Comment on page
How it Works
A behind the scenes look at Kaedim's 3D model pipeline
Kaedim's approach to generating 3D models combines both our in-house developed machine learning models for reconstruction and an in-house team of artists to refine the 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. This represents a 20x improvement from manual modeling times. But we're not stopping there. Every month we implement an improved version of our machine learning models to automate a greater portion of the total modeling process, leading to faster outputs for you and your team.
Once your image is submitted, it'll go through a series of preprocessing steps to help our AI generate an accurate representation of your image in 3D. The preprocessed image then enters our machine learning portion of the pipeline and exits as a 3D model. But here's the tricky part: it's not always perfect. Automated 3D model generation is brand new technology, so the models don't always look ready for production.
That brings us to our third step in the pipeline. A Kaedim artist checks your model, and makes any necessary edits for it to be production-ready. Not only does this ensure a great experience for our users (they get the model they need), 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.
Your model is now ready for download, texturing, and more! We provide a number of different tools to further improve your model. Go ahead and add textures, combine multiple generations, or export the model into your favorite modeling software to put your own spin on it.
The above process is fast, but it still takes some in-house work. So for the time being, we're primarily targeting enterprise customers that have a significant need for Kaedim's large-scale production capabilities. We're improving our system every day and we aim for a fully autonomous release in 2024. With a fully autonomous system, we will be able to provide plans that better target the prosumer use case.
Our other features take parts of the core pipelines to provide further enhancements to your models. Some of them, like texturing, are fully automated today. Others, such as model iterations, require an in-house team member to intervene and ensure optimal quality.
Last modified 1d ago