v10 showcase resolution 832×1536 / no Hires, no ADetailer ☺ a little more about v10 below 🔽
💌
A small thank you —~
to those who tinker with me,
the dreamers,
the enjoyers :)
I hope you're having fun exploring v10. 💗

Animij
Hey guys,
u asked abt it. 😊 Now I present to u: Animij ~ retro-modern soft cored. 4 different styles ~ moaarrrrrr possibilities?! 😋
For a long time I asked myself why merge an anime model ~ when there are a wealth of great models that serve this genre?
Ur voice is important to me, so I took the plunge. U know me, I like things different, and so Animij spawns as an anime style that may not suit everyone's taste ~ but maybe urs?
Animij is coming
soft
drawn
vibrant
retro touched
with a touch of eroticism
minimalistic, but detailed
and full of emotions
Tasty for more? Feel free to play around ~ every version comes with her own playground ☺
💌 NEW
✦ Version 10 —~ Core ID
💙 Animij is an ID model.
What does that mean?
💙Animij has character ;)
• clear subject separation
• soft isofocus / gentle background falloff
• light defines form
• soft-gloss skin highlights
• clean, controlled speculars
• color interaction (warm ↔ cool)
• background supports subjects
• playful / confident posing
• slight perspective presence
• restrained, accessible expressions
• calm / soft / slightly playful emotion
• clean edges + soft rendering balance
• semi-anime aesthetic
• lighting with behavior
• flexible character identity
💙 v10 was created from a feeling of gratitude.
So many heartwarming words touched me.
…they stayed.
This is how v10 emerged ☺
For this moment ~ I’m a little quieter in words
So I let the images speak.
I hope you enjoy.
—~ Reij ♥
Reij's Animij v10 Sweetspots
This is how I personally like to use Animij v10.
UniPC Karras
24–28 steps
CFG 5.0–5.5
832 x 1216 / 832 x 1536
clean / balanced / soft finish
DPM++ 2S a Karras
24–28 steps
CFG 5.0–5.3
832 x 1216 / 832 x 1536
more punch / stronger contrast / more energy
DPM++ 2M Karras
26–30 steps
CFG 5.1–5.6
832 x 1216 / 832 x 1536
stable / clean / slightly sharper
Euler a
20–24 steps
CFG 5.5–6.5
832 x 1216 / 832 x 1536
more playful / more variation
✦ Version 9
"Sparks Found in the Space Inbetween Shadows"
hey friends, ☺
i’m not a fan of comparisons, but what stands out with v9 is that it works both with various quality tags and negative prompts ~ and also without them.
Adetailer? optional.
in some showcase images i simply didn’t use it ^^
hires, i remain a fan of upscaling, just to give the model more room to unfold. but the model can be used without it.
steps? i prefer around 40, but 20 have their own charm too. entirely up to your taste ^^
for different hardware, whether onsite or local. i’ll only be able to test onsite later, so it was important to me to try the model locally first with smaller settings :)
sampler? feel free to experiment ~ currently i’m using Euler a and DPM++ 2M Karras.
✅ ladies? yes
✅ gentlemen? yes
✅ 🥵 nsfw? yes
✅ 🦊 furry? yes
i love it ♥
i hope you enjoy the new playground just as much :)
with ♥ Reij
✦ Version 7
| ♥ | Specs: Euler a / DPM++ 2M Karras · 30+ steps · 832 x 1216 · CFG 4-7 · Hires x 1,5
| ♥ | Quality Tags (option examples):
| + | masterpiece, high_quality, highres
| ~ | worst quality, bad quality,
✦ Version 1.2
hey friends,
your feedback is important to me ☺
That's why there is now the first Animij update ^-^
v1.2 retains the style of v1, is a little more refined in the rendering and spawns with more options to use quality tags or leave them out to suit your taste ☺
I hope you have a lot of fun being creative
with ♥ Reij
20+ steps
Euler a / DPM++ 2M Karras
CFG 2-7
832 x 1216
quality tags & negative prompt optional ~ use or omit as preferred
prompt examples:
+prompt: masterpiece, high_quality, highres
-prompt: worst_quality, bad_quality, poorly_detailed
✦ Version 6
Like the previous versions of Animij, v6 is a separate model, not an update ☺
~ blending anime, semi-realistic, and comic-inspired elements.
⚠️
♥ settings:
🛠️ General settings I like to use: Euler a, 30+ steps, and a resolution around 832 × 1216. But of course ~ u’re warmly invited to be creative here too! 🌸✨
During merging, special care was taken so you don’t need any quality tags for great results ~ and u can even leave out the negative prompt if u like.
Just create the way u love 💖
Most importantly: I want to give both prompt pros and new creative AI minds the same freedom to explore – just prompt whatever comes to mind and let the magic happen 💖
I am happy if u enjoy and stay curious abt ur results ♥ best regards Reij
✦ Version 5
hey friends ☺
V5 spawns soft kissed by V1 ~ not back to the roots but with retro shine and velvety soft ~ comes with more freedom in the quality prompt design ~ more user-friendly for beginners and prompt professionals, whether realism or toon, V5 stays true to her style.
