Hacker Newsnew | past | comments | ask | show | jobs | submit | echelon's commentslogin

> difficulty of generalizing chamfers/fillets. There's a reason that basically all FOSS CAD packages have struggled with it.

Could you decompile CAD, run it through an LLM, and call it a day?


Aren't there licesing issues with porting a proprietary implementation into an open-source one that could open up the project to legal issues with the proprietary vendor?

The Plasticity guy seems to be threading that needle.

That might seem to be the case, right up until he gets taken to court over it.

Never presume that a thing is legal (or will not later be punished) on the basis that someone is already doing it.


The only exciting leak would be the Opus weights themselves.

What's special about Claude Code? Isn't Opus the real magic?

Surely there's nothing here of value compared to the weights except for UX and orchestration?

Couldn't this have just been decompiled anyhow?


I think pi has stolen the top honors, but people consider the Claude code harness very good (at least, better than Cursor)

Pi is the best choice for experts and power users, which is not most people.

Claude Code is still the dominant (I didn't say best) agentic harness by a wide margin I think.


Pi really is amazing. It's as much or as little as you need it to be.

Not having to deal with Boris Cherny's UX choices for CC is the cherry on top.


"Approach Sonnet"...

So not even close to Opus, then?

These are a year behind, if not more. And they're probably clunky to use.


The Stripe horror stories are adding up, making me think startups will move to different platforms.

A friend in China recently got shut down by Stripe for a perfectly legitimate business. They moved to Creem.

Stripe cares about big business. Startups can't really be moving the needle much for them anymore.

Stripe's APIs have become too confounding for small business anyway. They care about big business shaped entities at the expense of smaller scrappy players. Easy things aren't easy. It sucks.

You have to build your own hooks and logic for upgrades and downgrades. The event types are mismatched yet you have to listen across several semantic classes to capture the right state changes. Absurd, legacy/big biz focused garbage.

It should be click a button and integrate one API and webhook and you're done.

Huge opportunity.


> Creem

I kinda liked the website, I feel like there are some good features within this website, the more competition the better, Kinda like its referral and split model within co-founders, especially like this for something like a course website/maybe even Patreon alternative, combined with Cloudflare workers/Hetzner. I am seeing a lot of competition within this space, Does anyone have a github awesome-list about these or should I create one?

On personal experience, Just recently, I still have 12$ of my money stuck within Pulsedmedia/their payment provider as I had done a (crypto) payment but their system has failed to recognize it and Pulsedmedia could do nothing about it, not cancel or accept the deal as after 24 hours or at this point close to 48 hours, yet no response by coinpayments or any response at all.

All of this happened because I had accidentally sent them the whole amount but just 60 cents less... let that sink in, Pulsedmedia says that they can't do anything but as a customer, I don't feel like recommend Pulsedmedia anymore because of my experience with their payment provider being so bad and they have said that they have no control over, not even refunding me. Man, a lot of the times, I feel like payment processing should be a solved problem but recent experience indicates otherwise as I felt restlessness from my money being stuck in limbo

Either refund me as soon as possible or allow me to have a service, having to open up mail multiple times and seeing no response feels really frustrating even if the money might be low, I would still like to have it back. I feel like your choice of payment provider matters quite a lot and it can be a differentiating factor even, for the end consumer.

I had heard some good things about pulsedmedia within the forums I browse but when I had raised tickets, they had responded to me with AI too :-/ It felt extremely weird typing everything out taking time only to be responded with I hear you--you are extremely right. I don't want to blame pulsedmedia here but man oh man, some payment processors make me feel so rageful, and I think that I am fairly patient in most cases, but having money stuck no matter how tiny definitely makes me a bit stressed.

If I learn anything from all of this, it's to not have a shit payment provider, heck I might even try becoming a customer and raise issues like payment getting stuck or see their customer support response times before buying them.


I like VR and immersion in theory. I like being able to look around, but I absolutely hate the movement controls.

I know some people complain of motion sickness, but that doesn't bother me. I just want controls like Mario or Zelda on a regular joystick. Why can't this be done?

It doesn't even have to be first person. I'd play a third person game like Mario or Zelda with a VR camera tracking them. I just want that kind of movement.

Pushing a button to teleport in short hops is annoying as hell. I hate everything about it.

I gave up trying in frustration.


I always thought a great compromise would be games that gave you an overhead “gods eye” third person perspective. People seem to be obsessed with making VR games first person, but that’s where the movement problems come in.

The game Moss did this well for a platformers. But it could also be really fun for realtime strategy/simulation games (StarCraft, sim city) or sports games like Madden.


That would be fantastic.

I think even being the Lakitu camera in Mario 64 would work.

And you're right, it would kick ass for strategy, sim, and sports games.

I wonder if folks have tried building the movement and camera before the main gameplay loop. Just to see what feels right.


What's your take on WorldLabs and Apple's splat models? Are there other open source alternatives?

How would editing work?

Do you think these will win over video world models like Genie?

Have you played with DiamondWM and other open source video world models?


My understanding is that those models create gaussian splats from a text prompt, kinda like a 3d version of nano banana. I'm not doing that (yet), what I'm doing is creating splats from a set of photos - aka "splat training" and then rendering the splat as a static (working on dynamism) on the Quest headset. This is pretty well-worn territory with a lot of good implementations, but I have my own implementation of a trainer in C++/CUDA (originally based in SpeedySplat, which was written in Python, but now completely rewritten and not much of SpeedySplat is left) and renderer in C++/OpenXR for the Quest (originally based on a LLM-made port of 3DGS.cpp to OpenXR, but 100% rewritten now), and I can easily integrate techniques from research.

It takes a lot to deliver value at velocity with a team of engineers that couldn't give a damn about the product and just want to get a paycheck, move up the ladder, etc.

LinkedIn is not a fun problem.

The UI, the design, the dark patterns - all of it sucks.

It's a job. Nobody particularly wants to be there. There's nothing sacred about the product. Engineers don't worship it.

It isn't a place you'd take a pay cut for the opportunity to work there.

Hence the bloat.


Before LLMs, it was a frequent haunt of fiction writers.

Work is delivering value.

Yes, we have craftsmanship, but at the end of the day everything is ephemeral and impermanent and the world continues on without remembering us.

I think both the IC and executive are correct in superposition.


Indeed. Even the ur-craftsman, John Carmack, says that delivering value to customers is pretty much the only thing that matters in development. If AI lets you do that faster, cheaper, you'd be a fool not to use it. There's a reason why it's virtually a must in professional software engineering now.

Right, but Carmack has always delivered. I need someone without Carmack’s ability to show me what value they can deliver with AI

They cant, its your job to train them or take their salary

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: