Update
The Pony version has an extended dataset and is a LoCON.
cb2077 is the actual trigger word, everything else such as "cyborg, prosthetic arm, etc." is flavor text so include/exclude as you see fit. Silverhand's silver hand is going to appear a lot, because of its overwhelming presence in the dataset.
This model is just a jumble of (admittedly, hand-picked) screenshots of the game, mods included, and so it aims to provide a general feel and vibe - a little bit of everything, a wide puddle if you will.
I could try and explore specific concepts of that universe because, personally, I believe the esthetic that CDPR achieved is awesome. Concepts like backgrounds, tech like guns, mantis blades, etc., characters, so forth.
Let me know if that sounds like a good idea and what you'd like to see.
This LoRA was trained on the in game images of my various playthroughs and characters in Cyberpunk 2077.
The trigger words that should be enough as a start are:
cb2077, science fiction, cyberpunk, sci-fi, neonand if you want Johnny Silverhand's silver hand you could go with:
cb2077, science fiction, cyberpunk, sci-fi, neon, cyborg, prosthetic arm Further details could be: cybernetic eye, artificial eye, gun, katana, aviator sunglasses, etc. - check the info [( i ) button in in upper right corner of the lora in webui] for all the tags.
I'm thinking of trying a similar approach and train a Lora on Baldur's Gate III now that there is a photo mode in the game.