Olivier Ozoux’s Post

I created these stones with holes in Cinema 4D 2024 using Boolean and volumetric modeling, then dynamically simulated the cloth in Marvelous Designer, and finally rendered the effects using Octane Render. It is worth mentioning that I created these stickers in Midjourney, then used Bigjpg to non-destructively enlarge the stickers, and finally applied them to the 3D scene. This is a combined creation of AI and 3D.

An image of a demon character for a face mask was generated in the Stable Diffusion neural network. Next, a 3D model was created for this image, corresponding to face tracking in the Lens Studio. SnapChat lens link: https://lnkd.in/ea99g4ue Implementation: Stable Diffusion - generating a fantasy character for a face mask; Blender - Low Poly modeling; Lens Studio - AR effect, face tracking, VFX effects. #augmentedreality #snapchatlens #ar #3dmodeling #neuralart

Taylor Swift model looking much more realistic this time. This work has a 3d to AI and AI to AI workflow. I modelled a simple 3d character with Zbrush and rendered with Maya, then generated the first sequence with the rendered images in the Stable Diffusion. Then I generated the SD generated sequence one more time with the Krea.ai and the final result is looking much more realistic that way. #3d #3dmodeling #cgi #vfx #stablediffusion #stabilityai #ai #aiart #aicommunity #aicccreators #1image #characterdesign #characterart #charactermodeling

This will change the process of architecture, urban design , landscape architecture and interior design.

What is the hash tag for procedural 3d modeling? How are the work flows and data streams managed for parts shopping lists and schedules?

I’ve recently seen some work like this being done in engineering. I’m fascinated by the intersecting of industries like this. Gaming - Engineering Enterprise software - space planning Online education - HR development Architecture - UX design Cyber security - insurance I’m convinced we spend so much time navel gazing at our own work that we don’t allow ourselves to be inspired by those adjacent to us. In the 15 years of Crema I’ve studied hundreds of companies. Startups - global enterprises Services - Product B2B - B2C Sports brands - financial services In some ways this is what AI has been attempting to do as well. Collect all the ideas and generate the concepts between them. We ought to do the same thing. This is why I still believe that consulting, design, and creativite firms have a bright future. Sure, enabled by ai. But ultimately helping companies connect dots of opportunities they don’t see before. Keep playing Keep learning Keep exploring Keep making.

It is all “intelligence modeling”. There is nothing artificial about it. Intelligence persists where it can. Bubbles come and go. https://lnkd.in/eFNjtnFS

Procedural tools have a lot more control built in too.

Finally had a chance to evaluate Postshot! The gaussian splats are better than a month ago. However, I REALLY struggled with the camera animator. Pro tip: stop the training at 20k. At 30k I was getting giant gaussians because pruning and densifying stopped well before that step. If I was to use this more for rendering, I would need a better toolset. Which brings me to my final point: who is going to make the great gaussian splat tool with camera animations outside of Luma AI and Nerfstudio? Neither of those solutions can use PLY or Splat files from other toolsets. #3d #AI #VFX #computervision

Guess what? - Aender Lara's method relies on Bone Constraints. - and consider kaedim3d.com for AI that turns sketches into 3D game assets in minutes #GDC #GDC2024 #a16z #GamingIndustry #GameNews #GameDevelopment #VideoGameDevelopment #IndieGame