Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

Microsoft open-sourced a 4B model that turns any image into a production-ready 3D asset in 3 seconds. It’s called TRELLIS.2, a fully textured, physically accurate 3D models with PBR textures out of the box. → Full PBR (base color, roughness, metallic, opacity) → Handles hair, cloth, glass, non-manifold geometry... show more

Superman

12,078 subscribers

283,528 views • 1 month ago •via X (Twitter)

Science & Technology

Anya Rossi• Live Now

Private livecam show

35 Comments

Superman1 month ago

Repo:

RedRising1 month ago

i have seen a few posts similar to this, and i decided to play around with all the AI tools to see how they can't make actual cad models. Not one has produced anything close to an object that you could actually use. Making a AI video like this isn't the same thing!CADcanIIan

Jett Insight1 month ago

topology is probably a mess

System of Catgirl Idealism e/acc1 month ago

that came out 7 months ago

androo1 month ago

This is a bullshit model that requires another GATED model DinoV3 that you need to get META to approve. Open my ass.

Derek Fossier1 month ago

*cries in

Riz1 month ago

Tell me you haven’t used it without tell me

Hopper1 month ago

Does it work for 3D printing?

Walid Aliouche1 month ago

That was released a while ago

Alejandro Maestre | AI1 month ago

TRELLIS.2 suena prometedor pero 3 segundos para un asset 3D con PBR completo es un salto enorme. Me pregunto qué tan bien escala con geometría compleja o si solo funciona para objetos simples.

Vlad Oreshkov1 month ago

The gaming industry will take a hit with this one. More and more people are testing AI’s on abilities to clone games. If this becomes the standard, I wouldn’t know the difference between hand made games and AI made games. Both will be amazing to play though.

Aaron Wacker1 month ago

This is Epic. Thanks @Microsoft Trellis is the best png to glb 2d to 3d model transformer and will be the rise to generative 3d models from image generation with #threejs and #aiuiuxjs apps and sims.

Magicfit1 month ago

open source is doing the heavy lifting

Johnny Utah1 month ago

hey cool thanks for sharing a 6 month old repo this is the equivalent of 2,000 human years in the world of AI

Kilgor Trout1 month ago

@threejs It is good before you see topology :)

K8r1 month ago

waiting on these things to get animated. I am good with tripo3d for now and meshyAI

Matthew Mombrea1 month ago

Prerequisites System: The code is currently tested only on Linux. Hardware: An NVIDIA GPU with at least 24GB of memory is necessary. The code has been verified on NVIDIA A100 and H100 GPUs. Let me just grab my $8k - 30k GPU quick.

星川 ✨ StarFlow1 month ago

Has anyone tried running this locally yet? Wondering how much VRAM it needs.

Abides1 month ago

@Spectromachina thought of you as soon as I saw this

Quantum Homeboy1 month ago

Production ready? Lmao. Guess china can just take a picture of a raptor engine and spacex edge is gone

Vyacheslav Ops1 month ago

"Runs locally, fully open training codebase" is the detail worth highlighting more than the speed — no cloud dependency, no data leaving your machine for asset generation. That's a meaningfully different trust model than most generative tools shipping right now

Modly1 month ago

This can be use directly in Modly 👀

Anthon Noire1 month ago

@SpaceXAI this is exactly the asset extraction feature I was suggesting for @grok imagine images and videos. Basically you begin with a prompt to produce an image and refine it. Each image is a sample run. The right image is then a reference point for an animation scene. The extrusion of the 2D asset to a timeline in motion produces extrapolated logical 3D geometry. That scene/video acts as the basis for multiple 3D assets which can be selected extracted and saved as .STL .obj .NC .gcode .ply .f3z .rvt and .dxf for example. (Autodesk, Gaussian splats, 3D printer, 360 fusion etc.) The combined outputs would be physicalization and phygital or virtualization products, such a chair, scale figure, or partially/wholly complete hypercar, scaling detail with iterative models. I'm surprised 4B already gets close.

Slew Shy1 month ago

goodbye to yet another profession and market

Zimeng Xiong1 month ago

this gets posted like every two weeks

kalos1 month ago

Can it do anatomy?

