Enhancing Photorealism Enhancement
Enhancing Photorealism Enhancement
Stephan R. Richter, Hassan Abu AlHaija, and Vladlen Koltun
Paper: https://arxiv.org/abs/2105.04619
Code and data: https://github.com/intel-isl/PhotorealismEnhancement
Project page: https://intel-isl.github.io/PhotorealismEnhancement/
We present an approach to enhancing the realism of synthetic images. The images are enhanced by a convolutional network that leverages intermediate representations produced by conventional rendering pipelines. The network is trained via a novel adversarial objective, which provides strong supervision at multiple perceptual levels. We analyze scene layout distributions in commonly used datasets and find that they differ in important ways. We hypothesize that this is one of the causes of strong artifacts that can be observed in the results of many prior methods. To address this we propose a new strategy for sampling image patches during training. We also introduce multiple architectural improvements in the deep network modules used for photorealism enhancement. We confirm the benefits of our contributions in controlled experiments and report substantial gains in stability and realism in comparison to recent image-to-image translation methods and a variety of other baselines.
Видео Enhancing Photorealism Enhancement канала Intel ISL
Stephan R. Richter, Hassan Abu AlHaija, and Vladlen Koltun
Paper: https://arxiv.org/abs/2105.04619
Code and data: https://github.com/intel-isl/PhotorealismEnhancement
Project page: https://intel-isl.github.io/PhotorealismEnhancement/
We present an approach to enhancing the realism of synthetic images. The images are enhanced by a convolutional network that leverages intermediate representations produced by conventional rendering pipelines. The network is trained via a novel adversarial objective, which provides strong supervision at multiple perceptual levels. We analyze scene layout distributions in commonly used datasets and find that they differ in important ways. We hypothesize that this is one of the causes of strong artifacts that can be observed in the results of many prior methods. To address this we propose a new strategy for sampling image patches during training. We also introduce multiple architectural improvements in the deep network modules used for photorealism enhancement. We confirm the benefits of our contributions in controlled experiments and report substantial gains in stability and realism in comparison to recent image-to-image translation methods and a variety of other baselines.
Видео Enhancing Photorealism Enhancement канала Intel ISL
Показать
Комментарии отсутствуют
Информация о видео
Другие видео канала
Tiny Baby Stoat Has The Best Reaction When She Meets Someone Like Her | The Dodo Little But FierceAny Song Sounds Better In Heavy Metal | Bill Bailey | Universal ComedyThe W124 Mercedes 500E was the world's most perfect sedan | Revelations with Jason Cammisa | Ep. 05VENOM 2 Official Trailer (2021)Free View SynthesisThe Man Who Drove McDonald’s out of Iceland | Hungerlust S2Is it possible to spin the last gear? (1:65000 gear ratio)How Fast Would an F1 Car Go at the Indy 500?TIMELAPSE OF THE FUTURE: A Journey to the End of Time (4K)How I hacked the Apple AirTagsI promise this story about microwaves is interesting.I Installed 1,000+ GTA 5 Mods and Here Is The Result - Better Than GTA 6?A.I. teaches itself to drive in TrackmaniaVladlen Koltun: Further Towards Photorealism (May 2021)Abandoned Puppy Followed Us Home And We Let Him InGTA 5: Stunning Photorealism With An Experimental AI ToolShaq using normal sized things and being a giant"Shut Down" Doesn't Actually Shut Down Your PCMortician Answers Dead Body Questions From Twitter | Tech Support | WIREDTRUE Limits Of Humanity – The Final Border We Will Never Cross