Hotaru Blend Anima is a model based on the latest Version base-v1.0 of the new Anima model since V3.0. Previous builds based on Preview 3.
It offers a clean Anime style and supports prompting in booru tags and natural language.
You need to download additional text encoder and VAE to use this model:
Text encoder: https://huggingface.co/circlestone-labs/Anima/blob/main/split_files/text_encoders/qwen_3_06b_base.safetensors
VAE: https://huggingface.co/circlestone-labs/Anima/blob/main/split_files/vae/qwen_image_vae.safetensors
Recommended Settings for V6.0:
Sampler / Scheduler: Euler A or ER SDE with Normal scheduler
Sampling steps: 20-35 (8-16 for Turbo)
CFG Scale: 2.5 - 3.5 (1.0 - 1.5 for Turbo)
Shift: 3-4.5
Recommended resolution: 1024px based or higher
Model tested with Forge Neo
Recommended Settings up to V5.0:
Sampler / Scheduler: Euler A or ER SDE with Simple or Beta
Sampling steps: 20-35
CFG Scale: 3-5
Shift: 3-4.5
Recommended resolution: 1024px based
Model tested with Forge Neo
Please refer to the official model page for detailed instructions: https://civarchive.com/models/2458426/anima-official
For all samples, the Forge Neo Add-on "Anima Detail Enhancer" was used to improve minor details. Below settings were applied.

I recommend using a post processing upscale instead of a usual High-res fix.
The preview images were 4x upscaled with 4x-ClearRealityV1_Soft.
Quality tags:
Positive:
masterpiece, best quality, score_9Negative:
score_1, score_2, score_3, blurry, jpeg artifacts,lowres, worst quality, low quality, bad hands, bad anatomy,signature, text, watermark, censor, bar censor, mosaic censorship, mosaic censor, sound effects, blubble speech, patreon logo, patreon username, twitter username, twitter logo, blur, multiple views or similars.Merged models (before V3):
Anima_chromalinemixNoobaiVPred_v10AnimaPreview3
Anima_animaika_v30
animaCatTower_v05
pornmasterAnima_preview2V1
Permission: no commercial use, no selling, use only on Civitai (please also check the sources for their respective permissions)
Description
Version 6.0 is a total overhaul with a more detailed and slightly cuter style and more vibrant colors. This release also comes with a Turbo lora baked in for faster image generation. There is also a int8 convrot version available, which is faster and requires less VRAM.
The Recommended Settings changed a bit compared to previous versions:
Recommended Settings:
Sampler / Scheduler: Euler A or ER SDE with Normal scheduler
Sampling steps: 20-35 (8-16 for Turbo)
CFG Scale: 2.5 - 3.5 (1.0 - 1.5 for Turbo)
Shift: 3-4.5
Recommended resolution: 1024px based or higher
Model tested with Forge Neo
FAQ
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.


