Majestic monuments to create fantasy cities and grandiose places.
Much easier to use with horizontal images (I made verticals for the Lora"s presentation)
There are more night images in the V2 training.
SHOW ME WHAT THIS LoRa IS CAPABLE OF!
This is the simpliest way to help me to improve the quality of a possible future model training and donate some BUZZ to train new ones on PonyXL or Flux with Civitai later.
I would be very grateful if you could donate some Tips to help me if you can. Officially, it would allow me to improve my GPU which is not powerful enough to training locally at the moment. Honestly I know I will never receive enough for this, but it will already be perceived as a very much appreciated comfort and encouragement from you.
Thanks for your interest, and have fun!
Description
Retrained with larger 1024px images and using some good results from V1.
Lower settings under 0.8 for smoother generations.
FAQ
Comments (2)
tbh it looks more gothic/reinassance france than fantasy :P fantasy i imagine more spires like numenor/elven realms
The word "fantasy" was used here in the "imaginary" sense of the term and that it is the result of a mixture of several styles; art deco, rococo, art nouveau, Victorian, renaissance, neoclassical...
Nothing prevents you from modeling the prompt until it looks like an elven city even if I grant you that the training was done a lot from straight lines and not much from nature, so it is quite difficult.
I admit that I published my V2 without having tested it too much on many different prompts (it was 4am when my training finished) and I posted some of my 10 first generations, but by default we have a mixture that can perfectly create "Fantasy" temples (I published examples).
I could do a specialized training for elvish architecture later, it's an idea that I can explore later.