Talkin' bout my (image) generation - Experimental Medicine Edition (Patreon)

Greetings dear Patron!

How's the weekend? Looking forward to the week ahead? Here's hoping for the best in all directions.  I wanted to share a few pictures I have summoned forth from the ether of artificially intelligent beings. This is going to end up being a sort of rant about book covers, AI-generated images, and a chance to drool a bit at all the pictures of the doctor I recently made.

So, first off, the elephant in the room among creatives, is the use of AI to create art. It has a harsh, harsh dividing line and me even showing you all these pictures I made pretty much pegs me on one side of the camp. Although its not black and white, or I'd like people to understand that. I am an artist myself. Not as in my writing, but actual pictures I drew. If you want to see some of the cringey schlock I have uploaded in the past, the links are quite available to you by my own hand if you look at my website. The great deviant art purging/upheaval/revamp will happen one day since there are a lot of people on there who seem to dig my work and its another (seemingly growing rare) avenue that allows flagrant erotica and hardcore goodness.

I still haven't answer the question or even posed it. Is using AI to create art something we should do? Especially when there are actual living breathing artists who would readily fulfil a similar role if given enough money and time and input from a client?

1. I think in an ideal world people who need/want original customized artwork would be able to commission artists easily and get what they want/need.

2. In those situations as in #1, artists would be fairly compensated and clients would have the means to easily give artists what they fairly and justly deserve for their efforts, creativity, and ability to please a client.

3. Sadly this isn't the case more often than not for a variety of reasons (Economically, Availability,  Urgency, etc.)

4. In those times where it isn't an ideal world, you have to make do with what you have and what is within your means.

5.  <m(_ _)m>   <- (This is me performing the Japanese 'dogeza', bow of unforgivable shame.)

It isn't a black and white situation. We all know this. I am repentant and don't want to disrespect artists, but I am also far from rich and it would take months of earnings from this patreon alone to get the kind of professional work that I'd love to put on the cover of my books. It would take weeks and maybe even months of back and forth – time and working with people – that I could be spending writing, or with friends and family, or sleeping. What if the commission I ask for isn't exactly what I look for? What if I want tweaks and changes beyond the polite/determined amount I paid for? Pay more money and get more tweaks? Spend more money and realize too late maybe this wasn't the artist for me? Then I have to bite the bullet and suffer my own tendency to be a people pleaser and thank them profusely for their (no doubt extremely high quality and well-earned) work, while feeling like it isn't an exact fit for me. Again, this isn't a black and white situation and what I just described is kind of a nightmare scenario.

I am pretty commissions are far from that situation when communication is strong enough. At least that is how it goes with the ones I do. I think a human-written story and artificially-generated one have plenty of differences and the same goes with art as well. There are pros and cons to both. In some cases both are free and in some cases both can be very costly. Does it all justify my use of AI to use in my book covers? I think in some ways it does, but in other ways it doesn't. Ignoring AI and not using it on principle doesn't feel like a sound argument on its own. I think we should celebrate the merits and differences of both and use one or the other when it is most appropriate. I think as a society we are all still figuring out when either or is more or less appropriate.

I also, believe it or not, feel that it takes a bit of skill and knowledge to get any of the various AI companions we have available on the internet nowadays, to even cooperate.  Look at the ton of stuff I generated. I typically have to generate a ton of images and find one I can tweak in a photo manipulation program before I even start to assemble the book cover itself. I have spread out my feelers for new AI and found a couple and kind of want to talk about a few which have piqued my interest. And a couple that I am looking forward to coming up.

I'm not a scientist (in real life at least!), but I managed to get stable diffusion running on my PC and thought it was amazing for a long time. I thought a lot of the online offerings at the time were weak and watered down. Also, because I have certain needs and desires, everything I wanted to make made the AI blush, or at least the little sissies who control it blush. This is why open source everything is so crucial. I don't care for censorship in almost any of its forms so its annoying. It still is annoying, but I digress. The first cover for Experimental Medicine, with the close up face of the doctor, was thanks to Stable diffusion. I had some lucky run of generations or something because that was a miracle. I haven't been able to get the thing to play nice with me since.

That isn't true actually. I have a handful of covers generated for Lucy, the 'Boss' of experimental medicine. Some of these are pretty good. I am so lazy and averse to making covers that I was thinking of just using the one half decent image I generated and only changing the numbers on them for each part. I am glad I didn't. But I was going to use the Lucy images for Experimental Business. But I have a feeling there will be a few more contenders going forward.

