Quick test with Topaz A.I. Gigapixel - Impressive, inconsistent results.

Started 1 week ago | Discussions thread
MauriceRR Regular Member • Posts: 410
Re: Excellent examples!

Erwann Loison wrote:

What I also meant is that most of the times I've been successful with topaz gigapixel was by using a much lower resolution version of my original files. Probably because the neural network has been trained most especially at that: inventing missing details (but not from excessive noise or moiré etc. Given the results I'm fairly certain they trained the network by producing themselves low rez photos from high rez ones.

Yes

So reduced photos tends to be pretty sharp. The key for me seems to be the sharpness. The algo works fine on sharp images, they have another algo for details enhancement which maybe would be more appropriate if the original photo was not enough sharp.

For exemple gigapixel works very fine on sigma foveon images 😉. That's logical the algo needs to make a difference beetween sharpness(for texture reconstruction) and just blurr (for not producing texture on original blur or not focused objects).

Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow