This should not be possible 🤯 I've enhanced this 8x from 1.2MP to 77MP (10.752x7168px) using Magnific ai. 64x more pixels than the OG image. You can do the same. Each super enhance takes 10-15 minutes and can run in the background. A thread

Here is the OG image from Midjourney.

Here is the upscaled version from Magnific (x will make it downres) So here is an URL to the image.

It's absolutely unbelievable that we've gone from like 512x512 top resolution to being able to output 10.000 x 10.000px images (100MP). The equivalent of the Hasselblad X2D 100C in terms of resolution.

It's all possible due to this. Javi and Emilio have been working hard to come up with this magic. Because it truly is magic.

Being able to process and enhance images from 1.2MP to 77MP is just opening up for so many use-cases. Now I can work on an image, enhance it, and have it be ready to use on large scale formats. I can even crop sections that I want from these large outputs. 🤯

It's still early days, and I can see that this will go far beyond where we are today really soon. Yes it takes time, but that will also be coming down rapidly. There is a really strong set of parameters here for making this real necessity for every creative AI workflow.

Hats off to @javilopen and @emailnicolas for making this all possible. Go check out

It's not upscaling, it's rebuilding the image.

I never wrote the word upscaling... I wrote enhancing. >Very different. So yes you are accurate.

Yeah but it’s totally hallucinating shit. It’s not very accurate with like complicated textures. It’s fairly obvious.

There are slider controls in Magnific, I had creativity set higher, so I want it to hallucinate