- TimeLens: Event-based Movie Framework Interpolation
- Diverse Generation from a Single Videos Authorized
- Competent Precipitation Nowcasting making use of Deep Generative Models of Radar
- The Cocktail Shell Difficulties: Three-Stem Music Split for Real-World Soundtracks
- ADOP: Rough Differentiable One-Pixel Point Rendering
- (Style)CLIPDraw: Coupling Content and Style in Text-to-Drawing Synthesis
- SwinIR: graphics restoration utilizing swin transformer
- EditGAN: High-Precision Semantic Picture Modifying
- Urban areaNeRF: Building NeRF at Area Scale
OpenAI effectively trained a network capable build images from text captions. It can be much like GPT-3 and graphics GPT and create incredible results.
Google utilized a customized StyleGAN2 structure generate an internet fitting space where you are able to immediately try-on any jeans or t-shirts you want only using an image of yourself.