Ben H

Two of our worst VC stories: 1. A Sequoia partner passed on Cloudflare because he didn’t think a woman could lead a security infrastructure company. Seriously. 🙄 2. I got introduced to @pmarca. Meeting got scheduled for a Monday, which should have been a clue. I thought it was just a casual meeting. He thought it was a pitch and brought the whole @a16z partnership team. Hilarity ensued. 🤪 At one point one of them said: “You don’t seem very prepared.” Which was true because I wasn’t. I framed the rejection letter they sent.

@TingwuWang hi! is there any plan to release the motionbricks weight with humanoid bones ? I've been trying to experiment (inside GODOT) with the current G1-based weights in game, but since it lacks feet/toes and clavicle, it's a bit sketchy. thanks!

@JonWoodyCleans @ritwikpavan It looks massively cheaper than 8sleep.

@ritwikpavan No subscription and air quality detector seems to be the only difference because pod 5 also will raise heads when snoring is detected. Getting away from a subscription is a huge selling point

NEW: This bed claims to add up to an extra hour of real sleep. Dreamspan just introduced an intelligent sleep system that reads your body, tunes your bed, and monitors air quality automatically every night. • AI thermoregulation adjusts heat, airflow, and elevation while you sleep • Raises your head when it detects snoring • Tracks heart rate, HRV, breathing, recovery, and snoring with nothing on your body • Dual-zone temperature so each side can be warm or cool • Monitors CO₂, TVOCs, humidity, and room temperature • Thermal and vibration alarms wake you without noise @dreamspanHQ is turning the bed into an active sleep system.

@HGD_Dev Tu as prévu de release le plugin quand ?

A comparison video of dynamic GI methods in #Godot : VoxelGI/SDFGI/my LPVGI plugin, is available on Youtube ! 👉 youtu.be/4q4v4BNtY00

LARRY ELLISON: AI IS RAPIDLY COMMODITIZING BECAUSE MOST MODELS ARE TRAINED ON THE SAME PUBLIC INTERNET DATA. THE REAL COMPETITIVE EDGE ISN’T THE MODEL ANYMORE — IT’S ACCESS TO EXCLUSIVE, PROPRIETARY DATASETS. THAT MAY BE THE ONLY MOAT LEFT.

Why Paris may be the most important AI city outside Silicon Valley techcrunch.com/2026/05/28/why…

@davidpwalter I would love for you to pursue the work started on the Nvidia RTX fork.. Nanite + path tracing ?

