🔍

pushing styles in generative images with LORAs in Stable Diffusion

The use of LORA’s is one key element to bring a custom style to each generative output. In the following are some samples how LORA’s effect a standard prompt.

POS: painting portrait traditional of caucasian androgyn male person in traditional room, trans short hair bald, traditional dress baroque, ( detailed skin:1.3), detailed stroke brush, hyper realistic, renaissance futurist style
NEG: naked, beard, (low quality, worst quality:1.4), (monochrome, dof:1.2), easynegative, bad anatomy, bad hand, extra digits, extra fingers, twisted fingers, long body, jpeg artifacts, signature, watermark, username, censor, text,

Disco Elysium LORA executed in different exposure
Bichu LORA executed in different exposure
PupUpParade LORA executed in different exposure
chinese ink LORA executed in different exposure – slight changes in prompt necessary!

Link to the Disco Elysium LORA model
Link to the Blindbox LORA model
Link to the Drawing LORA model

context and lora


zoomed portrait of androgyn introverted futuristic person ( male:0.1) in ( three quarter profile : 1.2) in (detailed white extravagant surreal oversized inflated sweeping design bio insect inspired royal crochet knitted macrame lace shaman ghost head mask and collar and knitted top shirt ritual jewelry long french and cords tentacles , shaman outfit :1.7), curvy detailed embroidery underwater animal bio flower (voronoi:0.5) like, (asymmetrical:0.8), ( traditional shaman textile knitting crochet patterns and veil :1.3), dust in the air, white photo border , sharp, detailed, strong rim lights, dirty, distressed, scraps, bright background ( futurist cyberpunk:2.0) ( rim lights :1.2 )
blurry, photo frame, ugly, blur, glasses, breast naked, puppet, drawing, illustration, mouth