Sampling different models - Example results to compare

[Kinomora]

I decided to download a bunch of the different models offered on https://rentry.org/sdmodels and test the quality of the results. I used a fairly limited set of tags to prevent running into too many issues with tags not being "well understood" and generating meaningless results.
The settings I used are as follows:
Steps: 26
Size 512x512
CFG scale: 11
X/Y Plot, x = seed, y = none
Set seeds: 1,2,3,4,5,6,7,8,9
Stop At last layers of CLIP model: 2

I used 2 similar sets of tags and generated 2 image batches to see the differences they produced.
Set 1_a used the following prompts:
highest quality, masterpiece, maid, thighhighs, elbow gloves, choker, long hair, medium breasts, cleavage, skirt, girl
Set 1_b used the following prompts:
highest quality, masterpiece, maid, thighhighs, elbow gloves, choker, long hair, skirt, crossdresser, boy

Additionally, both sets used the following negative prompts:
lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry, artist name, young, child, loli, shota, obese, fat, large breasts, huge breasts, inflation, pregnant

Each model will be listed by the model hash, then the provided name, then the test ID.
Please be aware that as many of these models are constantly changing, their model hash may change which could alter the results. I'll do my best to keep this updated with new releases and refinements of different models but I cant promise to keep on top of things. This is meant to be a rough guide to help determine which models produce what you may be looking for.

STABLE DIFFUSION
[81761151] v1-5-pruned-emaonly.ckpt
set_1_a

set_1_b

[7460a6fa] sd-v1-4.ckpt
set_1_a

set_1_b

[06c50424] sd-v1-4-full-ema.ckpt
set_1_a

set_1_b

WAIFU DIFFUSION
[45dee52b] wd-v1-2-full-ema.ckpt
set_1_a

set_1_b

[84692140] wd-v1-3-float16.ckpt
set_1_a

set_1_b

HENTAI DIFFUSION
[e2ec4647] HD-16.ckpt
set_1_a

set_1_b

LEWD DIFFUSION
[f4030c43] LD-70k-2e-pruned.ckpt
set_1_a

set_1_b

YIFFY
[50ad914b] yiffy_e-18.ckpt
set_1_a

set_1_b

FURRY
[323f8dd8] furry_epoch4.ckpt
set_1_a

set_1_b

R34
[6e45cf2c] r34_e4.ckpt
set_1_a

set_1_b

NOVELAI
[925997e9] nai.ckpt
set_1_a

set_1_b

FOR LIABILITY REASONS. NO I WILL NOT SEND YOU THE MODEL.CKPT FOR ANY PARTICULAR MODEL. I MAY OR MAY NOT HAVE THE MODEL. I CAN ONLY POINT YOU TO WHERE YOU MAY BE ABLE TO FIND THE MODEL.

[Kinomora]

If you would like me to download and test a particular model, or have tested one with the exact same configuration I have used (and confirmed it on an existing post) please let me know and I will be happy to include it.

[grexzen]

This is awesome, thank you. I was wondering how the different models would do, since I am still working only with 1.4.

[enn-nafnlaus]

Next time, can you test with something other than softcore porn? :Þ

[Kinomora]

people use this for things other than porn? :p
Jokes aside, I could run another set of tests with different tags if you'd like to offer a suggestion :)

[Manimap]

Makes me actually curious on what improvement did SD1.5 bring over SD1.4.

[briancullinan2]

I've been running test on Juggernaut that blow these out of the water. I tried to imagine the first 16 pages of Neuromancer here on the "scenes" tab:
https://docs.google.com/spreadsheets/d/19lHurXLcmpfyWlmmJJWhRMqSAyIh9yfjrQUbnP7MG20/edit?usp=sharing
And because I have no imagination or soul, I made an x-rated version in the column next to it.

[FilipeF12]

@Kinomora Cool! TY