I have not posted for a while, so here is something I played with again on Nightcafe AI. The prompt I used dates back quite some time, I resurrected it again for a new interpretion with different models.
These images are currently untitled and unpublished, all based on the same prompt.
This was the first use of the 'resurrected' prompt. It is based on a actual photo of friends around a table at my favorite hang-out. The model I used was Flux Kontext Dev which basically is a editing model. Running the prompt as the editing input together with the photo gave me a interesting result. Only one person in front is actually recognisably the same as in the photo. From there, the prompt took over.
This is the prompt I used for this, as well as all the other images that follow:
The Hero's Surreal Journey; Digital Painting by Paulo Cunha in a post apocalyptic wonderland, in heaven everything is fine by lady in the radiator and brothers quay, inspired by Otto Rapp, H.R. Giger, Hieronymus Bosch, Art Nouveau, intricately detailed, hyperrealism, fantasy, masterpiece oil painting, Bogomils Universe, imperial colors
The 'Paulo Cunha' in the prompt is one of my artist friends, who is also my 'prompt-pong buddy' on Nightcafe, that is, we share and improve on each others prompts.
Now for results with this prompt with the 'regular' models (I give the name of the model in the description):
The images are created with either Flux or Flux Schnell, but using different formats and prompt weights, the results are also different. This could be repeated endlessly.
Now getting back to what I started with, the photo of my friends at my favorite tavern. Once I 'clued in' on how the editing version of Flux Kontext Dev worked, I took a couple of crops from the initial photo and used different prompts to put them into different situations.
The prompt for this was: Enhance the images portraits of people and change the background to a sunny mountain outdoor scene
The prompt for this was: Enhance the images portraits of people and change the background to a sunny scene at the beach
I guess I was on a roll after this - seems when you are having fun with it, it is hard to stop. A while ago I had trained a Lora for Selfies. So I thought I use the 'Hero' prompt with my Selfie Lora. This are the results:
Because the way the Lora functions, most times the images with multiple persons in the seed image turn out like clones of myself. What follows next was still with my Selfie Lora, but without a seed image.
This was with the selfie Lora, but at a higher prompt weight - on the next one, I reduced the prompt weight, and got a 'selfie' again:
And one more time, Selfie Lora with the input image, but at a high prompt weight:
Looks a bit like me. Since all these were Flux based (including my selfie lora), I create one with a different model, namely HiDream I1 Dev:
To put the basic idea of the prompt into context, check the wikipedia article about
Quote: In narratology and comparative mythology, the hero's quest or hero's journey, also known as the monomyth, is the common template of stories that involve a hero who goes on an adventure, is victorious in a decisive crisis, and comes home changed or transformed.
I know I went all over the place with this, but hope you enjoyed this, one of my infrequent posts.
