Did you know that in a way the Cleric is the first hero ever to appear in the original Pixel Dungeon?
Pixel Dungeon had a different default hero sprite for about 6 days after its first release in 2012. This was back before the game even had hero classes, and this sprite was quickly scrapped and replaced with the Rogue's sprite. I saw a screenshot of this design back in 2015, and it was my initial inspiration for the Cleric in Shattered Pixel Dungeon! The design has changed a lot since, but it's neat to think that in a way this hero is about 12 years in the making!
Also, I now have an account on Bluesky! While it isn't my first choice, I have been happy to see that Bluesky has been replacing Twitter for many people in the last little while. If you'd like, you can now find me there as well: https://bsky.app/profile/shatteredpixel.com
(Image Description: A screenshot from the original Pixel Dungeon circa 2012. Several graphical elements are outdated compared to the current version of Pixel Dungeon, and compared to Shattered Pixel Dungeon. Most notably, the hero sprite is completely different in this screenshot, using a design that was scrapped shortly after.)
What software is that, out of curiosity? Either you're not referring to the same type of AI as the type of generative machine learning algorithms that are commonly used today (which is a fair misunderstanding, given all the naming is marketing BS and not at all descriptive of the technology) or you're referring to a locally run generative model, which still needs a huge amount of training that was presumably done by it's developer before distributing the model to run locally.
It is not internet based. It is my own scripted code that calls on Lanczos, B-Spline, and Bicubic mathematical formulas that use white light frequency enlargement and then filtered through the tgp upscale to remove pixelation and interlacing. Most of these are available from IrfanView and Adobe.
This use of formulas to enlarge is considered A.I. as a complex system of processes is used to examine, evaluate, and "decide" the best result without human interaction.
While it is true that I did not create the mathematical formulas, to ban derived images is like saying we should ban JPEG's if one doesn't acknowledge every member of the Joint Photographic Experts Group, or JPEG whose work we've used.
As a programmer of more than 4 decades, believe it or not, I am able to enlarge a picture without relying upon a online server farm.
In that case you might want to reconsider the use of the term 'AI' in your description of such images, which unfortunately means something very specific when it relates to images on the internet right now. Is there any reason why you wanted to use this apparently very sophisticated custom approach when something extremely simple like a nearest-neighbor upscale already works very well for pixel art?
I prefer to think for myself and figure it out on my own. Too, often I am without internet. Having my own tools I don't have to depend on online.
I began working on them for some Atari 2600 work I am trying to expand upon. I worked with pixel art long before the phrase was coined. I created the Image and Greeting carts that display digital photos. The tools are used mainly for that project.
Don't believe it. Prove it.