@ScriptedAlchemy@ericbouck Since 5.5 its good at refactoring. Back then openai made it so the 5hour window of a plus plan continue task below 0% until next compaction (~22min). So with a script I always opened at 1% 72 cli and gave them same prompt just different number and got 72 files refactored.
@kekzploit Its the darkish images, buttons, the text structure: (tiny text, blow big white text, smaller gray text, colored link text)and the svg especially the round svg next to explore freely
@jh3yy I mean internet was still doing noises when dialing in. I kinda like this more than what we have now but same happened to the world. Gyms, mcdonald, etc.
@viewsfrom02108 The sc2 machine learning bots played Centurys against each other and pushed themselve to a level far beyond human.
At some point there might be a breaktrough that does the same with AI
@LeoBuilds_@MnFounder Its a phase. When midjourney was new everyone was posting cool pics now its not so interesting anymore. Openai is not only b2b but also b2c where the making something is the „game“
Popular opinion: If you pay for ads like on insta pls don‘t just chatgpt imagegen it. It looks awful! You didn‘t develop taste and 9/10 of those ads look identical and overloaded.
@letstri I have a list of ready made big prompts where I change just some things at the end. And yes one is for refactor, it tells it for example to avoid cheap facade pattern and the Target lone count etc.
@tobias_petry@SheriefFYI Never in my mind would I have static allocated everything. I always tried to avoid static since it tends to create bad structure, like singletons.
I think I learned something new here.
@dangreenheck What I wanted to say is that they don't need to steal from your videos. Probably some do, I can imagine that but it's actually better if they don't do it, looks more unique.
But yeah thinking is partly outsourced and impsoter syndrome grows bigger by the day.
@dangreenheck I have really really well made (darkcreamy) water too, not even astra. Its just for a download page. I made a mockup fed it into minmax, split to png sequence (not too many) and pointed at errors and it collected research papers like nividas for ggx and gerstner waves.