UnrealSpice Generation Guide: click here

*not much info yet, to be updated further with part 2...

Update 26/02/2026

IL v3 is an attempt to upgrade lighting, texture and background & attempt to shift from 2.5D realism toward more 3D realism.

Update 15/11/2025

IL v2 is a slight upgrade to IL v1, it is my 1st attempt on using supermerger, which open up much more possibilities but also time consuming. Main fixes primarily focus on improving realism, lighting and composition. Overall my own experience is an improvement. But occasionally will get one bad image (i.e. bad white spot in mouth part, one additional finger that is badly shaped and out of place), this issue can occurs in 1 out of maybe 10 image i guess, it is quite random and i am too lazy to fix it due to the amount of time spent on v2.

Summary of my experience on all version:

realism: ILv3 > ILv2 > IL v1 = Pony v6.1

prompting: ILv3 = IL v2 = IL v1 > Pony v6.1

aesthetic: ILv3 > ILv2 > IL v1 = Pony v6.1

I am just starting to use Illustrious checkpoints, so I am still not able to utilize Illustrious potential fully. Overall, i would treat it as a slight upgrade to Unreal Spice Pony v6.1 as it open up more potential in terms of concept.

A model i merged using various checkpoints in civitai. The main purpose of this checkpoint is for my own purpose of achieving unreal nsfw content in realistic setting.

Notes:

1. adetailer is highly recommended to be used in conjunction with this checkpoint, especially for full body image as face generation will be very bad without it.

2. I use Reforge WebUI with various settings enabled (adetailer, FreeU Integrated, "APG's now your CFG" for reForge).

3. Image showcased attached with the model itself without any shown settings (except adetailer), but you might not generate the same image in civitai as there is possibility of difference due to webui (same applies to Forge / A1111/ ComfyUI webui)

4. For lora and image enhanced with settings, may refer to image showcase at bottom of page.

5. I won't be updating often on the checkpoint, i only update if I feel i can achieve greater results with more merging. (I have zero knowledge in merging model and this is purely out of luck)

My Generation Setting

Super Fast Generation (with DMD2 speed lora in CivitAI)

Samplers: LCM Exponential

Steps: 4-8

CFG Guidance: 1 - 2

Resolution: 1024 x 1024 (square) / 832 x 1280 (portrait) / 1280 x 832 (landscape)

Standard Generation

Samplers: SDE2++ Karras / Euler A Simple

Steps: 20-40

CFG Guidance: 5-7

Resolution: 1024 x 1024 (square) / 832 x 1280 (portrait) / 1280 x 832 (landscape)

Standard prompt to get started:

Positive: realistic,absurdres,cinematic,masterpiece, intricate detail,8k,uhd,

Negative: monochrome,grayscale,overexposure,overbrightness,blur,blurry,out of focus,cropped image,bad proportion,text,watermark,signature,username,bad hand,too many fingers,missing limb,lazy eyes,bad eyes,blanked eyes,tan line,tanned body,skin pores,imperfect skin,skin spot,

*Might upload a detail guide on generating when i have the mood to do it.

Lastly, thanks to all the civitai creator for consistently creating and sharing your works. Everyone feel free to share your own creation for others as reference of actual generation of this checkpoints.

Feel free to tip me some buzz if you are enjoying this checkpoint.