While I love cloud services like these, I always wonder about it for the high-end accounts. With pricing at $3.60/hour of encoded video, it seems building your own render server seems to pay out fairly soon.
For http://OneFrameOfFame.com I was looking for a video encoding service that would take a bunch of images & sound file as input, and would encode it to a video file.
Since none of the cloud video encoding parties I looked at (Encoding.com, Heywatch, ankoder) offered this, I ended up renting a linode VPS and encoding myself with ffmpeg. Which turned out to be not that hard. If you need the management tools, there are packages like http://pandastream.org/
Our pricing decreases as volume increases, so at a higher level, an hour of video is currently $1.68, not $3.60. At scale, we have other options we can offer (like dedicated servers at a fixed monthly price). We feel pretty strongly that even someone doing 10M videos/month is better off using a service like ours rather than investing in a custom solution. But we're biased, of course. :)
> We feel pretty strongly that even someone doing 10M videos/month is better off using a service like ours rather than investing in a custom solution. But we're biased, of course.
Speaking as a company that reluctantly made a decision to offer this for our customers when we couldn't find (a few years ago) an affordable volume solution to outsource, I agree with you completely.
Given the rate of change and state of tools, running even an in-house encoding solution is likely to be a much bigger undertaking than expected.
In-house IT staff would need to understand distributed computing, dynamic load based provisioning, web services software development, and somehow, blended with that, the black art of producing good looking video at low bitrates using minimal CPU time. Any one of these areas is a full discipline on its own.
If I understand correctly, Posterous got funding, and is pushing this part out the door to Zencoder. That's a smart decision. Focus on what you do.
The video industry needs more Zencoders and fewer people trying to do it in house. There are still far more backlogs of unencoded video than there are movies online.
The API is just http + json, so it works with any language. But we also hope to have an open source Python integration library very soon. (Anyone want to help us with that?) :)
To your point, perhaps flat rate encoding pricing is better for high volumes.
Our flat rate transcoding pricing (http://www.advection.net/pricing/transcoding-pricing) is based on the idea of an encoding queue with bursting. Under the hood, it's a similar cloud approach, but more economical for those with volumes or steady encoding work. Customers--or their end users--can upload at any rate (volumes of upload sessions at once through gateways located around the country for low latency), or ship drives. The queues process at a steady predictable rate, delivering content directly to the network.
In the course of processing countless libraries for clients, we've found they tend to have such unique ideas about the end result that a "one size fits all" web service often doesn't meet their needs out of the box.
So, we also offer Windows Media (WMV), Silverlight (VC-1), Ogg Theora, and cross-player H.264 encoding, adaptive bitrate encoding for Apple devices, iPhone video rotation, automatic thumbnails, web postbacks so you don't have to poll, incoming files of up to 4.7GB in size, and arbitrarily complex encoding requirements.
To your point about images + sound, for example, some clients supply metadata as XML, and using the metadata we generate pre-roll and post-roll video footage that stitches onto the transcoded video.
(We also have a consumer service that bundles both encoding and storage into a single "per minute" price. Consumers shouldn't need to know about gigabytes.)
I agree. Also the industries that have large volumes of video to be encoded, who can afford that rate, are ones like "adult" -- and I believe Zencoder won't touch adult video conversion.
I know they have their reasons, and I respect them, but I do think it is a mistake and a limitation.
i might be wrong but they even had higher prices at launch
I was thinking they were targeting film and media agencies (high-end market); instead of the thousands little bootstrapping startups.
zencoder should look into colocating servers and having customized servers with a video card capable of video transcoding... their costs would lower dramatically
This interests me for ThinkCode.TV. We tried half a dozen companies and were overall disappointed with their results. (Ogg Theora in particular was challenging for most.)
This is great. I've had clients years ago want to offer video sharing on their site and I basically have to tell them it's a major project so they might as well have their users upload it to YouTube instead. This is exactly the kind of service that makes video UGC possible, even easy.
Hi ripb. Yep, our sales site needs a page that covers this. If you sign up and play around with the API, you'll see this info. At the moment, we do:
Input: just about everything
Output: H.264 video, AAC audio, and MP3 audio, in a MP4 or Flash 9+ container format. Ogg/vorbis/theora are coming soon, as are a few others that we can't talk about yet. We've decided to take the "narrow" approach for now - 99% of the demand that we see is for these five codecs (H.264, AAC, MP3, Theora, Vorbis). If there's anything in particular that you're looking for, let us know.
Soon! It's in the works. In the meantime, we can of course do non-streaming video for iPad/iPhone/etc. (Email us if you're interested in early access to iPhone streaming.)
The problem with the non-streaming is that Apple doesn't approve apps anymore that do non-streaming over 3G (for over 5 min/10 MB video's). Will email you if I need it tho!
Oh, and then a harder one: Could you take a bunch of images & soundtrack file, and render it to a movie?
Yep, we're AWS at the moment. We're ultimately cloud agnostic, though, and can support other cloud environments (Rackspace Cloud, etc.) as well as our own servers. EC2 is pretty compelling for transcoding, though, because (1) so many people use S3 for video storage and delivery, and (2) Amazon's fastest servers (XL High CPU) are pretty fast - faster than most cloud servers.
At that time we had just released Flix Cloud. We have spent the last year supporting that product and learning from our customers. This release of Zencoder is the result.
As for the last question: when you're offered a spot in YC, you don't turn it down!
For http://OneFrameOfFame.com I was looking for a video encoding service that would take a bunch of images & sound file as input, and would encode it to a video file. Since none of the cloud video encoding parties I looked at (Encoding.com, Heywatch, ankoder) offered this, I ended up renting a linode VPS and encoding myself with ffmpeg. Which turned out to be not that hard. If you need the management tools, there are packages like http://pandastream.org/