--- Log opened Fri May 16 00:00:44 2025
00:51 < hprmbridge> kanzure> for sprite sheets you can generate a 3d model, and then capture different poses into your sheet
00:57 < fenn> yes
00:58 < fenn> that's what the colorful skeletons are in the above link
00:59 < fenn> it won't necessarily work for animal body plans
01:00 < fenn> huh there is at least a quadruped controlnet https://github.com/abehonest/ControlNet_AnimalPose
01:00 < fenn> one of the bazillion AI papers i read in the last year was able to do arbitrary body plans, it was some chinese thing based on IPadapter i think
01:02 < fenn> not really necessary for this use case since you can use depth or edge
01:05 -!- darsie [~darsie@84-113-82-174.cable.dynamic.surfer.at] has joined #hplusroadmap
01:19 -!- darsie [~darsie@84-113-82-174.cable.dynamic.surfer.at] has quit [Quit: Avoid fossil fuels and animal products. Have no/fewer children. Protest, elect sane politicians. Invest ecologically.]
01:19 -!- darsie [~darsie@84-113-82-174.cable.dynamic.surfer.at] has joined #hplusroadmap
02:17 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has quit [Read error: Connection reset by peer]
03:11 < hprmbridge> cryogenator> Fenn, check your email.
03:20 < fenn> uh whoops
04:18 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has joined #hplusroadmap
05:06 < NewtonTrendy> L29Ah: india, africa,    china?
05:08 < L29Ah> NewtonTrendy: do you think they can afford perfect healthcare in the first place?
05:53 < NewtonTrendy> L29Ah: I realise they generally dont have healthcare. I think your actual point might be beyond my level of intelligence though. I'm trying to work it out. Maybe you think medicine is country based. Maybe you missed the comment asking for greater understanding and acceptance before wishing harm in general, using ai designed algorithsms, so that healthcare can work without overpopulation. At
05:53 < NewtonTrendy> then end of the day we dont need to win their elected government, we just need them to sign up.
05:53 < NewtonTrendy> sorry about my annoying typos
07:05 < kanzure> i don't know where i saw this but someone proposed a "one shot prompt" LLM hackathon, instead of a multi-round LLM hackathon.
08:45 -!- darsie [~darsie@84-113-82-174.cable.dynamic.surfer.at] has quit [Quit: Avoid fossil fuels and animal products. Have no/fewer children. Protest, elect sane politicians. Invest ecologically.]
08:46 -!- darsie [~darsie@84-113-82-174.cable.dynamic.surfer.at] has joined #hplusroadmap
09:31 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has left #hplusroadmap []
11:54 -!- TMM [hp@amanda.tmm.cx] has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]
11:55 -!- TMM [hp@amanda.tmm.cx] has joined #hplusroadmap
11:57 -!- darsie [~darsie@84-113-82-174.cable.dynamic.surfer.at] has quit [Ping timeout: 248 seconds]
11:57 -!- darsie [~darsie@84-113-82-174.cable.dynamic.surfer.at] has joined #hplusroadmap
12:35 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has joined #hplusroadmap
15:59 -!- darsie [~darsie@84-113-82-174.cable.dynamic.surfer.at] has quit [Ping timeout: 252 seconds]
16:26 < hprmbridge> nmz787> I was talking about tool batteries today and how adapters for one brand to work with another are annoying to think about making, and they mentioned this photo to STL converter https://www.meshy.ai/
17:41 -!- stipa_ [~stipa@user/stipa] has joined #hplusroadmap
17:43 -!- stipa [~stipa@user/stipa] has quit [Ping timeout: 245 seconds]
17:43 -!- stipa_ is now known as stipa
18:00 < fenn> i have a hard time believing that meshy can do precise enough models from a photo in order to interface with a mechanical part, but it could be useful for modeling the rest i guess
18:02 < fenn> LLMs do much better on one-shot prompts to start with. like 90% benchmark score with all the info up front vs 60% on multi turn conversation. you're supposed to back and forth with the LLM to generate a spec that contains all the necessary info and rationale, and then dump that into a new context
18:02 < fenn> it's not an inherent limitation just how they've been trained
18:03 < fenn> .t https://arxiv.org/abs/2505.06120
18:03 < saxo> [2505.06120] LLMs Get Lost In Multi-Turn Conversation
18:04 < fenn> maybe i'm misunderstanding
--- Log closed Sat May 17 00:00:45 2025