What does AI 3D Modeling Mean?
What you see is what you get. A smart AI algorithm is used to model photographs, without the need for a separate scanner. Just take a picture and upload it via App then KOKONI will automatically generate a model prototype and print it. We're going to bring you a brand new and amazing printing experience.
Create a unique 3D model or avatar by simply snapping or uploading photos to the App, and let your creativity shine! Now, you can make anything from personalized avatars of your favorite anime or movie characters to complex models for art or school.
It's not Magic! It's Tech!
We live in a 3D world and our eyes have the incredible ability to perceive depth information. When we see an object with both eyes, we already have a 3D model of the object in our minds.
Enabling machines to create 3D models works on a similar principle, our photo modeling method is based on the recent development of deep learning technology. The algorithm first learns from numerous sample data to obtain an understanding of the 3D object’s appearance in the real world. This process has been perfected over the past two years, as we collected tens of thousands of 3D model samples and their corresponding pictures to optimize our machine learning algorithm so that it can generate high-fidelity realistic 3D models.
Avatar Modeling - How can we turn a selfie into a 3D avatar model?
In "Avatar Modeling", we only need to input 1 picture, and the algorithm uses deep neural networks to extract the 3D features. Then the neural network performs nonlinear processing of the features according to the learned method (at this point the neural network already has the ability to generate a realistic model), and finally, a realistic frontal 3D face is produced. As for the back of the avatar that the picture didn’t capture, the algorithm has already seen numerous 3D models of the "avatar" (as priori knowledge), so it can consistently complete the remaining features based on the recognized front features. The final result of the algorithm is a complete model that can be printed in 3D.
Object Modeling - How can we convert multiple pictures into a 3D model?
In "Object Modeling", we need more images for the algorithm to extract more information and these images need to be taken in a certain order. For each image, the algorithm will estimate the 3D depth information using the adjacent angle images taken, similar to how the human eye perceive depth using both eyes. We then use the estimated 3D depth as a further clue to generate a complete 3D model based on the global semantic information. Finally, secondary optimization is performed to correct features that do not match the picture information, and the final output of the algorithm is a complete manifold that can be printed in 3D. We combine the benefits of deep learning with the traditional optimization process - unlike traditional optimization-based photogrammetry, our deep learning-based approach captures global semantic information and combines prior knowledge of lighting and texture to make reasonable initial estimates, and also exploits the benefits of the optimization method to maximize the use of multi-view images taken by the user to target the synthesis of high-fidelity and 3D-printable models.
4 Steps to Print
With four easy steps, KOKONI is simple to set up. 3D printing just got easier. Most of operations can be completed within the application.
Extensive Models Gallery, 1-Click Printing
Thousands of creative models are ready for you, enjoy the fun of 3D printing with ease. The model library is constantly updated.
Visualize 3D Model Generation
You can adjust the model size and precision, visualize the model generation, and the APP will automatically check the status of filaments before printing.