Today was to be a big day – an exciting one for many reasons. Finally, the kids were going to get to make something in Blender, and as an additional surprise I had received a rough draft of the transformer scene Richard had been working on. It was already awesome!
But the day was not to be a problem free one. In fact, as with the previous session, things went off-piste almost immediately.
Blender keeps crashing!
The school, like many in the UK I’d imagine, uses a third party contractor to manage its IT infrastructure. I’d taken the step of making contact with them well before Week 1, and had been through the process of having Blender installed onto every PC in the ICT suite. This hadn’t been without its problems, because the school’s equipment is a bit last season, and Blender was crashing out after the initial Python console start-up window appeared.
After a bit of Googling around I’d been able to narrow that down to a few candidate root causes and it turned out that the issue was related to the video drivers installed on the PCs. Tech support had kindly installed, via their remote connection, the correct equipment and I’d even been back into the school, again before Week 1, to test them. And they had worked. Everything had been ready!
But, best laid plans and all that.
Here we were at the start of Week 3, logging in and starting Blender – to a succession of “Blender’s crashing, Blender won’t start, Blender’s broken!”.
I made a note to self – “Must get that sorted!”, and pressed on with Plan B.
We see how the pros do it, and move to ‘Plan ‘B’
It was thanks to my wife the evening before that I’d thought things through a bit more thoroughly. “What if Blender doesn’t actually work tomorrow?”, she’d asked. Oh, how I hate when she’s right all the time! So a Plan B was decided upon then and there and thank goodness we’d had the conversation, because at school in the ICT lab next afternoon I was able to switch the group fairly seamlessly from Blender-anticipation mode to “Let’s do some more tracking footage!”. Which sounds boring, but we actually made it quite fun.
Being on a steep learning curve of my own here, I had come to understand that tracking can be ridiculously easy, or (if you film a bunch of kids in a playground without adequate preparation and with little or no planning) tracking can be really quite hard!
Since Week 1 I’d been obsessively going over my playground shot, trying to obtain a solve error below 0.4. No matter what I did it just wasn’t working out for me. The best I’d managed had been 1.5.
Fortunately, I did have a very strong ally in Richard van der Oost, the creator of the original open source transformer animation, so (although I had agreed to pass him tracked footage) I ultimately had to pass him over some rather badly tracked footage. He was very nice about it though, and as you’ll see in the highlights reel he was able to succeed where I had failed, and had placed an object into the scene which really did look stuck to the ground.
Not only that he had also placed the transformer animation into the scene and I had a draft to show off to the kids. And oh boy, they loved it!
So anyway, to set up my Plan B (using the knowledge that my tracking powers basically sucked, but not wanting to leave this chapter on a fail), I made a more concerted effort to learn how tracking should be done and that’s when I stumbled upon the phenomenal 24 chapters of Sebastian Koenig’s “Track, Match, Blend” series. By the time I was on Chapter 9 I had enough new knowledge to have confidence that I could reproduce some tracked footage and make something stick to it.
So, I shot a quick test feature on my phone, opened it in Blender, and – success! I was able to track, add a scene, add objects and they were sticking to my shot. This was now my Plan B.
We shoot some tracking video
So there we were in the ICT suite, with nothing but a camera and a piece of A4 paper covered in crosses.
The kids were given the task of creating their own tracking sheets and once they were ready I filmed each of them individually, saying a line of their choosing, like Scott here, who boldly claimed to have captured Captain America himself!
This exercise took us to the end of the session. All that remained was for me to check the footage at home, which I’m pleased to say was a very lightweight task and I had been able to run test tracking across each of the videos within just a couple of hours.