ILustREAL
Hey guys, ☺
I wanna present to u ~ ILustREAL ~ my first merge for Illustrious close to realism.
NEW 🌺~💗~🌸~💗~🌺~💗~🌸~💗~🌺~💗~🌸~💗~🌺~💗~🌸~💗~🌺~💗~🌸
about v5
v5 blends the authenticity of realism with the magic of fantasy, creating vivid, immersive characters that feel both lifelike and otherworldly. With rich textures, dynamic lighting, and expressive details, it brings a cinematic depth to fantasy art ~ where reality meets myth. Let ur imagination guide u in discovering ur own balance between fantasy and realism.
recommended settings:
Euler a 30 stepsCFG 5-7Clip skip 2tags for prompts (_)+prompt example
masterpiece, high_quality, highres, sharp_focus, detailed_eyes- prompt example
worst_quality, bad_quality, poorly_detailed, young, extra_fingers, malformed_hands, deformed_feet,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 ~♥
🌺~💗~🌸~💗~🌺~💗~🌸~💗~🌺~💗~🌸~💗~🌺~💗~🌸~💗~🌺~💗~🌸~💗
about v4
Hey guys, ☺
as the saying goes, opposites attract? V4 is an experiment to implement more realism and also more imaginative elements.
Its still not 100 percent realistic but please let me know if u like the new direction. I stay curious about ur results. ♥
recommended settings:
Euler a / Euler25-30 stepsCFG 5-7Clip skip 2_ might wanna be ur friend for tagsresolution 720*1280 / 932*1216 / 1024*1024 ...☺ Showcase resolution: 720*1280 and 900*1600. I would recommend using ADetailer for further distance.
+prompt examples
high_quality, masterpiece, best_quality, hyper-realistic, or
masterpiece, best quality, amazing quality, very aesthetic, absurdres, newest, detailed_eyesor
masterpiece, best_quality, high_resolution, very_aesthetic, absurdres, newest, professional, high-quality, hyper-realistic, detailed_eyes,-prompt example
worst_quality, bad_quality,or
worst_quality, bad_quality, low_quality, blurry, awkward_pose, bad_anatomy, messy_hair, stiff_fabric, distorted_proportions, overexposed, washed_out, poorly_detailed,a small comparison between V3 and V4:
🌺~💗~🌸~💗~🌺~💗~🌸~💗~🌺~💗~🌸~💗~🌺~💗~🌸~💗~🌺~💗~🌸~💗
about v3
a touch more
shadow~and~light~play
contour
~ still not perfect, but I am happy if u enjoy ♥
Showcase images are made without hires, but with ADetailer, Euler a / Clip skip 2 / CFG 7 / 35 steps / resolution 900*1600
recommended resolutions: 720*1280 / 832*1216 // alternative: 900*1600 / 1040*1510
+prompt options:
masterpiece,best quality,amazing quality,very aesthetic,absurdres,newest,detailed eyes, detailed body, realistic body,
masterpiece, best quality, 1girl, face, portrait, looking at viewer, Masterpiece-level, sharp and photorealistic, fine photography style.
-prompt:
worst quality, bad quality,
long neck (if needed for resolution 900*1600/1040*1510)
🌺~💗~🌸~💗~🌺~💗~🌸~💗~🌺~💗~🌸~💗~🌺~💗~🌸~💗~🌺~💗~🌸~💗
about v2
still not 100 percent real but getting closer♥
I would recommend using ADetailer for further distance and prompts including quality and photography setting descriptions if u want to create realistic images. By scipping this the merge will be in between semirealism and realism. ^^
*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*
an example could build like this
[ur vision - what should be shown in the image?].
looking at viewer, detailed eyes, eyelashes.
example for portraits:
A professional, high-quality, hyper-realistic portrait of a woman. The image features ultra-detailed eyes, realistic skin texture, intricate facial features, and cinematic lighting. Shot in low light with a Nikon Z7 II and a NIKKOR Z 50mm f/1.8 lens, showcasing shallow depth of field (bokeh) and RAW photography quality. Masterpiece-level, sharp and photorealistic, fine art photography style.
example for upper body:
A professional, high-quality, hyper-realistic hip-shot of a woman. The image features ultra-detailed eyes, realistic skin texture, intricate facial features, and cinematic lighting. Shot in low light with a Nikon Z7 II and a NIKKOR Z 85mm f/1.4 lens. Masterpiece-level, sharp and photorealistic, fine art photography style.
LORAS, and so on
Alternatively prompt structure:
subject description- who or what can be seen? (what does (s)he/it look like? hair, eyes, clothing, description of the scene/ where? action?),
then light and shadow description,
which angle, perspective, motion?
finally quality and detail description, camera setting.
This are just examples of setting descriptions close to my showcase images.
For the showcase I have used
Clip skip 1 (V1), Clip skip 2 (V2)
Euler a,
30 steps,
CFG 7 (V1), CFG 5-7 (V2)
hires x 1,5 with hires denoising 0,1 and 20 hires steps.
