Roni Paiss

Roni Paiss, Hila Chefer, Lior Wolf · meta-ai

The application of zero-shot learning in computer vision has been revolutionized by the use of image-text matching models. The most notable example, CLIP, has been widely used for both zero-shot classification and guiding generative models with a text prompt. However, the zero-shot use of CLIP is unstable with respect to the phrasing of the input text, making it necessary to carefully engineer the prompts used. We find that this instability stems from a selective similarity score, which is based only on a subset of the semantically meaningful input tokens. To mitigate it, we present a novel explainability-based approach, which adds a loss term to ensure that CLIP focuses on all relevant semantic parts of the input, in addition to employing the CLIP similarity loss used in previous works. When applied to one-shot classification through prompt engineering, our method yields an improvement in the recognition rate, without additional training or fine-tuning. Additionally, we show that CLIP guidance of generative models using our method significantly improves the generated images. Finally, we demonstrate a novel use of CLIP guidance for text-based image generation with spatial conditioning on object location, by requiring the image explainability heatmap for each object to be confined to a pre-determined bounding box.

Vimal Thilak, Etai Littwin, Shuangfei Zhai et al. · apple-ml

The grokking phenomenon as reported by Power et al. ( arXiv:2201.02177 ) refers to a regime where a long period of overfitting is followed by a seemingly sudden transition to perfect generalization. In this paper, we attempt to reveal the underpinnings of Grokking via a series of empirical studies. Specifically, we uncover an optimization anomaly plaguing adaptive optimizers at extremely late stages of training, referred to as the Slingshot Mechanism. A prominent artifact of the Slingshot Mechanism can be measured by the cyclic phase transitions between stable and unstable training regimes, and can be easily monitored by the cyclic behavior of the norm of the last layers weights. We empirically observe that without explicit regularization, Grokking as reported in ( arXiv:2201.02177 ) almost exclusively happens at the onset of Slingshots, and is absent without it. While common and easily reproduced in more general settings, the Slingshot Mechanism does not follow from any known optimization theories that we are aware of, and can be easily overlooked without an in depth examination. Our work points to a surprising and useful inductive bias of adaptive gradient optimizers at late stages of training, calling for a revised theoretical analysis of their origin.

Roni Paiss, Ariel Ephrat, Omer Tov et al.

Large vision-language models (VLMs), such as CLIP, learn rich joint image-text representations, facilitating advances in numerous downstream tasks, including zero-shot classification and text-to-image generation. Nevertheless, existing VLMs exhibit a prominent well-documented limitation - they fail to encapsulate compositional concepts such as counting. We introduce a simple yet effective method to improve the quantitative understanding of VLMs, while maintaining their overall performance on common benchmarks. Specifically, we propose a new counting-contrastive loss used to finetune a pre-trained VLM in tandem with its original objective. Our counting loss is deployed over automatically-created counterfactual examples, each consisting of an image and a caption containing an incorrect object count. For example, an image depicting three dogs is paired with the caption "Six dogs playing in the yard". Our loss encourages discrimination between the correct caption and its counterfactual variant which serves as a hard negative example. To the best of our knowledge, this work is the first to extend CLIP's capabilities to object counting. Furthermore, we introduce "CountBench" - a new image-text counting benchmark for evaluating a model's understanding of object counting. We demonstrate a significant improvement over state-of-the-art baseline models on this task. Finally, we leverage our count-aware CLIP model for image retrieval and text-conditioned image generation, demonstrating that our model can produce specific counts of objects more reliably than existing ones.

Hila Chefer, Shiran Zada, Roni Paiss et al.

Customizing text-to-image (T2I) models has seen tremendous progress recently, particularly in areas such as personalization, stylization, and conditional generation. However, expanding this progress to video generation is still in its infancy, primarily due to the lack of customized video data. In this work, we introduce Still-Moving, a novel generic framework for customizing a text-to-video (T2V) model, without requiring any customized video data. The framework applies to the prominent T2V design where the video model is built over a text-to-image (T2I) model (e.g., via inflation). We assume access to a customized version of the T2I model, trained only on still image data (e.g., using DreamBooth or StyleDrop). Naively plugging in the weights of the customized T2I model into the T2V model often leads to significant artifacts or insufficient adherence to the customization data. To overcome this issue, we train lightweight $\textit{Spatial Adapters}$ that adjust the features produced by the injected T2I layers. Importantly, our adapters are trained on $\textit{"frozen videos"}$ (i.e., repeated images), constructed from image samples generated by the customized T2I model. This training is facilitated by a novel $\textit{Motion Adapter}$ module, which allows us to train on such static videos while preserving the motion prior of the video model. At test time, we remove the Motion Adapter modules and leave in only the trained Spatial Adapters. This restores the motion prior of the T2V model while adhering to the spatial prior of the customized T2I model. We demonstrate the effectiveness of our approach on diverse tasks including personalized, stylized, and conditional generation. In all evaluated scenarios, our method seamlessly integrates the spatial prior of the customized T2I model with a motion prior supplied by the T2V model.

Omer Bar-Tal, Hila Chefer, Omer Tov et al.

We introduce Lumiere -- a text-to-video diffusion model designed for synthesizing videos that portray realistic, diverse and coherent motion -- a pivotal challenge in video synthesis. To this end, we introduce a Space-Time U-Net architecture that generates the entire temporal duration of the video at once, through a single pass in the model. This is in contrast to existing video models which synthesize distant keyframes followed by temporal super-resolution -- an approach that inherently makes global temporal consistency difficult to achieve. By deploying both spatial and (importantly) temporal down- and up-sampling and leveraging a pre-trained text-to-image diffusion model, our model learns to directly generate a full-frame-rate, low-resolution video by processing it in multiple space-time scales. We demonstrate state-of-the-art text-to-video generation results, and show that our design easily facilitates a wide range of content creation tasks and video editing applications, including image-to-video, video inpainting, and stylized generation.

