This one was really damaged. I couldn’t do any better. There’s still a slight AI-like quality when you zoom in. But when not zoomed in, I think it’s fine.
Flux2 Klein 9B + Wonder 3.5 and Subtle Redefine with Prompt.
Here, I also used Flux2 Klein 9B + Wonder 3.5 and subtle Redefine.
That sea lion #2 is a beautiful display of AI capabilities. Incredible.
By the way, thanks a lot for the long walk in Pairi Daiza. A unique zoo. It must have taken you hours and hours of work. 
That can happen in so many different kinds of things! 
Speaking of challenges, I think we need to come up with a good name for this thread to reflect parts of the names of the main participants. But I’m having a tough time combining Harald and Unrelent… 
Along the lines of “Bennifer” (for Ben Afleck and Jennifer Lopez - well, once upon a time).
Thanks 
.
You’re welcome . Yeah, I love this park. I still have some photos from the 2019 album to show you, hahaha. And I might even go there again today.
At least two full weeks for the complete 2018 album, which includes 275 photos. And approximately four days for the 2019 album, as it only requires a few corrections and contains 68 photos.
Hahaha, yes I think it’s going to be hard
The sequel to the 2019 album
Female European fallow deer picture 1
Well, it’s not the best, but compared to the source, it’s still an improvement.
This one is much improved and I like the framing I had done when I took the picture. I also added a slight depth of field to accentuate the gaze on the main subject.
And I close the 2019 album with the last photos with snow leopard.
Snow Leopard Picture 1
Snow Leopard Picture 2
Okay, the head isn’t positioned exactly like in the original, but it’s not always easy to get exactly what you want. Maybe I should have made a variation with a closer shot of the animal so the AI could do it better 
. Oh well, that’ll be for the next photo.
Snow Leopard Picture 3
I like this one, but again, for the next ones, I’ll have to do a cropped version centered on the main subject so the AI hallucinates as little as possible. As you can see, it added a few extra black spots, and the head position is slightly different.
Snow Leopard Picture 4
As for the photo of the Male European fallow deer, the background was nicely blurred, so the antlers stood out beautifully.
Thanks . Yeah, that was the goal of adding a little depth-of-field blur to make the animal stand out better.
Another photo that I’ve slightly improved to correct the overall motion blur.
I took this photo in 2024. I then edited it using Lightroom Classic and Gigapixel AI at the time. But I couldn’t completely remove the motion blur, especially on the ground. Now, thanks to Flux2 Klein 9B, it’s possible.
A wonderful parrot. Its rainbow feathers are really nice. And the truth is that AI-editors have come a long way in the last two years alone (including Topaz). So sometimes it pays to improve the originals again. Or not to improve and just wait and wait for ever more perfect versions 
?
Thanks . But I think that this time, there won’t be any further improvement to the photo. I’m quite satisfied with the result . But it’s true that as the years go by, the models will get better. And they’ll do even better restorations on a whole other level.
Just an improved seagull with Flux2 Klein 9B. Then enlarged with Wonder 3.5 and subtle feather improvement with Subtle Redefine.
Photo from 05/03/2025.
A red dragonfly visited me again. It is not as shy as the blue one, so I easily took a picture of it. Unfortunately, it was sitting on a nasty piece of concrete, which is not exactly a picturesque background for the photo, and most importantly: The concrete is clearly visible through the translucent wings, so it disrupts the veining on the wings. My goal was not to perfectly capture the structure of the concrete , but to weaken the disruptive effect of the concrete through the wings and strengthen their veining. For the third time, with the support of the prompt in Nano Banana 2 (and then Gigapixel Wonder 3.5 4x), it was somewhat successful, although not perfect (the concrete under the wings still remained strong, but I think the wings and their veining are better now). Overall, the dragonfly looks nice, I dare say.
What’s good here is the removal of the blur on the wings and the addition of depth of field to focus the eye more on the main subject 
.
A few hours ago, I fixed the offset issue. It now works equally well for 16:9 (left image) and 4:3 (right image) aspect ratios.
An attempt to create a (pseudo-)macro shot of a red dragonfly’s head from a photo that had already been cropped. My main goal was to get better detail on the eyes—since dragonflies have compound eyes made up of many facets, and the original image (posted here somewhere previously, also as a crop) hadn’t captured that well. Nano Banana 2 (using a prompt) helped with the initial eye enhancement, while the Wonder and Standard Max models in Gigapixel were used to fine-tune the result. (I had to downscale the final image for publication.) I’ve read that dragonflies have a nearly 360-degree field of vision, allowing them to see in any direction—even backwards. That explains why my attempts to sneak up on them for a photo are usually in vain .
