross-attention map of the low-resolution pass predicts which prompt tokens belong in each high-resolution patch (Eqs. 9 to 12).
    Sec. 4.1 grounds the entire patch-prompt module on this correspondence; it is validated only qualitatively (Figs. 4 and 5).
  • domain assumption Canny edges of the upscaled low-resolution image are a correct local-structure prior for the high-resolution image (Eq. 13).
    Sec. 4.2 assumes the low-resolution structure extrapolates to high resolution; a distorted or compositionally different low-resolution draft would bake errors into the ControlNet anchor.
  • ad hoc to paper Some cross-sample coupling exists during standard denoising, otherwise the permutation in Eqs. 16 and 17 cancels with its inverse and the interaction module does nothing.
    Sec. 4.3 claims samples exchange information but never specifies a coupling mechanism in the U-Net; the equations as written implement the identity.
  • domain assumption The low-resolution generation is clean enough to serve as a content blueprint for the full pipeline.
    The progressive recipe upscales the low-resolution result at every stage; failures at low resolution propagate through all three modules.
  • standard math Standard latent diffusion equations (Eqs. 1 to 6) hold as given.
    Background from DDPM, DDIM, and LDM is taken as a black box, which is appropriate for a training-free method.
invented entities (1)
  • Position-wise bijection f for window interaction (Eq. 16)
    purpose: Claimed to let dilated global samples exchange information before each denoising step, producing smoother global semantics.
    No falsifiable handle outside the paper, and by the paper's own equations the bijection is an identity operation when denoising is applied per sample, so the entity does no work.

how reviews work

0 comments
Cite this review

Pith. "Pith review of AccDiffusion v2: Towards More Accurate Higher-Resolution Diffusion Extrapolation." pith.science (2026). https://pith.science/paper/NMHUZPLS

@misc{pith2026241202099,
  author       = {Pith},
  title        = {Pith review of: AccDiffusion v2: Towards More Accurate Higher-Resolution Diffusion Extrapolation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NMHUZPLS}},
  note         = {Machine review of arXiv:2412.02099}
}
read the original abstract

Diffusion models suffer severe object repetition and local distortion when the inference resolution differs from its pre-trained resolution. We propose AccDiffusion v2, an accurate method for patch-wise higher-resolution diffusion extrapolation without training. Our in-depth analysis in this paper shows that using an identical text prompt for different patches leads to repetitive generation, while the absence of a prompt undermines image details. In response, our AccDiffusion v2 novelly decouples the vanilla image-content-aware prompt into a set of patch-content-aware prompts, each of which serves as a more precise description of a patch. Further analysis reveals that local distortion arises from inaccurate descriptions in prompts about the local structure of higher-resolution images. To address this issue, AccDiffusion v2, for the first time, introduces an auxiliary local structural information through ControlNet during higher-resolution diffusion extrapolation aiming to mitigate the local distortions. Finally, our analysis indicates that global semantic information is conducive to suppressing both repetitive generation and local distortion. Hence, our AccDiffusion v2 further proposes dilated sampling with window interaction for better global semantic information during higher-resolution diffusion extrapolation. We conduct extensive experiments, including both quantitative and qualitative comparisons, to demonstrate the efficacy of our AccDiffusion v2. The quantitative comparison shows that AccDiffusion v2 achieves state-of-the-art performance in image generation extrapolation without training. The qualitative comparison intuitively illustrates that AccDiffusion v2 effectively suppresses the issues of repetitive generation and local distortion in image generation extrapolation. Our code is available at https://github.com/lzhxmu/AccDiffusion_v2.

Figures

Figures reproduced from arXiv: 2412.02099 by the authors.

