ComfyUI just did something quietly huge. The open-source tool that built its name on node-based image generation has officially stepped into the third dimension. The latest releases around v0.31.0 turned the native 3D Viewer on by default and wired up an entire family of export nodes, which means creators can now generate, preview, and ship 3D assets inside a tool that runs entirely on their own hardware. No subscriptions, no cloud lock-in, no black box.
The moment you open the new 3D Viewer is genuinely exciting. Hit generate, and instead of a flat grid of tiles you are staring at an actual scene you can orbit, zoom, and inspect from every angle. For a community that has spent years stitching together image and video workflows, this feels like the missing piece finally snapping into place.
ComfyUI now speaks the language of 3D
This is not another wrapper around some closed 3D API. ComfyUI shipped the infrastructure natively. The Save 3D node family went from a power-user trick to a first-class citizen, giving you direct control over how your generated geometry leaves the tool.
- Save 3D (Advanced): export refined 3D scene assets with precise control over the pipeline
- Save Splat: drop your result straight into the 3D Gaussian splatting format that game engines and real-time renderers love
- Save Point Cloud: publish point-cloud data for scanning, mapping, and spatial pipelines
In plain terms: you can now take a text prompt, turn it into a 3D asset, and export it in formats the rest of the industry actually uses. For indie game developers, product designers, and VFX artists, that is the difference between a toy and a real workflow.
Where the momentum shows up
The timing is everything. ComfyUI has spent 2026 systematically absorbing every content format. Native LTX 2.5 brought open-weight video, Wan models pushed animation forward, and the MiniMax H3 nodes opened up 2K video with sound. The music tools landed too, turning your GPU into a studio. Now 3D layered on top of all of it.
- Image: the original strength, still the baseline
- Video: LTX 2.5 and Wan model support built right into the nodes
- Audio and music: native text-to-audio tooling in recent releases
- 3D: live viewer, splat export, and point-cloud support
When one open-source tool can carry you from a blank canvas to a finished 3D scene, the creative pipeline stops being a pile of disconnected tools and becomes a single canvas. That is the real breakthrough here, and the community is already salivating over it.
Investors believe it too
This is not just developers being enthusiastic. Earlier this year ComfyUI announced a $30 million financing round at a $500 million valuation, bringing its total funding to $48 million. The round was led by Craft, with participation from Pace Capital and others. Money like that does not chase hype; it chases a platform that creators genuinely cannot stop using.
Pair that funding with the new 3D pipeline and you can see the strategy. ComfyUI is not content to be the best image UI anymore. It is becoming the local-first home for all of generative creation, and 3D is the final frontier it needed to crack.
What to try right now
The best way to feel the shift is to open the newest ComfyUI version and watch the 3D Viewer light up. Drop in a splat workflow, orbit the result, and export a point cloud. You will understand in ninety seconds why the community is calling this the start of something big.
ComfyUI went 3D natively, and it is every bit as exciting as it sounds. The canvas just grew a new axis, and open-source creators are the ones holding the brush.
Comments