Should I keep my Godot fork as separate or try to merge? TLDR: probably going to keep it as a fork (read below) but please let me know if you have an opinion. I've been thinking about this for a long time. I've implemented: 1. Godot WebGPU: mobile and forward+ renderers (strong beta level - 2months of work) 2. Godot nanite prototype: gpu driven renderer for opaque, transparency and lighting coming soon (<1 week to prototype, probably needs another 2 week to get to strong beta-level) These are huge Godot changes. Tens of thousands of lines of code. With effort I'm confident I could get webgpu merged. But on nanite: Juan Linietsky (the original creator of Godot) has said explicitly that he doesn't support visibility buffers as a viable GPU driven rendering path for Godot (I use visibility buffers in my implementation). Visibility buffers are way better than what he proposes, but among other reasons, the issue is they won't work on Godot's compatibility renderer. So it's probably almost guaranteed that the Godot team won't accept the Godot-nanite changes (I also rewrite a bunch of the core shaders to get it working). I'm not mad about it. I understand why they won't. I'm just stating the facts. Why do we need the compatibility renderer? for 2 big reasons: 1. some older hardware (e.g. 2015 and before) 2. web But WebGPU doesn't use the compatibility mode anyway. ~83% of browsers now support WebGPU with a few caveats. Chrome and safari support it. Threejs made webgpu their default renderer and many others are moving on to webgpu. It's here. The whole point is we are now in the future. AI is helping us develop games, it's helping us develop engines, making them faster, making them better. Maybe 5 years ago the compatibility renderer still mattered but with webgpu and time, it's probably less than 5% of godot game users if not less: just ones using really old android devices and that use firefox on linux. Apple devices are there, newer android devices are there, new desktops are all there, and almost all browsers are there. How can I possibly succeed at this?? Using AI to not just develop but test the engine fork. I've already done this extensively. I have an extensive set of tests that run the changes on many scenes, and I keep adding scenes. I've added unit tests and I've added fuzz test, and I'm just getting started. No hate to the Godot team, but I don't understand how they've made it this far with the minimal CI (continuous integration) pipeline that they have right now. AI development for something as complex as an game engine is only viable with extensive tests, but that's not an issue. You can and likely should spend 20% of your tokens developing features and 80% testing and verifying features. In my case, the 80% has been automatic as I develop features because as I develop, things don't work, so I add tests to get them working, then I keep the tests and the working features at the end. AI driven development is both the train and the train tracks, vs human development is like riding a bike. It has risks, but ultimately AI driven dev is here. I have a choice: 1. put in months of extra effort to convince the Godot team to get my PRs reviewed, verified, and merged 2. continue to improve my testing suite for the fork so it's basically undeniably working, I have AI review the code over and over, I have it continue to improve the testing suite, laying down the tracks, then the AI choo choo train chugs along. 3. I do nothing and someone does (2.) anyway. Godot, Unity, Unreal, all of them are DONE within 5 years if they refuse to use AI. Seems like Unreal and Unity are trying but they're too big and are too averse to changing legacy code. We have a chance to move forward into the future. We can choose to have better performing game engines, extensive and legendary testing suites. On top of this: Godot has 3 renderer levels: 1. compatibility: OpenGL, essentially legacy at this point 2. mobile: tile-based rendering in a single rendering pass, lacks fancy lighting features 3. forward+ fancier AAA-level features for lighting and more. If I move forward with this fork, my plan is to: 1. remove the entire compatibility renderer which is about 50% of the Godot codebase! 2. merge the mobile and forward+ renderers so there's only 1 renderer. And developers simply choose their feature set without choosing their renderer: this also removes another ~10% of the engine's code Then what we are left with: - Single renderer forward-merged - WebGPU for web - GPU Driven renderer (nanite) huge performance benefit - Improvements to global illumination that I haven't talked about in this post yet - huge testing suite improvements to prevent regressions moving forward - AI-driven development momentum that will not stop 🚂 - Most Godot projects will still work in this fork that use the mobile or forward+ renderers. With all of this, I think the best decision is to keep a fork of Godot and move forward. If the Godot team wants to use any of my improvements, they are very welcome to but if they aren't on board with my AI driven development strategy then that's totally fine and I think them for being the giant shoulders that I can stand upon. With that in mind, I can be convinced either way so Please comment below if you think I should or shouldn't keep a fork of Godot or if you have any other ideas related to this!

Real-time neural rendering model @moonlake Works on from any physics/game engine to close the sim-to-real gap. If this is relevant to your team, I'd love to chat (my DM is open)!

@__microdancing What’s the tech behind the face ?

sin instalador de 20 GB, sin precompilar shaders, un solo ejecutable de 200 mb, totalmente open source, 60 fps sin inmutarse en una macbook M1. godot señoras y señores. un invento ARGENTINO.

@spheno_xiphoid I understand the principle, but then, if you are constantly training, especially strength training/heavy lifting, you do actually need to work flexibility and stretching on tight muscles, or you'll lose it over time. What do you think ?

Forcibly stretching a tight muscle functions like an eccentric contraction, allowing calcium ions to leak into muscle cells. This can further stress the already taxed muscle, further contributing to its dysfunction. Avoid stretching chronically tight muscles when possible.

In 2023, a study tracked 24,000 professional athletes across 150 years of records. Published in Science Advances. The discovery: how fast an athlete’s performance declines predicts how long they live. Not during their career. Decades after retirement. The speed of decline is itself a biomarker for aging. Tonight, the @enhanced_games are publishing every drug protocol, running medical teams, and measuring performance to the hundredth of a second. 42 athletes. All young. All on testosterone and growth hormone. But imagine the same format applied to aging. A retired sprinter. 55 years old. Personal best of 10.3 seconds from 25 years ago. Currently running 11.8. You put them on a rejuvenation protocol, published, supervised, same infrastructure. They run 10.9. In one race, you’ve demonstrated what the longevity field has been trying to prove for a decade. And everyone watching understood it without reading a single paper.

