👋 If you like what I do and want to support the development, feel free to buy me a coffee:

💫 Want to try it before downloading?

🌐 Test UtopiaXL right in your browser
🌀 Click & create instantly – no install required

Hi! Here's a quick introduction to the model:

You probably don't want to read the whole thing, right? In short, with just a few steps (14-25) and (1-6) CFG Scales, you already have a quality image.

PS: Don't forget to add these triggers if you use cfg Scale 1

Positive Prompt: hdr:2.5
Negative Prompt: flat color:2.5

(However, with Loras, use fewer CFGs; that will greatly help with interpretation. If you find it too opaque, check out the negative prompts I recommended.)

PS: This version should be 100% compatible with SDXL-based Loras (PDXL, NOOBAI, IL).

📦 UtopiaXL V4.0 — More Stable Than Ever

💀 Goodbye v3… Hello v4

If you’re bored, curious, or just hoping for something interesting from this model — good news.

After juggling v3 (yes, the one that seemed great… until it just stopped working), I discovered there was something wrong with the fusions. It didn’t matter what LoRA I added or how much weight I used — the model just wouldn’t improve.

The cause? A bad merge.
The fix? None — I didn’t have a save point for that version and couldn’t recall all the steps to rebuild it.

So I decided to go back to the roots: v2.

V2 is way more stable than v3 and much more LoRA-compatible. To refine it, I had to remove some messy fusion remnants and various PDXL model fragments.
And yes… that means no ponies (please use LoRAs — or stick with v3 if you know how to make it shine).

In the end, that cleanup led to this version:
📌 v4 — more stable, more compatible, and much more complete.

🛠️ What’s New?

🔨 Fully rebuilt from v2 (the truly stable one).
🎯 Fully compatible with multiple LoRAs — styles, characters, poses, etc.
🎨 Realistic illustration (don’t expect photorealism without help from prompts or LoRAs).
🧪 Fixed saturation (even CFG 16 won’t turn neon).
🧩 Works with SDXL LoRAs like pdxl, illustrious, noob, etc.
🧠 Includes 5000+ artists for prompting.
🧯 Tip: Use low CFG when using LoRAs — trust me.

🧪 How I Tested It:

Up to 1800px — no HiResFix.
24 steps / CFG 5 = solid results (especially with artist tags).
Looks decent starting from 18 steps.
Pro tip: shiny is your friend. Use negatives if it gets too glossy.
Adjust negatives depending on the effect you want (examples below).

🤷‍♂️ In Summary...

v3 was pretty — but unstable.
v4 is actually usable, more stable, and highly compatible — perfect for experimenting.

My goal has always been to make a model that’s useful even without custom LoRAs.
And this time, I think I finally nailed it.

Thanks for using it.
And remember: this model was made for playing and experimenting.

Is it worth it? That’s up to you.
But I’m already having fun with it — and it works.

I always wanted something that could help create new styles and train LoRAs — and now, I think it’s finally possible.
Honestly, I had a lot of fun making this.

🎁 Thanks for using it!

⚙️ Optimal Settings Guide (Same as v3.0)

🧭 Steps: 14–30 (Sweet spot: 20–25)
🎯 CFG Scale: 3.5 for soft style, 5+ for crisp details
📈 Samplers: DPM++ 2M Karras, Euler a, DPM++ SDE
🖼️ Resolution: 1024×1024 and up – 1152×1344, 1536×896 work well
🔁 Upscalers: R-ESRGAN 4x+ Anime6B / 4x-UltraSharp
📌 Note: The VAE is already integrated; no need to load an external one.

🛡️ Safety Labels

This model may generate content of various types. Users are expected to correctly categorize its use into the following categories:

General / Sensitive / NSFW / Explicit

Please consciously add "NSFW" to negative labels if you wish to avoid explicit content.