Figure 1
Figure 1. Comparison of GPU memory and qualitative results for existing [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Image-content-aware prompt v.s. Patch-content-aware prompt [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. The framework of AccDiffusion v2 simplified by setting the denoising step [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (11 more)
Figure 4
Figure 4. Figure 4: This repetition typically falls into two categories: [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 4
Figure 4. Figure 4: Results of higher-resolution image generation. (a) MultiDiffusion [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 7
Figure 7. Figure 7: Results of 4× higher-resolution image generation. (a) The result of dilated sampling without window interaction. (b)The result of our dilated sampling with window interaction. Best viewed by zooming in. Dilated Sampling Reshaped Recover Denoise 2 2 2 2 2 2 2 2 2 2 2 2 …
Figure 8
Figure 8. Figure 8: Illustration of dilated sampling with window interaction: [PITH_FULL_IMAGE:figures/full_fig_p006_8.png]
Figure 9
Figure 9. Figure 9: Qualitative comparison of our AccDiffusion with existing training-free image generation extrapolation methods [8]–[11], [17], [18], [24]. We [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]
Figure 11
Figure 11. Figure 11: Results of AccDiffusion v2 on other stable diffusion variants: (a) [PITH_FULL_IMAGE:figures/full_fig_p009_11.png]
Figure 10
Figure 10. Figure 10: More selected results of AccDiffusion v2 at various resolutions. [PITH_FULL_IMAGE:figures/full_fig_p009_10.png]
Figure 12
Figure 12. Figure 12: Ablations of Patch-content-aware prompts ( [PITH_FULL_IMAGE:figures/full_fig_p010_12.png]
Figure 13
Figure 13. Figure 13: Visual results of different threshold c, prompted by “A cute corgi on the lawn.” The repetitive objects are highlighted with a red box and the detail degradation is stressed with a blue box. The best trade-off between object repetition and detail degradation is highli…
Figure 14
Figure 14. Figure 14: Ablation study on the number of ControlNet steps. Local distortions are highlighted with red boxes. The inference time is measured on one [PITH_FULL_IMAGE:figures/full_fig_p011_14.png]
Figure 15
Figure 15. Figure 15: Failure cases of AccDiffusion v2. The bad details are high [PITH_FULL_IMAGE:figures/full_fig_p012_15.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

47 extracted references · 41 canonical work pages

  1. [1]

    Denoising diffusion probabilistic models,

    J. Ho, A. Jain, and P . Abbeel, “Denoising diffusion probabilistic models,” in Adv. Neural Inform. Process. Syst., 2020

    work page 2020

  2. [2]

    Denoising diffusion implicit models,

    J. Song, C. Meng, and S. Ermon, “Denoising diffusion implicit models,” in Int. Conf. Learn. Represent., 2021

    work page 2021

  3. [3]

    Diffusion models beat gans on image synthesis,

    P . Dhariwal and A. Nichol, “Diffusion models beat gans on image synthesis,” in Adv. Neural Inform. Process. Syst., 2021

    work page 2021

  4. [4]

    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,” in IEEE Conf. Comput. Vis. Pattern Recog., 2022

    work page 2022

  5. [5]

    Stable diffusion v1-5 model card,

    P . E. Robin Rombach, “Stable diffusion v1-5 model card,” 2022. [Online]. Available: https://huggingface.co/ stable-diffusion-v1-5/stable-diffusion-v1-5

    work page 2022

  6. [6]

    Stable diffusion v2-1 model card,

    ——, “Stable diffusion v2-1 model card,” 2022. [Online]. Available: https://huggingface.co/stabilityai/stable-diffusion-2-1

    work page 2022

  7. [7]

    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,” inInt. Conf. Learn. Represent., 2024

    work page 2024

  8. [8]

    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,” in Adv. Neural Inform. Process. Syst., 2023

    work page 2023

Show all 47 references
  1. [9]

    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,” in Int. Conf. Learn. Represent., 2024

    2024

  2. [10]

    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,” in Int. Conf. Mach. Learn., 2023

    2023

  3. [11]

    Demo- fusion: Democratising high-resolution image generation with no $$$,

    R. Du, D. Chang, T. Hospedales, Y.-Z. Song, and Z. Ma, “Demo- fusion: Democratising high-resolution image generation with no $$$,” in IEEE Conf. Comput. Vis. Pattern Recog., 2024

    2024

  4. [12]

    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,” in AAAI, 2024

    2024

  5. [13]

    Difffit: Unlocking transferability of large diffusion models via simple parameter-efficient fine-tuning,

    E. Xie, L. Yao, H. Shi, Z. Liu, D. Zhou, Z. Liu, J. Li, and Z. Li, “Difffit: Unlocking transferability of large diffusion models via simple parameter-efficient fine-tuning,” in Int. Conf. Comput. Vis. , 2022

    2022

  6. [14]

    Syncdiffusion: Coherent montage via synchronized joint diffusions,

    Y. Lee, K. Kim, H. Kim, and M. Sung, “Syncdiffusion: Coherent montage via synchronized joint diffusions,” in Adv. Neural Inform. Process. Syst., 2023

    2023

  7. [15]

    Upsample guidance: Scale up dif- fusion models without training,

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

    2024 arXiv

  8. [16]

