SuperAPI AI Directory
Partner with us中文AI newsFavorites

All AI tools

35,708 results · Page 876 of 992

StableVideo

AI video

StableVideo is a text-driven video editing tool that achieves consistency-aware editing based on diffusion models, released by the paper authors at ICCV 2023. The project is open-sourced on GitHub, and the code implements the ability to edit video content through text descriptions, with a focus on maintaining inter-frame consistency. Before use, pretrained models and example videos need to be downloaded, and it may depend on components such as ControlNet. The project is mainly intended for research purposes and provides CPU and GPU running options. Users can run app.py to launch the interface for editing.

Read more

Stacks

AI productivity

Stacks is an AI workspace tool positioned as replacing multiple apps such as Google Drive, bookmarking tools, and Notion with a single intelligent workspace, helping users save, search, and share web content and knowledge. The official website claims it has over 50,000 users and claims to save 2 hours per day. It supports contextually capturing notes, bookmarks, links, and files while browsing, using AI to search saved content and discover hidden connections, and also provides public or private sharing controls and real-time team collaboration. Users can train the AI based on their own digital footprint to build a personal knowledge system. It is suitable for scenarios such as paper writing, entrepreneurship, and niche knowledge collection. The official materials provided do not specify the exact fees, pricing, or mobile support.

Read more

star

AI developer platforms

STAR (STacked AutoRegressive Scheme) is an open-source framework for unified multimodal learning, corresponding to the paper "STAR: STacked AutoRegressive Scheme for Unified Multimodal Learning." By stacking homogeneous autoregressive modules, it decomposes multimodal learning into three stages: understanding, generation, and editing. While keeping the parameters of the frozen base autoregressive model fixed, it progressively expands capabilities to alleviate cross-task interference and performance trade-off issues. The project also introduces high-capacity VQ to improve the granularity of image representation and adopts an implicit reasoning mechanism to improve generation quality under complex conditions. Experiments show that it achieves current state-of-the-art performance on GenEval (0.91), DPG-Bench (87.44), and ImgEdit (4.34). The repository provides model checkpoints, environment configuration, instructions for downloading pretrained models, and a Gradio-based interactive demo, supporting three types of tasks: image understanding, text-to-image generation, and image editing, under the Apache 2.0 license.

Read more