You have 6.7 milliseconds. Spend them.
Your team works through this in a breakout room. Everything runs in this browser tab, nothing to install. About 20–25 minutes, in two rounds.
One submission per team. Every room gets a different brief, decided by your team name — that is deliberate, do not compare notes with the room next door yet.
Your brief
Your camera capture, VIO and image tracking already spend roughly 10ms of every frame before your own code runs, whatever frame rate you are aiming for — that pipeline runs at its own pace, it does not shrink just because you asked for fewer frames a second. What is left over is yours to spend. Pick your target below and the exact number is computed for you. Every realism feature has a cost, and you can go over — the bar on the side turns red and shows it — but you cannot lock in until you cut back down. That is deliberate: in a real project you overspend first and optimise second, not the other way round.
Choose your target frame rate
This is a real decision, not a formality. A steadier target buys you headroom. A faster one buys you almost nothing to spend, because the fixed 10ms overhead eats a bigger share of a smaller frame.
Pick a target above to see your Round 1 budget.
You will pick from five categories: grounding, lighting, audio, legibility and rendering. Spend it the way your specific brief needs, not the way that seems cleverest in the abstract. There is no single right answer — there is a right answer for your brief.
Fill every mandatory category first.
Five minutes in.
Your budget for the rest of this session just shrank. Your Round 1 picks are still there — you can keep, swap or drop anything, as long as the new total fits.
Could you have designed for this from the start?
Fit under the new budget, and tell us which way you would have planned ahead.