构建不可能之事

2020-12-10 00:00·2031天前

AI 摘要

Runway 从2019年的机器学习平台发展为新一代创意工具，依托生成式机器学习、新分发渠道和Web技术三大驱动力，已支持用户训练超5万个AI模型、上传2400万资源。平台通过AI视频编辑、绿幕工具等技术缩小好莱坞与TikTok创作者的内容差距，服务IBM、Google及RISD等院校，致力于让创意实现从想法到执行的即时转化，重新定义AI时代的创作流程。

原文 · 未翻译

A glimpse into the future of Runway.

In June of 2019, we released the beta version of Runway. Propelled by an avalanche of open research in computer vision and generative networks, we launched with one goal in mind: build a platform that helps creatives use and make sense of machine learning.

Since then, Runway has been used by a range of creative professionals that includes filmmakers, designers, VFX and CGI professionals, artists, coders, musicians, students, and educators. We’ve helped them understand and incorporate machine learning models into their workflows. Runway has also catalyzed discussions over the role of generative modeling in architecture, AI-driven techniques in industrial design, the role of synthetic media in content creation, and the automation of filmmaking, among other topics.

Today, Runway is somuch more than a general-purpose platform for AI.Creatives come to Runway to accomplish things that are impossible to achieve anywhere else. From video-driven high-resolution animations to natural text generation, artificial intelligence brings automation at every scale,introducing dramatic changes in how we create. To fully support this new creative paradigm, we need an entirely new creative suite, one that’s fundamentally different from the outdated creative tools we use today, which are based on old media paradigms and ancient distribution channels. By bringing AI closer to creators we’ve laid the foundation to develop a meaningful and radical change in how we create. Now,our goal is to build the next generation of creative tools, allowing more creatives to do impossible things.

The new creative standard.

There are three forces driving our mission to bring AI into everyday creative processes:

Generative Machine Learning: Deep neural networks for image and video synthesis are becoming increasingly precise and realistic. In just a couple of years, we’ve gone from128x128 blurry black and white images to high-resolution photorealismallowing for the rise ofsynthetic media. At the same time,neural renderingtechniques are bringing computer graphics up to date with the latest developments in deep learning. This means faster and more precise rendering techniques, more accessible graphics pipelines, and automated tools that are easier to use,such asour new Green Screen tool. These advancements make one thing increasingly clear:the gap between content created by Hollywood and a TikToker is becomingrapidly smaller. New Distribution Channels: We not only consume faster than before, we consume it differently. The rise of platforms like YouTube, Twitch, TikTok, Patreon, and Substack has given birth to a new generation of creators, necessitating a new generation of creative tools. AI-powered tools will dramatically reduce the cost of creating content for independentcreators, studios, and brands looking to monetize their distribution channels. The Web: The future of media creation should be collaborative and accessible; for that reason, it needs to happen on the web, and be based on open web technologies. Web and cloud-native creative applications are scarce but growing fast. The rise of accelerated graphics and machine learning frameworks on the web — with WebGL, WebAssembly, and soon with WebGPU — will catalyze a revolution in real-time graphics and machine learning on the browser. Furthermore, cloud processing significantly reduces the need for sophisticated hardware to run large deep learning models, broadening the accessibility of machine learning.

Runway：News（网页）

导出 Markdown

构建不可能之事

2020-12-10 00:00·2031天前

阅读原文· runwayml.com

AI 摘要

原文 · 保持原样，未翻译

A glimpse into the future of Runway.