I am happy if u enjoy and stay curious abt ur results
with ~♥ Reij
♥ setting examples V5
Clip skip 2 Euler a 30+ stepsCFG 4-71024*1024 resolutions +++prompt:
-prompt:
worst_quality, bad_quality, poorly_detailedRegarding the quality tags, u are cordially invited to let off steam creatively or to give more freedom ~ according to ur taste ☺
✦ Version 4
from minimalist to fine details ~ ur prompt, ur fantasy ~ remains anime-based even with realistic attention. Through self-determined prompting more play in style variations: from matt to bright, accentuated or rich in detail, there is always a touch of detail, whether as a classic 2D or with more depth.
I am happy if u enjoy and stay curious abt ur creations
with ♥
~Reij
♥ setting examples V4
Clip skip 2 Euler a 30+ stepsCFG 71024*1024 resolutions +++prompt for 2D anime:
masterpiece, best quality, gradient, flat color,+prompt for more depth:
masterpiece, best quality, gradient, realistic skin, depth of field,-prompt
bad quality, worst quality, lowres, jpeg artifacts, bad anatomy, signature, watermark, censored✦ Version 3 ~ drawn symbiosis
v3 is the most drawn version so far. The focus is on the connection between AI and creative intuition ~ a quiet, painterly collaboration.
Born digital, but with the spirit of analogue: Every stroke appears as if it were drawn by hand ~ clear, accentuated, deliberately placed. It's not paper, but a canvas. Not a brush, but still a brush stroke.
v3 brings the “Drawn Touch” back to the fore ~ as a homage to drawing itself and as an expression of a new aesthetic: where AI does not replace, but accompanies.
✨ The Art of Animij ~ where details speak ✨
In Animij, details whisper of emotion ~ a stroke of storytelling. Eyes shimmer with depth, reflecting unspoken dreams. Light and shadow dance across soft gradients, giving each face a warmth that feels almost alive.
The magic lies in the small things: the delicate blush of a cheek, the glint of light on a ring, the way colors breathe with life. Nothing is static~ every detail wanna invite u: wanna come deeper? making each piece feel like a moment suspended between fantasy and reality.
✨ Wanna look closer? The story is in the details.
♥ setting examples V2
Clip skip 2 Euler a 20+ stepsCFG 4+1024*1024 resolutions+prompt
masterpiece, high_quality, highres, -prompt
worst_quality, bad_quality, poorly_detailed,V2 can be used without -prompt
♥ setting examples V1
Clip skip 2 Euler a 30 stepsCFG 61024*1024 resolutions_ tags ♥ anime lookalike 2D
+prompt
masterpiece, high_quality, highres, flat_colors, gradient, -prompt
worst_quality, bad_quality, poorly_detailed,♥ semireal lookalike 2.5D
+prompt
masterpiece, high_quality, highres, sharp_focus, detailed_eyes, retro, 3D, realistic, realistic_body, sharp_eyes,-prompt
worst_quality, bad_quality, poorly_detailed, young, nude, 2d, sketchanime style examples V1
♥ watercolor_anime
♥ pencil_scetch_anime
(pencil_sketch_anime), monochrome, black_and_white,
♥ soft_shading_anime
♥ neon_anime
♥ grunge_anime
♥ pastel_dreamy_anime
♥ 90s_anime
♥ 80s_anime
♥ pastel_retro
♥ retro
♥ Isekai Fantasy Style
♥ Isekai Fantasy Style & Grunge
♥ Kawaii_anime

I am happy if u enjoy and stay curious abt ur results ~♪~
If u like what I do consider leaving me a like or a comment, much appreciated and helps me getting a feeling for tinkering ~♥
Description
VAE included
merge: Animij baked from my own ressources and Lora trained especially for Animij
FAQ
Comments (47)
what hi-res fix settings do you recommend?
my local hires settings (forge/832*1216) morstly are: Upscale by x2 / Hires steps 10-20 / Denoising strength 0,1 (max 0,2) / Hires CFG 1. ☺
This looks fun, I'll give it a whirl next time I fire up SD!
thank u very much 4 ur kind comment ^-^ curious 2 see ☺
It looks interesting, it makes sense to try , I'll describe the experience later . (Google translator )
thank u ^^ have fun trying it out. I'm looking forward to ur creations
beautiful and aesthetic
thank u very much ☺ happy if u like the lookalike ^.^
Love the style, will definitely be giving it a try soon and looking forward to some nice results and seeing what everyone else makes. Thanks for sharing!
thanks alot, I'm happy if Animij brings u joy ☺ and u have fun being creative ☺
An excellent checkpoint, very easy to use.