🏷️ Prompting Tips

🔹 Positive Prompts:

(masterpiece), (best quality), (ultra-detailed), amazing quality, very aesthetic, perfect composition, intricate details, absurdres, incredibly absurdres, 8K, novel illustration, cute face, pretty, ultra-detailed-eyes, ultra-detailed, extremely detailed eyes, detailed face, clear skin, shiny hair, shiny, greatscore, dynamic angle, dynamic pose, high resolution, absurdres, newest, best illumination, depth of field, complex background, detailed background, official wallpaper

Optional Enhancers:

Others More: anime, cartoon, shiny skin, shinym slutty, sexy, slut:2.1, cute, eyelashes,
pretty, Creepy, watercolor, abstract, dark theme

photorealistic,(REALISTIC:1.15),(REALISIM:1.15), detailed eyes:1, masterpiece, best quality, very aesthetic, high resolution, ultra-detailed, absurdres, newest, extremely detailed eyes, realistic,

score_9, score_8_up, score_7_up, Score_6_up, Score_5_up, Score_4_up,  source_anime, hi res, masterpiece, best quality, detailed, amazing quality, detailed background, detailed eyes, clean eyes, best quality, highres, source_cartoon, shiny,

🔸 Negative Prompts:

bad quality,worst quality,worst detail,sketch, censor, patreon username, patreon logo, watermark, artist name, signature, logo, anatomical nonsense, bad anatomy, ugly, worst aesthetic, old, mismatched pupils, dirty face, extra limbs, bad hands, bad fingers, text, normal quality, low quality, web address, email address

Others More:

(worst quality), (low quality), (bad quality), unaesthetic:2.10, censored, (watermark), big nose, (extra digits:1.4), (extra fingers), (bad hands), missing fingers, poorly drawn face, simplified eyes, flat color:2.5, flat colors:2.5, blurry:2.5, washed out:2.1, muddy colors:2.5, dull colors:2.5, faded, overexposed, underexposed, low saturation:2.5, low contrast:2.5, bad lighting, lack of shadows, poor shading, simple shading, flat shading, boring background

📦 AGGA’s UtopiaXL v3Q – The Chimera Update 🧬

⚙️ What is UtopiaV3Q?

This is not a traditional update.
UtopiaV3Q is a parallel refined version, designed to improve compatibility with LoRAs from the following ecosystems:

PonyDiffusion
Illustrious 0.1 / 1.0 / 2.0
SDXL 1.0
Noob-AI

…without compromising the core visual identity of the model.

To achieve this, carefully selected LoRA extracts were fused using a low total alpha of 0.25. This was just enough to synchronize the internal architecture with the activation patterns expected by those environments — without significantly altering the style or tone of Utopia.

🧠 What changed?

The original V3 version had issues with LoRAs trained on Illustrious 0.1, which only worked properly with very low weights (~0.5).
Now, in V3Q, those same LoRAs can be used comfortably with weights up to 0.8+, without distortion or degradation.

🐎 Bonus:

The model now also recognizes tags and triggers from PonyDiffusion, allowing character and concept LoRAs from that ecosystem to function properly.
Visually, not much has changed — but conceptually, the model understands a lot more.

In short:
UtopiaV3Q is a refined chimera, made not to overwrite the style, but to understand more LoRA dialects.

⚙️ Optimal Settings Guide (Same as v3.0)

🧭 Steps: 14–28 (Sweet spot: 20–24)
🎯 CFG Scale: 4.5 for soft style, 5.5+ for crisp details
📈 Samplers: DPM++ 2M Karras, Euler a, DPM++ SDE
🖼️ Resolution: 1024×1024 and up – 1152×1344, 1536×896 work well
🔁 Upscalers: R-ESRGAN 4x+ Anime6B / 4x-UltraSharp
🧬 VAE: Built-in, no external VAE needed

🧠 Prompting Tips

Positive:
(masterpiece), (best quality), ultra-detailed, greatscore, score9, anime style, cute, absurdres, detailed lighting, shiny skin, cinematic, dynamic pose

Negative:
bad anatomy, worst quality, extra limbs, sketch, watermark, signature, cropped, shiny skin (if unwanted), lowres, blurry, oil skin, artist name

🧪 Fun fact:

You can now combine NoobAI-style LoRAs with Pony-style and even standard SDXL LoRAs in a single generation.
It shouldn’t glitch unless weights are too extreme — play around with alpha 0.4–0.7 for safest results.

