More AI – it’s surreal!

[sberry]

For a number of reasons, I have started recently to explore the uses of AI for creating or converting pictures. One inspiration was the so aptly named thread AI in all of its wonder by steve_pickstock – just look at those floating islands!

For some experiments I used a photo from my recent Tibur project:



and fed it to Stable Diffusion. Setting the output style to “Cinematic”, I got this result:



This is quite fancy, because not only the mood of the scene has changed – I also got some additional people there, who even look a bit Roman (but have some anatomy problems…).

So, the next round I used the style “Surrealism”, which yielded this gem:



Well, we now have something like a “Salvador Dali clock” here, which is not that surprising given the requested style. But one thing really is really weird – the AI has turned the rectangular temple by 90 degrees and created a whole new front view with a door. Wow!

[MABO]

This is another very interesting discussion about AI in our hobby. I look at your results and Steve's results and can only say that I have no idea how to deal with this, as I haven't looked into it yet. But I'm sure the time will come when I'll get involved in a project like this.

[sberry]

This is another very interesting discussion about AI in our hobby. I look at your results and Steve's results and can only say that I have no idea how to deal with this, as I haven't looked into it yet. But I'm sure the time will come when I'll get involved in a project like this.

The good thing is that one can experiment with these picture generators for free. One needs an account, but usually you can log on with an existing account at Google, Apple etc.
So it is really simple to try a few things, but of course the free versions have certain limits. For me it is of interest to compare a number of different platforms, because I am considering getting a paid subscription in the future, which would enable professional picture generation (rates are about 10 ... 40 € per month).

[sberry]

This stuff can become quite addictive, and thus I decided to try just one additional pic, from a different project this time. It is already nine years old (oh my, time flies), showing the adventures of my Roman Space Marines:



And this is what I got in return from Stable Diffusion, setting the style to “Horror”:



I have no idea what RCANND WIREES means (I hope it’s not something inappropriate), but I like the atmosphere of the transformed photo.

Next, I tried the effect of setting the style to “Boudoir Photography” (whatever that is), and the scene changed quite a bit:



It seems the AI has spent a lot of time with Warhammer 40K, I’d say. But if you consider now enlisting with the Roman Space Marines, please note that this is still an AI-generated fake photo!

A style setting that resulted in almost no changes of my pic was “Alien Sci Fi Photography”:



The colors have changed, and two additional marines appear (who seem to be strangely ignorant of the alien next to them). But otherwise, it seems that my original template was already quite Alien Sci Fi Photography-like, which is kind of reassuring.

And here we have the result when the style is set to “Chinese Art”:



That looks pretty decent, I would say. I like the idea of that alien coming out of a bamboo forest, and the picture got even those typical red seals marking it as an important piece of art! But what do these Chinese signs mean – perhaps the translation of RCANND WIREES?

[Bessiere]

At least for boys/young men I can them experimenting with this "Boudoir" setting endlessly instead of staring at girlie magazines. Not even real women. God knows what kind of mental issues this is going to cause. To be clear I'm not against AI but I think it's creating a financial bubble with too many putting too much faith in it. The resulting mass unemployment and competition for electricity and water will create problems of their own. Despite saying that I have 2 sons very well employed in the AI field.
Cheers,
Bessiere

[steve_pickstock]

Love the images - to be honest I hadn't thought of this. I've done this for friends, but not models.

The way I am reading the news at the moment - it appears that the rapid expansion of companies offering AI applications has got a lot of people worried that it is a rapidly expanding bubble that is about to burst.
The biggest effect will be that people that have invested in it will lose money. Some services will disappear and others will remain and pick themselves out of the rubble.
I have always expected this.
When AI first appeared as simply an image generator people went mad producing lots of the 'boudoire' images (among others) flooding the internet with masses of pictures, some were bad and some were atrocious. My feelings were that this phase would pass. It is a tool, a new tool and people were using it willy-nilly.
My reasoning was that eventually the new adopters would get tired of it and stop using it, the novelty would wear off. This was based on my experiences in the mobile telecoms industry - Picture messaging on mobile phones being a good example - a new product comes along, it gets used wildly at first then settles down. At that point people find uses for the application (that may or may not still incorporate boudoire) and 'serious' users would be the main users and it would become just another tool - like cameras on mobiles.

Anyway here - in the spirit of the thread a picture of some Linear A Amazons and their war-mammoth as they storm a ruined gate in a war game, and the same after Chatgpt has done with it.

[Santi Pérez]

There really are vast fields for AI applied to our hobby. I love the versions it makes of the pictures of your dioramas and figures, Stephan and Steve.

For better or worse, I consider myself too traditional and inept to embark on these technological innovations.

Santi.

[sberry]

I understand why some people are skeptical of AI, and some problems have already been mentioned: the enormous resource requirements for electricity and microchips, the associated costs, the fact that no one can predict the long-term impact of these systems on our societies, and so on.

I think it's especially important to remain skeptical of all the exaggerated promises about how AI will solve all of humanity's problems within the next few years, and so forth. Most people here on the forum will be old enough to remember the introduction of the internet more than two decades ago, which came with the very same false promises. Before that, it was the same with the computer, and even the introduction of television, and before that, radio, was celebrated as a breakthrough for the dissemination of universal knowledge and wisdom to all people worldwide…

I therefore think it's best to find a middle ground between hype and condemnation, using these AI tools appropriately while always keeping the drawbacks in mind. Personally, for example, I would never create fake photos with real people, and I find the creation of "boudoir photos" or worse, featuring actresses and other women without their knowledge, absolutely disgusting.

The most important application I envision for our hobby is using AI-generated backgrounds for photographing my dioramas. I'm always looking for a landscape photo that fits my scene to create a cohesive overall look. I often find wonderful images on platforms like Creative Commons, but sometimes there's simply nothing suitable.

And I must admit that I really like the AI version of these Amazons on a mammoth; it looks so dramatic!

[Iceman1964]

I agree with you Sberry, the AI revolution will be like the previous Computer, cell phones and Internet revolutions I experienced in my day-by-day life (radio and TV revolutions were before me ) .
Many things will change, but not as they are stating now, no one knows how will be in few years, not even AI itself

[MABO]

This stuff can become quite addictive, and thus I decided to try just one additional pic, from a different project this time. It is already nine years old (oh my, time flies), showing the adventures of my Roman Space Marines:



And this is what I got in return from Stable Diffusion, setting the style to “Horror”:



I have no idea what RCANND WIREES means (I hope it’s not something inappropriate), but I like the atmosphere of the transformed photo.

Next, I tried the effect of setting the style to “Boudoir Photography” (whatever that is), and the scene changed quite a bit:



It seems the AI has spent a lot of time with Warhammer 40K, I’d say. But if you consider now enlisting with the Roman Space Marines, please note that this is still an AI-generated fake photo!

A style setting that resulted in almost no changes of my pic was “Alien Sci Fi Photography”:



The colors have changed, and two additional marines appear (who seem to be strangely ignorant of the alien next to them). But otherwise, it seems that my original template was already quite Alien Sci Fi Photography-like, which is kind of reassuring.

And here we have the result when the style is set to “Chinese Art”:



That looks pretty decent, I would say. I like the idea of that alien coming out of a bamboo forest, and the picture got even those typical red seals marking it as an important piece of art! But what do these Chinese signs mean – perhaps the translation of RCANND WIREES?

[MABO]

Amazing Picture!