I admit, I was i little bit distracted in the last two weeks.I decided to have a look at that part of the virtual world, that is named Artificial Intelligence (AI). Before last saturday I never realised, that artificial intelligence maybe could add an interesting new world to the blog. Only because this blog is retrospective, it isn't neccessary only to look back, sometimes it make sense to look forward instead. And after having the first serious contact followed by playing with my first AI tools on last saturday (an interactive chat with a machine), I couldn't stop, because there were hidden some interesting places in that brave new world.
I tried out a couple of AI picture generators and learned that it isn't easy to tell a machine with only words, how to create pictures in the right way. I've learned within a week, how to manipulate the machine to get the ladies on the pictures chlothed with nearly real looking 70s denim vests. This isn't easy, a 70s denim vest or 70s denim waistcoat isn't a concept the machines understand. And if you look at the button lines of the denim waistcoats of the ladies, you see, that there are still many things to be fixed in the future.
When the weekend started on Friday, I thought, the AI post would show some of the pictures I had generated while the last week. But yesterday evening I read a short report in the WWW about a new technology on the website: https://hailuoai.com/video
After I opened that door, everything changed. The three pictures before are not made with an AI Image generator. These photos are stills from videos, I made on the Hailuo AI webpage. At the actual moment (September 2024) it is possible to create silent videos with a length of 5-6 seconds on the page, only by telling the machine some words. I hope, it is possible to show you here, what I'm talking about. And I'm sorry about all the blonde women with braided pigtailes. I had some kind of a Scandinavia Flash and animated many many young women from Sweden or Denmark today. Here is the video from the basement party, where I took the still used for introduction at the beginning of this post.
Faszinating and disturbing at the same time or not? It took only three or four handfull of words to create that silent 5 seconds long party video. I saved some of the phrases I used, but not all. The exact phrase from this video doesn't exist anymore, because I changed it while the creative process a couple of times. But to create the lovely party people, the input for the Hailuo engine must have looked like this:
a color movie filmed 1976, in a party room in the basement of a house, 7 cute young blonde Swedish women with braided pigtails are dancing and smiling. They all keep eye contact to the camera man the whole time. There are also some boys in the background.
All girls wear blue jeans and formal denim waistcoats.
I think I remember, it took about three times to get the result postet here. The first two videos generated with that words looked similar, but the denim waistcoats were not the same. It took three times to get denim waistcoats looking close to my idea of 1970s denim waistcoats. The machine has a tendence to the kind of denim vests looking like denim jackets with cut off sleeves. My ideal denim vest looks like close to a menswear suit vest cut. To show you the last steps of the process to create that video, here are three videos, that came short before the final dancing video, it were also the last videos from the kissing circle. The bad quality of the basement filming was set by me in that way on purpose to create realistic looking videos from the past, not hD stuff.
A problem, I'm always fighting with is too keep the quality of the denim vest choice of the engine, because the engine never gives you two similar denim waistcoats again and again. There are always changes and if you are not lucky, it goes into a total wrong direction. This is the video I recieved from the engine two steps after the final basement video:
After that result I left the basement party and started a new party outside in the garden in the dawn to get a kind of reset:
This result remembers me to an American movie, Dazed and Confused from the early Ninetees, that also got a post here.
I think that was enough excurs to show you the little problems you will meet when you try to create videos by yourself. Always keep in mind: one good video and five - ten average / bad videos.
Before someone comes to the false conclusion after seeing all the blue clothes, the blue eyed woman with the blond hair and the braids. I also tried other directions, but not as long as the Scandinavian direction. This is a still from another video:
Now back to the Video stills: here is the video from the women in the Tram. To get one really interesting video, you have to start normally 5 to 10 videos, that are average. At the moment you can influence the machine only a little bit in outfit questions, there is still some luck necessary. Maybe it will be easier in forthcoming versions of the webpage:
I brouth the ladies to an outdoor lake somewhere in Sweden at sunrise:
After that I thought, it would be a good idea to let them ride on the rooftop of a train:
After the train we made a stop at another place for travelling near the capital city of Sweden:
Last step was on the rooftop of the largest church in Stockholm. A good place for a goodbye. I think, you will meet more AI influenced postings here too in the future. But the retrospective part with the real past will go on too. I have to find out a way to give both directions a fair share in the blog now.
I hope, you find the journey to new horizons as interesting as I found it and it was ok for you to get a fake reality and past this time. At the end one of the pictures I made with an image generator while the week. And - no, the pictures I generate are not always as sexy as here. And I think, it was a fictive Germany, where these Ladies are at home.
