Using Adobe Media Encoder to create iPad video

Video production in Sydney, corporate, communications, media, events, motion graphics, interviews, camera work, editing & post production services.

Using Adobe Media Encoder to create iPad video

Having just bought my own iPad, I can say that video really does look good played back on this larger screen. However, despite the fact that Apple (and everyone else, for that matter) seems to be focusing on its ability to play back high-definition video, it seems as though we’ve overlooked the fact that it’s a 4:3 screen.

I see that there are a number of sites offering advice on how to encode video for use on an iPad, but all of these seem to be using 1280×720 as their baseline – which will playback letterboxed and downscaled. If you want to create video that will actually fill the iPad’s XGA (1024×768 pixels) screen without rescaling, then you need to step back to a 4:3 aspect ratio. It’s a bit old-school, but this may prove really useful for software tutorials, or even digitising older footage.

Sadly, Adobe Media Encoder doesn’t have any presets for creating videos that natively fit the iPad screen, so I’ve put a together a couple for you. They both use h.264 and AAC codecs and use the iPad’s native resolution. They use 3.1 profile, with a VBR between 2-3Mbps. The only difference between the two is the frame rate. As a PAL user, all of my footage is captured and output to 25fps so that’s the one I’ll be using – if you’re in the US, then you’ll be better off with the 30fps version.

To set these up, download the ZIP file and unpack it onto your computer, then open up Adobe Media Encoder and select the import preset function to copy them to your presets list.

I’ve tested the 25fps version with some fairly tricky scenes for compression (vignetting, monochrome, high-motion), but you can always increase the bit rate if you find your results need a bit of extra bandwidth to help them on their way. A six-minute video turned out around 90MB, so it’s a pretty conservative bit rate that won’t suck up all your storage space like the typical 720P HD presets will.

Hope it’s of some use!

Have you tried YouTube’s Transcribe Audio Beta?

Screen grab of YouTube Transcribe in action

Not quite ready for release, yet.

I was checking out the status of my tutorials on YouTube the other day and noticed that Google has added a new beta feature called Transcribe Captions (it takes a while to kick in – the latest video I posted doesn’t have it enabled yet, but the earlier two videos do). I’m a massive fan of the automatic caption generation tool that they currently provide for English-speaking videos where the owner has a .txt transcript of the video they’ve uploaded. It takes ten minutes and generates a series of closed captions that are timed perfectly with the speech – perfect for keeping video delivery in line with accessibility standards.

The Auto Transcribe function, however, seems to have a few issues to deal with (it may be better at transcribing US-accented speech), getting about one word in forty right. It puts me in mind of the transcription tool that Adobe built into Soundbooth in CS4 (I’ll be checking out the CS5 version to see if it’s got any better). Given that most people have come to the conclusion that speech recognition (not to be confused with voice recognition) is virtually impossible to achieve effectively, why are heavyweights like Adobe and Google trying to get it to work?

The answer, as is so often the case these days, is search. And advertising.

Imagine if you could create an accurate text transcript of a video, complete with timecode data. You could use it as metadata for the video, making every part of it searchable – type a name into Google for example, and not only could you see videos about that person, you could also find points in the video that mention that person’s name. Take it a step further, and you can deliver context-sensitive advertising on that video at exactly the point where the relevant metatag occurs.

If Adobe or Google (or anyone else, for that matter) can nail this technology down, it would be a massive coup – content/contextual searching of video. Don’t hold your breath, though – if current indicators are anything to go by, we’ve got a long way to go!