Developer makes Nvidia DLSS 5 run in browsers, macOS also supported

The solution works on macOS devices.The project provides an online demo and plans to release the source code to GitHub.

The current implementation runs slower and is not suitable for real-time gaming.DLSS 5 was originally planned as a exclusive tech for NBA 2K27, but after the technology was cracked, many developers ported it to various 3A games.

This time, MAAN takes it a step further by bringing it into the browser environment, bypassing the DLSS integration for applications.

Nvidia currently does not offer an official browser DLSS solution.The company's DLSS SDK documentation mainly targets native applications based on DirectX or Vulkan, usually integrated through the Nvidia NGX interface or Streamline SDK.

The Streamline sits between the application and the graphics API, requiring developers to provide motion vectors, depth data, and rendered frames.

These resources and processing pipelines do not directly correspond to the browser's rendering pipeline.Nvidia's official documentation does not list WebGL or WebGPU as supported graphics interfaces for DLSS.

Since MAAN has not yet disclosed the technical details of the project, it is currently unclear how the solution connects the rendering resources in the browser to DLSS 5.

Possible implementation methods include using native bridging to call Nvidia's existing binary components, converting browser-generated rendering resources for external program processing, or other methods.

With the project's source code to be publicly released on GitHub later, the specific implementation mechanism is expected to become clearer.

From the current demo experience, the project runs relatively slowly, closer to an early concept verification scheme, and is difficult to be used for actual games.

The demo mainly provides various 3D scenes and "cinematic," "natural" style presets, allowing users to adjust intensity, local tone, local structure, and skin structure parameters.

However, its running load is quite high;even if users use an RTX 4090, loading the neural rendering model takes about 1 to 2 seconds.

Although it can run on macOS devices as well, it cannot achieve smooth operation at the moment.In reality, it is more suitable for "building visualization," "real-time 3D model preview" scenarios.

Although this technology is still far from real-time gaming applications, in the absence of an official browser support from Nvidia, being able to run DLSS 5 through WebGPU in the browser itself is already a significant technical achievement.

开发者成功令英伟达 DLSS 5 在浏览器里运行，苹果 macOS 也可体验

IT之家 9 月 18 日消息，据外媒 TweakTown 报道，近期开发者 MAAN 在 X 平台展示了一项概念验证项目，成功让英伟达 DLSS 5 在 WebGPU 环境下直接处理浏览器中的 3D 图形，相应方案可在 macOS 设备上运行。目前项目已经提供在线演示（ 点此访问 ），开发者还计划将源代码发布至 GitHub，不过从实际体验来看，其运行速度仍然较慢，暂时不适合实时游戏场景。 IT之家获悉，DLSS 5 原本计划作为《NBA 2K27》的独占技术推出，但在相关技术被破解后，大量开发者将其移植至各种 3A 游戏，而此次 MAAN 更进一步，将其带入浏览器环境，绕开了基于应用的 DLSS 集成方式。 事实上，英伟达目前并没有提供官方支持的浏览器 DLSS 方案。该公司的 DLSS SDK 文档主要针对基于 DirectX 或 Vulkan 的原生应用，通常通过英伟达 NGX 接口或 Streamline SDK 进行集成。其中 Streamline 位于应用程序与图形 API 之间，开发者需要向其提供运动矢量、深度数据以及渲染帧等资源，而这些数据和处理流程并不能直接对应到浏览器的渲染管线中。英伟达官方文档目前也没有将 WebGL 或 WebGPU 列为 DLSS 支持的图形接口。 由于 MAAN 尚未公布项目的技术细节，目前还无法确定该方案究竟如何将浏览器中的渲染资源连接至 DLSS 5。可能的实现方式包括通过原生桥接调用英伟达现有二进制组件、将浏览器生成的渲染资源转换后交由外部程序处理，或者采用其他方式。随着项目源代码后续在 GitHub 公开，具体实现机制预计将会更加清晰。 从目前的 Demo 体验来看，该项目运行较为缓慢，更接近早期概念验证方案，难以用于实际游戏，Demo 主要提供了多种 3D 场景以及“电影感”“自然”等风格预设，同时允许用户调整强度、局部色调、局部结构和皮肤结构等参数。不过其运行负载相当高，即使用户使用 RTX 4090，加载神经渲染模型也需要约 1 至 2 秒；尽管其在 macOS 设备上同样可以跑动，但暂时无法实现流畅运行，实际上更适用于“建筑可视化”、“实时 3D 模型预览”等场景。 尽管目前这项技术距离实时游戏应用还有明显距离，但在英伟达没有提供官方浏览器支持的情况下，能够让 DLSS 5 通过 WebGPU 在浏览器中运行，本身已经是一项颇具技术展示意义的尝试。

This report is categorized under AITOP's "How-to" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include DLSS 5, WebGPU, browser, AI rendering.

Source attribution: curated from IT之家.