Video: Can Cloud Encoding Deliver the Same Quality as Hardware-Based?

Watch Sean Longworth's full presentation from 流媒体 East, 基于云的编码的5个神话, on the 流媒体 Conference Video Portal.

Read the complete transcript of this clip:

肖恩·朗沃思: Is it true that quality is not as good in the Cloud as it is in a hardware-based encoder? The answer to that question is: It depends. 它不一定是对的或错的. It depends on what kind of software you use. The truth of the matter is the high majority of encoding software built today was built with a fixed hardware mindset. It was built with the assumption that it's going to go on appliance or in some type of fixed mode within the Cloud. If you're an engineering team or an operations team and you are focusing on building a highly efficient encoder that has a controlled environment like an appliance, the majority of your innovation and your engineering focus is going to be on trying to make that machine work as efficiently as possible. You're trying to see how much encoding you can squeeze out of a single GPU or a set of GPUs or a CPU. 你要专注于此. That's the majority of the software that's out there today.

If you take that software that is highly optimized for a specific environment and you pour it and put it into the Cloud, 它不会运行得很好. I think that's what most people's experience is with Cloud encoding: they've gotten in situations like that and they've compared it to the hardware encoding, 这就不那么好了. 自然, it's not going to work as well if you do that, but not all software stacks are created equally.

Bitmovin's been around for seven years. 我们从来没有电器. We've always done Cloud-based encoding. We've always thought about encoding from a Cloud perspective. Our innovation and our engineering and our focus has not been around how can we optimize the codec of 10% more for this particular GPU and really get the best out of it. Our innovation and what we've always done is thinking about how we can leverage the Cloud more efficiently in order to drive better performance. 具体地说, how can we take files and split them up into the appropriate amount of segments and run them off of a variety of resources and then stitch them back together in the most efficient way possible that drives quality? That's just a different track for an engineering team than an appliance team.

我们的软件, if you took the software that we design and how it works and you put it on an appliance, 它的效果会差得多, because we designed all of our engineering and engineering about the ability to use the Cloud. Whether or not something performs better or worse in the Cloud depends on the software and how it was written and how it was architected.