Not loaded
Download
1 variant available
fp16 SafeTensor
und3rw4t3r_c1ty-i.safetensors
Half precision, best balance • 217.88 MB
Verified: 18 hours ago
260 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9
00 1 2 3 4 5 6 7 8 9
Not loaded
(5)
Oct 15, 2026

358.3K0 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9.0 1 2 3 4 5 6 7 8 9K
5.7K0 1 2 3 4 5 6 7 8 9.0 1 2 3 4 5 6 7 8 9K
61.5K0 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9.0 1 2 3 4 5 6 7 8 9K

License:
Illustrious LicenseENGLISH
und3rw4t3r c1ty — Gold beneath the abyss
I designed this background LoRA concept around an art-deco underwater metropolis that still shines long after its inhabitants have disappeared. Black marble, brass, riveted steel and warm golden windows meet the deep blue ocean: a monumental, theatrical city with a quiet melancholy beneath its grandeur.
My 45-image dataset explores eleven spaces: the grand dome, shopping promenade, bathysphere dock, hydroponic gardens, residential gallery, filtration works, observatory, panorama restaurant, diving workshop, engine hall and memorial hall. A final upward view brings the metropolis together. Each space has its own props, floor, lighting and camera compositions.
I kept the backgrounds empty: no people, skeletons or mannequins. The diving suits are unoccupied equipment, and the observatory contains only a faint indistinct shadow far outside the glass. The city is generic, without brands, franchise references or readable text.
Trigger: und3rw4t3r c1ty
I suggest starting the scene prompt with the exact trigger, then naming the space, viewpoint, materials and lighting. Keep exclusions in the negative prompt or the separate generation constraints.
Examples:
und3rw4t3r c1ty, monumental grand glass dome, wide establishing shot, black marble floor with brass inlay, dry illuminated fountain, golden lighting, deep blue ocean beyond the glass
und3rw4t3r c1ty, hydroponic gardens, low camera angle, tiered grow trays, hanging vines, tiled walkway, magenta and blue grow lights
und3rw4t3r c1ty, engine hall, diagonal steel grille catwalk, colossal pistons and flywheels, steam, red and amber machinery glow
Suggested starting points only, not verified results: after training a LoRA for a matching SDXL or Illustrious base, try 24–32 steps, CFG 4–6 and LoRA weight 0.6–0.9, then adjust through comparison tests. SDXL and Illustrious require their respective compatible training and loading workflow. For Krea, compatibility and any supported LoRA workflow must be checked separately; these numerical settings should not be assumed to transfer. I have not trained or tested a LoRA from this dataset yet, so I am not claiming model compatibility or confirmed generation settings.
arrssenne
FRANCAIS
und3rw4t3r c1ty - L'or sous les abysses
J'ai concu ce concept de LoRA background autour d'une metropole sous-marine art deco qui continue de briller apres la disparition de ses habitants. Marbre noir, laiton, acier rivete et fenetres dorees rencontrent le bleu profond de l'ocean : une ville monumentale et theatrale, avec une melancolie discrete sous sa grandeur.
Mon dataset de 45 images explore onze lieux : grand dome, promenade commerciale, dock de bathysphere, jardins hydroponiques, galerie residentielle, installations de filtration, observatoire, restaurant panoramique, atelier de plongee, salle des machines et memorial. Une derniere vue vers le haut rassemble la metropole. Chaque lieu possede ses accessoires, son sol, sa lumiere et ses compositions.
J'ai garde les backgrounds vides : aucune personne, aucun squelette ni mannequin. Les scaphandres sont des equipements inoccupes et l'observatoire montre seulement une ombre lointaine et indistincte derriere la vitre. La ville est generique, sans marques, references a des franchises ni texte lisible.
Trigger : und3rw4t3r c1ty
Je suggere de commencer le prompt de scene par le trigger exact, puis de preciser le lieu, le cadrage, les materiaux et la lumiere. Les exclusions restent dans le negative prompt ou dans les contraintes de generation separees. Les exemples anglais ci-dessus peuvent servir de structures de prompt.
Points de depart suggeres seulement, sans validation : apres entrainement pour une base SDXL ou Illustrious correspondante, essayer 24 a 32 steps, CFG 4 a 6 et un poids LoRA de 0.6 a 0.9, puis ajuster par comparaison. SDXL et Illustrious exigent chacun un workflow d'entrainement et de chargement compatible. Pour Krea, la compatibilite et un eventuel workflow LoRA doivent etre verifies separement; ne pas supposer que ces valeurs numeriques se transferent. Je n'ai encore entraine ni teste de LoRA issu de ce dataset : aucune compatibilite ni aucun reglage ne sont confirmes.
arrssenne


