Many bugs in 2.4.1

Can’t detect horizontal faces :
Topaz photo AI does not detect faces in a horizontal foreground position. I can send the corresponding photo to improve the software, but it cannot be reused without my explicit agreement.
It’s a dance sport photo.

Saving enhanced Image, Photo AI change its orientation:
In a specific case : just after the 2.4.1 update, cropping image in the same orientation, this one is saved In landscape and display only half the vertical image.
After closing and reloading Photo AI the problem is gone.

I was also very happy with the 2.4.0 interface, this new one is less well.
Thanks for your work.

Would you mind sending in some more information to get a better understanding of the problem you are having with this particular type of image? Please send the following information:

  1. The original image file.
  2. The processed image.
  3. A screenshot or description of the settings panel so I can see what adjustments were turned on.

You can securely submit your image(s) to my Dropbox using the link below.

Dropbox File Request

Please be sure to send me a note to let me know you sent something.

Hi John.
Thanks for your reply , three files have been sent :
1- Image saved in landscape while the cropping was in portrait orientation.
2 and 3 : 2 original photo with dancer who have the head in horizontal position.

I managed to partially bypass the lack of facial recognition using denoise enhancement .

Thanks for sending in those files, I was able to replicate this issue with Face Recovery and detecting this subject’s face. Our developers are aware of this and are working on options for that feature to help with this case.

For the cropping issue, I was not able to replicate the output image you sent in. Would you mind sending in a screenshot of the crop you were using so I can see what type of crop caused this problem for you?

The cropping problem has disappeared for generated files, but the preview thumbnail, before saving the enhanced image, is still in landscape orientation instead of portrait. This problem only appears when the image is cropped in portrait orientation.