Derek Fossier1 month ago

@loktar00 may be helpful with your characters.

Pearl AI1 month ago

fine tuning on your own assets is the real flex

VeyrAshka1 month ago

Necessary. Now we can divide logic and tasks up to engage multiple resins types and tools completing a synchronized project with a data centre, functioning as a digital brain running a gyroscopic masterpiece of biomimicry. Modeled after a tree, an octopus, a spider…or fungi

α Sagittarius ✨1 month ago

In then meantime they also shipped Windows 11 and Cortana.

Bally_AgenticAI1 month ago

Image-to-3D is a reasoning problem, not pixel generation - the model infers geometry it never sees. Shipping it at 4B open-weights is the real story: it drops into a local creative pipeline with no cloud round-trip. #GenAI #AITools @Microsoft

Camaleón Raro1 month ago

Open-weights 4B models running locally on consumer GPUs disrupt traditional asset pipelines. Sub-second inference with PBR textures eliminates the 0-00 per-asset API cost from proprietary cloudservices. How is VRAM consumption holding up during batch processing?

Devil Cloud1 month ago

Ah older news

Gluteus Maximus 🇹🇯🇰🇬🇺🇿🇰🇿🍣🥩🏋️1 month ago

Can’t wait for @machineviolence to go 3d 😁

Nils1 month ago

pretty cool

Related Videos

MICROSOFT DROPPED A 4B PARAMETER MODEL THAT TURNS ONE IMAGE INTO A 3D ASSET IN 3 SECONDS and it's open source TRELLIS.2 fully textured, physically accurate 3D models with PBR textures out of the box not a rough mesh..not a placeholder roughness, metallic, opacity the kind of detail that makes things look real under any lighting and it handles the weird stuff too..open surfaces, hollow interiors, geometry that breaks every other tool the model doesn't know the word "limitation" apparently demo is live on hugging face right now

MICROSOFT DROPPED A 4B PARAMETER MODEL THAT TURNS ONE IMAGE INTO A 3D ASSET IN 3 SECONDS and it's open source TRELLIS.2 fully textured, physically accurate 3D models with PBR textures out of the box not a rough mesh..not a placeholder roughness, metallic, opacity the kind of detail that makes things look real under any lighting and it handles the weird stuff too..open surfaces, hollow interiors, geometry that breaks every other tool the model doesn't know the word "limitation" apparently demo is live on hugging face right now

Vaishnavi

355,121 views • 4 months ago

Meshy V7 is now available on fal - Turns an image into a textured, export-ready 3D model in seconds - From image to animated character in one pass - Auto-rigging and ready-to-use animation - Full PBR textures included production-grade materials and clean topology

Meshy V7 is now available on fal - Turns an image into a textured, export-ready 3D model in seconds - From image to animated character in one pass - Auto-rigging and ready-to-use animation - Full PBR textures included production-grade materials and clean topology

fal

34,300 views • 1 month ago

Someone just built a tool that turns any object image into a fully coded, animation-ready 3d model in your browser. just drop a photo and it sculpts the object in pure Three.js code, animation-ready with pivots, physics anchors, and PBR materials. 100% free.

Someone just built a tool that turns any object image into a fully coded, animation-ready 3d model in your browser. just drop a photo and it sculpts the object in pure Three.js code, animation-ready with pivots, physics anchors, and PBR materials. 100% free.

Superman

37,450 views • 1 month ago

someone open-sourced a tool that turns any phone video into a full 3D world in real time. you record a walk around your house, drop the mp4 in, and the entire scene comes back as a walkable 996k-point world at 60fps on your local GPU. no cloud. no subscription. runs 100% locally. 100% open source.

someone open-sourced a tool that turns any phone video into a full 3D world in real time. you record a walk around your house, drop the mp4 in, and the entire scene comes back as a walkable 996k-point world at 60fps on your local GPU. no cloud. no subscription. runs 100% locally. 100% open source.

Superman

818,816 views • 1 month ago

How to split a 3D model into printable parts in one click? Here’s the full workflow with Meshy Auto Split: - Generate a draft model with Image-to-3D - Hit Auto Split. In 40 seconds, you get parts that follow the model’s natural structure - Download and open your slicer Every part comes out clean, watertight, and ready to print. Connectors? We hear you. Coming soon 👀

How to split a 3D model into printable parts in one click? Here’s the full workflow with Meshy Auto Split: - Generate a draft model with Image-to-3D - Hit Auto Split. In 40 seconds, you get parts that follow the model’s natural structure - Download and open your slicer Every part comes out clean, watertight, and ready to print. Connectors? We hear you. Coming soon 👀

MeshyAI

71,403 views • 2 months ago

Automated PBR Textures. Flat color vs PBR. Normal 3D scans only capture raw color. They bake in shadows and have zero material data. Drop them into a modern lighting engine, and they look lifeless because they don’t know what’s shiny or matte. So we built an automated PBR export. Delit Base Color, Roughness, and Metalness—all output alongside your geometry. Is it 100% flawless every time? No. But it gives you a massive, creative head start. With some tweaking and layering, get the exact look you want in minutes instead of hand painting for hours.

Automated PBR Textures. Flat color vs PBR. Normal 3D scans only capture raw color. They bake in shadows and have zero material data. Drop them into a modern lighting engine, and they look lifeless because they don’t know what’s shiny or matte. So we built an automated PBR export. Delit Base Color, Roughness, and Metalness—all output alongside your geometry. Is it 100% flawless every time? No. But it gives you a massive, creative head start. With some tweaking and layering, get the exact look you want in minutes instead of hand painting for hours.

KIRI Engine - 3D Scanner App

12,302 views • 3 months ago

We're releasing ACE-Step-v1.5(2B), a fast, high-quality open-source music model. It runs locally on a consumer-grade GPU, generates a full song in under 2 seconds(on an A100), supports LoRA fine-tuning, and beats SUNO on common eval metrics. GitHub: Key traits: Quality: beats Suno on common eval scores Speed: full song under 2s on A100 Local: ~4GB VRAM, under 10s on RTX 3090 LoRA: train your own style with a few songs License: MIT, free for commercial use Data: fully authorized plus synthetic The music AI space lacks commercial-grade open models. Many creators are forced to rely on closed-source services, and can’t fully own, run locally, or fine-tune their own models. We want to help change that.

We're releasing ACE-Step-v1.5(2B), a fast, high-quality open-source music model. It runs locally on a consumer-grade GPU, generates a full song in under 2 seconds(on an A100), supports LoRA fine-tuning, and beats SUNO on common eval metrics. GitHub: Key traits: Quality: beats Suno on common eval scores Speed: full song under 2s on A100 Local: ~4GB VRAM, under 10s on RTX 3090 LoRA: train your own style with a few songs License: MIT, free for commercial use Data: fully authorized plus synthetic The music AI space lacks commercial-grade open models. Many creators are forced to rely on closed-source services, and can’t fully own, run locally, or fine-tune their own models. We want to help change that.

ACE Music

1,058,127 views • 7 months ago

I had an early access to GPT-6 Astra and I can say, everyone in the world now has a 3D designer at their fingertips. I gave it an image of a house and asked to create it in 3D with all the details including toys, appliances and furniture. This a full 3D model reconstruction in Blender with geometry that you can manually tweak and run 60fps as a locally rendered "game" on device.

I had an early access to GPT-6 Astra and I can say, everyone in the world now has a 3D designer at their fingertips. I gave it an image of a house and asked to create it in 3D with all the details including toys, appliances and furniture. This a full 3D model reconstruction in Blender with geometry that you can manually tweak and run 60fps as a locally rendered "game" on device.

Tom Krcha

1,435,333 views • 10 days ago

img2threejs 1.5.1 is out 🚀 Checkout: This release introduces a new approach to image → 3D → Three.js. Image → Tripo → img2threejs → reconstructed geometry → Three.js Instead of using the generated 3D model as the final asset, img2threejs uses it as a reference to rebuild the character into a representation that can be generated and rendered directly in Three.js. No Blender workflow, and no GLB required as the final output. The animation in the demo is still a bit rough because this release is focused on reconstruction, not animation. I’ll be improving that in the next release. Still experimental, but this is a big step toward where I want img2threejs to go.

img2threejs 1.5.1 is out 🚀 Checkout: This release introduces a new approach to image → 3D → Three.js. Image → Tripo → img2threejs → reconstructed geometry → Three.js Instead of using the generated 3D model as the final asset, img2threejs uses it as a reference to rebuild the character into a representation that can be generated and rendered directly in Three.js. No Blender workflow, and no GLB required as the final output. The animation in the demo is still a bit rough because this release is focused on reconstruction, not animation. I’ll be improving that in the next release. Still experimental, but this is a big step toward where I want img2threejs to go.

