clm_1e417801By decomposing the image formation process into a sequential application of denoising autoencoders, diffusion models (DMs) achieve state-of-the-art synthesis results on image data and beyond.
arXiv 2112.10752 · cs.CV
Read on arXiv · arXiv non-exclusive license
Claims were extracted automatically by reducer-1 on 2026-11-12. The paper's license does not allow republishing it, so long claims are shortened to a quotation. Read the full text on arXiv.
clm_1e417801By decomposing the image formation process into a sequential application of denoising autoencoders, diffusion models (DMs) achieve state-of-the-art synthesis results on image data and beyond.
clm_df997f3cOur latent diffusion models (LDMs) achieve new state-of-the-art scores for image inpainting and class-conditional image synthesis and highly competitive performance on various tasks, including text-to-image synthesis, unconditional image generation and … (excerpt)
clm_34d6f5b4We propose latent diffusion models (LDMs) as an effective generative model and a separate mild compression stage that only eliminates imperceptible details.
res_fb12bdc9(iii) We show that, in contrast to previous work [93] which learns both an encoder/decoder architecture and a score-based prior simultaneously, our approach does not require a delicate weighting of … (excerpt)
res_ec51a755Revisiting the analysis above (Fig. 1 and 2) we attribute this to i) leaving most of perceptual compression to the diffusion model and ii) too strong first stage compression resulting … (excerpt)
res_e0d76d5aLDM-{4-8} outperform models with unsuitable ratios of perceptual and conceptual compression.
res_6c3d37c7Especially compared to pixel-based LDM-1, they achieve much lower FID scores while simultaneously significantly increasing sample throughput.
res_ea426177In summary, LDM-4 and -8 offer the best conditions for achieving high-quality synthesis results.
res_c92e6a2fOn CelebA-HQ, we report a new state-of-the-art FID of 5.11, outperforming previous likelihood-based models as well as GANs.
res_10338e2fWe also outperform LSGM [93] where a latent diffusion model is trained jointly together with the first stage.
res_678df2acWe outperform prior diffusion based approaches on all but the LSUN-Bedrooms dataset, where our score is close to ADM [15], despite utilizing half its parameters and requiring 4-times less train … (excerpt)
res_62fea507Evaluation of text-conditional image synthesis on the 256× 256-sized MS-COCO [51] dataset: with 250 DDIM [84] steps our model is on par with the most recent diffusion [59] and autoregressive … (excerpt)
res_7678a86bHere we outperform the state of the art diffusion model ADM [15] while significantly reducing computational requirements and parameter count, cf .
res_e9f222a2Our qualitative and quantitative results (see Fig. 10 and Tab. 5) show competitive performance and LDM-SR outperforms SR3 in FID while SR3 has a better IS.
res_b1780ee7Overall, we observe a speed-up of at least 2.7× between pixel- and latent-based diffusion models while improving FID scores by a factor of at least 1.6×.
res_cafd047dThe comparison with other inpainting approaches in Tab. 7 shows that our model with attention improves the overall image quality as measured by FID over that of [88].
clm_d18ca28eWe have presented latent diffusion models, a simple and efficient way to significantly improve both the training and sampling efficiency of denoising diffusion models without degrading their quality.
clm_c3723637Based on this and our crossattention conditioning mechanism, our experiments could demonstrate favorable results compared to state-of-the-art methods across a wide range of conditional image synthesis tasks without task-specific architectures.
res_bd9cb664Our COCO model reaches the performance of recent state-of-the art models in layout-toimage synthesis, when following their training and evaluation protocol [89].
res_79417bd2Our OpenImages model surpasses the results of Jahn et al [37] by a margin of nearly 11 in terms of FID.
res_ba93be6eLDM-8 requires significantly fewer parameters and compute requirements (see Tab. 18) to achieve very competitive performance.
res_0f15ad9dThe results of this comparison are shown in the last two rows of Tab. 11 and demonstrate that LDM achieves better performance while allowing for significantly faster sampling.
clm_be587db3Compute during training in V100-days, numbers of competing methods taken from [15] unless stated differently; ∗: Throughput measured in samples/sec on a single NVIDIA A100;†: Numbers taken from [15] ;‡: … (excerpt)
Record your result in the Mearc app on your own machine, then use Publish this result with this paper's id (2112.10752) and the claim id shown next to the claim. Mearc creates a file that holds only what you choose to publish. Add it to the public claim data by pull request; a maintainer other than you reviews it before it counts.