Face
Create a digital human of any face just from a single photo with complete natural movements.
Face Swap
Face Swap is a cutting-edge technology that seamlessly replaces a specific person's face of a photo or video with the desired face.
With just a single photo, our Face Swap technology can transform the original image or video into a natural-looking and expressive face.
This advanced technology represents our core expertise in generating various contents through precise facial transformations, including facial expressions.
Result
Head Swap
Klleon's Head Swap technology replaces an entire person's head in a photo or video with just a single head image.
Unlike Face Swap, our Head Swap technology transforms not only the face but also the hairstyle, capturing even the nuances of facial expressions.
This remarkable technology creates a realistic effect, making it appear as if the you have actually participated in the desired scene.
Result
.png)
Target
Result1
Result2
Face Generation
Face Generation technology replaces a specific person's face in different photos or videos with the desired face.
By effectively separating facial features and utilizing 3D-based motion information, various applications can be achieved.
Result
Original
Tilting
Turning
Motion generation can be achieved with just a single photo, without the need to record actual videos.
Comparison
Original motion (Video)
Target face (Image)
Make-up Transfer
Make-up Transfer technology applies make-up onto a face using just a single desired make-up photo.
Without the need to physically apply makeup, one can experiment with various colors and styles on their face.
Result
We have developed our own generation model that allows the features of diverse images to be encoded into separate vector spaces.
Additionally, we have created an image encoder corresponding to the generation model to accurately embed given images into latent vectors.
This enables precise encoding of the given images into the desired vector space.
Comparison
Original face image
Make-up image
Company P
University T
LPIPS
0.013 / 0.003 / 0.075 / 0.036
68%
0.018 / 0.006 / 0.095 / 0.046
56%
0.015 / 0.004 / 0.088 / 0.040
65%
ID Similarity
Result
Our technology excels in make-up transfer compared to traditional models, with a low LPIPS (makeup distortion rate) and high ID similarity (facial resemblance).
This preserves both the desired make-up style and the facial appearance, enabling seamless and accurate transformations.
Related Papers
Our research team has written research papers documenting the results of our study and submitted them to various international computer vision conferences.
IntereStyle (ECCV 2022)
Encoding an Interest Region for Robust StyleGAN Inversion
HeadSwap (ICCV 2023)
Towards High-Fidelity Head Swapping with Chroma Keying
etc
1. Fix the Noise: Disentangling Source Feature for Controllable Domain Translation (CVPR 2023)
2. Fast and Realistic Face Swapping with Triple Adaptive Normalization (WACV 2022)
3. Preserving High-Frequency Details via Wavelet-based GAN Inversion and Editing (ICCV 2023)
4. LFS-GAN: Lifelong Few-shot Image Generation (ICCV 2023)
Want to learn more about our tech?
Discover our groundbreaking technologies and unlock a world of limitless possibilities.