Sign In

GDP Illustrious

72
696
22
22
Updated: Mar 3, 2025
base model
Verified:
SafeTensor
Type
Checkpoint Merge
Stats
386
22
Reviews
Published
Mar 3, 2025
Base Model
Illustrious
Training
Steps: 452
Epochs: 10
Usage Tips
Clip Skip: 2
Hash
AutoV2
44C699A89E
default creator card background decoration
JO
joe78

Description

I enjoy making images that are more illustrated/comic style, more realistic than standard Illustrious or Pony, but less realistic than the "realistic" models.

My Pony model GDP_PonyXL attempted to do all I was trying to do, and I still enjoy using it. This is an attempt to achieve similar results, but with the better prompt adherence and detail of Illustrious.

So, I took three absolutely amazing Pony Illustrious models and merged them (which was way easier than I'd thought it was, seriously, I learned the basics in like 10 minutes on YouTube).

Ingredients

This is a merge of several amazing Illustrious models:

The primary model is an excellent realistic model, NovaRealityXL. This model produces fantastic images, please check it out!

Nova was then merged with the awesome WAI-NSFW_Illustrious. I love this checkpoint!

The resultant checkpoint was then merged with another fantastic model that is more 2.5D, Illustrij.

I highly recommend consulting the descriptions on the three above models to get the best results.

Parameters

Sampler: Euler A

Steps: 25 - 30

CFG: 5 - 7

VAE: sdxl_vae

Clip Skip: 2

Prompting

In the positive prompt, I almost always include some quality tags that I borrowed from this excellent article by @Cicatriz. If you haven't checked it out, you should!

masterpiece, best quality, absurdres, amazing quality, very aesthetic, newest, perfect quality, absolutely eye-catching, ambient occlusion, raytracing, subject focus, character concept art,

I've noticed Illustrious is more sensitive to negative prompting, so this might be overkill. I usually use the following in the negative prompt:

child, kid, childish, childlike, loli, asian, lowres, (bad), bad anatomy, bad hands, extra digits, too many fingers, multiple views, fewer, extra, missing, text, error, worst quality, jpeg artifacts, low quality, watermark, unfinished, displeasing, oldest, early, chromatic aberration, signature, artistic error, username, scan, 

An example prompt:

masterpiece, best quality, absurdres, amazing quality, very aesthetic, newest, perfect quality, absolutely eye-catching, ambient occlusion, raytracing, subject focus, 1girl, depth of field,
quality highlights, 8k detail, detailed background, dynamic view, ultra-detailed, 
sexy adult woman, character concept art,  
beautiful, sexy, skinny, 25 years old,  
posing, seductive, highly detailed face, 
full body, standing, from below, from behind, twisted torso, bent over, knees together feet apart, looking back, short blue latex skirt, tight skirt, skin tight, black highheels, black latex sleeveless shirt, cross-laced thighhighs, center cross lace latex top, g-string, ass support, 25 years old, sexy, skinny, fair skin, glossy hair, {black|blonde|brown|gray|white|red|auburn|blue} hair, {brown|blue|grey|green} eyes, high ponytail, hoop earrings, 
very narrow waist, 
dramatic lighting, cinematic, perfect shading, concept art, dark background, high contrast,

Negative prompt:
child, kid, childish, childlike, loli, asian, lowres, (bad), bad anatomy, bad hands, extra digits, too many fingers, multiple views, fewer, extra, missing, text, error, worst quality, jpeg artifacts, low quality, watermark, unfinished, displeasing, oldest, early, chromatic aberration, signature, artistic error, username, scan,