From Anime Beauties to 2.5D Women: Creating Diverse Art with Stable-Diffusion – The World of darkseal

This site uses affiliate links.

Beyond expectations: darkseal’s generative art

darkseal’s journey as a free-spirited artist

Merry Christmas! I’m kojirom, the loner.

In my previous post, I introduced the “The Ani-Loop” model. I was so impressed with it that I decided to download all of darkseal’s works and give them a try.

In this post, I’ll share my experience and results with you.

[Author = darkseal]

DarkFruit Mix – DarkBerry_
DarkFruit Mix – DarkBerry_ | Stable Diffusion Checkpoint | Civitai

DarkGoblin – DarkGoblin
DarkGoblin – DarkGoblin | Stable Diffusion Checkpoint | Civitai

DarkToon – DarkToon
DarkToon – DarkToon | Stable Diffusion Checkpoint | Civitai

Gag Harness – GHV7
Gag Harness – GHV7 | Stable Diffusion Checkpoint | Civitai

Half and Half – v1.0
Half and Half – v1.0 | Stable Diffusion Checkpoint | Civitai

Homosapian – Homosapian
Homosapian – Homosapian | Stable Diffusion Checkpoint | Civitai

MasterSeal – v1.0
MasterSeal – v1.0 | Stable Diffusion Checkpoint | Civitai

Pinocchio – Pinocchio
Pinocchio – Pinocchio | Stable Diffusion Checkpoint | Civitai

Pinocchio 1.6 – Pinnochio 1.6
Pinocchio 1.6 – Pinnochio 1.6 | Stable Diffusion Checkpoint | Civitai

The Ani-Loop – 1.5
The Ani-Loop – 1.5 | Stable Diffusion Checkpoint | Civitai

The Loop – The Loop 1.1
The Loop – The Loop 1.1 | Stable Diffusion Checkpoint | Civitai

Well, what kind of images should I generate today? Since it’s December 25th, “Christmas” is the obvious theme!

For prompts, I chose “1girl”, “Christmas”, and “wearing a swimsuit” and rotated through them.

My usual negative prompts were not enough this time. The models’ depictions can be quite risque at times, so additional restrictions were necessary. I filled the negative prompt box with as many adult-blocking spells as I could, but some of these naughty girls just wouldn’t listen!

But honestly, that daringness is part of their charm too.
So for today, I have fewer images than usual to share, but please enjoy them slowly.

DarkFruit Mix

Support Darkseal by buying cat food at For thicker lines, try using a Hypernetwork like SamDoes at about 20-30% (see models 2657/sdaa3 and 1658/incasea3). For VAE, I often use 840000 ema pruned, though its impact on this anime model is uncertain (more info at Explore more of my work as Darkseal or DarksealStudios on the web, including at

(Site Summary)

〈This is an advertisement〉


Support Darkseal by buying cat food at The DarkBerry Mix and GobGirlz combination, weighted at 0.3, creates the DarkGoblin. No hypernetwork was used for the preview images. See examples of this and other models at

(Site Summary)


Support darkseal at The DarkFruit model is merged with mistoonAnimeV10 for style, but still has issues with duplication and body distortion. Corrections are currently managed by loras and locons without direct fixes to body parts. Good luck to everyone!

(Site Summary)


Gag Harness

※ I decided that it would be inappropriate to post the generated images here. Please infer from the model name.

Half and Half

Support me by visiting my DeviantArt shop: The ‘Half and Half’ model is based on model.ckpt, enhanced with custom-made loras at a low rate, aiming for ‘good hands’. It transitions towards a photorealistic model, derived from a nai base. Enjoy and support me using the link above!

(Site Summary)

〈This is an advertisement〉


Support darkseal by buying cat food at The ‘Homosapian’ model is a mix of models that align with my vision, with a touch of anime. Use embeddings, Ti, Hypernetworks, and Loras to tailor your style. The model helps in conjuring the items and composition you need. Support me on Gumroad at and see more at

(Site Summary)


MasterSeal is a model created from the base cartoon “model” and fed loras created by 2 people using the lowest learning rates around to produce an insane model. Open the masterseal and dive into anime madness.

(Quote from a website)


  • Pinocchio model mixes various models, used supermerger to increase photorealism and improve head/hands
  • Continuation of Homosapian model which has more 2D anime style
  • If images look bad, lower CFG scale to 5 or less
  • No Dream mixer or hires fix used in most samples
  • Embeds used sometimes for light/mood
  • General use model focused on photorealism, but can get other styles by adding prompts like illustration, 3D render etc.
(Site Summary)

Pinocchio 1.6

Note: For more information on this content, please visit the following link. Due to copyright concerns, we have refrained from publishing the image, but we respect the author’s efforts and talent. Thank you for your understanding.

The Ani-Loop

This amazing model was featured in my previous article, which can be found at the link below.

The Loop

Welcome to the Loop. You want to be in the loop. You Need to be In The Loop!

(Quote from a website)

Revealing the Configuration Parameters of Stable Diffusion Web UI

Steps: 80, Sampler: DPM++ SDE Karras, CFG scale: 5, Seed: 207331821, Size: 512×512, Model hash: 4d99c5679c, Model: theAniLoop_15, VAE hash: 0b204ad0ca, VAE:, Denoising strength: 0.55, Clip skip: 2, Hires upscale: 1.5, Hires steps: 40, Hires upscaler: R-ESRGAN 4x+ Anime6B, Version: v1.7.0

All Illustrations Are Commercially Usable

Some illustrations may have incomplete details such as hands and fingers. We kindly ask for your understanding.

When using these images, please feel free to crop or modify them as needed.

We encourage you to fully express your creativity and use these illustrations with a free and imaginative approach.

The illustrations featured on this site are available for commercial use.

Related Articles

Illustrated by DALL-E 3