Nick

117,298 views • 21 days ago

There's a free, open-source music generator that claims to land between Suno v4.5 and Suno v5 in quality, and it'll run locally on a GPU with under 4GB of VRAM. It's called ACE-Step 1.5. → Free to try instantly, no GPU, no install → Generates a full song in under 10 seconds on a regular RTX 3090 → Train it on your own sound: feed it about 8 of your own songs and it learns your style as a LoRA in about an hour Paste a genre and mood, get back a finished song with vocals, in any of 50+ languages.

There's a free, open-source music generator that claims to land between Suno v4.5 and Suno v5 in quality, and it'll run locally on a GPU with under 4GB of VRAM. It's called ACE-Step 1.5. → Free to try instantly, no GPU, no install → Generates a full song in under 10 seconds on a regular RTX 3090 → Train it on your own sound: feed it about 8 of your own songs and it learns your style as a LoRA in about an hour Paste a genre and mood, get back a finished song with vocals, in any of 50+ languages.

Simplifying AI

16,221 views • 20 days ago

2D is what the camera sees.👀 3D is what the world is. 🌍 Today, we’re proudly open-sourcing QuerySplat — an open-source feed-forward 3D model that turns a few unposed images into a navigable 3D scene in seconds (run on 4090🪶). Your input: a few unposed images Our output: a sharp, navigable 3D scene — generated in seconds! 🚀 🧠 organize scenes with 3D queries, not pixel-bound Gaussians 📷 no camera poses or manual calibration required ⚡ reconstruct new scenes in a single forward pass ✨ preserve sharp geometry and high-frequency appearance 🏆 achieve state-of-the-art results on DL3DV-Evaluation 🔓 code and model weights are now open-source From pixels to scenes, we are making 3D generation scalable. Try our free iOS app — it’s a lot of fun! 😆 🔗 Try it on the Web: 🔗 Paper: 🔗 Code & weights: 🔗 Project page: 🔗 #AIGC #3DGS #AI #Computervision #3D #artist #3dartist #3d

2D is what the camera sees.👀 3D is what the world is. 🌍 Today, we’re proudly open-sourcing QuerySplat — an open-source feed-forward 3D model that turns a few unposed images into a navigable 3D scene in seconds (run on 4090🪶). Your input: a few unposed images Our output: a sharp, navigable 3D scene — generated in seconds! 🚀 🧠 organize scenes with 3D queries, not pixel-bound Gaussians 📷 no camera poses or manual calibration required ⚡ reconstruct new scenes in a single forward pass ✨ preserve sharp geometry and high-frequency appearance 🏆 achieve state-of-the-art results on DL3DV-Evaluation 🔓 code and model weights are now open-source From pixels to scenes, we are making 3D generation scalable. Try our free iOS app — it’s a lot of fun! 😆 🔗 Try it on the Web: 🔗 Paper: 🔗 Code & weights: 🔗 Project page: 🔗 #AIGC #3DGS #AI #Computervision #3D #artist #3dartist #3d

InSpatio｜影溯

29,806 views • 1 month ago

New open-source 3D world-generation model. I'm rendering a couple of worlds in the video, so check it out. You'll find the GitHub and the Hugging Face links to the model below. This is a multi-modal world model that you can use for a bunch of things: • To generate new worlds • To reconstruct worlds • To simulate 3D interactive worlds from a prompt, images, or a video You can edit the 3D outputs in Unity and Unreal Engine (they export as meshes, 3DGS files, and point clouds). You can also generate 3D characters in the world and walk around. Pretty fun stuff!

