Page cover image

💡How it Works

A behind the scenes look at Kaedim's 3D model pipeline

Kaedim's approach to generating 3D models combines our state-of-the-art machine learning algorithms with our in-house art team. Our in-house team of artists help refine our 2D to 3D reconstruction outputs and improve future AI-generated models. This guarantees users get a production-ready 3D model, every time.

Today, the process takes an average of 15 minutes from user submission to a model ready for download, a 20x improvement from manual modeling times. But we're not stopping there - each month, we implement multiple improvements to our machine-learning models with the goal of automating a greater portion of the pipeline while maintaining our high-quality standards, leading to faster outputs for you and your team.

Inputs

You can input an image or text to generate your model. Before making your first submission, please check out input guidelines to ensure your output will look as expected.

Automated Generation

Once the inputs are submitted, they'll go through a series of preprocessing steps to help our AI accurately represent them in 3D. The processed images then enter our 2D to 3D reconstruction step of the pipeline and exit as a 3D model. But here's the tricky part: it's not always perfect. 3D reconstruction is brand new technology, so the models are not always ready for production.

Artist Refinement

That brings us to our third step in the pipeline. A Kaedim artist checks the 3D reconstruction and makes any necessary edits to ensure it's production-ready. This guarantees a great experience for our users since our outputs are of the highest quality, and it also helps train our machine-learning models. We're gathering generations and their improved versions in real time, creating a feedback loop for our AI to improve itself.

Model Download

Your model is now ready for download, texturing, and more! We provide a number of different tools to further edit your model. Go ahead and add textures, iterate the geometry, or export the model into your favorite modeling software to put your own spin on it.


A Note on Fully Autonomous Generations

The above process is fast, but it still takes some in-house work. The Kaedim team is working towards achieving better and better 3D reconstructions, automating more and more object categories. This work is aiming towards making our software accessible to more prosumers and indie projects.

We're improving our system every day and aim for a fully autonomous release in 2024. Currently, we're primarily focusing on enterprise customers that have a significant need for Kaedim's large-scale production capabilities.


Other Features

Our other features use parts of the core pipelines to enhance your models further. Some, like UV unwrapping, Levels of Detail and Texturing, are fully automated today. Others, such as model iterations, require an in-house team member to intervene and ensure optimal quality.

Last updated