Distortion-Free Wide-Angle Portraits on Camera Phones

YiChang Shih
Wei-Sheng Lai
Teaser (a) A group selfie taken by a wide-angle 97° field-of-view phone camera. The perspective projection renders unnatural look to faces on the periphery: they are stretched, twisted, and squished. (b) Our algorithm restores all the distorted face shapes and keeps the background unaffected.


YiChang Shih, Wei-Sheng Lai, and Chia-Kai Liang, Distortion-Free Wide-Angle Portraits on Camera Phones, to appear in SIGGRAPH 2019

Paper: High resolution PDF (142 MB) | Low resolution PDF (13 MB)


Photographers take wide-angle shots to enjoy expanding views, group portraits that never miss anyone, or composite subjects with spectacular scenery background. In spite of the rapid proliferation of wide-angle cameras on mobile phones, a wider field-of-view (FOV) introduces a stronger perspective distortion. Most notably, faces are stretched, squished, and skewed, to look vastly different from real-life. Correcting such distortions requires professional editing skills, as trivial manipulations can introduce other kinds of distortions. This paper introduces a new algorithm to undistort faces without affecting other parts of the photo. Given a portrait as an input, we formulate an optimization problem to create a content-aware warping mesh which locally adapts to the stereographic projection on facial regions, and seamlessly evolves to the perspective projection over the background. Our new energy function performs effectively and reliably for a large group of subjects in the photo. The proposed algorithm is fully automatic and operates at an interactive rate on the mobile platform. We demonstrate promising results on a wide range of FOVs from 70° to 120°.


Supplemental Materials

  • See our method on 167 images from Flickr and comparisons to other methods at here.
  • Details in lens distortion correction and user study at the supplemental document.
  • Acknowledgements

    We thank the reviewers for numerous suggestions on user study and exposition. We also thank valuable inputs from Ming-Hsuan Yang, Marc Levoy, Timothy Knight, Fuhao Shi, and Robert Carroll. We thank Yael Pritch, David Jacobs, Neal Wadhwa, Juhyun Lee, and Alan Yang for supports on subject segmenter and face detector integration, Kevin Chen and Sung-fang Tsai for GPU acceleration. We thank Sam Hasinoff, Eino-Ville Talvala, Gabriel Nava, Wei Hong, Lawrence Huang, Chien-Yu Chen, Zhijun He, Paul Rohde, Ian Atkinson, and Jimi Chen for supports on mobile platform implementations. We thank Weber Tang, Jill Hsu, Bob Hung, Kevin Lien, Joy Hsu, Blade Chiu, Charlie Wang, and Joy Tsai for image quality feedbacks, Karl Rasche and Rahul Garg for proofreading. Finally, we give special thanks to Denis Zorin and all the photography models in this work for photo usage permissions and supports on data collection.