📦 UtopiaXL V3.0

So, I’ve been testing Utopia v2, and while it was great for concepts and poses, the base style sometimes broke if you didn’t use quality prompts. (Honestly, it was pretty random unless you added an artist tag—but when you did, it worked quite well.)

Anyway… this version represents several days of work, over a dozen revisions… and many restarts (not that I’m complaining 😅).

At first, my plan was to continue building on ILv0.1. I even created a lighter 2.5 version (less than 5 GB, nearly 4), which was aimed at cleaning up some issues.

It worked well, but I wasn’t sure if removing certain data would break it further if I planned to update it again. So, after a lot of trial and error, I got tired and… decided to start from scratch.

This time, I started from ILv2.0, merging NoobAI and all the models I had previously used—plus some new ones (I’m sorry, I don’t have a full list). Of course, I included the key data from earlier versions and added new concepts, characters, and styles. My goal was to make it as compatible and adaptable as possible, so I refined a lot of things to avoid the usual errors from the previous version—like glitches when not using styles or quality triggers such as anime, cartoon, shiny, or shiny skin, or when missing recommended negatives.

I also fixed conceptual styles like pixel art, realism, 3D, PVC, and semi-realism, which barely worked before.

In short, AGGA’s Utopia is now optimized enough to generate detailed illustrations in 15 steps or less. It’s versatile, flexible, and much more consistent… exactly what I was aiming for.

I would’ve loved to add even more content, like additional VTubers or alternate versions of certain characters, but the submission deadline was getting close, so I chose to release it now.

If everything goes well, this should be the final version of the project. There’s a lot I set aside—maybe it’ll see the light in a future model (waaay in the future).

I hope you enjoy using it as much as I enjoyed creating it! See you ~ (´･з･)/

⚙️ Optimal Settings Guide

Designed for flexibility and quality across multiple styles.

🧭 Steps: 14–28
Efficient results from 10+ steps. Best quality around 20–24.

🎯 CFG Scale: 3.5 – 7
Try 4.5 for flexible results or 5.5 for sharper definition.

📈 Samplers: DPM++ 2M Karras, Euler a, DPM++ SDE
Works well with most modern samplers. Choose what fits your style.

🖼️ Recommended Resolution:

Start from 1024x1024 or larger for better structure and detail.
Model scales well to 1536x896, 1152x1344, and portrait/landscape layouts.

🔁 Upscaling Recommendations

To retain fine detail while boosting resolution:

Upscaler: R-ESRGAN 4x+ Anime6B / 4x-UltraSharp
Denoising strength: 0.35 – 0.5
Steps (if re-rendered): 20
Scale: 1.4 – 2.0

🧬 VAE Support

Note:
The model already includes an internal VAE — no need to load an external one.

🛡️ Content Labeling

This model can produce content across categories.
Ensure you mark your generations responsibly:

General / Artistic / Sensitive / NSFW / Explicit

To reduce undesired adult content, add "nsfw" or "explicit" to the negative prompt.

🏷️ Prompting Tips

🔹 Positive Prompts:

sexy_perspective, (masterpiece), (best quality), (ultra-detailed), amazing quality, very aesthetic, illustration, perfect composition, intricate details, absurdres, incredibly absurdres, shiny skin, shiny hair, clear skin, ultra-detailed-eyes, greatscore, extreme perspective, dynamic angle, dynamic pose

Optional Enhancers:

Others More: anime, cartoon, shiny skin, shinym slutty, sexy, slut:2.1, cute, eyelashes,
pretty, Creepy, watercolor, abstract, dark theme

shiny skin, abstract elements, glowing effect, magical aura, intense gaze, cinematic background, surreal scenery, high realism

anime style, cartoon, soft shading, detailed skin, stunning eyes, dynamic lighting, best quality, masterpiece, intricate detail, expressive face, dramatic pose, cinematic lighting, depth of field, amazing composition, elegant design, vibrant colors

🔸 Negative Prompts:

bad quality,worst quality,worst detail,sketch,censor, low quality, anatomical nonsense, bad anatomy, lowres, normal quality, worst aesthetic, ugly, mismatched pupils, dirty face, extra limbs, bad hands, bad fingers, watermark, artist name, signature, text, patreon username, patreon logo, web address, email address,

