Identifying and Solving Conditional Image Leakage in Image-to-Video Diffusion Model

• URL: http://arxiv.org/abs/2406.15735v1
• Date: Sat, 22 Jun 2024 04:56:16 GMT
• Title: Identifying and Solving Conditional Image Leakage in Image-to-Video Diffusion Model
• Authors: Min Zhao, Hongzhou Zhu, Chendong Xiang, Kaiwen Zheng, Chongxuan Li, Jun Zhu,
• Abstract summary: I2V diffusion models (I2V-DMs) tend to over-rely on the conditional image at large time steps, neglecting the crucial task of predicting the clean video from noisy inputs. We introduce a training-free inference strategy that starts the generation process from an earlier time step to avoid the unreliable late-time steps of I2V-DMs. We design a time-dependent noise distribution for the conditional image, which favors high noise levels at large time steps to sufficiently interfere with the conditional image.
• Score: 31.70050311326183
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Diffusion models have obtained substantial progress in image-to-video (I2V) generation. However, such models are not fully understood. In this paper, we report a significant but previously overlooked issue in I2V diffusion models (I2V-DMs), namely, conditional image leakage. I2V-DMs tend to over-rely on the conditional image at large time steps, neglecting the crucial task of predicting the clean video from noisy inputs, which results in videos lacking dynamic and vivid motion. We further address this challenge from both inference and training aspects by presenting plug-and-play strategies accordingly. First, we introduce a training-free inference strategy that starts the generation process from an earlier time step to avoid the unreliable late-time steps of I2V-DMs, as well as an initial noise distribution with optimal analytic expressions (Analytic-Init) by minimizing the KL divergence between it and the actual marginal distribution to effectively bridge the training-inference gap. Second, to mitigate conditional image leakage during training, we design a time-dependent noise distribution for the conditional image, which favors high noise levels at large time steps to sufficiently interfere with the conditional image. We validate these strategies on various I2V-DMs using our collected open-domain image benchmark and the UCF101 dataset. Extensive results demonstrate that our methods outperform baselines by producing videos with more dynamic and natural motion without compromising image alignment and temporal consistency. The project page: \url{https://cond-image-leak.github.io/}.

Related papers

• Self Forcing: Bridging the Train-Test Gap in Autoregressive Video Diffusion [70.4360995984905]
  We introduce Self Forcing, a novel training paradigm for autoregressive video diffusion models.<n>It addresses the longstanding issue of exposure bias, where models trained on ground-truth context must generate sequences conditioned on their own imperfect outputs.
  arXiv  Detail & Related papers  (2025-06-09T17:59:55Z)
• Continuous Video Process: Modeling Videos as Continuous Multi-Dimensional Processes for Video Prediction [43.16308241800144]
  We introduce a novel model class, that treats video as a continuous multi-dimensional process rather than a series of discrete frames.<n>We establish state-of-the-art performance in video prediction, validated on benchmark datasets including KTH, BAIR, Human3.6M, and UCF101.
  arXiv  Detail & Related papers  (2024-12-06T10:34:50Z)
• Buffer Anytime: Zero-Shot Video Depth and Normal from Image Priors [54.8852848659663]
  Buffer Anytime is a framework for estimation of depth and normal maps (which we call geometric buffers) from video. We demonstrate high-quality video buffer estimation by leveraging single-image priors with temporal consistency constraints.
  arXiv  Detail & Related papers  (2024-11-26T09:28:32Z)
• Fast constrained sampling in pre-trained diffusion models [77.21486516041391]
  Diffusion models have dominated the field of large, generative image models. We propose an algorithm for fast-constrained sampling in large pre-trained diffusion models.
  arXiv  Detail & Related papers  (2024-10-24T14:52:38Z)
• FrameBridge: Improving Image-to-Video Generation with Bridge Models [23.19370431940568]
  Image-to-video (I2V) generation is gaining increasing attention with its wide application in video synthesis. We present FrameBridge, taking the given static image as the prior of video target and establishing a tractable bridge model between them. We propose two techniques, SNR- Fine-tuning (SAF) and neural prior, which improve the fine-tuning efficiency of diffusion-based T2V models to FrameBridge and the synthesis quality of bridge-based I2V models respectively.
  arXiv  Detail & Related papers  (2024-10-20T12:10:24Z)
• Truncated Consistency Models [57.50243901368328]
  Training consistency models requires learning to map all intermediate points along PF ODE trajectories to their corresponding endpoints. We empirically find that this training paradigm limits the one-step generation performance of consistency models. We propose a new parameterization of the consistency function and a two-stage training procedure that prevents the truncated-time training from collapsing to a trivial solution.
  arXiv  Detail & Related papers  (2024-10-18T22:38:08Z)
• One-Step Image Translation with Text-to-Image Models [35.0987002313882]
  We introduce a general method for adapting a single-step diffusion model to new tasks and domains through adversarial learning objectives. We consolidate various modules of the vanilla latent diffusion model into a single end-to-end generator network with small trainable weights. Our model CycleGAN-Turbo outperforms existing GAN-based and diffusion-based methods for various scene translation tasks.
  arXiv  Detail & Related papers  (2024-03-18T17:59:40Z)
• AdaDiff: Adaptive Step Selection for Fast Diffusion Models [82.78899138400435]
  We introduce AdaDiff, a lightweight framework designed to learn instance-specific step usage policies. AdaDiff is optimized using a policy method to maximize a carefully designed reward function. We conduct experiments on three image generation and two video generation benchmarks and demonstrate that our approach achieves similar visual quality compared to the baseline.
  arXiv  Detail & Related papers  (2023-11-24T11:20:38Z)
• Decoupled Diffusion Models: Simultaneous Image to Zero and Zero to Noise [53.04220377034574]
  We propose decoupled diffusion models (DDMs) for high-quality (un)conditioned image generation in less than 10 function evaluations. We mathematically derive 1) the training objectives and 2) for the reverse time the sampling formula based on an analytic transition probability which models image to zero transition. We experimentally yield very competitive performance compared with the state of the art in 1) unconditioned image generation, textite.g., CIFAR-10 and CelebA-HQ-256 and 2) image-conditioned downstream tasks such as super-resolution, saliency detection, edge detection, and image in
  arXiv  Detail & Related papers  (2023-06-23T18:08:00Z)
• Preserve Your Own Correlation: A Noise Prior for Video Diffusion Models [52.93036326078229]
  Off-the-shelf billion-scale datasets for image generation are available, but collecting similar video data of the same scale is still challenging. In this work, we explore finetuning a pretrained image diffusion model with video data as a practical solution for the video synthesis task. Our model, Preserve Your Own Correlation (PYoCo), attains SOTA zero-shot text-to-video results on the UCF-101 and MSR-VTT benchmarks.
  arXiv  Detail & Related papers  (2023-05-17T17:59:16Z)
• Dynamic Dual-Output Diffusion Models [100.32273175423146]
  Iterative denoising-based generation has been shown to be comparable in quality to other classes of generative models. A major drawback of this method is that it requires hundreds of iterations to produce a competitive result. Recent works have proposed solutions that allow for faster generation with fewer iterations, but the image quality gradually deteriorates.
  arXiv  Detail & Related papers  (2022-03-08T11:20:40Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.