Our app enables people to design their dream space with the help of AI.

4 features

1) AI inspirations - This feature allows the player to take a screenshot and send it to receive inspiration based on the room's layout.

2) AI Texture - The user can create an ideal texture or image using prompts, select and apply the preferred result to the furniture.

3) AI Assistant - The user can wake up the AI assistant, and are able to ask the AI assistant interior design/fengshui related questions

4) Auto layout and placement - Users can create an empty room outline, and based on the room's edges, the optimal furniture placement is automatically generated.

Challenges

AI Inspiration

  • Testing the screen capture requires a rebuild each time, which is time-consuming.
  • Additionally, configuring the Android manifest file has been problematic.
  • There are still issues with outputting high-resolution images, and while there's a potential solution, there's not enough time to implement it.

AI Texture

  • The issue/challenge we are still having now is to make sure the uv mapped correctly.
  • For texture generation such as ‘wooden’ texture, we have thought about a workaround: to add the key word ‘seamless’ in the prompt.

AI assistant

  • The first challenge is the chat box UI
  • Since every messages are instantiated and the chat box has specific boundary which needs to contain the text, thus breaking text into lines was challenging.

Auto Layout & Placement

  • We designed a system to align items correctly while ensuring logical placement without overlaps.
  • While there are still some challenges to refine, the placement accuracy has significantly improved from the initial version.
  • We faced challenges with the MRUK plugins, leading us to adjust our approach and set aside some features.
  • Despite these hurdles, the current placement functionality is much more reliable and user-friendly.

Built With

  • adobe
  • aikit
  • c#
  • controlnet
  • dalle2
  • gpt40
  • math
  • meta
  • screenshotloader
  • speechtotext
  • texttospeech
  • unity
  • winner
Share this project:

Updates