At 100 likes I will launch version 3
🧬 What is JRD_AnimaFusion?
JRD_AnimaFusion is a soft‑realistic, neutral‑style AI model designed to produce clean, elegant, high‑quality images without forcing any specific artistic direction. It naturally delivers:
smooth, luminous skin
soft and controlled lighting
clean color palettes
natural 2.5D depth
zero noise, zero painterly artifacts
stable anatomy and consistent rendering
It adapts to the prompt without imposing a style, making it ideal for portraits, half‑body shots, animals, and simple environments.
⚙️ Recommended Settings (Official Preset)
Sampler
ER_SDE (native, best overall results)
Alternative: SIMPLE or Normal (more contrast)
Steps
30
CFG
4
Denoise (img2img)
0.45 – 0.55
🧾 Positive Prompt (Official Base Prompt)
masterpiece, best quality, score_9, score_8, score_7,
year 2025, newest, highres, absurdres, very aesthetic
From here, add ONLY:
subject
one light source
background
neutral details (hair, clothing, animal, location)
No styles. No “realistic”, “painterly”, “anime”, etc.
🚫 Negative Prompt (Highly Recommended)
extra digits, conjoined fingers, wrong hand, extra legs,
worst quality, low quality, early, old, score_1, score_2, score_3,
cartoon, graphic, painting, crayon, graphite, abstract, glitch,
deformed, mutated, ugly, disfigured, long body, bad anatomy,
bad hands, missing fingers, extra fingers, extra digits, fewer digits,
cropped, very displeasing, artist name, blurry, jpeg artifacts,
lowres, censor, text bubble, speech bubble, child, kid, young
This negative prompt stabilizes the model and prevents:
anatomical issues
cartoon drift
painterly artifacts
low‑quality outputs
unwanted styles
🎨 How JRD_AnimaFusion Works
JRD_AnimaFusion interprets prompts in a neutral and adaptive way. It does not impose a style; instead, it enhances:
soft lighting
clean textures
natural depth
balanced colors
subtle highlights
It excels with one clear light source and simple backgrounds, producing images that feel modern, elegant, and aesthetically consistent.
🏷️ Model Name
JRD_AnimaFusion A fusion of your signature LoRA identity and a neutral, adaptive rendering engine.
Description
JRD_AnimaFusion_Reload is the upgraded and refined version of my ANIMA checkpoint. This release delivers a noticeable improvement in both rendering quality and prompt adherence, making it more stable, more consistent, and more expressive than the previous version.
🔥 What’s improved
Higher rendering quality Cleaner skin, sharper materials, smoother lighting, and an overall more polished, premium look.
Much better prompt adherence The model now follows outfits, poses, accessories, and scene details with far greater accuracy — even in complex prompts.
Improved anatomical stability Fewer distortions, more natural proportions, and more reliable full‑body compositions.
More consistent color grading No more random color shifts or oversaturation. The output is more cinematic and controlled.
Clearer visual identity ANIMA now has a recognizable, stable aesthetic that feels intentional and professional.
In short: more quality, more control, more fidelity to your ideas.
⭐ How to use it
This is a full checkpoint, no trigger word required.
Works smoothly on any frontend (Civitai, ComfyUI, A1111, Mage, etc.).
Recommended CFG: 4–6
Recommended steps: 28–36
Best samplers: res_multistep, euler, er_sde
Best scheduler: simple
Ideal for portraits, full body shots, fashion, fantasy, OCs, and modern aesthetics.
FAQ
Comments (6)
Due to the size (5.24GB), I assume this doesn't need a text encoder nor vae?
File size does not determine whether an ANIMA model requires text encoders and VAE.
ANIMA uses specific components (Qwen text encoder + dedicated VAE), so even a large checkpoint still needs its original files to function correctly
@JRD_2212 Got it, from what I've seen, checkpoints this size normally are AIO and don't need anything else, so, thanks for confirming this is not the case.
@OTAKKx exactly... there are checkpoints that don't need them; but not this one :), thanks to you