Hila Chefer, Sagie Benaim, Roni Paiss et al.

We make the distinction between (i) style transfer, in which a source image is manipulated to match the textures and colors of a target image, and (ii) essence transfer, in which one edits the source image to include high-level semantic attributes from the target. Crucially, the semantic attributes that constitute the essence of an image may differ from image to image. Our blending operator combines the powerful StyleGAN generator and the semantic encoder of CLIP in a novel way that is simultaneously additive in both latent spaces, resulting in a mechanism that guarantees both identity preservation and high-level feature transfer without relying on a facial recognition network. We present two variants of our method. The first is based on optimization, while the second fine-tunes an existing inversion encoder to perform essence extraction. Through extensive experiments, we demonstrate the superiority of our methods for essence transfer over existing methods for style transfer, domain adaptation, and text-based semantic editing. Our code is available at https://github.com/hila-chefer/TargetCLIP.

David Junhao Zhang, Roni Paiss, Shiran Zada et al.

Recently, breakthroughs in video modeling have allowed for controllable camera trajectories in generated videos. However, these methods cannot be directly applied to user-provided videos that are not generated by a video model. In this paper, we present ReCapture, a method for generating new videos with novel camera trajectories from a single user-provided video. Our method allows us to re-generate the reference video, with all its existing scene motion, from vastly different angles and with cinematic camera motion. Notably, using our method we can also plausibly hallucinate parts of the scene that were not observable in the reference video. Our method works by (1) generating a noisy anchor video with a new camera trajectory using multiview diffusion models or depth-based point cloud rendering and then (2) regenerating the anchor video into a clean and temporally consistent reangled video using our proposed masked video fine-tuning technique.

Daniel Garibi, Shahar Yadin, Roni Paiss et al.

We present TokenVerse -- a method for multi-concept personalization, leveraging a pre-trained text-to-image diffusion model. Our framework can disentangle complex visual elements and attributes from as little as a single image, while enabling seamless plug-and-play generation of combinations of concepts extracted from multiple images. As opposed to existing works, TokenVerse can handle multiple images with multiple concepts each, and supports a wide-range of concepts, including objects, accessories, materials, pose, and lighting. Our work exploits a DiT-based text-to-image model, in which the input text affects the generation through both attention and modulation (shift and scale). We observe that the modulation space is semantic and enables localized control over complex concepts. Building on this insight, we devise an optimization-based framework that takes as input an image and a text description, and finds for each word a distinct direction in the modulation space. These directions can then be used to generate new images that combine the learned concepts in a desired configuration. We demonstrate the effectiveness of TokenVerse in challenging personalization settings, and showcase its advantages over existing methods. project's webpage in https://token-verse.github.io/

Jingwei Ma, Erika Lu, Roni Paiss et al.

Panoramic image stitching provides a unified, wide-angle view of a scene that extends beyond the camera's field of view. Stitching frames of a panning video into a panoramic photograph is a well-understood problem for stationary scenes, but when objects are moving, a still panorama cannot capture the scene. We present a method for synthesizing a panoramic video from a casually-captured panning video, as if the original video were captured with a wide-angle camera. We pose panorama synthesis as a space-time outpainting problem, where we aim to create a full panoramic video of the same length as the input video. Consistent completion of the space-time volume requires a powerful, realistic prior over video content and motion, for which we adapt generative video models. Existing generative models do not, however, immediately extend to panorama completion, as we show. We instead apply video generation as a component of our panorama synthesis system, and demonstrate how to exploit the strengths of the models while minimizing their limitations. Our system can create video panoramas for a range of in-the-wild scenes including people, vehicles, and flowing water, as well as stationary background features.

Alex Trevithick, Roni Paiss, Philipp Henzler et al.

Novel-view synthesis techniques achieve impressive results for static scenes but struggle when faced with the inconsistencies inherent to casual capture settings: varying illumination, scene motion, and other unintended effects that are difficult to model explicitly. We present an approach for leveraging generative video models to simulate the inconsistencies in the world that can occur during capture. We use this process, along with existing multi-view datasets, to create synthetic data for training a multi-view harmonization network that is able to reconcile inconsistent observations into a consistent 3D scene. We demonstrate that our world-simulation strategy significantly outperforms traditional augmentation methods in handling real-world scene variations, thereby enabling highly accurate static 3D reconstructions in the presence of a variety of challenging inconsistencies. Project page: https://alextrevithick.github.io/simvs

Ronen Kamenetsky, Sara Dorfman, Daniel Garibi et al.

Large-scale text-to-image diffusion models have become the backbone of modern image editing, yet text prompts alone do not offer adequate control over the editing process. Two properties are especially desirable: disentanglement, where changing one attribute does not unintentionally alter others, and continuous control, where the strength of an edit can be smoothly adjusted. We introduce a method for disentangled and continuous editing through token-level manipulation of text embeddings. The edits are applied by manipulating the embeddings along carefully chosen directions, which control the strength of the target attribute. To identify such directions, we employ a Sparse Autoencoder (SAE), whose sparse latent space exposes semantically isolated dimensions. Our method operates directly on text embeddings without modifying the diffusion process, making it model agnostic and broadly applicable to various image synthesis backbones. Experiments show that it enables intuitive and efficient manipulations with continuous control across diverse attributes and domains.