¡¡
¡¡
¡¡
Yiqian Wu, Jing Zhang, Hongbo Fu, and Xiaogang Jin
GImage and shape samples generated by EG3D models trained with the same training strategy but using different datasets (our new dataset LPFF and FFHQ for (a) and FFHQ for (b)). The generators are conditioned by the average camera parameters. Shapes are iso-surfaces extracted from the corresponding density fields using marching cubes. Our dataset helps reduce distorted, ¡°seam¡±, ¡°wall-mounted¡±, and blurry artifacts exhibited in (b).
The creation of 2D realistic facial images and 3D
face shapes using generative networks has been a hot topic in recent
years. Existing face generators exhibit exceptional performance on faces
in small to medium poses (with respect to frontal faces) but struggle to
produce realistic results for large poses. The distorted rendering
results on large poses in 3D-aware generators further show that the
generated 3D face shapes are far from the distribution of 3D faces in
reality. We find that the above issues are caused by the training
dataset¡¯s pose imbalance. In this paper, we present LPFF, a large-pose
Flickr face dataset comprised of 19,590 high-quality real large-pose
portrait images. We utilize our dataset to train a 2D face generator
that can process large-pose face images, as well as a 3D-aware generator
that can generate realistic human face geometry. To better validate our
pose-conditional 3Daware generators, we develop a new FID measure to
evaluate the 3D-level performance. Through this novel FID measure and
other experiments, we show that LPFF can help 2D face generators extend
their latent space and better manipulate the large-pose data, and help
3D-aware face generators achieve better view consistency and more
realistic 3D reconstruction results.