The purpose of this model is to recreate the style of the game/experience "LSD Dream Emulator".
The models was trained on 40 screenshots of the game.
If you don't know the game or haven't played it, I highly recommend to check the english translation out. Its fairly weird, but an exciting experience imo.
Description
Works best on Pony base model to also get that PS1 look, but can be used on other models too. From my experience it would only take over the colors and objects, while leaving as an example for cyberrealist v6 the photorealistic style.
FAQ
Comments (5)
How's the English translation? I recall it had minimal UI interaction, so I small guide would probably be enough; I know most of the gameplay is just walking and touching surfaces to change dream environments. Oh also, The Gray Man.
it doesnt change too much. its mainly the texts after the dreams, which are translated as also some other texts. still for none native speakers, its the best way to play apart from the port imo :)
Holy shit do not sleep on this lora. You can get some amazingly fucked up results
How important are captions when training a lora like this? I would love to do something similar
@mongot from my experience, the image quality is more important combined with some base trigger words for the overall concept or style you want to create. like here "lsdemulora". i most of the time work with the civitai tagger and add details if its missing and important and got nice results overall for most models. for further examples, i normally attach my datasets to the models. Check it out and you should find all tags and how they are set :)