Shutterstock adds generative 3D as Getty speeds up AI images

Ask any 3D artist about the most tedious part of their week, and they will probably tell you it is modeling background forks or hunting down an environment map that matches the lighting.

Now, Shutterstock and Getty Images are stepping up their generative AI games with help from NVIDIA. Both platforms just launched major upgrades powered by NVIDIA Edify architecture and NIM microservices, focusing strictly on models trained on commercially safe, licensed data.

3D assets in seconds and 16K environment maps

Shutterstock has launched its Generative 3D service in commercial beta. Designers can punch in a text description or upload an image to generate a single 3D asset preview in as little as 10 seconds.

Once approved, that preview converts into a production-ready 3D asset complete with clean geometry and physically based rendering materials like wood, leather, or concrete, fully compatible with standard digital content creation software.

At SIGGRAPH, Shutterstock plans to showcase a Blender workflow generating assets right inside the viewport, while HP will display 3D prints made from these generated models. Ad giant WPP is already hooking it into virtual production pipelines.

Shutterstock also rolled out 360 HDRi generation. Instead of hauling multi-camera rigs on location to capture lighting data, creators can prompt text to get a 16K resolution panoramic HDRi background. Dassault Systèmes and CGI studio Katana are already preparing to wire the API into their pipelines.

Getty Images gets faster and adds pro camera dials

Meanwhile, Getty Images supercharged its own generative image engine. The upgraded Edify model generates images twice as fast, churning out four images in about six seconds and scaling them up to 4K resolution.

Creatives also get much tighter steering wheel controls. Users can adjust focal length and depth of field, or feed in sketches and depth maps to lock down composition.

Crucially, Getty now supports enterprise fine-tuning. Brands can train the commercially safe foundation model on their proprietary assets to ensure outputs stay strictly on-brand, a feature already in use by marketing giant Omnicom.

Why it matters

For 3D artists, game developers, and marketing agencies, these updates do not replace core artistry. Instead, they vacuum up the boring grunt work: populating background tables, testing prop variations, and finding matching scene illumination in seconds.

We think the real play here is legal peace of mind. Many corporate marketing teams still ban popular AI image generators because of murky training data and copyright lawsuits. By anchoring their models to licensed catalogs, Shutterstock and Getty are clearing the biggest roadblock to enterprise adoption.

That said, the actual geometry clean-up remains an open question. While clean meshes are promised, whether generative 3D meshes can drop seamlessly into animation rigs without manual retopology will decide if studios use this for final shots or just fast prototyping.