Where's my flashlight?

If you have any request or would like a commission, feel free to comment or message me here or on discord, I will see what I can do!

If you enjoy my work and you would like to support me, how about a coffee? ❤️

Concept LoRA to strengthen and improve darkness of images. It will also improve images with better directional lighting if you prompt specific light source. Composition of image should stay the same.

This concept also stabilises image and controls amount light sources. (usually you get one candle not 5 randomly placed)

LoRA weirdly enough works a little bit like slider, the higher weight the stronger darkness will be without changing composition much.

Use dark to make image with none or little light from light source.

Use dim lighting and type of lighting to make light coming from of the light source

You can mix both for stronger effect

Please avoid using too many LoRAs as the effect will weaken. (will still darken the image but not as much)

Anima :

Anima "dark" prompting works similarly to Illustrious, sometimes base triggers are not enough but it's generally more precise. For example you can use stuff like candles are the only light source to focus only on one lighting direction

The less colours you use in your prompt the darker image will get. Colours also influence ambient lighting, if you ask for green eyes you might get green lighting. Counter it with negatives like green theme etc.

Same idea as before try to imagine how colour look in darkness or ask for dark colour (dark pink etc. )

Illustrious :

Some tags might add unusual amount of brightness.

Notes from testing :

• keeping prompt short gives best result (up to 75 tokens)

• very dark - will trigger very dark skin tag

• night is usually brighter than standard dark environment

• sometimes brightness is tied to a seed so you might want to change it

• usage of colours affects darkness greatly, try to think what colour would look like in darkness, or add dark to every colour you can, for example dark blue eyes, dark pink dress, dark green hair

• shiny skin is an unrealistic common problem with a lot of models, adding it to negatives will shade skin better (shiny skin in darkness is just silly unless oiled up, sweaty or wet...)

• overly unnecessary negatives will destroy darkness, keep it simple and work out from there

• forge/a1111 adetailer is not working too well with this LoRA and it causes brighter box, so be wary and I do recommend to use face detection models instead of box ones.

LoRA weight : 0.1-1

Main trigger :

• dark

• dim lighting

• dark background

Secondary triggers:

• dark [colour] - will make better shading for example : dark blonde hair, dark green eyes,dark pink lingerie

• night

• glowing

• darkness - will make image darker but it's also tag for "dark magic" so you can expect a lot of shadowy stuff)

• any light source to use with dim lighting, for example :

  • lamp

  • phone

  • torch

  • moon

  • glowing eyes

  • pryokinesis (fireball)

• silhouette

• shaded face

• shadow

• backlighting

• underlighting

• sidelighting

Recommended negatives :

• backlighting

• underlighting

• sidelighting

• glowing [X] (hair etc.) - glowing tag can cause too much stuff to glow

• shiny skin (will shade skin better)

• and anything else that gets in a way

.