    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. Chen et al., “Make a cheap scaling: A self- cascade diffusion model for higher-resolution adaptation,” in Eur. Conf. Comput. Vis., 2024

    2024

  9. [17]

    Hid- iffusion: Unlocking higher-resolution creativity and efficiency in pretrained diffusion models,

    S. Zhang, Z. Chen, Z. Zhao, Y. Chen, Y. Tang, and J. Liang, “Hid- iffusion: Unlocking higher-resolution creativity and efficiency in pretrained diffusion models,” in Eur. Conf. Comput. Vis., 2024

    2024

  10. [18]

    Diffusehigh: Training-free pro- gressive high-resolution image synthesis through structure guid- ance,

    Y. Kim, G. Hwang, and E. Park, “Diffusehigh: Training-free pro- gressive high-resolution image synthesis through structure guid- ance,” arXiv preprint arXiv:2406.18459, 2024

    2024 arXiv

  11. [19]

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

    G. Kim, H. Kim, H. Seo, D. U. Kang, and S. Y. Chun, “Beyondscene: Higher-resolution human-centric scene generation with pretrained diffusion,” in Eur. Conf. Comput. Vis., 2024

    2024

  12. [20]

    Cutdiffusion: A simple, fast, cheap, and strong diffusion extrapolation method,

    M. Lin, Z. Lin, W. Zhan, L. Cao, and R. Ji, “Cutdiffusion: A simple, fast, cheap, and strong diffusion extrapolation method,” arXiv preprint arXiv:2404.15141, 2024

    2024 arXiv

  13. [21]

    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,” arXiv preprint arXiv:2406.07251, 2024

    2024 arXiv

  14. [22]

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

    M. Haji-Ali, G. Balakrishnan, and V . Ordonez, “Elasticdiffusion: Training-free arbitrary size image generation through global-local content separation,” in IEEE Conf. Comput. Vis. Pattern Recog., 2024. 13

    2024

  15. [23]

    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,” in Int. Conf. Comput. Vis., 2023, pp. 3836–3847

    2023

  16. [24]

    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,” in Eur. Conf. Comput. Vis., 2024, pp. 38–53

    2024

  17. [25]

    Improved denoising diffusion probabilistic models,

    A. Q. Nichol and P . Dhariwal, “Improved denoising diffusion probabilistic models,” in Int. Conf. Mach. Learn., 2021

    2021

  18. [26]

    Photorealistic text-to-image diffusion models with deep language understanding,

    C. Saharia, W. Chan, S. Saxena, L. Li, J. Whang, E. L. Denton, K. Ghasemipour, R. Gontijo Lopes, B. Karagol Ayan, T. Salimans et al. , “Photorealistic text-to-image diffusion models with deep language understanding,” in Adv. Neural Inform. Process. Syst. , 2022

    2022

  19. [27]

    Make-an-audio: Text-to-audio generation with prompt-enhanced diffusion models,

    R. Huang, J. Huang, D. Yang, Y. Ren, L. Liu, M. Li, Z. Ye, J. Liu, X. Yin, and Z. Zhao, “Make-an-audio: Text-to-audio generation with prompt-enhanced diffusion models,” in Int. Conf. Mach. Learn., 2023

    2023

  20. [28]

    Text-to- audio generation using instruction-tuned llm and latent diffusion model,

    D. Ghosal, N. Majumder, A. Mehrish, and S. Poria, “Text-to- audio generation using instruction-tuned llm and latent diffusion model,” in ACM Int. Conf. Multimedia, 2023

    2023

  21. [29]

    Make-a-video: Text-to-video generation without text-video data,

    U. Singer, A. Polyak, T. Hayes, X. Yin, J. An, S. Zhang, Q. Hu, H. Yang, O. Ashual, O. Gafni et al., “Make-a-video: Text-to-video generation without text-video data,” in Int. Conf. Learn. Represent., 2023

    2023

  22. [30]

    Imagen video: High definition video generation with diffusion models,

    J. Ho, W. Chan, C. Saharia, J. Whang, R. Gao, A. Gritsenko, D. P . Kingma, B. Poole, M. Norouzi, D. J. Fleet et al. , “Imagen video: High definition video generation with diffusion models,” arXiv preprint arXiv:2210.02303, 2022

    arXiv

  23. [31]

