What it is
Generating full 3D models (characters, props, environments) from text or a single image — with textures, sometimes with retopology and rigging, ready for Unity/Unreal/Blender/Godot. This is the trend that blew open game dev's narrowest bottleneck: modeling by hand.
Where it came from
Early text-to-3D produced blobs. In 2024-25 Meshy and Tripo pulled both quality and pipeline up (2D to 3D, textures, low-poly, engine export), alongside a wave of image-to-3D. The industry marks 2025 as "the year 3D AI went to production".
Why it took off
- It removes the main jam. One character by hand is dozens of hours; an AI prototype is seconds.
- Within reach of a solo developer. The free tiers of Meshy and Tripo make entry nearly free.
- The whole cycle in one place: model, textures, rig, export.
How to use it now
- Prototype an asset: Meshy/Tripo — from text or by uploading a concept image (which you can generate in Midjourney or FLUX).
- Low-poly mode for real-time engines; export to Unity/Unreal/Blender/Godot.
- Clean up geometry and UVs by hand in Blender (AI gives you a blank, not a final). What you get: a starter set of 3D assets for a game prototype or visualization, many times faster.
What to watch out for
- Topology and UVs are often messy — production quality still needs manual cleanup.
- Licensing for commercial use and for model training — read the service terms.