Haloween is too early?
Whad did I ?
- Claude Chat Desktop and Blender MCP (earlier I did claude code, run python script but no MCP)
- I was able install DLSS as Blender Viewport which is much faster than Optix for real time judgment
- it took 2 hours, more or less.
Whad did I ?
- Claude Chat Desktop and Blender MCP (earlier I did claude code, run python script but no MCP)
- I was able install DLSS as Blender Viewport which is much faster than Optix for real time judgment
- it took 2 hours, more or less.
1
0
0
21
Tomas Maldonis retweeted
You can also use the video decoder hardware as image transcoder. AVIF is the i-frame format of AV1. Just make your texture set a stream of i-frames. Then feed those iFrames in GPU VRAM to @castano’s Spark (ludicon.com/spark/) or a similar real time GPU ASTC/BC7 compressor.
Nvidia Nsight now supports Vulkan video (still not DX12), here is a frame from Wicked Engine using async decode of multiple videos:
2
10
148
6,575
Tomas Maldonis retweeted
Entirely true. It's the second-rate programmers that are having trouble adapting to AI, and screaming about it.
The A-list guys are delighted. Listen to Linus Torvalds. Or DHH. Or, a little outside my usual community, Uncle Bob Martin. We can get so much more done now!
The A-list guys are delighted. Listen to Linus Torvalds. Or DHH. Or, a little outside my usual community, Uncle Bob Martin. We can get so much more done now!
I agree with everything OP says except "watching all of those years go down the drain."
AI might have made it almost meaningless to write code by hand anymore. But the people who have spent years/decades writing code are now in the best position to leverage AI coding tools.
Learning to code was a stepping stone toward learning to decompose and solve complex problems. I have worked with a lot of different people with different levels of coding experience. And I can say with confidence that, in general, the more coding experience you have, the better you're positioned to think about long-horizon tasks and to organize and tackle problems with multiple moving parts.
All those skills directly translate to using AI coding agents. And I'm seeing it happen in real time. People with long software engineering experience have a totally different way of using AI coding agents (given that they set their ego aside and accept that this is the new norm). They are more organized, know how to steer the agent like guiding an intern or mentoring a junior dev.
People who don't have coding experience are using agents in a random and chaotic way (the approach has its outside-the-box-thinking merits but doesn't have the same reliability as the experienced SWE approach).
131
112
1,944
179,487
Tomas Maldonis retweeted
I feel like we're looking at something so much BIGGER than just better and more detailed SVGs.
SVGs are essentially images made entirely from code. Instead of storing millions of pixels like a PNG or JPEG, it uses math instructions to describe shapes, paths, colors, gradients, etc....
and Quiver is now generating these incredibly detailed illustrations that look almost like polished 3D renders, except they're actually editable vector graphics.
But what happens when these models eventually get so good that they can generate SVGs that are indistinguishable from photorealistic images? At that point, why would we need image models at all?
and if you can create any image entirely through code, you could theoretically generate every frame in a video in the same way and suddenly, instead of generating a fixed sequence as an mp4, you would be able to generate fully editable programmatic scenes.
I don't know what this means for the future, or if it will even work in this way, but holy molly, it's gonna be interesting
SVGs are essentially images made entirely from code. Instead of storing millions of pixels like a PNG or JPEG, it uses math instructions to describe shapes, paths, colors, gradients, etc....
and Quiver is now generating these incredibly detailed illustrations that look almost like polished 3D renders, except they're actually editable vector graphics.
But what happens when these models eventually get so good that they can generate SVGs that are indistinguishable from photorealistic images? At that point, why would we need image models at all?
and if you can create any image entirely through code, you could theoretically generate every frame in a video in the same way and suddenly, instead of generating a fixed sequence as an mp4, you would be able to generate fully editable programmatic scenes.
I don't know what this means for the future, or if it will even work in this way, but holy molly, it's gonna be interesting
These are all SVG!
Something really cool is coming to @QuiverAI around icons, asset families, and styles.
Reply for early access.
36
47
995
45,478
Topology is quite complicated which #opus didnt understand immediately. I had to create a skill. Hope I can keep improving the bad model to better one.
0
0
0
12
why do i think that Linux is the future OS?
Do you remember meme where one sells finger "holder" to held Macbook lid open ?
I can modify 100% OS except kernel which I leave to Linus. I could travel on tram and write a prompt on phone when the laptop is closed and tucked. #linux
Do you remember meme where one sells finger "holder" to held Macbook lid open ?
I can modify 100% OS except kernel which I leave to Linus. I could travel on tram and write a prompt on phone when the laptop is closed and tucked. #linux
0
0
0
69
Tomas Maldonis retweeted
The levels of shader math (in my opinion):
0. Linear, Multiply color by a brightness scalar
1. Steps, Floor coordinates for pixelation
2. Vectors, Center and correct aspect ratios
3. Geometric, Dot product for lighting
4. Exponential, Square distance falloff
5. Trigonometry, Revolving light source
6. Matrix, Coordinate space transformations
7. Bitwise, Bitpacking G-buffers
8. Derivatives, tangent space normal mapping
9. Integration, analytic volumetric fog rendering
0. Linear, Multiply color by a brightness scalar
1. Steps, Floor coordinates for pixelation
2. Vectors, Center and correct aspect ratios
3. Geometric, Dot product for lighting
4. Exponential, Square distance falloff
5. Trigonometry, Revolving light source
6. Matrix, Coordinate space transformations
7. Bitwise, Bitpacking G-buffers
8. Derivatives, tangent space normal mapping
9. Integration, analytic volumetric fog rendering
19
27
609
14,829
some video.
Still lots to do. My goal is realism. Also provide interaction for the lovers of Architecture.
