Catalog
Lyra 2.0: single image to 3D asset
Nvidia's Lyra 2.0 takes one photo and hands back a 3D asset, which cuts out a lot of the manual work that usually sits between a reference image and something you can actually use in a scene.
What it is. An image-to-3D model from Nvidia. You give it a single photograph or render, and it generates a 3D asset from that one view.
Why it matters. Getting from a flat reference to a usable 3D object has always meant either a lot of manual modeling time or a pipeline stitched together from several tools. A model that takes one image and produces geometry directly shortens that gap in a meaningful way for motion designers, art directors building scenes, or anyone who needs a quick 3D pass on a visual idea.
What it improved on. Earlier image-to-3D approaches often required multiple views of an object, careful camera setups, or significant cleanup after generation. Lyra 2.0 is built around the single-image case, which is closer to how reference images actually exist in a working creative’s folder.
Strengths. Single-image input keeps the workflow simple. The weights are available on Hugging Face, so it fits into a local or self-hosted pipeline rather than locking you into a cloud service. Nvidia backing means the research is serious and the model has been built with real compute behind it.
Weaknesses. Single-image 3D reconstruction is genuinely hard, and the model will struggle with occluded geometry, the back of an object it cannot see. Output quality will vary with how well-lit and well-framed the input image is. It also wants capable hardware to run at a reasonable speed, and like most models in this space, the results will likely need some cleanup before they are production-ready.
More Models
New A fine-tuned LoRA on FLUX.2-klein that takes a black-and-white manga page and a reference image and colors the line art to match the palette and style of that reference.
LoRA for coloring manga pages by reference image
JUN 10
New Roblox released open weights for a diffusion model that turns text into 3D meshes, which puts game-ready asset generation in reach for anyone with the hardware to run it.
Roblox's open diffusion model for text-to-3D mesh
MAY 27
Black Forest Labs' 4-billion-parameter model handles both text-to-image and image editing in a lighter package, which makes it worth a look if the full FLUX models have been too heavy for your setup.