What we are building
- We are implementing a trained AI model that controls the computer (manages the cursor and keyboard - as a person usually does) in the 3D modeling software Onshape
- Input: prompt. ("Make a basic LEGO-style block. It should have 8 round cylinders on top, in 2 rows. Keep the shape simple and clear.")
- Input -> converts text to a 3D model drawing in isometric through Nano Banana 2 API
- The trained AI model expects a drawing and not a prompt, so after the user clicks "Approve", inference begins.
- The research paper is attached
Requirements:
- Implement the existing code in this repository - https://github.com/ghadinehme/VideoCAD
- After implementation, connect a chat window that converts text to images through Nano Banana 2
- After creating the chat, add the "Approve" function, where the user has the option to accept or reject (change the prompt or indicate to Nano Banana 2 what is wrong with the drawing) the proposed 3D model drawing
- The finished model must work on my Mac, but if technically impossible - this requirement is waived
- It needs to be done by March 26 - this is my strict deadline
Other
- I will pay for the entire inference (after the project or before it starts).
- It is difficult to estimate the cost of the project, so I want to discuss the advance payment before the project and the second installment after completion