#threejs
Still lots to do. My goal is realism. Also provide interaction for the lovers of Architecture.
#threejs
0
0
0
69
Constantly rebuilding the house takes my energy ... I need have some fun sometimes. I did decide to build a garage, then drive a car and put it inside the house.
hashtag#blender hashtag#threejs hashtag#web hashtag#creative
hashtag#blender hashtag#threejs hashtag#web hashtag#creative
0
0
0
25
Tomas Maldonis retweeted
Unitree Introducing: H2-X Kung Fu Legend Guardian🤖
Powerful performance to serve and guard you, with a maximum endurance of approximately 8 hours.
Please use the robot safely and in a friendly manner.
Powerful performance to serve and guard you, with a maximum endurance of approximately 8 hours.
Please use the robot safely and in a friendly manner.
1,833
3,673
27,267
11,947,139
Tomas Maldonis retweeted
Introducing tsrs: a Rust port of the TypeScript compiler 🦀
We are not forking TypeScript!
◆ In close collaborating with the TypeScript team
◆ Up to 8x faster than TS 7.0 & ~70% less memory
◆ Runs in WASM, so it works in your browser & on Workers
◆ Soon part of the Oxc toolchain
npx tsrs
blog.cloudflare.com/typescript-rus…
We are not forking TypeScript!
◆ In close collaborating with the TypeScript team
◆ Up to 8x faster than TS 7.0 & ~70% less memory
◆ Runs in WASM, so it works in your browser & on Workers
◆ Soon part of the Oxc toolchain
npx tsrs
blog.cloudflare.com/typescript-rus…
145
536
4,990
625,490
Tomas Maldonis retweeted
Linux. Git. SSH. MySQL. IRC.
Almost all started by individual Finns. 🇫🇮
The key here is individuals. The more organizations or broader environment get involved, the worse it gets.
It’s nice to be proud of where these things came from. But what matters more is what you do with them.
European individuals can invent. But America is, ironically, a much better collective environment for trying new things and turning them into something impactful.
That’s why so many these ambitious people move there.
Building something great is and needs to be a shared ambition. This has been missing from Europe maybe the last 70-100 years. Ambition is very little at all. Shared ambition even less.
Especially now with AI, you can raise your ambition a lot. Build the best products in the world. Make rockets, satellites, drones, robots, software, drugs, whatever.
Remove councils. Fire the grey hair boards. Refuse to raise from the EU financier VCs. Incorporate in Delaware if you have to. Put talented individuals in charge.
Averaging everyone’s ideas doesn’t produce great work and working with people stuck in the past is not worth it.
Reset your thinking.
What would it take for you to be the best of your generation?
What’s the greatest thing you could possibly build?
How could you build fast?
Make it matter.
Almost all started by individual Finns. 🇫🇮
The key here is individuals. The more organizations or broader environment get involved, the worse it gets.
It’s nice to be proud of where these things came from. But what matters more is what you do with them.
European individuals can invent. But America is, ironically, a much better collective environment for trying new things and turning them into something impactful.
That’s why so many these ambitious people move there.
Building something great is and needs to be a shared ambition. This has been missing from Europe maybe the last 70-100 years. Ambition is very little at all. Shared ambition even less.
Especially now with AI, you can raise your ambition a lot. Build the best products in the world. Make rockets, satellites, drones, robots, software, drugs, whatever.
Remove councils. Fire the grey hair boards. Refuse to raise from the EU financier VCs. Incorporate in Delaware if you have to. Put talented individuals in charge.
Averaging everyone’s ideas doesn’t produce great work and working with people stuck in the past is not worth it.
Reset your thinking.
What would it take for you to be the best of your generation?
What’s the greatest thing you could possibly build?
How could you build fast?
Make it matter.
We gave you C++.
We gave you DHH, RoR and Omarchy.
We gave you C#, PHP and TypeScript.
Friggin’ Bluetooth.
It’s. All. Danish. 🇩🇰
Now tell me again how technologically irrelevant Europe is.
I’ll wait.
What did you even ship??!
25
33
470
24,317
Tomas Maldonis retweeted
1. World models are good at cyberpunk apartments
2. World models are cyberpunk apartments
3. Cyberpunk apartments are world models
2. World models are cyberpunk apartments
3. Cyberpunk apartments are world models
21
105
1,219
51,522
Tomas Maldonis retweeted
you can give claude an entire 3d asset library for $0
2,300+ models, materials and lighting environments from poly haven.free api. no api key. commercial use allowed.
give it api.polyhaven.com/assets and ask:
“find a softbox studio lighting setup, download it and scene”
2,300+ models, materials and lighting environments from poly haven.free api. no api key. commercial use allowed.
give it api.polyhaven.com/assets and ask:
“find a softbox studio lighting setup, download it and scene”
14
48
776
43,163
Tomas Maldonis retweeted
You should see his house.
One of the most creative people I have met.
One of the most creative people I have met.
New follower? I’m Greg Madison. I’ve been pioneering tangible interfaces in XR for decades: making digital illusions part of the physical world, keeping us grounded in it.
A thread on the ideas behind my work. 🧵
Here’s an early exploration from 2009.
fr.slideshare.net/slideshow/the-…
10
30
505
28,189
Tomas Maldonis retweeted
Vibe coding, DIY hardware, and 3D printing will lead to a renaissance in arcade-style gaming.
391
1,850
20,578
715,419
Tomas Maldonis retweeted
Blenderを自分では一度も触らずにAIのみでBlenderシーンを作るテストの昼間バージョン
昼間のライティングへの変更もコンポジットもAI任せ
ライティングやフォグの扱いは以前に比べて遥かに良い
昼間のライティングへの変更もコンポジットもAI任せ
ライティングやフォグの扱いは以前に比べて遥かに良い
50
231
2,557
187,745