@bryan_johnson @enhanced_games @james_maggie91 Bryan be like : don’t eat fast food it’ll kill you and then endorse a guy that injects himself with trenbolone - a steroid made for cows that is among the most toxic and unhealthy thing you can ever put into your body. Damn.

This may be the fastest race in swimming history… James "The Missile" Magnussen is 6'6", 35 yrs old, 2x world champion, and Olympic medalist. I'm commenting the @enhanced_games tomorrow night where James is competing. Insane stats on him... #1 His resting heart rate is 28 bpm (cleared by his medical team). For context, elite endurance athletes are typically 40–50 bpm. Mine is around 39-42 bpm. #2 Last year he was 253 lbs. This was after his first cycle of performance-enhancing-drugs. He got too big. He was sinking in the water. He's down 40 lbs this year to try and hit the sweet-spot of muscle and buoyancy. #3 He's so big he can barely fit into his enhanced suit. Enhanced Games swimmers wear full-body polyurethane "supersuits" that World Aquatics banned in 2010. These are the same suits that broke 43 world records in 2009. #4 He came out of a 7-yr retirement for this. He retired from competition in 2019. Tomorrow night, I'll be reviewing their protocols and measurements live.

@kommanderkarl That shader ported to GODOT would be so useful ! Are you going to share the code ?

No doubt. Parallax shaders in particular are one of my favorites as well. I'm the environment lead on this indie project, built this scene with Blender & Unreal 5. I built a custom landscape shader with 4 unique paintable material types, all using POM for fake 3D. I still remember being blown away by parallax mapping in Elder Scrolls: Oblivion (2004), though a much simpler form of it.

@SebAaltonen Is it open source ?

AI 3d gen produces super crappy UV charts (thousands of small islands) and geometry is not clean either. Impossible to LOD. Our pipeline remeshes them and calculates new optimal UV charts. Looks almost the same, runs much better. LODs much better. 100,000 -> 500 triangle box:

@LudensLudonauta Sounds interesting but are they all for 2D games ?

Wachu you think? I'm VERY THRILLED to start working on that. I've been refining this framework of mine for years and I think I can finally turn it into something useful for the society, indie game developers in particular. Let me know your thoughts.

I think I finally managed to come up with a great game design service to offer for indies. I call this "Gameplay Map", a collection of mechanics that build up a game. Each entry classifies the feature in my taxonomy framework and gives insights on when to use it and synergies.

@slimbuck7 @misaki00258 @willeastcott Hi @slimbuck7 I would love some advices on this - had some issues too with splat transform , I’ve dmed you

@misaki00258 @willeastcott Perhaps I can help @misaki00258 ? splat-transform needs flags based on the scene to generate good collision.

3DGSのメッシュ変換プログラム実装して見ました｡ #3dgs

Blackrock CEO Larry Fink said: "The United States 🇺🇸 is short power, short compute, short chips. There are going to be shortages in all three. We just don't have enough compute power right now."

@willahmed @WHOOP Hey @willahmed i'm a long term whoop user and I love the product, but maybe instead of whining about people copying you, you should actually build new - better features, and get better hardware, you're late. Also:stop gate-keeping features behind higher tiers, that's ridiculous.

engraved on every circuit board

@bilawalsidhu I'm thinking the method used by Hyworld 2.0 (see image), might be a good solution. They only released stage 4 for now so i'm waiting to test it..

@amentialinked You have world labs and spAItail as the obvious candidates - but you will be semi manually expanding or stitching scenes together to get something of this scale

Big indoor scan - fully explorable in real-time on a browser. Distribution of 3DGS is pretty much solved. Blocker is still large scale capture - you need an expensive LiDAR + RGB scanner to get results like this. 360 video is still hard to pose indoors.