I recently found a site called flowgpt. It is pretty impressive and is a collection of chats that people train or encourage to do this or that and some of them generate images. Some of them do it really well and others do it tremendously well. ChatGPT is famous for being heavily censored among AI lovers I think. And without much trouble I am able to get it to generate topless photos here and there. There are plenty of times where it just doesn't generate anything or says something like 'We here at blah blah blah and we want to foster a responsible blah blah blah'. Just have people click the 'I'm 18' check box before joining up and give everyone the goods.

So anyway. FlowGPT is pretty good and there are a handful of good image generators on there which I could use to get some quality work done. It is so reluctant to give me the sizes and shapes of girls I write about, but with enough encouragement and some brute force regenerating countless times, it gets there. Of the maybe couple hundred images I had it make, many of these are the cream of that crop.

If you look at the file names, I basically went through a few 'eras' of generation. Doctor and Sexy doctor were the early ones. Then busty doctor and finally ultra busty doctor when the AI started to understand that I wasn't trying to dip a toe in the waters and wanted to plunge to the depths of what it had to offer. Makes me wonder when the guys and gals who have a fully unbridled AI image generator can get up to. You see some of these tremendous images coming out and I can only wonder how they get them to do that at this stage in my career as a prompt engineer. You can tell a lot of the pics all have similar faces and I think that is because one model or another is based on that image and it is loathe to step away from it. I typed 'Asian' and 'Japanese' and all that in there, but they refused to give me any faces other than the one it started with for the most part. The one image it was based on actually wasan Asian model I think, so that was fortunate.

So that was Stable Diffusion, and flow GPT, and there is one more. There are 5 images I shared which have a more blueish tint overall and are all extremely similar. That was a new AI model I found after reading about it dominating others and the article I read was quite convincing. So I was looking at this article:

https://www.geeky-gadgets.com/stable-diffusion-3-vs-midjourney-6-vs-dalle-3-ai-art-generator-prompts-compared/

And I realized my stable diffusion setup was jenk at this point and resolve to uninstall/delete it shortly. I had so much luck with flowgpt and the newest AI I stumbled across. Ideogram:

https://decrypt.co/219776/ideogram-is-a-new-ai-image-generator-that-obliterates-the-competition-outperforming-midjourney-and-dall-e-3

You know how our google masters do it. You visit an AI website and soon your news feed is filled with AI news and all that. I was thankful for this one though because ideogram is out of this world. You have to wrestle with and reason with and beg and plead other AI to get them out of their comfort zone and keep them on task and prevent them from producing some silliness that isn't getting anybody anywhere. Ideogram just worked and the images I posted here are from three generations of 4 images a pop. THREE times I pressed the enter button and got 12 images total. These 5 are the best ones and honestly feel miles beyond all the other stuff mixed in there. What do you think?

Either way, having a full body and a face that I can put on a cover of a book is so huge. Every book cover is the 'from under the nose down to just below the knees' and you hope to god the arms are doing something you like. But some of these more anime-ish ones that are still kind of realistic paintings could work. They aren't perfect by any means, but you blur out a funky semi-finger and make sure the title goes over that strange overlap of this or that and you are good to go. I am not a pro at making book covers, and have tried a variety of more traditional means and gotten some results as well. You have seen my efforts on that front no doubt. Some of it isn't that good and most of you said in a poll you want me to spend some money on fiverr and get a cover made for me. Very fair points.

Here is another tricky thing. Of all the AI I play with, almost NONE of them want to give me dudes. I am trying to get dudes with huge bulges going on and AI isn't playing nice at all. I got somebulge action going on, but we are talking smaller than I imagine most of the characters start with. When I am looking for like a big fat squishy softball-sized package like some European biker who pumps daily AND injects silicone. That is the kind of crazy thick sack I want to see. And I KNOW that the AI coulddo it, but there are too many limiters on. I tried to make some futanari as well, but didn't have much luck there. Even when I googled for what other people made it felt like it was less prominent.

I am going to hold out for stable diffusion 3 before I start to knock on those doors again. Or mix AI generations with some traditional media and good old fashioned photo shopping and air brushing.

So that was a super mega long rant, but I wanted to share some of the stuff I have been playing around with and had this plethora of sexy images and thought where better to share the goodness than with my beloved Patrons.  I will put out a poll later too.

Hopefully this windfall of images of Dr. Cuunis, according to a variety of AI image generators, titilates your early week self. If you liked them, maybe I can share the images I made of Little Lin? I had the AI getting all upset with me when I talk about a girl in a chinadoll dress giving a massage to another woman. Wasted more than a few of my 'daily prompt allowance' just to get told I made the robots' handler's blush. Some of those would have been choice af though. Alas.

Ta ta~

~Michelle