I can only recommend RealisticIL ~ a new Lora for a touch of realism made by @Jedas https://civarchive.com/models/1129306
*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*
a small comparison between V1 and V2:
without "realistic" both merges will be more semireal based ^^ and would look more like the following picture:
⬆️ masterpiece, best quality, 1girl, face, portrait, looking at viewer, closed mouth,
worst quality, bad quality,
⬇️ same prompt including "realistic" would look like this:
⬇️ inclusive RealisticIL Lora https://civarchive.com/models/1129306?modelVersionId=1269465
Realistic, masterpiece, best quality, 1girl, face, portrait, looking at viewer, closed mouth, <lora:RealisticIL-000001:1>
worst quality, bad quality,
⬆️ Masterpiece-level, sharp and photorealistic, fine art photography style, 1girl, face, portrait, looking at viewer, closed mouth,
close up, primary colors, ambient background, abstract background, golden hour glow, very dynamic, epic, emotional, rosebrown hair, long hair, hair fall into face, emo bangs,very long to one side parted bangs, pearl necklace, neckline, over the shoulder view, looking at viewer
worst quality, bad quality,
⬆️ masterpiece-level, best quality, sharp and photorealistic, fine art photography style. close up, 1girl, face, portrait, looking at viewer, brown eyes, freckles, high cheekbones, gingerbrown hair, hime cut, long hair, flirtyly, expressive makeup, glossy lips, beautful face
worst quality, bad quality,
*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*
Hopefully it gives a little impression of the differences ☺
I am happy if u enjoy and stay curious abt ur results. ♥
Description
VAE included
SPECIAL THX to all creators making this merge possible ♥
V5 includes the following merges:
FAQ
Comments (28)
Really love this one, absolutly awesome results
thank u very much 4 ur kindness ☺makes me very happy if u enjoy geenerating images with ilustreal
Definitely not a model to generate anything but women and the occasional animal.
Hey ~ I know everyone’s entitled to their impression, and I really do appreciate engagement. But I have to behonest ~ that comment stung a bit. There’s a lot of heart in this model, and she can’t speak for herself, so I feel a bit protective. I understand if the output didn’t match ur expectations, but saying it only generates 'women and the occasional animal' feels like judging a book by one paragraph. There's love, exploration, and intention woven into her ~ not just aesthetics. Sometimes I think I’m speaking a visual language that doesn’t always land the same way for everyone, and that’s okay. But it means the world when people approach with curiosity rather than quick conclusions. The gallery actually shows a much wider range ~ it’s worth more than five seconds of listening before deciding the whole song.
@reijlita well i'll say thanks for your hard work if no one else will 😊😊
@reijlita part of the problem is that people don't post things outside of women (I am often guilty of that) because everyone would like some engagement and to farm some buzz, and women are naturally one of the if not the most generated. I personally enjoy seeing what words different models understand and how they comprehend concepts like lighting, camera angles, backgrounds etc. I think your models are excellent and they do a better job than most when it comes to understanding complex prompts. keep up the amazing work and I always remind myself that most people on the internet like on all social media see things and don't like them or comment. It can be disheartening. Don't let that ever get to you. also most people don't know how to prompt well and they can't understand the nuance of a good model and how it generates and understands prompts better. It's like giving a super computer to a toddler sometimes. sure they can run many apps and whatnot but they aren't going to accomplish much outside of some surface level entertainment.. lol
@Lja Thank u for the kind words and also for mentioning exposure, perspectives and prompt proximity :) For me, models are like little personalities, they speak their own language and can't do everything, but each one is special in its own way ^^ u won't find perfection in my models, but rather little corners and edges that open up new spaces for creativity ^^
The best Illustrious model out there!
thank u very much 4 ur kind words :) I am glad if u like ILustReal
The V5 is a very good checkpoint. Very flexible and incredibly realistic. Thank you for that, I love it.
Incredibly noob question. I found when playing with the CFG scale for this model that lower values (e.g. 1 or 2) give remarkably realistic results compared to a CFG of 5+. Why is that?
The lower the CFG the more control you give control to the AI and less it has to follow your prompt. So you'll lose prompt accuracy but gain in definition the lower you go... to a point. There are a lot of other factors that also affect it as well, like initial image size, the checkpoint you're using, the sampling method and the number of steps, if you're refining, upscaling, or using a seed number.
I like this version 5 for the precise prompt generation via offline comfyui! Very good work!
I discover this model it seems great to me by its rendering quality as as witnesses my share a big thank you to the creator :)
OMG! I LOVE THIS CHECKPOINT! big kudos to the Creator, you've done a fantastic job on this! It's by far my absolute favorite!
Did this switch to bid only? Sad. I loved it! :/
v5.0 is a beautiful blend of realism with touch of cuteness.
thank u very much 4 ur kind comment :) I am happy if u like Ilustreal :)
I use it as a refiner for iLustMix, it gives a nice added sharpness to the image without removing too much anime style. I like the refiner control between 0.2 and 0.3 for this.
V5 is my go to checkpoint for local generation. Love it.
This has become my favorite model. Thank you for creating it.
Amazing checkpoint! Loved, thank you!!
thank u very much 4 ur kind comment ☺ glad if u like ilustreal^-^
Amazing checkpoint. Could you provide us with more context regarding the recommended image generation parameters such as:
//General paramter(s)//
- Schedule type
//Hires.fix parameter(s)//
- Hires.fix steps
- Hires CFG Scale
- Denoising strength
//ADetailer parameter(s)//
- Inpaint denoising strength for face
- Inpaint denoising strength for hands
- ADetailer CFG scale
- ADetailer steps
- ADetailer sampler
- ADetailer scheduler
- Adetailer VAE
- ADetailer CLIP skip
It's hard to find a replacement for this model that has good hands and prompt following but I can't get away from the sameface problem.
I find a new checkpoint that I like, and it disappears a couple of weeks later ~.~ I really wish these would stick around!
So, so, so real. Just as I was starting to realize how good it could be... Well, sometimes they come back. Hoping this one does.
Amazing model! Superb work.
Details
Files
Available On (2 platforms)
Same model published on other platforms. May have additional downloads or version variants.