With the increasing prevalence of the image generating AI programs, there is now a whole host of AI generated content. Is it self-aware? Of course not, behind the scenes of all these different projects is a lot of human cultivation and creativity, the AI is merely a tool in the artist’s hand. A very impressive tool nonetheless.
Here are some songs where there is an AI generated image for each line of lyrics:
Who here remembers this song and it’s original flash video?
More music videos can be found on SolarProphet’s YouTube channel.
I posted this story earlier but will link it here again since it’s relevant to this thread:
And i got an idea, so we created a metahuman version of her, thats more in her prime and then took real photos of her for reference but swapped out the Environment and her, with the de aged prime version of her for what im calling the world tour,
It’s really impressive, and creative. And I’m intrigued by the way you’ve done it. Step us through the process please. What’s the workflow, what tools did you use? I can’t even figure out how to get reference images into Metahuman. Maybe that’s because I’ve only ever messed around with the online tool. Do I have to go install Unreal 5 and do it in the latest offline version?
From what I’m understanding, you took pictures of the Mrs. Uploaded them as reference images into Metahuman Creator using it to create a digital version of her, and de-aged her a bit, and then put her into poses in different locations?
Almost everything is done primarily in googles ai studio.
So i have been using googleai for a while to display the tshirts i design, on a virtual model that google created according to the requirements i gave it
And all i have to do to is upload that model everytime i have a a new tshirt design made along with the new tshirt to do a photographic pose because google stores that model on its server, propably anonymously but every time u upload it to use it, it checks facial data and pulls it from the server and i prompted the standard language model for that cause i said i wanted consistency when i do listings
Thats why i have the same virtual model for all my tshirts i advertise online
Google doesnt allow you to create virtual models of real people , cause for ethical reasons they say they cant do manipulation with real humans but they have an app called fit check in ai studio that literally creates a meta human when you upload a photo of someone, so we uploaded a younger photo of her to fit check, and it created a virtual version of her that u can use to try on different clothes, but the detail isnt high quality when u save the final render so we imported that into nano banana and asked it to increase the details to 8k resolution and it will be used to stage tshirt prints and we gave it a reference photo and it got the details close but not accurate, so we trained it for about 2 months as a side project in the evenings with a few photos of her in the same ai studio acccount, just prompting the regular gemini model to rate the photos and it would give critique butthen we got to a point and asked it how it can improve the photos and it suggested improvements and we said then do it and it rendered us a virtual model which i could clearly see was a 3d model of the initial photo that we asked it to improve, and i just took it from there iterating on that and we created various models first until we got to the current one
So to use the model accurately you need a reference photo for the model to coly so we take reference photos, asked gemini 3 pro to create an detailed prompt based on that reference pic, then we tweak it slightly once we have the pose we use that new posed model with a new reference to Stage a scene
So the goal is to create the most realistic wallpaper worthy pictures, that are indistinguishable from her real ones, also i think i can just add this to my resume also, an ai digital artist
Mirrors edge cosplay, Since i have access to a detailed meta human, why not, the reference data i used was a real hiking photo thats why she is on the mountain and not on top of a building which would have been preferable.
