Exploring the Limitless Possibilities of Generative Fill: Unveiling AI’s Photographic Wizardry

Don’t get bored yet! With Photoshop Beta — Generative Fill, you can now expand the original image. I tried enlarging a group of non-photoshopped photos by 300% using Photoshop, just to see the results.

Honestly, the original photo is already perfect because each photographer captured the angles perfectly.

I used these photos to test if AI can recognize that they are non-edited photos taken from various angles.

These photos might have unique angles, proportions, and special moments that I don’t think there are many similar images in the database to simulate.

But from my testing, AI can indeed recognize them. It knows which parts come from which areas and can simulate the overall atmosphere quite accurately. For example, the first photo clearly shows that it comes from a single shot, even though the original image (in the white frame area) looks like it’s made up of a grid of four images.

In summary, it exceeded expectations. It captures the overall atmosphere, lighting conditions, textures, and various elements remarkably well.

I’m impressed once again.