<lora:Inconceivable Flux:1.0> A dwarf with a cool steampunk gun riding a griffin through mountain skies, lens flare

<lora:Inconceivable Flux:1.0> A man with a cool augmented reality glasses sipping a steaming mug of coffee

<lora:Inconceivable Flux:1.0> A woman with a cool steampunk hat

<lora:Inconceivable Flux:1.0> A Paladin with blue and silver armor, longsword

A stormtrooper at an abandoned snowy mountain outpost <lora:Inconceivable Flux:1.0>

<lora:Inconceivable Flux:1.0> A stormtrooper at an abandoned jungle outpost

<lora:Inconceivable Flux:1.0> A creature that is a mix of a tiger and a bat, correct anatomy, hyper real

<lora:Inconceivable Flux:1.0> A wolf dragon hybrid, correct anatomy, hyper real

<lora:Inconceivable Flux:1.0> A wolf dragon hybrid

St_Moses @St_Moses

Inconceivable Flux

Name: Inconceivable Flux LoRA (Flux)
Author: St_Moses

LoRAFlux·Published a year ago ·161 creations·0 comments

#digital-art #illustration #painting

Adds lots of details. Lends itself to fantasy. Does well with a color in prompt.

This model was trained on many images that used the Glowing Runes and NC Hyperreal Loras.

Most images in the dataset included something along these lines:

A creature that is a mix of a tiger, an owl, and a bat, glowing orange runes, <lora:Glowing Runes:1.0> hyper detailed digital painting, hyper detailed background <lora:NC Hyperreal:1.0> perfect anatomy

You’ll likely need to include something about “long limbs” or “long necks” in your negative prompt.

If your Lora tag is at the beginning, it tends to generate sharper images, with oversaturated colors. I suggest a Lora weight of 0.5 - 0.8 and increasing your negative prompt to at least -0.6.

If Lora tag is at the end, tends to generate softer edges and functions ok with full Lora weight and default negative prompt.

Upscaling it at the end will also give it a very polished, refined look. Happy prompting 😁👍

Usage guide

I suggest Lora tag at end and default negative prompt. If at beginning, drop Lora weight to 0.5 - 0.8 and boost negative to -0.6. See model page for more info. Tends to default to a darker background unless specified.