평범한 필기장

Paper : https://arxiv.org/abs/2407.21720 Detecting, Explaining, and Mitigating Memorization in Diffusion ModelsRecent breakthroughs in diffusion models have exhibited exceptional image-generation capabilities. However, studies show that some outputs are merely replications of training data. Such replications present potential legal challenges for model owners, espearxiv.orgAbstract문제 : 생성모델의 몇 o..

Project Page : https://rf-inversion.github.io/ Litu Rout1,2 Yujia Chen2 Nataniel Ruiz2 Constantine Caramanis1 Sanjay Shakkottai1Wen-Sheng Chu2 1 The University of Texas at Austin,    2 Google ICLR 202" data-og-host="rf-inversion.github.io" data-og-source-url="https://rf-inversion.github.io/" data-og-url="https://rf-inversion.github.io/" data-og-image=""> RF-InversionSemantic Image Inversion and ..

Paper : https://arxiv.org/abs/2309.06380 InstaFlow: One Step is Enough for High-Quality Diffusion-Based Text-to-Image GenerationDiffusion models have revolutionized text-to-image generation with its exceptional quality and creativity. However, its multi-step sampling process is known to be slow, often requiring tens of inference steps to obtain satisfactory results. Previous attemparxiv.orgAbstr..

Paper : https://arxiv.org/abs/2210.02747 Flow Matching for Generative ModelingWe introduce a new paradigm for generative modeling built on Continuous Normalizing Flows (CNFs), allowing us to train CNFs at unprecedented scale. Specifically, we present the notion of Flow Matching (FM), a simulation-free approach for training CNFs basearxiv.org(성민혁 교수님 강의 자료 참고 : https://www.youtube.com/watch?v=B4F..

Paper : https://arxiv.org/abs/2304.08465 MasaCtrl: Tuning-Free Mutual Self-Attention Control for Consistent Image Synthesis and EditingDespite the success in large-scale text-to-image generation and text-conditioned image editing, existing methods still struggle to produce consistent generation and editing results. For example, generation approaches usually fail to synthesize multiple imaarxiv.o..

Paper : https://arxiv.org/abs/2310.01506 Direct Inversion: Boosting Diffusion-based Editing with 3 Lines of CodeText-guided diffusion models have revolutionized image generation and editing, offering exceptional realism and diversity. Specifically, in the context of diffusion-based editing, where a source image is edited according to a target prompt, the process comarxiv.orgProject Page : https:..

Paper : https://arxiv.org/abs/2401.11739 EmerDiff: Emerging Pixel-level Semantic Knowledge in Diffusion ModelsDiffusion models have recently received increasing research attention for their remarkable transfer abilities in semantic segmentation tasks. However, generating fine-grained segmentation masks with diffusion models often requires additional training on anarxiv.org0. Abstract Diffusion m..

Paper : https://arxiv.org/abs/2408.06070 ControlNeXt: Powerful and Efficient Control for Image and Video GenerationDiffusion models have demonstrated remarkable and robust abilities in both image and video generation. To achieve greater control over generated results, researchers introduce additional architectures, such as ControlNet, Adapters and ReferenceNet, to intearxiv.orgGithub : https://g..

Project Page : https://style-aligned-gen.github.io/ StyleAlignStyle Aligned Image Generation via Shared Attention CVPR 2024, Oral Amir Hertz* 1 Andrey Voynov* 1 Shlomi Fruchter† 1 Daniel Cohen-Or† 1,2 1 Google Research 2 Tel Aviv University *Indicates Equal Contribution †Indicates Equal Advising [Paper]    style-aligned-gen.github.ioPaper : https://arxiv.org/abs/2312.02133 Style Aligned Image Ge..

Paper, Project Page, Github GitHub - inbarhub/DDPM_inversion: Official pytorch implementation of the paper: "An Edit Friendly DDPM Noise Space: Inversion anOfficial pytorch implementation of the paper: "An Edit Friendly DDPM Noise Space: Inversion and Manipulations". CVPR 2024. - GitHub - inbarhub/DDPM_inversion: Official pytorch implementa...github.com해결하려는 문제 본 논문에서는 기존 DDIM latent가 아닌 DDPM la..