I discovered that most icons and logos are already well understood by the flux model thanks to the highly skilled people who trained Flux. I only needed to train the LoRA on the dystopian concept which in my testing it was already understood but not in the way I wanted, It would generate a regular cookie monster walking down a dilapidated street. I wish I could say it's an original idea many great artists have used it for social commentary for a while and their work inspired the images I created to use as training data.

There is something bleak about the dystopian ruins of earth when there is a ruined structure or monument to cultural icons we know so well, but it is also exactly what I would expect to find. It offers answers to where we will go wrong and I like to think about the future of people without access to information that have no idea what these icons or logos were and the stories that they make up about them. Religions would probably sprout up around Superman, Batman, ect as if they were the Greek/Roman gods future humans may question where Superman/Batman went to, why they left us, and if they were coming back.

apologies for such a long answer and I'll try to restate the answer now as short as possible: The model is already great at logos anything else that it is already great at will also work good with this LoRA. All of my test images came from my propensity / expectation for that to appear in dystopia. I do hope that whatever you do with the LoRA may open a new door in my brain to concepts I've never imagined.