Luca Matteis

@NegativeVibrat1 This will sell WAY more than any AI generated painted stuff. We need more craftsmen in the digital world.

All textures in our game and hand painted, including wind particles ☺️ I believe that every manual brushstroke carries an artistic expression, making the visuals truly human." #screenshotsaturday

@unclebobmartin Why Iran over say North Korea, Eritrea, Turkmenistan, Syria, Belarus, etc?

My thoughts on the Iran issue. Thank God that there are leaders who understand the threat and are taking the necessary action to address it. I've been waiting for this for 47 years. I remember, all too well, the day the American hostages were taken, and the 444 days of their captivity. I remember, all too well, the Iranian diaspora, and the horrors that the regime imposed upon it's own people and upon the world. And when it is gone, and buried, with a stake through its heart, I will smile and say: Good Effing Riddance.

Jean-Claude Mézières (1938–2022) was a legendary French comic artist and one of the most influential visionaries in science fiction illustration. His aesthetic helped shape the visual universe of Luc Besson’s 1997 film The Fifth Element, where he worked as a concept artist. Many of the film’s iconic elements—layered megacities, flying traffic lanes, and eccentric extraterrestrial environments—echo the visual language he had developed decades earlier in his comics.

@ryanflorence Or maybe a different graph showing Top 10 runtimes? Javascript can run on Java ;)

I still can't believe they measure TS and JS as different languages for this graph. JS has been on top for a decade

@IsThisA3DModel @andreintg Although one could technically encode all the polygons of a 3D model in a 2D image 😅 But that concept will probably make your brain 🤯

@IsThisA3DModel @andreintg Because it's unclear what you're arguing. We identified that it's using 3D info from the scene (like depth and normal maps). That was the entire scope of the discussion. That's it. Nobody is arguing that an image is a 3D model.

the question in the video is coming up often🤔 As kids we didn't create by prompting a chat and waiting for random stuff to come out. Hands dirty with clay, paint and magic happened. 3D AI should feel the same, damn it!!

@IsThisA3DModel @andreintg You're messing with the context of the discussion. You made an analogy to how not every image is a 3D scene. The normal map is not a 3D scene (duh?), but a piece of information that can be derived from the SDF 3D scene.

@lmatteis @andreintg yes we are ya goof

@IsThisA3DModel @andreintg You're probably in the camp of people that thinks the only way to do 3D is via polygons. Nobody tell him about SDFs

@andreintg Doesn't matter. ControlNet does not take 3D object or scene data as an input nor does it produce it as an output.

@IsThisA3DModel @andreintg We were just discussing whether the new Unbound AI render was using any 3d info, and it clearly is. Nobody is talking about 3d models or scenes...

@lmatteis @andreintg This is such an incredibly silly argument. No every image every created is not a 3D scene and no ControlNet is not processing anything in a 3D scene like raytracing would.

@IsThisA3DModel @andreintg Call it whatever you want. Bottom line is that the AI model is using 3 pieces of information (the 3d vector). Not 2.

@lmatteis @andreintg ALL pixels are 3D vector data. But we don't call your profile picture a 3D model. This is such a silly argument. Its fucking image2image. God damn lmao

@IsThisA3DModel @andreintg A normal map is a 3D concept encoded inside a 2D texture. The data it stores is 3D. Each pixel stores a 3D vector. So while the container is 2D, the value at each pixel is 3D.

@lmatteis @andreintg A normal map is a texture map. Texture maps are 2D images lmao

@IsThisA3DModel @andreintg There's literally a normal map example in the README... that's 3D

@andreintg @lmatteis ControlNet is image to image. Its not taking 3D data as input. Its image to image like we said.

@IsThisA3DModel @andreintg As @andreintg mentioned, their AI toolset does use the 3d scene information (SDFs) to steer the AI. In that sense they're both using the 3d scene.

They *are* different. Raytracing works by bouncing light rays around actual 3D scenes. Img2Img does absolutely nothing with a 3D scene and runs denoising algorithms on a flat image and turns it into a different image. This is like saying a lion and a chair are the same thing because they are both things you can sit on and support themselves with 4 legs

@IsThisA3DModel @andreintg Curious, what makes you think they're different? They're both technologies based on algorithms that artists use to enhance their art. I should note that some composition algorithms in your favorite 3d app also use prior art/data to more realistically render the thing.

@lmatteis @andreintg Its not rendering 3D and it's not even remotely the same as raytracing

@IsThisA3DModel @andreintg Why though? I agree with @andreintg AI is just a tool. Similar to how raytracing is a tool you use to make your 3d rendering more realistic

@andreintg Renders are representative of a 3D model that actually exists. GenAI is not. If you tried to show off one of these GenAI outputs and claim it is a rendering of 3D work that would be incorrect and a red flag about the users credibility.

@unbound_io Question: couldn't one create that SDF model today with Unbound, take a screenshot, and feed it into any of the hundreds of AI image generation tools readily available to generate the more realistic image?

So close we can taste it...

Finished porting @unbound_io's SDF core to @WebGPU. Two days, 6h + 4h. Crazy complex code. Open source @Clinebot FTW! It spared me from learning the massively unnecessary WGSL language in depth. 🤓

@rauchg So "use vulnerabilities" also works with forks?

We've identified, responsibly disclosed, and confirmed 2 critical, 2 high, 2 medium, 1 low security vulnerabilities in Cloudflare's vibe-coded framework Vinext. We believe the security of the internet is the highest priority, especially in the age of AI. Vibe coding is a useful tool, especially when used responsibly. Our security research and framework teams are extending their help and expertise to Cloudflare in the interest of the public internet's security.

@andreintg People are just going a bit crazy because it's unclear what problem this solves. I think perhaps providing better understanding of this new AI future will calm the storm

this is me after a day talking with people about why we are doing AI

@IsThisA3DModel @andreintg All these tools available to 3d artists are "augmenting" my creativity because there's no way I could've made something so realistic via hand-drawing. Even just the ligthing is a huge automated creative decision done automatically for me.

3D modeling software does not augment your creative intention. Zbrush doesn't make a model for me. It doesn't make ANY creative decisions whatsoever. Everything I make in Zbrush is an accumulation of thousands of individual creative microdecisions I directly execute in a creation process. AI makes literally every creative decision in the execution of something.

Skilled artists have developed their eye to a high enough level where they can see how awful GenAI output is. The people who keep glazing AI on here every day don't realize they've just been consistently announcing to the world for the last few years how inexperienced they are.