An independent game developer planning an underwater adventure might need a coral-covered treasure chest, a sea creature and a few plants just to see whether the setting works. Instead of modeling each object by hand, the developer can use artificial intelligence to generate rough 3D versions from a description or reference image.
Meshy is one example of a browser-based tool that offers text-to-3D and image-to-3D generation. Both approaches give creators a model they can rotate and inspect before committing to a design. The difference is what they start with: a description of something they want to explore or an image of something they already have in mind.
Starting with a prompt or an image
A text prompt can spell out details such as shape, material and style. “A weathered brass observatory with circular windows and a telescope on the roof” gives the tool more to work with than “make a building.” The result gives the creator a way to assess the proportions and silhouette, though details may not come through as intended.
An image can be more useful when shape is difficult to describe. A designer working on a desk accessory with an unusual curved base, for example, could start with a sketch and compare the generated model with the original drawing. Where a workflow accepts multiple images, references from different angles can show features a single view leaves out.
Adding textures and preparing characters for animation
Meshy also offers tools for texturing and automatic rigging. Texturing lets a creator try different colors or surface finishes without rebuilding the shape. A designer could compare how the same lamp looks in brushed metal or painted wood before choosing a direction.
Rigging adds a digital skeleton so a character can move. Automatic rigging can give an animator something to test, but the character still needs to be checked in motion. A model that looks fine standing still may bend awkwardly at the shoulders or knees once animated.
Taking a model into a game or to a 3D printer
For a small game team, rough models of benches, signs and luggage can help establish the look of a railway station during early playtesting. Before shipping the game, the team may still need to simplify the geometry, fix textures or adjust the model so it runs well in the game engine.
A model for 3D printing needs a different set of checks. An STL file may open in slicing software, but that does not mean the object is ready to print. Its dimensions, wall thickness and geometry still need to be checked. Other formats, such as OBJ, GLB or FBX, may be better suited to a 3D editor or game engine, depending on what the receiving software supports.
Where browser tools and an API fit
Meshy also provides browser tools for viewing and converting 3D files, along with an STL repair utility. Those tools can help with a quick inspection or file conversion. Developers can use the company’s API to connect supported generation features to their own applications or automate parts of an asset workflow.
Before using a generated model commercially, creators should check the tool’s licensing terms and confirm they have the rights to any reference images they upload. The practical test is whether generating the first draft saves time after accounting for the cleanup needed to use it.