1/3
2/3
3/3
Extending the love of tails into the digital realm.
- Type
- Internal
- Client
- Unknown
- Models
- Category
- Development
- Year
- Fall, 2023
Multi-Modal Virtual Character Synthesis: Robustness Testing
• Tests robustness using low-quality, ambiguous images.
• Generates high-quality Pixar-style characters from blurry input.
• Successfully creates a profile picture of the character.
• Shows resilience to challenging data and produces coherent outputs.
• Generates high-quality Pixar-style characters from blurry input.
• Successfully creates a profile picture of the character.
• Shows resilience to challenging data and produces coherent outputs.
An intended test on the robustness of photo analysis by inputting low-quality, ambiguous images. The model successfully generated consistent, high-quality Pixar-style virtual character scenes from a blurry input image. It was then asked to create an profile picture for the same character. Model demonstrated its ability to handle challenging data and produce coherent outputs.