🪿🪿🪿 GooseMix 🪿🪿🪿
A dark, atmospheric Anima LoRA merge.
Welcome to the Goose-verse! GooseMix is my first public Anima-based model, meticulously crafted to capture the rich, cinematic, and slightly unhinged beauty that defines high-end anime illustration.
I purposely avoided forcing too rigid of a style onto the checkpoint during the merge. This ensures the model remains highly flexible and incredibly responsive to artist tags, allowing you to steer the aesthetic exactly where you want it.
🌟 What It Excels At
GooseMix truly shines in moody, dramatic settings with high-contrast, dark backgrounds. It has an exceptional grasp of:
Volumetric Lighting & Depth: Incredible handling of god rays, heavy contrast, and dramatic rim lighting.
Vibrant Glow Effects: Perfect for glowing magical elements, floating bioluminescent particles, and neon accents.
Rich Color Palettes: Colors tend to be deep, saturated, and vibrant against dark backdrops.
Versatility: While it thrives in the dark, it can handle traditional, bright anime illustration styles with ease.
⚙️ Recommended Settings
To get the absolute best results out of GooseMix, try starting with these parameters:
ParameterRecommended SettingSamplers DPM++ 2M SDE GPU / Euler / Euler a
Steps 30 – 50
CFG Scale 4 – 7 (Lower values allow for more creativity)
⚠️ CRITICAL REQUIREMENT: This model requires the Qwen base encoder and the Qwen image VAE to function.
Happy generating, and welcome to the flock! 🪿
Description
Cleaner images, removed some randomness. Adjusted colors.
FAQ
Comments (4)
I'd be silly not to try this checkpoint!
Absolutely! Imagine missing out on Goose Magic!
Been waiting to comment on this one, had to give it a few days.
In comparison to other models, this is pretty superb.
Under 5gb and has most semi-real detail.
I (very) dislike for anime though... for semi-real, this might be the standard 🦄
Ultimately, this is one of the best models I tried to get the best "creative" looking image with it's deep contrast and colors.
Tested it against many other models and this one always came up on top for it's visuals and different results compare to the more "basic" looking ones.