GPU 或具有 API 的基于硬件的 h.264 编码器？
我们希望根据服务器上生成的帧将 h.264 视频流式传输到浏览器。这些帧将是合成的，即不是来自另一部“真实电影”、mpeg 等。我们将单独生成它们作为模拟的一部分，可能使用 OpenGL/DirectX 来渲染各个帧。在我天真的近似中，我们会在内存中预留一个位图，渲染它，告诉编码器执行其任务，然后重复。最终用户将看到连续的流媒体视频。
我知道有几种基于非 GPL 软件的 h.264 编码器，例如 Intel IPP，但了解它们通常无法实时压缩 h.264。在理想的世界中，我们的电影将是几百万像素，例如两台高清电视。我知道我们需要专用硬件来实现这一点。人们是否知道具有此功能的扩展卡、专门连接的服务器等？
We'd like to stream h.264 video to a browser based on frames generated on the server. The frames would be synthetic, i.e. not coming from another "real movie", mpeg, etc. We'd be generating them separately as part of a simulation, potentially using OpenGL/DirectX to render individual frames. In my naive approximation, we'd have a bitmap set aside in memory, render to it, tell the encoder to do its business, and repeat. The end user would see a continuously streaming video.
I'm aware of several non-GPL software-based h.264 encoders, e.g. Intel IPP, but understand that these are often not capable of h.264 compression in real time. In an ideal world, our movies would be several megapixels, e.g. two HD TVs. I understand we need dedicated hardware for that. Are people aware of expansion cards, specially-connected servers, etc. that are capable of this?
如果你对这篇内容有疑问，欢迎到本站社区发帖提问 参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱，如果其他用户或者作者回复了您的评论，将不能在第一时间通知您！
发布评论
评论（1）
只有两个字——他们很糟糕。
GPU avc 编码器根本不支持 RDO。如果没有 RDO，它就不是编码器。实际上，如果您采用 x264 并使用一些像样的 I7，它将在同等质量下击败任何 GPU 编码器。
just two words - THEY SUCK.
GPU avc encoders are incapable of RDO at all. And without RDO it is not an encoder. Actually if you take x264 and use some decent I7 it will beat out any GPU encoder at comparable quality.