Tabellion1, Aleksander Holynski1, Ben Poole2, Janne Kontkanen1 1Google Research 2Google DeepMind *Equal contribution 2024/Aug/3 第61回 コンピュータビジョン勉強会@関東 CVPR2024読み会(後編) 2
and end frame and conditioning the super resolution model on the original high-resolution frames without additional parameters unlocks high- fidelity results • Classifier は使わず、start and end frame を使う • start and end frame : conditioning 2024/Aug/3 第61回 コンピュータビジョン勉強会@関東 CVPR2024読み会(後編) 8
• concatenates each (naively upsampled) low-resolution conditioning frame to the noisy high-resolution frames along the channel axis 2024/Aug/3 第61回 コンピュータビジョン勉強会@関東 CVPR2024読み会(後編) 14 start end