    Magic3d: High- resolution text-to-3d content creation,

    C.-H. Lin, J. Gao, L. Tang, T. Takikawa, X. Zeng, X. Huang, K. Kreis, S. Fidler, M.-Y. Liu, and T.-Y. Lin, “Magic3d: High- resolution text-to-3d content creation,” in IEEE Conf. Comput. Vis. Pattern Recog., 2023

    2023

  24. [32]

    Dream3d: Zero-shot text-to-3d synthesis using 3d shape prior and text-to-image diffusion models,

    J. Xu, X. Wang, W. Cheng, Y.-P . Cao, Y. Shan, X. Qie, and S. Gao, “Dream3d: Zero-shot text-to-3d synthesis using 3d shape prior and text-to-image diffusion models,” in IEEE Conf. Comput. Vis. Pattern Recog., 2023

    2023

  25. [33]

    Dreamfusion: Text-to-3d using 2d diffusion,

    B. Poole, A. Jain, J. T. Barron, and B. Mildenhall, “Dreamfusion: Text-to-3d using 2d diffusion,” in Int. Conf. Learn. Represent., 2023

    2023

  26. [34]

    A computational approach to edge detection,

    J. Canny, “A computational approach to edge detection,” IEEE Trans. Pattern Anal. Mach. Intell., 1986

    1986

  27. [35]

    Deeppose: Human pose estimation via deep neural networks,

    A. Toshev and C. Szegedy, “Deeppose: Human pose estimation via deep neural networks,” in IEEE Conf. Comput. Vis. Pattern Recog. , 2014

    2014

  28. [36]

    Depth map prediction from a single image using a multi-scale deep network,

    D. Eigen, C. Puhrsch, and R. Fergus, “Depth map prediction from a single image using a multi-scale deep network,” in Adv. Neural Inform. Process. Syst., 2014

    2014

  29. [37]

    Segment anything,

    A. Kirillov, E. Mintun, N. Ravi, H. Mao, C. Rolland, L. Gustafson, T. Xiao, S. Whitehead, A. C. Berg, W.-Y. Lo et al. , “Segment anything,” in Int. Conf. Comput. Vis., 2023

    2023

  30. [38]

    Prompt-to-prompt image editing with cross atten- tion control,

    A. Hertz, R. Mokady, J. Tenenbaum, K. Aberman, Y. Pritch, and D. Cohen-Or, “Prompt-to-prompt image editing with cross atten- tion control,” in Int. Conf. Learn. Represent., 2023

    2023

  31. [39]

    Soille et al

    P . Soille et al. , Morphological image analysis: principles and applica- tions. Springer, 1999, vol. 2

    1999

  32. [40]

    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. Hochre- iter, “Gans trained by a two time-scale update rule converge to a local nash equilibrium,” in Adv. Neural Inform. Process. Syst., 2017

    2017

  33. [41]

    Improved techniques for training gans,

    T. Salimans, I. Goodfellow, W. Zaremba, V . Cheung, A. Radford, and X. Chen, “Improved techniques for training gans,” in Adv. Neural Inform. Process. Syst., 2016

    2016

  34. [42]

    Learning trans- ferable 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. Clark et al. , “Learning trans- ferable visual models from natural language supervision,” in Int. Conf. Mach. Learn., 2021

    2021

  35. [43]

    Any- resolution training for high-resolution image synthesis,

    L. Chai, M. Gharbi, E. Shechtman, P . Isola, and R. Zhang, “Any- resolution training for high-resolution image synthesis,” in Eur. Conf. Comput. Vis., 2022

    2022

  36. [44]

    Laion-5b: An open large-scale dataset for training next generation image-text models,

    C. Schuhmann, R. Beaumont, R. Vencu, C. Gordon, R. Wightman, M. Cherti, T. Coombes, A. Katta, C. Mullis, M. Wortsman et al. , “Laion-5b: An open large-scale dataset for training next generation image-text models,” in Adv. Neural Inform. Process. Syst., 2022

    2022

  37. [45]

    Controlnext: Powerful and efficient control for image and video generation,

    B. Peng, J. Wang, Y. Zhang, W. Li, M.-C. Yang, and J. Jia, “Controlnext: Powerful and efficient control for image and video generation,” arXiv preprint arXiv:2408.06070, 2024

    2024 arXiv

  38. [46]

    Visual instruction tuning,

    H. Liu, C. Li, Q. Wu, and Y. J. Lee, “Visual instruction tuning,” in Adv. Neural Inform. Process. Syst., 2023

    2023

  39. [47]

    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,” in Int. Conf. Mach. Learn., 2023

    2023

Pith tools

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