New open-source 3D world-generation model. I'm rendering a couple of worlds in the video, so check it out. You'll find the GitHub and the Hugging Face links to the model below. This is a multi-modal world model that you can use for a bunch of things: • To generate new worlds • To reconstruct worlds • To simulate 3D interactive worlds from a prompt, images, or a video You can edit the 3D outputs in Unity and Unreal Engine (they export as meshes, 3DGS files, and point clouds). You can also generate 3D characters in the world and walk around. Pretty fun stuff!

Santiago

65,446 views • 4 months ago

China open-sourced a model that reconstructs any scene in 3D from a regular video, in real-time. one camera. no LiDAR. 10,000+ frames without falling apart. just walk around with your camera and watch the entire world get rebuilt in 3D at 20 fps. → runs at ~20 FPS on a single GPU → Stable over 10,000+ frames → Beats optimization-based methods on benchmarks → Works on drone footage, driving videos, indoor walkthroughs 100% open source.

China open-sourced a model that reconstructs any scene in 3D from a regular video, in real-time. one camera. no LiDAR. 10,000+ frames without falling apart. just walk around with your camera and watch the entire world get rebuilt in 3D at 20 fps. → runs at ~20 FPS on a single GPU → Stable over 10,000+ frames → Beats optimization-based methods on benchmarks → Works on drone footage, driving videos, indoor walkthroughs 100% open source.

Yasir Ai

254,590 views • 1 month ago

China open-sourced a model that reconstructs any scene in 3D from a regular video, in real-time. one camera. no LiDAR. 10,000+ frames without falling apart. just walk around with your camera and watch the entire world get rebuilt in 3D at 20 fps. → runs at ~20 FPS on a single GPU → Stable over 10,000+ frames → Beats optimization-based methods on benchmarks → Works on drone footage, driving videos, indoor walkthroughs 100% open source.

China open-sourced a model that reconstructs any scene in 3D from a regular video, in real-time. one camera. no LiDAR. 10,000+ frames without falling apart. just walk around with your camera and watch the entire world get rebuilt in 3D at 20 fps. → runs at ~20 FPS on a single GPU → Stable over 10,000+ frames → Beats optimization-based methods on benchmarks → Works on drone footage, driving videos, indoor walkthroughs 100% open source.

Superman

1,456,075 views • 1 month ago

Small Language Models (SML) are the future of AI. "Small" (SML) instead of "Large" (LLM). These small models are highly specialized models with superhuman abilities on specific tasks. Here are two techniques to build these models: • Spectrum • Model Merging I give you a short introduction in the attached video, but here is a quick summary: Spectrum helps us identify the most relevant layers to solve one specific task. We can ignore everything else and focus on fine-tuning these layers. Using Spectrum, we can fine-tune models in a heartbeat. Model Merging combines multiple models into a unique, much better model than any of the individual input models. You can also combine models specialized in different tasks and get a model with multiple abilities. This is the state of the art of productizing models. It's what Arcee.ai's platform does behind the scenes. Arcee collaborated with me on this post and is sponsoring it. There are three main steps to produce a model for your particular use case: 1. You create a dataset by uploading your data. 2. You train a model. At this step, Arcee uses Spectrum and Model Merging to produce a highly specialized model for your task. 3. You can deploy that model to any environment you want. Three important notes: • Training process is 2x faster and 2x cheaper than regular fine-tuning. • Resultant models are smaller and have higher accuracy. • They create these specialized models from open-source models. Check this site so you can fully appreciate how this works: If you want to fine-tune an open-source model, consider Arcee's platform. This is the state of the art.

Small Language Models (SML) are the future of AI. "Small" (SML) instead of "Large" (LLM). These small models are highly specialized models with superhuman abilities on specific tasks. Here are two techniques to build these models: • Spectrum • Model Merging I give you a short introduction in the attached video, but here is a quick summary: Spectrum helps us identify the most relevant layers to solve one specific task. We can ignore everything else and focus on fine-tuning these layers. Using Spectrum, we can fine-tune models in a heartbeat. Model Merging combines multiple models into a unique, much better model than any of the individual input models. You can also combine models specialized in different tasks and get a model with multiple abilities. This is the state of the art of productizing models. It's what Arcee.ai's platform does behind the scenes. Arcee collaborated with me on this post and is sponsoring it. There are three main steps to produce a model for your particular use case: 1. You create a dataset by uploading your data. 2. You train a model. At this step, Arcee uses Spectrum and Model Merging to produce a highly specialized model for your task. 3. You can deploy that model to any environment you want. Three important notes: • Training process is 2x faster and 2x cheaper than regular fine-tuning. • Resultant models are smaller and have higher accuracy. • They create these specialized models from open-source models. Check this site so you can fully appreciate how this works: If you want to fine-tune an open-source model, consider Arcee's platform. This is the state of the art.

