Using Gigapixel AI Experiment

[Lowjiber]

A couple of days ago, I posted an image that covered almost the entire NE quadrant of the sun.  The image contained a couple of decent proms close to each other and another little "bump" that I decided to leave without cropping it out.  I've been experimenting with a fairly new application recently called Gigapixel AI by Topaz Labs, and decided to see what it could do.

I took my original shot (TIFF) that was 800x643 pixels and doubled its size to 1600x1286 pixels.  I then cropped out what I wanted to keep to a "new" image that was 673x859.  Gigapixel did a good job at "stretching" everything.  However, there were a few "speckles" that sort of resembled noise, so I stuck it back in Photoshop for a very little fine tuning...

Thanks for the peek.

Clear Skies

John

 

[Altocumulus]

John, could you put the original in this thread as well, so we can compare.

[Kitsunegari]

i am convinced there is a secret technology out there for image processing that we do not get to play with as civilians.

about 13 years ago , i distinctly remember some random someftware prototype "thing" i found on the net called dragon.  It was developed by microsoft or assimilated by them.   

 

The gist of this was also using a sort of "AI" funtion to rebuild an image.     First you input something, then it automatically sourced ;literally; every image on the internet accessed by a neural network system with 1000's of cores.

 

So you took a picture of the eyeball of a statue in a random museum, and the AI would find the eyeball posted on the internet by millions of  humans that actually visited the statue and posted a photo of it on the internet.

 

The AI and Super core algorithm takes crops and pixels from literally every photo on the internet, and rebuilds a mosaic.    The mosaic turns out so well, and so many people have photographed stuff in the museum the mosic becomes startling real and three dimensional.  

 

Every angle of that statue ever photographed and every picture ever uploaded to the internet of that statue, is sourced into one new image.   After a few hours you have a complete scene and a puddle of drool on your desk from the awe-struct state you are left in upon gazing at the first public demonstration of   "dragon"

 

anyways, it was the most amazing thing ive ever seen that escaped the public eye   ( i am assuming because of the inteligence agencies and satellite technology/ patents etc)

There certainly are super secret imaging techniques and tools waiting to be harnessed by average Joe.

https://futurism.com/microsofts-ai-image-generator-construct-life-like-pictures-pixel-by-pixel

https://www.customvision.ai/

https://en.wikipedia.org/wiki/Super-resolution_imaging

https://en.wikipedia.org/wiki/Deep_Zoom

https://en.wikipedia.org/wiki/Photosynth

https://en.wikipedia.org/wiki/Image_Composite_Editor

https://www.microsoft.com/en-us/research/product/computational-photography-applications/image-composite-editor/

 

Edited September 27, 2020 by Kitsunegari

[Lowjiber]

19 hours ago, Altocumulus said:

John, could you put the original in this thread as well, so we can compare.

Sure!  Here it is... I also sent you a PM with both attached.

Edited September 27, 2020 by Lowjiber