Do we have any information on how the choice of codecs (both video and audio) affects battery life on the tablets?

I can relatively easy compare codecs (and settings like bitrate etc.) in terms of file size and (subjective) quality.

It's more tedious and time consuming to do testing on battery life; so maybe some of you already have made experiences they can share.

My assumption is that the more processor power is needed during playback, the sooner the battery will be empty. A google search for terms like 'video codec cpu battery life' tells me that this is probably true. The question is how much influence a CPU usage of, say, 75% vs. 50% has on battery life...

It would be worth investingating if I could gain 30min or so by using codecs that result in larger files or worse quality. If the difference is only 5min or 10, it's not relevant at all.

Any ideas?

The easy thing to do would be to encode based on your knowledge at the lowest acceptable standard. Since you've already found that lower quality may result in longer battery life. The assumption you can then make is that you are already using the least amount of power that you can.

My suggestion however, if battery life is very important would be to get an extended battery as you can then get excellent quality videos and longer battery life.

You can also install CPU monitoring software that will tell you the CPU load while playing videos. If the CPU is being maxed out by your video, you can be sure that there will be a battery life penalty. On the other hand it may not be possible the get acceptable quality video while not running the CPU over 60%.

Oh... so I just encode a short video in 4 variants and simply look at the output of battery-status before and after? How frighteningly simple.
Good idea.

sounds respectable, maybe you can post the results?

but i've read too many posts about the quirks of the battery's ability to monitor its own charge, and/or the (in)accuracy of any battery monitoring software. my recommendation: encode 4 variants, yes, but watch the CPU load (like dbec10 said) for each one, then post your results. (couldn't hurt to do both, though, i guess.)