Others More:

low quality, bad anatomy, blurry, distorted face, messy hands, long neck, censored, multiple pupils, duplicate limbs, cropped, extra fingers, overexposed, boring composition

worst quality, bad quality, very displeasing, displeasing, bad anatomy, artistic error, lowres, bad hands, signature, artist name, variations, old, oldest, missing fingers, excess fingers, extra hands, excess hands, oil skin, wet skin, shiny, shiny hair, shiny skin

📦 UtopiaXL V3.0

📦 UtopiaXL V2.0

This is a model I created inspired by some of my testers (like CuteMaster CKPT, various LoRAs, and a mix of styles and concepts). NOW SEVERAL 100% NSFW CONCEPTS ADDED!
I recently refined it with the help of Illustrious, and although it was originally meant for personal use — as I use it to generate new styles and concepts — it was also built with several community resources and ended up performing better than expected. That’s why I decided to polish it and share it.

In short: I experimented with many things until I finally got the result I was looking for. It might not be perfect, but I really liked how it turned out.

Hope you enjoy it!

This version includes a bigger number of concepts and tries to rescue some styles that were lost in the previous version (I only added 200 styles since I'm not sure how many illustrious base includes), the ones I've tested work well so there aren't many problems there, it should also include the concepts I added below, but I'm still testing the model to see how many were merged, if it's as I expect, there should be more than 200 concepts (although I only added them to make my life easier, in short... I hope you like it~)

PS: Many of the new concepts added are things already included but improved, such as the hairjob (I actually don't remember almost anything about what I added, but I do have it written down. Perhaps you'll see it in the examples I uploaded or should upload. Most of them will be succubus, so if you see them, those will be the concepts).

⚙️ Recommended Settings:

Steps: 15–30
CFG Scale: 2.5–7 (4.4 or 5.5 recomended).
Sampler: Euler a, DPM++ 2M, etc.
Resolution: Larger than 1024x1024 to take advantage of its original resolution.
Recommendation:
• Clip Skip 1 | CFG 4.5–6 | Greater accuracy in anatomy and clothing, less clutter | Fine-tuning, details, bug fixes.
• Clip Skip 1 | CFG 5.5–7 | Balanced between detail and stylized expression | General use, complex prompts.
• Clip Skip 2 | CFG 4–6 | Smoothed, sometimes dull, less expression | Stability on heavy models.
• Clip Skip 2 | CFG 3.5–7 | Strong style, more saturation or deformations | Artistic effects, exaggerated style.
```
PS: Try Clip Skip 1 + low CFG (4–6) if you're looking for precision.
```
```
PS: Use Clip Skip 2 + higher CFG (4–7) if you're looking for more style or artistic effects.
```
Recommended Upscaling:
• Scale: 1.5
• Steps: 20
• Upscaler: R-ESRGAN 4x+ Anime6B
• Denoising strength: 0.35 – 0.5

📌 Note: The VAE is already integrated; no need to load an external one.

🛡️ Safety Labels

This model may generate content of various types. Users are expected to correctly categorize its use into the following categories:

General / Sensitive / NSFW / Explicit

Please consciously add "NSFW" to negative labels if you wish to avoid explicit content.

🏷️ Prompt Tips

Positive Prompt:

sexy_perspective, (masterpiece), (best quality), (ultra-detailed), amazing quality, very aesthetic, illustration, perfect composition, intricate details, absurdres, incredibly absurdres, shiny skin, realistic,

Others More:

anime, cartoon, shiny skin, shiny, slutty, sexy, slut:2.1, dynamic angle, dynamic pose, dark_art_style, abstract element, abstract background, foreshortening, extreme perspective, high resolution, ultra-detailed, extremely detailed eyes, detailed face, newest, best illumination, depth of field, complex background, detailed background

Negative Prompt:

worst quality, old, early, low quality, lowres, signature, username, logo, bad hands, mutated hands, artist name, sketch,  , watermark, logo, patreon username, bad details

Others More:

bad quality,worst quality,worst detail,sketch, censor, patreon username, patreon logo, watermark, artist name, signature, logo, anatomical nonsense, bad anatomy, ugly, worst aesthetic, old, mismatched pupils, dirty face, extra limbs, bad hands, bad fingers, text, normal quality, low quality, web address, email address,

📦 UtopiaXL V1.5

This version reduces the strength of the base style and adds some details, it doesn't change much, it makes it easier to see some perspectives.

