Pros and Cons: The New Features of the aiTransformer Cartoonizer and Video Cartoonizer: Celeba Distill, Face Paint 1, Face Paint 2

We added three more new cartoon styles: Celeba Distill, Face Paint 1, Face paint 2. 

These styles work very well on "big head" photos with only one face.

The new styles also work fine for big head photos with more than one face.

But, here comes the BUT part, if the faces are not big enough, the results can be distorted.

What will happen to pictures without any faces? It's fun to try.

                                       

      The same rules apply to the Video Cartoonizer. They work well with big head videos. Beware the new styles take much longer to process videos. 

                                           Celeba Distill

                                           Face Paint 1

                                            Face Paint 2

If the face is not big enough, transformed videos can be distorted.    

                    

                               

       Face paint 1 and pace paint 2 have similar issues, probably because the AI those new styles are based on was trained with big head images.                               

       Besides the new styles, we also updated the Paprika style, which tended to change the color tone of the picture (making it reddish), we found a better model and updated it together with the 3 new styles. Compare the results:

       Lastly we'd like to remind users that the 3 new styles and the updated Paprika style may look fancy for "big head" photos, the execution speed is slower than previous styles, and the difference is huge on videos. So if you want to transform videos using the 4 styles, please try to keep the video short, say, within 30 seconds, if you don't want to wait for too long.