Dev Blog #16: No plan survives first contact

I expected the series of demo videos I'm making at the moment to be a distraction from getting real work done. If anything, they have proved the opposite.

Making the first video forced a complete rewrite of the @time propagation algorithm to make it something worth demonstrating. The second video obliged me to complete the process of porting all the graphical rendering to distance field textures, and making the third revealed a deep-seated and profound limitation of the way LISA was parsing notes, prompting a near-full rewrite of a substantial chunk of her algorithm (As a bonus, the new system is ~50% faster processing the thousand-note benchmark)

As a result, the app is now in much better shape despite me taking three full evenings 'off' to edit together video clips. Incidentally, I can very much recommend Lightworks for that purpose: even the free edition is a complete beast. A bit intimidating and not especially intuitive (the first time I used it it seemed to want to just chew up everything I did with it) but once you wrap your head around it it's an incredibly fast and precise bit of kit.

One downside of pulling things together in a hurry for the purpose of a video is that the code now needs a bit of a spring-clean, so that's what I'll be doing this morning. After that, the Doctree system is going to be made ready for its closeup...

Comments

Popular Posts