Things to keep in mind with this version:

Don't use this model with cutelatte, as this model includes 200% of that data, and that would break the entire model by forcing overfitting, even though it was reduced in version 1.5 (it makes it look unfinished like a scribble). Simply use the descender to force its appearance.
My idea with this is to include several of my loras so the next version will have a larger flow of data in terms of poses or concepts. I'm still not sure whether to use it as a base or clean up some of the folders, or change the base model by going back to Illustrious 0.1

⚙️ Recommended Settings:

Steps: 15–30
CFG Scale: 2.5–7 (4.5 or 5.5 recomended).
Sampler: Euler a, DPM++ 2M, etc.
Resolution: Larger than 1024x1024 to take advantage of its original resolution.
Recommended Upscaling:
• Scale: 1.5
• Steps: 20
• Upscaler: R-ESRGAN 4x+ Anime6B
• Denoising strength: 0.35 – 0.5

📌 Note: The VAE is already integrated; no need to load an external one.

🛡️ Safety Labels

This model may generate content of various types. Users are expected to correctly categorize its use into the following categories:

General / Sensitive / NSFW / Explicit

Please consciously add "NSFW" to negative labels if you wish to avoid explicit content.

🏷️ Prompt Tips

Positive Prompt:

anime, cartoon, shiny skin, shiny, masterpiece, best quality, amazing quality, very aesthetic, high resolution, ultra-detailed, extremely detailed eyes, detailed face, absurdres, newest, best illumination, depth of field, complex background, detailed background

Others More: slutty, sexy, slut:2.1, dynamic angle, dynamic pose, dark_art_style, abstract element, abstract background, foreshortening, extreme perspective,

Negative Prompt:

bad quality,worst quality,worst detail,sketch,censor, big nose, 
bad anatomy, ugly, worst aesthetic, old, mismatched pupils, dirty face, extra limbs, bad hands, bad fingers

📦 UtopiaXL V1.0

My baking method is a bit more involved than just picking models and merging them with Supermerge (although I do use Supermerge too).

I usually convert models into folder format and manually replace the UNet, while sorting or merging the text encoders—assisted by scripts, of course; I’m not a masochist.

Naturally, the easiest way to control the base style is by creating an external LoRA, baking it in heavily, and using it as a control layer… though I went a little overboard with this version and ended up overpowering some other elements. I’ll be fixing that in the next few days.

⚙️ Recommended Settings:

Steps: 15–30
CFG Scale: 2.5–7 (2.5, 4.5 or 5.5 recomended).
Sampler: Euler a, DPM++ 2M, etc.
Resolution: Larger than 1024x1024 to take advantage of its original resolution.
Recommended Upscaling:
• Scale: 1.5
• Steps: 20
• Upscaler: R-ESRGAN 4x+ Anime6B
• Denoising strength: 0.35 – 0.5

📌 Note: The VAE is already integrated; no need to load an external one.

🛡️ Safety Labels

This model may generate content of various types. Users are expected to correctly categorize its use into the following categories:

General / Sensitive / NSFW / Explicit

Please consciously add "NSFW" to negative labels if you wish to avoid explicit content.

🏷️ Prompt Tips

Positive Prompt:

anime, cartoon, shiny skin, shiny, masterpiece, best quality, amazing quality, very aesthetic, high resolution, ultra-detailed, extremely detailed eyes, detailed face, absurdres, newest, best illumination, depth of field, complex background, detailed background

Others More: slutty, sexy, slut:2.1, dynamic angle, dynamic pose

Negative Prompt:

bad quality,worst quality,worst detail,sketch,censor, big nose, 
bad anatomy, ugly, worst aesthetic, old, mismatched pupils, dirty face, extra limbs, bad hands, bad fingers

Thanks so much for your support! ♥

Files with SHA256: 809355522850b5543731428172f7350d97e4980ebef0ea14e1ea8b4dba794bf2