While travelling through the endless spaces of videos generated by using artificial intelligence and using the newer "image to video"-tool I have learned, that the more interesting results can be recieved, if you don't use only the !image to video"-tool alone. Use it together with the "text to video"-tool at the same time and there will be surprising and satisfying photos / videos be generated, that could have been taken or videographed in the 1970s.
Video Generator: https://hailuoai.com/video (a pay website in the meantime)
Sometimes you will meet anatomical surprises too. People with more than two arms or legs, hands without bodies, people that can turn their bodies in impossible ways. The AI of hailuo is able to generate very realistic avatars of human bodies, but you still meet some bugs and mistakes time by time. Same with the denim vest. The AI is able to generate some accurate physical models of denim vests, but it isn't able to understand the button line architecture and the buttoning process in the right way. Maybe later in the development process.
Mama Marlich is a women from Venezuela, photographed in the 1970s together with her small son in a harbour. I used only the part of the picture with the women.
I used this black and white photo to create some interesting colorful 1970s stuff by cross-overing the photo with a text prompt. The AI generated videos are all unique. Every time you start a new generating process, you get a new video.
The following results are all generated with the same photographic picture and the same text prompt.
According to the video generation algorithm of hailuo sometimes the results stay in the picture frame of the used picture and sometimes they leave the frame. So the results appear more different.
