About 3D car synthesis(display details)
〇Synthesis procedure
- The feature vectors W (base) and W (style) are extracted from the two car bodies (base and style) to be combined using Encoder.
- Create an initial body mesh that will be the skeleton of the new design.
- Adjust the percentage of base / style ratio added to your new design.
- Feature vectors W (base) and W (style) are added to the body mesh using Decoder.
- Perform 5 addition steps to create a new car design.
〇Synthetic notes
- Encoder / Decoder for 3D shape recognition uses an AI model in which a large amount of vehicle body shapes are trained in advance.
- The 5 steps means 5 feature groups unique to the AI model.
- You can create various designs by giving parameters (ratio) to apply the style car to the base car in 5 steps.
Encoder
Decoder
- The file contains only coordinate and surface information, no information indicating the shape type.
- Please select a link of Base/Style. When you select a link, it will be displayed in the object view accordingly.
- The 3D object of the selected base / style car is displayed here.
- You can rotate the car by dragging left mouse button and move the car by dragging right mouse button.
- You can zoom in and out on the car with the mouse wheel.
- Adjust the ratio of applying the style car to the base car.
- You can adjust all parameters at once by "All Parameters" slider bar.
- You can adjust individually by "Parameter 1" to "Parameter 5" slider bar.
- If the parameter is set to 0% or 100%, a car body with the features of the base car or style car only will be generated.
(The application ratio is gradually increased to 20%, 40%, 60%, 80%, and the synthesis process is performed 4 times.)
- The combined base / style car 3D object is displayed here.
- You can rotate the car by dragging left mouse button and move the car by dragging right mouse button.
- You can zoom in and out on the car with the mouse wheel.