XR_EXT_render_model was the one that caught my eye earlier! Looking forward to chucking out a bunch of api-specific code next project. :)

That's a Jerry Mouse run cycle! Maybe the 'Naruto run' should just be called the 'Reverse Jerry' instead...

Keyring as uberkey, except for passwords, which are pure knowledge so you can just guess the number if you remember your last playthrough

Related - I love how comprehensive the log is; it captures *every* conversation and info screen read so you can always read back to find that password you forgot

The Steam Controller works *so* well for mouse-y things but in pad form. I played through the entirety of Stardew Valley with one before the game had pad support and it was great.

I tend to think in terms of depth complexity / overdraw / fillrate, which is not strictly what those three items represent, but useful in terms of what I need to optimise or change.

There's also quad overdraw, but for mobile VR I find we're not pushing enough polys for that to be important.

You can get cat collars with space for a slip of paper, then write a note explaining it's coming around and ask the owner to contact you and leave your phone number. I've done this previously and had an appreciative owner call me back. Plus you get to find out their real name!

And by 'feature complete' I mean the APIs just have big holes in because either someone didn't think about backwards/forwards compatibility, didn't think anyone would ever want to query that thing, or wrapped everything up in a baby-proof-api and hid important details.

The entire VR stack has been simultaneously overengineered, under-specified, fragmented and pseudo-consolidated that the early days of OpenVR feel like a breath of fresh air.

In our current project I've got an OpenVR, OVR and OpenXR backends, and OpenVR is the only sane, feature complete one.

This is one of those cases where I can't tell if the hoop-jumping to avoid branches actually comes out as an overall win, so something to benchmark later. In particular 'ceil' seems like a wildcard that could go either way, although Shader Playground suggests it's quite cheap.