Sign In

CalicoMix

3.4k
28.4k
1.6m
389
Verified:
SafeTensor
Type
Checkpoint Merge
Stats
1,344
65,454
Reviews
Published
Jun 8, 2024
Base Model
SD 1.5
Usage Tips
Clip Skip: 2
Hash
AutoV2
0666CC88F7
default creator card background decoration
First Birthday Badge
Kybalico's Avatar
Kybalico

Calico is the tri-color blending cat, 🐾CalicoMix is dedicated to combining shaded anime with elements of both semi-realism and surrealism.

This model was initially intended to produce cute anime characters, but it has also been discovered that it has the capability to produce NSFW content.

What's new

CalicoMix 8.5 builds on a focus on details. It sacrifices some hand quality in favor of improving overall quality.

CalicoMix 7.5 builds on the cuteness of the original series, but with a softer and more stylized approach. It traded some background details for character details.

**Please note that you are solely responsible for any images created using this model.

🐾CalicoMix is officially hosted for online generation on:
- TensorArt

Recommend: (masterpiece),

Negatives: (low quality, worst quality:1.2),

CFG Scale: 6-8
*Clip skip: 2

Sampler: DPM++ SDE Karras, DPM++ 2M SDE Karras

Sampling steps: 20+, 30 (recommended)

Resolution:
image compose change by resolution
384x512 (available, but may result in less detail)
512x768 (recommended with hi.res fix)
768x1152 or above (recommended, more details)
For landscape images, reverse the width and height values accordingly.

Hires step: from half to equal of the sampling steps. 12-30 (recommended)
Hires.fix: R-ESRGAN 4x+Anime6b, 4x-UltraSharp, 4x-AnimeSharp

Denoising strength: 0.4recommended

VAE:

Version 1: No VAE, recommend to use with VAE.
Version 2-6: Baked (vae-ft-mse-840000-ema-pruned)
Version 7.5+: Baked CalicoVAE

Change log:

Version 8.5:

Enhanced aesthetics, contrast, and details.
NegativeTI is no longer required, but it remains compatible

Version 7.5:

Restores and improves the cuteness, soften style.

Version 6.0 (Gen 2):
Major improvements, Improve versatility, Change style, Improve LoRA support.
Improved background, details, outfits.

Version 5.0:
Improved background, details, outfits, clothes wrinkles

Version 4.0 H:
major improved Hentai.
can draw male

Version 3.0:
more teen look and improved H-positions.
Enhanced hand and fingers, background, details, and color.

Version 2.6:
improved complexity shading, enhanced details, and a better composition. However, the finger shape quality is slightly lower compared to version 2.5.

Version 2.5:
fine tuning

Version 2.0:
improved background and enhanced details, modified style

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

I merged models and created example images using my poor laptop with a GeForce GTX 1650 (4GB memory).


If you like this model, you can support me on Ko-fi

Your support will help me afford new equipment that will allow me to create better models.

If there's anything you need, please don't hesitate to let me know.

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Special thanks to: