A stand-alone, workflow-oriented comparison between five tools that have overlapping functionality for both 3D modeling and game asset creation.
It is an ongoing dilemma for game developers to achieve a balance between speed and control. The tool could create a very nice-looking model, but the game development process also needs an asset that can be textured, edited, optimized, and implemented.
In this round-up, we will limit our scope to tools that share the same functionality as Meshy in the realm of artificial intelligence-assisted 3D modeling and game asset creation. We will not include other related software products that are specialized in creating skyboxes, browser-based 3D environments, and artwork.
1. Meshy AI : Extensive workflow for asset generation and preparation
Meshy AI includes both text-to-3D and image-to-3D technologies, as well as AI texturing, remeshing, auto-rigging, and animation. Moreover, their product documentation describes export and workflow options for game engines and 3D software. This might be useful if you need to go from asset conception to engine test without piecing together every step manually.
Useful for: asset prototype creation, asset variation creation through texturing, testing whether your asset concept can work in the engine.
Remember about: checking polygon count, UVs, materials, scale, and deformation in your game engine. Wide functionality does not mean that all generated models are ready to use right away.
2. Tripo: Asset iteration using text and images
Tripo aligns with Meshy in terms of creating 3D models based on prompts and references. This is useful for game teams that experiment with assets at speed, and may act as an additional tool for generating initial meshes and variations. Prior to making it a part of the pipeline, be sure to analyze its export capabilities and preparation capabilities in relation to your game.
Best for: fast iterations of asset concepts and alternatives.
Important to note: validity of the mesh generated, and rigging/export tools available in your plan of use.
3. Rodin by Hyper3D: High-resolution 3D model creation
Rodin is similar to Meshy in text- and image-based modeling, in that its product marketing focuses on control over the geometry and topology of the model, something that can be relevant if the shape is the focal point. In game development terms, measure the output mesh against the budget for your target platform, and see what work your team will have to do.
Use it for: detailed assets where geometry fidelity is critical.
Be aware of: planning the work needed down the line, particularly if deformations and animation are required.
4. Sloyd: Flexible assets for game development workflows
Sloyd fits in the list as it is another tool that creates 3D assets in games. In addition to supporting flexibility, it can be used for creating variations of existing asset types, whereas Meshy can be used to create asset types in a prompt-based process. If consistency is important, test if the asset types offered by Sloyd match those needed in your game.
Try it for: stylized props and variations within supported categories.
Look out for: if the controls provided by Sloyd allow for sufficient creative freedom.
5. Kaedim: A production-oriented way to take concepts to assets
Kaedim offers the same functionality as Meshy: it helps teams convert ideas into game-ready 3D assets. The main difference lies in the approach Kaedim uses to achieve this purpose - it is a production-oriented tool, offering teams the opportunity to revise and refine their work. This might make Kaedim a valuable consideration for those studios who need a managed asset process. For smaller teams, the process and price will have to be weighed against the team’s own capabilities.
Suitable for: game development teams interested in taking concepts towards production assets.
Look out for: how well its service approach fits in with your timeline and budget.
Choosing the right one
Focus on your requirements for the assets you will be using in your game. Create a prototype of your prop and your character and analyze it in the same engine under the same limitations. Compare silhouette, texture uniformity, topology, export process, and the amount of effort from your artists that goes into making the assets useful.
The actual test should be able to answer the following questions:
- Can the tool generate assets and style you are looking for?
- Can you optimize and edit them to fit your performance limitations?
- Is the export process clean and works without issues in your engine and pipeline?
- Are licensing and usage policies and limitations appropriate for your game?
In order to conduct a truly comparative test, use the same prompt or picture in the process for each model, as well as measure both generation time and cleanup time. The model, which might take more time for generation, might end up being a better choice as it requires less clean up.
Review the asset in the context; do not limit the review to turntable pictures only. Place the asset next to your current game assets, apply your project's lighting, and check the asset at the distance it would be viewed by the players. This way you can identify issues with scale, texture, silhouette, or level of detail.
For characters, make sure to have a simple deformation test in your review process: look at the joints and animate it if the rigging of the character is required. For props, consider collision needs, pivots, and reusability or scalability options.
Using Meshy AI as your first tool in such test is completely reasonable if you need several generation and preparation steps in your workflow. Tripo, Rodin, Sloyd, and Kaedim can be tested alongside with Meshy AI if their way of generating assets is more suitable for your project. The best solution is always the one that requires less production effort.
In what stage of a game development pipeline would AI-generated assets come in handy?
They are most helpful at the early stages of game development when designers could turn ideas into props, level designers could place items in blockouts, and artists could experiment with shapes or textures before creating an asset. That is why generated assets could be used for prototypes and testing even if the final version of an asset would have to be created from scratch or modified significantly.
An actual workflow might look something like that: generate a possible asset, review its silhouette and style, retopologize it in your 3D software if necessary, make sure you know about materials and scale, and then import it into the game engine and test it for performance and lighting. You must save the prompt or reference that was used to generate the asset along with it.
“Game Ready” definition for your team
It’s different for each project. In one project, it could mean a textured placeholder for a prototype, while in another project, you need efficient geometry, proper UVs, a standardized materials setup, proper scale/pivots, collision, and rigging for animation. List your project’s requirements before benchmarking the toolset; otherwise, a refined preview could be confused with production-quality content.
Conclusion
While AI 3D generators could certainly be useful for game studios in terms of testing and implementing asset concepts in shorter periods of time, the actual comparison lies in the full path from the initial idea or reference to an actual asset fitting into the particular artistic direction, engine, and performance requirements. Test out one of the tools on a small scale, tally the number of clean-up steps, and go with the generator that truly supports the process of creation.
Top comments (0)