I managed to use it locally in ComfyUI with Hyper SDXL 8-step Lora (cfg 2 - 3) without having picture quality problems, which is essential for any potato PC with 4 Gb VRAM.
Thank u for the feedback and kind words ^-^ I am very pleased that u are satisfied and ur feedback has invited me to test the model again with fewer steps^.^
I have to warn you that HyperSDXL tends to give 3D look to Illustrij (and i guess to other checkpoints with 2.5D option). On Animij it gives only a bit of painted volumetric effect, as you can see from my images. As for Illustrij, i found out that it gives good results on 12 steps even WITHOUT Hyper Lora in my local generations in ComfyUI.
@Tyomanator thanks alot again for ur feedback, good to know ^^ I noticed that there is a subtle soft 3D semi touch, but the results look surprisingly good ^^
@reijlita You are welcome, i am happy to be of little help with my feedback, thank you again for amazing checkpoints :)
I am looking forward for making some more experiments tomorrow with Animij (who already became my favourite) and with your Loras
@Tyomanator tyvm ^.^ curious 2 c
@reijlita Thanks for your interest ))
First one is that i was able to reduce generation steps in ComfyUI to 12 and even 8 without accelerator Loras and still get great results:
https://civitai.com/posts/14229406
Useful for trying new concepts, then i can increase steps when i have something worth impoving, and see if it helps with picture quality ))
So no more need to mess with HyperSDXL
@Tyomanator Thank u very much for diligently trying out the models with lower step options^^
@reijlita Well it is a kind of obsession for me, as it is the difference between generating one image in 2 or in 5 minutes ))
But result with your checkponts on 12 steps is really impressive, i am truly amazed!
@reijlita I spent a day adjusting picture quality nuances via prompt and am really happy about it )). Your chackpoint is truly amazing!
And thanks for the tip! ))
This is my favorite checkpoint so far! So easy to generate exactly what you want and the quality is amazing!
thank u very much 4 ur kindness ^.^ happy if u like generating images with animij ☺
It was so beautiful, but now it's doesn't work 😕
thank u very much 4 ur kind words ♥ I long for the day when it might be possible again.
+1
Now it's available 🥰
@ikarirh I'm so happy that it can currently be used for on site generation ♥ and u all wouldn't believe how happy I am about ur creations ^-^ I'm so grateful that we can still enjoy the moment together ☺
This is a great checkpoint. I know you have a soft-spot for 2.5D and semi-realism, but this is hands down one of my favourite flat/2d models. Especially when it comes to color. Other checkpoints often feel over-saturated or add blue shine to everything. But this one has rich colors but good balance.
Side note: can you say what illu versions this is based on? I'm so unclear about what makes something v-pred / v-3 / noob / etc and when it comes to choosing certain loras, it's been confusing the hell out of me.
Thank u very much for ur kind words and ur feedback ☺ I'm very happy if u like Animij. Precisely because of my passion, Animij is a little adventure for me too. But since ur voices are important to me, I dared to do it ^-^ I also love that u mentioned the colors ~ one of the points that is particularly important to me. For Animij I wanted a variety of colors and eye comfort :D I trained a Lora especially for this and I'm very happy that u like this little element.
To be honest, I can't tell u exactly what the merge inside the merge before the merge inside the merge based on, because I can only merge with my hardware and can't train myself. Nowadays I mostly only use my own models and loras to merge new merges, like with Animij. most newest models mostly consist of my own ressources, but some of these merges also contain some merges which include merges from others. Since it is often not stated what exactly it is based on, I can only speculate here ^^ So with Loras all we have left is the wild adventure of trying things out ^^
Perfect checkpoint
Thank u very much 4 Ur kindness, I am happy if u like generating images with animij :)
It seems like the image style will be affected by the background
Unless you specify the background in your prompt ;)
Are the underscores actually required in the Danbooru tags for this model, or can spaces be used instead?
u can use animij with _ but also without. ☺ Depending on the prompt, a different image can be created. Basically, I would only recommend using tags instead of complete sentences ☺
By far my favorite model on site - and can't wait to try out more of your work!
Thank u very much for the kind words ~ I'm very happy that u like it ☺
That's my new fav Model , will be using for a while now haha !
now u have made me smile. I'm very happy if u like Animij and have fun generating images with 😊
Finally I get to create with this absolutely wonderful checkpoint. I love it!!!!
Thank u for the kind words Rando 😊 I'm very happy if u like Animij and have fun with
@reijlita I've been going back through just about every pic I've ever created that I liked and recreating with this model. The results overall have been amazing. I love it. It might just become my default on CivitAI.
@RandoTheBarbarian I'm really glad that u like the model and that u're having fun trying out a lot of things ☺
@reijlita but now I'm bummed that their is a v2 and I can't use it. Lol. Maybe next week.
@RandoTheBarbarian I hope so 2 ^-^ Im very curious how it looks onsite
perfection
thanks alot 4 ur kindness ♥😊 I am glad if u like Animij






