Psssst... over here... you know what? Livebound. 👀 Yeah... don’t miss it 😉
An exploration of light, shadow and the space between photography and illustration.
Hollowlight is a cinematic style LoRA exploring darker atmospheres, pronounced contrast and expressive lighting.
It tends to deepen shadows, influence color relationships and introduce a greater sense of depth and texture. Depending on the subject, the result may take on a more photographic appearance while retaining elements of its original artistic style.
From photography to watercolor, anime and painterly illustrations, Hollowlight brings its own interpretation of light, color and atmosphere.
The result is a blend of cinematic aesthetics, dreamlike qualities and stylistic experimentation.
If Hollowlight finds a place in your workflow, share your first image through Add Post on this page. Seeing what it becomes in other hands is the best support.
The Look
Cinematic Contrast: Deeper shadows, stronger tonal separation and more pronounced lighting.
Atmospheric Colors: Cooler-looking shadows and shifting color relationships, often with warm highlights remaining prominent.
Photographic Influence: Textures, materials and lighting may take on a more dimensional, photographic appearance.
Dreamlike Atmosphere: A visual direction shaped by shadow, light and atmospheric depth.
Stylistic Blending: Illustrative, painterly and photographic qualities can coexist, with the balance varying between images.
Where It Belongs
Cinematic portraits and character studies
Fantasy, surrealism and dreamlike scenes
Anime and stylized illustrations
Watercolor, painterly and digital artwork
Landscapes and atmospheric environments
Photography and product imagery
Hollowlight can produce anything from atmospheric adjustments to substantial visual transformations, depending on the subject, prompt, model and strength.
Trigger and Strength
Trigger word: None
Hollowlight's influence varies across models and subjects. Lower strengths generally offer a lighter interpretation, while higher strengths may produce more pronounced changes to lighting, colors, textures and composition.
Stronger settings can also alter proportions, object relationships and fine details. Higher strength does not necessarily produce a more desirable result.
For recommended strengths and inference settings, please refer to the notes for the specific model version.
PLATFORM RESTRICTION
This model may be used locally, on privately controlled compute infrastructure or on Civitai. Uploading, hosting or using this model through third-party AI model hosting or image generation platforms such as TensorArt, SeaArt or similar services is not permitted. Personal use on your own hardware or on rented compute infrastructure under your exclusive control is permitted. This restriction concerns the hosting and use of the model itself. Rights for generated images, merges and redistribution are governed by the Civitai permissions shown on this model page. Any shared merge based on this model must retain the permissions required by the original model.
Description
Trained on: Krea 2 Base
Tested with: Krea 2 Turbo FP8
Selected checkpoint: Epoch 8
Trigger word: None
Tested Generation Settings
Sampler: Euler / Euler a
Scheduler: Simple
Sampling Steps: 6–8
CFG: 1
Suggested starting strength: 0.50
Common working range: 0.35–0.75
Version Notes
This version was primarily tested with Krea 2 Turbo FP8 across photographic, illustrative, painterly and stylized subjects.
It tends to introduce darker shadows, pronounced contrast and changes in lighting, colors and surface appearance. Some illustrations may develop a more photographic visual character.
Strengths around 0.50 offer a useful starting point. Higher values can produce more substantial transformations, including changes to framing, proportions and subject details.
The settings above reflect the tested inference setup rather than guaranteed optimal values for every workflow.