qchan realistic Mix

There are currently 5 models (27 versions) in the qchan Mix series, including derivatives.

"qchan realistic Mix" is for semirealistic to realistic type images.

v1.1, v1.0 : Realistic Asian style

v2.2, v2.0 : Realistic Western style

v3.0 : Elegant realistic Asian style

The top images were generated using the same prompts and conditions so that you can easily compare these models.

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

v3.0

I really like the elegace of Kakarot series by vay_kakarot, so I tried merging them with my realistic models together with other good realistic models.

The models used in the merge are as follows (VAE is baked):

Kakarot 2.8D Color https://civitai.com/models/201673?modelVersionId=217212

epicrealism_pureEvolutionV5 https://civitai.com/models/25694?modelVersionId=134065

qchan realistic Mix v1.1 https://civitai.com/models/258175?modelVersionId=328241

qchan realistic Mix v2.2 https://civitai.com/models/258175?modelVersionId=452814

yayoi_mix v2.5 https://civitai.com/models/83096?modelVersionId=178711

Kakarot 2.8D V2024 https://civitai.com/models/182723?modelVersionId=458684

(The ratio of Kakarot 2.8D Color is highest of the merged models, but the composition was adjusted so as to keep consistency with other qchan realistic models.)

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

v2.2

The reality of v2.0 has been improved by adding nice realistic models.

The models used in the merge are as follows (VAE is baked):

qchan realistic Mix v2.0 https://civitai.com/models/258175?modelVersionId=294668

The Truality Engine TRUALITY ENGINE (PRO) https://civitai.com/models/158621?modelVersionId=288888

sborka random v3.0 https://civitai.com/models/183568?modelVersionId=319852

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

v2.0

v1.0 tends to generate Asian faces, so I decided to merge a more Western style model.

The models used in the merging are as follows (VAE is baked):

qchan Mix v1.0 https://civitai.com/models/255778?modelVersionId=288338

epicrealism_pureEvolutionV5 https://civitai.com/models/25694?modelVersionId=134065

cyberrealistic_v41BackToBasics https://civitai.com/models/15003?modelVersionId=256915

kawaiiRealisticEuropean_v05 https://civitai.com/models/90694?modelVersionId=262227

kawaiiRealisticAsian_v03 https://civitai.com/models/83766?modelVersionId=100925

(kawaiiRealisticAsian_v03 was merged with minus weight, since it is included in the base qchan Mix)

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

v1.1

The reality of v1.0 has been improved by adding nice realistic models.

The models used in the merge are as follows (VAE is baked):

qchan Mix v1.0 https://civitai.com/models/255778?modelVersionId=288338

Brav5 https://civitai.com/models/25494?modelVersionId=63786

majicMIX realistic v7 https://civitai.com/models/43331?modelVersionId=176425

yayoi_mix v2.5 https://civitai.com/models/83096?modelVersionId=178711

VoidnoiseCore R1311 https://civitai.com/models/131565?modelVersionId=166019

chilled_remix_v2 https://huggingface.co/sazyou-roukaku/chilled_remix/tree/main

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

v1.0

This is my third merge model. I merged some good realistic models to my first model, qchan Mix.

The models used are as follows (VAE is baked):

qchan Mix v1.0 https://civitai.com/models/255778/qchan-mix

yayoi_mix v2.5 https://civitai.com/models/83096?modelVersionId=178711

VoidnoiseCore R1311 https://civitai.com/models/131565?modelVersionId=166019

The new model is also highly versatile and can generate a variety of images.

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

I usually enjoy random image creation using the "OneButtonPrompt" extension with the following conditions. (My PC isn't high spec so I haven't tried higher resolutions yet):

Sampler: Euler A

Steps: 40

Width x Height: 512 x 768

CFG scale: 7

Denoising strength: 0.3

Clip skip: 1

Hires upscaler: 4x-UltraSharp

Hires upscale: 2

Hires steps: 10

Denoising strength: 0.3

OneButtonPrompt strength: 10

Common prompt (prefix): 1girl

Base negative prompt: bad-hands-5, FastNegativeV2

No auto generate negative

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

For negative prompts, bad-hands-5 and FastNegativeV2 seem to work with this model.

bad-hands-5 https://civitai.com/models/116230?modelVersionId=125849

FastNegativeV2 https://civitai.com/models/71961?modelVersionId=94057