Santiago

164,162 views • 2 years ago

Run your coding agent on a local model, with QVAC. OpenCode now plugs into a model on your own machine through our official plugin. That changes two things. Privacy by default. Your repo, your prompts, and every command the agent runs stay on your device. No third-party model ever sees your code. For an agent that reads your codebase and runs commands, that is the point, not a nice-to-have. New workflows. With no meter and no rate limit, the agent does not have to stop. Let it loop on a task, refactor a whole repo overnight, or run a watch-build-fix loop that keeps going. Setup takes a few minutes: one line in opencode.json and OpenCode runs on a local QVAC model. In the video demo, it handles three real developer tasks back to back, fully offline.

Run your coding agent on a local model, with QVAC. OpenCode now plugs into a model on your own machine through our official plugin. That changes two things. Privacy by default. Your repo, your prompts, and every command the agent runs stay on your device. No third-party model ever sees your code. For an agent that reads your codebase and runs commands, that is the point, not a nice-to-have. New workflows. With no meter and no rate limit, the agent does not have to stop. Let it loop on a task, refactor a whole repo overnight, or run a watch-build-fix loop that keeps going. Setup takes a few minutes: one line in opencode.json and OpenCode runs on a local QVAC model. In the video demo, it handles three real developer tasks back to back, fully offline.

QVAC

11,382,936 views • 2 months ago

384GB of VRAM. On your desk. Not in a datacenter. Autonomous, the company behind SmartDesk, just open sourced complete build guides for a personal AI computer that runs open models on hardware no one can switch off. It's called Autonomous Computer. Three configurations, one philosophy: a model you rent can be cut off overnight. A model running in your own house can't. → Home build: 2x RTX 5090 workstation → Team build: 4x RTX PRO 6000 Blackwell, 384GB VRAM → On-prem business build: 8x RTX 4090/5090, up to 256GB VRAM Every config ships with the full bill of materials, 3D printable and CNC housing files, wiring diagrams, BIOS tuning, and assembly photos. Software side covers OS setup, NVIDIA drivers, and serving open models through Ollama, vLLM, and llama.cpp. Fork it. Change it. Build it. Sell it. MIT License. 100% Opensource. Repo:

384GB of VRAM. On your desk. Not in a datacenter. Autonomous, the company behind SmartDesk, just open sourced complete build guides for a personal AI computer that runs open models on hardware no one can switch off. It's called Autonomous Computer. Three configurations, one philosophy: a model you rent can be cut off overnight. A model running in your own house can't. → Home build: 2x RTX 5090 workstation → Team build: 4x RTX PRO 6000 Blackwell, 384GB VRAM → On-prem business build: 8x RTX 4090/5090, up to 256GB VRAM Every config ships with the full bill of materials, 3D printable and CNC housing files, wiring diagrams, BIOS tuning, and assembly photos. Software side covers OS setup, NVIDIA drivers, and serving open models through Ollama, vLLM, and llama.cpp. Fork it. Change it. Build it. Sell it. MIT License. 100% Opensource. Repo:

Harman

48,002 views • 2 months ago

LETS GOO! Generate full songs (4 min) with vocals in less than 10 seconds - open weights model! 🔥 It's crazy how much you can achieve from open models as of today! - going right for Suno and likes! VAE + Base model combined is < 2.5GB Open weights on the hub and a space to play around with it! 🤯

LETS GOO! Generate full songs (4 min) with vocals in less than 10 seconds - open weights model! 🔥 It's crazy how much you can achieve from open models as of today! - going right for Suno and likes! VAE + Base model combined is < 2.5GB Open weights on the hub and a space to play around with it! 🤯

Vaibhav (VB) Srivastav

64,204 views • 1 year ago