Meshy AI
Turn a prompt or a photo into a textured, game-ready 3D model on Meshy AI. Then put that asset on screen, with invideo's agent handling the prompting, routing, and technical work for you.
Why serious creatives choose Meshy AI
A sentence or a photo becomes a mesh
Describe the object or hand it a reference image, a photo, a sketch, a piece of concept art, and Meshy returns a textured 3D model in well under a minute. The thing that used to be an afternoon of box-modeling is now the first thirty seconds of the job.
Textures that arrive with the model
Generation includes a full PBR set: base color, normal, metallic, and roughness. The asset lands looking like a material rather than a gray blockout, and there is a higher-detail texture pass when a hero object needs to hold up close to camera.
Topology you can actually control
Low Poly Mode returns clean geometry at a polygon budget you set, and adaptive decimation lets you dial the same asset from hero density down to background prop. That is the difference between an asset you can ship into a real-time engine and one you have to rebuild.
Characters that come rigged and moving
Auto-rigging handles humanoid and quadruped characters in seconds, and a large library of motion presets, walk cycles, idles, jumps, gestures, applies straight to the rig. A character goes from prompt to performing without a skeleton ever being drawn by hand.
One generation, every format
GLB, glTF, OBJ, FBX, STL, and 3MF all come out of a single generation, so you pick the format the destination wants instead of regenerating for each one. FBX for the engine, GLB for the web, STL or 3MF for the printer.
Print-ready, not just render-ready
Meshy checks for watertight, manifold geometry and repairs what it finds, splits complex objects into printable parts, and exports multi-color 3MF for filament printers. The asset is ready to slice, not just ready to screenshot.
How Meshy AI works with invideo Agents
Meshy makes the asset. It does not make the shot, and Meshy's own recommended pipeline says so: generate the model, assemble and light it elsewhere, then hand a render to a separate AI video tool to get it moving. On invideo, that handoff is the part that is already built. Here is what that looks like in practice.
The agent picks the right model intelligently.
Meshy sits on a roster of 200+ models, and the agent routes work to it where it wins: props, weapons, furniture, environment set dressing, vehicles, product forms, and characters that need a rig more than they need a close-up. You never choose a model unless you want to, and if you have routing rules of your own, they hold for the whole project.
Your asset becomes a shot.
This is the seam that usually costs a day. A finished mesh still has to be framed, lit, moved, and cut before it is footage. The agent takes the asset from Meshy and carries it into the video models on the roster, so the object you described arrives as a shot in your timeline rather than a file in your downloads folder.
Your art direction rides along on every generation.
Approve your look, your palette, and your level of stylization once, and the agent saves them as the project's memory, then writes them into every Meshy prompt it sends. Asset twenty matches asset one, which is the whole game when the assets have to sit in the same scene.
One roster, one bill.
Meshy is on the same roster as the video and image models, under the same subscription. You are not paying for a 3D tool, an image tool, and a video tool and then moving files between three tabs to get one shot finished.
You always stay in control.
You set how much the agent does on its own: let it generate everything, ask before generations, or see every prompt before it runs. That setting is yours to make, and yours to change.
Who is Meshy AI for?
Meshy AI supports fast 3D asset generation across games, product, and motion workflows. With invideo Agents underneath it, each team gets the asset and the shot without moving between tools.
Game developers and indie studios
Create the props, environment dressing, and secondary characters that eat a schedule without ever being the reason anyone plays. Meshy gets a level populated at prompt cost instead of headcount cost, and the trailer gets cut in the same place.
Brand and product marketers
Create product forms, packaging, and concept objects you can turn, light, and put in a campaign before the physical sample exists. Useful when the product needs to be seen from every side and the render queue is the bottleneck.
Animators and motion designers
Create rigged characters and staged objects that arrive already moving, then take them straight into shots. Meshy gets a look tested at generation cost instead of pipeline cost.
Helping creatives stay creative
Multiplayer mode
Collaborate in real time with live cursors to show what everyone's working on.
Storyboarding
Turn any script or idea into a shot-by-shot plan, then tweak as needed before generating.
Script writing
Write your script inside invideo, and ask an AI co-writer for help if you'd like.
Timeline editor
Picture Premiere Pro with full AI.
Build your own agents
Create custom agents to fill specific roles like cinematographer, music designer, and more.
From solo creatives to creative enterprises
World-class investors stand behind invideo.
Backed by the firms behind Stripe, Spotify, Flipkart, and ByteDance.
Loading pricing…
Meshy AI FAQs
What is Meshy AI?
Meshy AI is a 3D model generator that turns text prompts or reference images into textured, downloadable 3D assets in under a minute. It covers modeling, PBR texturing, auto-rigging, and animation in one pass, and exports to the formats game engines, browsers, and printers expect. On invideo, it runs inside the agent's roster: you describe the asset in plain language and the agent writes the instructions.
What can the Meshy AI 3D generator make?
Props, weapons, furniture, vehicles, environment set dressing, decorative objects, product forms, and characters. Hard-surface and inanimate objects come out closest to finished. Detailed human anatomy and unusual creatures are where you should still expect to open a modeling package.
How does Meshy 3D generation work?
You give it a prompt or an image, it generates the geometry, then it generates and applies the PBR textures on top. Characters can then be auto-rigged and given motion from a preset library. Each stage is separate, so you can regenerate a texture without rebuilding the mesh.
What is Meshy image to 3D?
Meshy image to 3D takes a photograph, sketch, or piece of concept art and reconstructs it as a full 3D model rather than a flat extrusion. It is the mode to use when the thing already exists somewhere, as a product photo, a reference shot, or a drawing your art director has already signed off.
How is Meshy AI text to 3D different from image to 3D?
Meshy AI text to 3D is for things that do not exist yet, and it gives the model room to interpret, which is useful early and risky late. Image to 3D is for matching something specific, and it holds much closer to the reference. Early exploration wants text; anything that has to match an approved design wants an image.
Does Meshy AI generate animation?
Yes, for characters. Auto-rigging fits a skeleton to humanoid and quadruped models in seconds, and a preset motion library applies to that rig. It does not direct a scene, which is the part the video models on invideo's roster handle.
What formats does Meshy AI export?
GLB, glTF, OBJ, FBX, STL, and 3MF, all from a single generation. You do not regenerate to change format.
What are Meshy AI's limitations?
Fine anatomy is the weak spot: fingers, faces, and non-human creatures usually need manual work. Topology, while much improved and controllable, can still deform badly under heavy animation, so hero characters that carry a performance are not the use case. It is built for indie games, prototyping, and volume asset work rather than high-end productions where an artist needs control of every vertex. On invideo, the agent routes work that depends on those things elsewhere.
Do I need to know 3D modeling to use Meshy AI on invideo?
No. Meshy already removes the modeling, and the agent removes the prompting and the handoff. You describe the object and where it needs to end up, and the asset and the shot both come back.