dmarcos

Splats can indeed come with view dependent colors to represent lighting, reflections… Those are captured in values called spherical harmonics of different levels 0, 1, 2, 3 representing different aspects like diffuse, directional lightning, highlights… Not all splat reconstructions come with spherical harmonics though. And they add rendering overhead.

Besides spherical harmonics splats are also good at representing high frequency detail that is harder with traditional meshes: Hair, clouds, vegetation… notice how trees in these reconstructions already look much better than in google maps / earth 3D views.

Principled: open source is the right thing to do for the greater good

Tactical: Open source is the best tool to advance whatever goals I have at the moment but that could change.

What’s evidence Xi and China are principled when it comes to open source / weights?

What user above said is exactly what I was trying to convey

I don’t know. thinking machines perhaps?

A lot of things have to come together: talent, capital that is not looking to maximize returns, ambition to compete in the marketplace. At peak Mozilla it was clear to everyone that we needed an undeniably successful product to be able to influence the market. Advocacy alone is not very effective.

Only way would be an actual open source model: code + training data

All countries see AI as a geopolitical concern. An open source strategy is smart. It’s effective and also builds good will / soft power. Many see Chinese AI companies as the good guys and root for them. I’m pro open source and also celebrate but still aware this is most likely just the means to an end.

I used to work at Mozilla. I think we are missing a player in the market with a more principled open source approach.

The outcome is plausible. Open weights models though look like a tactical more than a principled play by Chinese companies to overcome the disadvantage and difficulties to access western markets. Two issues:

1. If market conditions change they might decide to close down like Meta did.

2. If as you said models keep getting more expensive to train, is an open weights strategy financially sustainable?

edit: typo

I’m a big fan of Musk. One of the few criticisms I have is how xAI is also inconsistent with original OpenAI mission. I had imagined xAI as en effort to correct and fully embody all original values of OpenAI and that Elon says they betrayed. That makes his criticism weaker and I understand why some can think it was all about control. In his words:

"I'm the reason OpenAl exists. I came up with the name. The name OpenAl refers to open source... The intent was - what was the opposite of Google? It would be an open source non-profit."

I sometimes feel xAI wants to live up to those open values so I always celebrate when they decide to engage in open source. They still don’t fully embrace it. Perhaps because they think is not practical or will make them less competitive?

I mostly agree with you but, how do you square the use of other types of automation that are not AI? e.g computer animation vs hand-drawn, CAD software for engineering and architecture… Are those lesser art forms or we’re just used to them?

Oak Lab mission statement: A trillion-parameter agent that learns and plans in real-time with 20 watts of energy.

Seems the kind of goal that suits Carmack perfectly. I wonder what the disagreement was

Thanks so much! Really happy to hear. It makes all sweat, blood and tears of open source maintenance worth it.

I created and still maintain A-Frame (aframe.io). It’s been a gentle gateway to learn 3D graphics for a decade. Cool community if I can say so ha ha. Web is a great way to share stuff as you learn, collect feedback and get visibility. Many cases in the community of people that ended up doing 3d graphics professionally.

This is the most common scenario. That’s why can’t compare Vision Pro with the iPhone. First iPhone was limited yeah but those who bought it loved and used it all the time. Word of mouth setup the success of later versions.

Disclaimer: I haven't been involved in the game development but uses SparkJS (https://sparkjs.dev/) An open source 3D gaussian splatting rendering engine I help maintain.

I have the opposite impression. I have consumed tons of hours of interviews with Elon and he comes off as more technically sound and speaks more substance than any other large org CEO I can think of. Karpathy worked directly under him. I haven’t worked for Elon but know people that did and experience is similar. What makes you think you are right?

edit: typo

I was part of the temple build last year and cleanup is extremely serious. We spent two days cleaning after the burn with magnetic rakes looking for minute pieces of metal. We take samples of dirt at different spots and count the number of MOOP fragments to measure progress

How come? We give IP law / copyright legitimacy but it’s not clear to me the more I think about it. If you draw something you def own the physical drawing but owning the idea of the drawing during your lifetime feels strange to me. It’s also a very recent invention and humans created art before and will create after.

I spent most of my career in the open source world and doesn’t bother me models are trained on my output. Should I feel differently? It seems there’s a kind of ego or emotional attachment to the output that is more common among artists than devs? Perhaps abundance vs scarcity mindsets?

Thanks for the kind words. I helped on the brain rendering side of things but I’m not an expert in the field he he. You seem to be way more knowledgeable than me. Do you have any of your work out there? I would love to check it out.

TRIBE v2 (Trimodal Brain Encoder), a foundation model trained to predict how the human brain responds to almost any sight or sound.

I helped implement the demo

They offered a version a few months ago with 4x5090 for 25k

https://x.com/__tinygrad__/status/1983917797781426511

Stopped due to raising GPU prices:

https://x.com/__tinygrad__/status/2011263292753526978

Is it still your claim that people spending 60k on Cybertruck don’t want it? How do you know? Given the lack evidence feels like motivated thinking. You don’t like Elon and can’t accept that tons of people actually like him and his products.

The motivation to buy something is always because you want it. That a product doesn’t meet your needs or expectations later is a different story. What’s your evidence to claim that people spending 60k in a cybertruck don’t want it? What’s your evidence to make a similar claim or the opposite for any other purchase? Without evidence it feels you are making baseless claims about peoples motivations.

Just quoting the above

“An artificial boost is given to stuff nobody wants, but at a lower price can be convinced to buy”

So people spent 60k on a cybertruck that they didn’t want? Is that the claim?

So people spent 60k on a cybertruck they don’t want? Do you believe that?

Your claim is that people that bought the cybertruck at a lower price don’t actually want it?

FWIW it looks there’s now a demand surge with the introduction of the new cheap cybertruck variant. delivery dates pushed out to the fall of 2026.

I wish they would also publish the source images used to generate the 3D representation so people can recreate with other techniques.

I'll be around but don't have a GDC pass. Any interesting AI + gaming meetups or side-events I should go?

What's your favorite game of this type?