nostr:npub15fkerqqyp9mlh7n8xd6d5k9s27etuvaarvnp2vqed83dw9c603pqs5j9gr nostr:npub1w7x9m8xutlezyzmu7d78d8ygzk6sdpncfuuds8jsz2wnwk8f52dqkrpc4w yeah that was back when RNG was insanely expensive to do in realtime. It still is if you're doing it for graphics (noise functions). Ultimately though it's hard to verify that the precomputed list is any better. It MIGHT be but I wouldn't personally know how to test that in a way that isn't insanely placebo effecty

I had some references about this but thanks to youtube being full of trash now it's hard to find them

though for the purposes of a game, urandom isn't really what we want if we want to avoid clumping. seems to me you'd want to start with the random numbers from whatever source, then remove the clumps with some transform

even if you're burning lots of numbers per frame and have to generate on the fly, I'd rather massage the numbers toward "feels good" gameplay than rely on the PRNG output directly. which means batch producing numbers, then transforming them, instead of making discrete calls as needed. could do an RNG batch every frame and throw it away at the end of the frame, that would be pretty cheap for game state updates, and if you have a rare frame where you have to do more state updates than you have numbers for, just go back to the beginning of the array. nobody's going to notice that in a single frame

Reply to this note

Please Login to reply.

Discussion

nostr:npub15fkerqqyp9mlh7n8xd6d5k9s27etuvaarvnp2vqed83dw9c603pqs5j9gr nostr:npub1w7x9m8xutlezyzmu7d78d8ygzk6sdpncfuuds8jsz2wnwk8f52dqkrpc4w yeah you'll need to massage it but I don't think precomputatuin is necessary. I've gotten extremely good results just from the realtime probably skewing. 5-7 streaks would previously be a regular occurrence and now getting up to 3 is pretty darn rare (if you're talking about 50/50 odds)