# waifudiffusion trinart aiイラスト stablediffusion oppai xhfanart xdinaryheroes xdinaryheroesfanart gaon overload ai画像 stabledifussion anything_v3 artistontwitter vtubers vtuber artistsconnect aiart aiartwork dreamstudio novelaidiffusion novelai novelaiaidiffusion

3. different settings
→There are several important settings for image generation that affect the generation, Step, Sampler, and scale are especially important. The settings used to generate the following images are described in ALT, so please try the same settings.

0 2
2022-11-14

ボツ絵の供養。

MODEL: sd-v1-5-fp16.ckpt + wd-v1-3-float16.ckpt(7:3) (https://t.co/YOBxW1dR2L)
Steps: 20, Sampler: Euler, CFG scale: 11
Seed: 1441089226, Face restoration: GFPGAN
Size: 1024x1408 Denoising strength: 0.2 First pass size: 0x0

0 2
2022-11-14

オラ、でかちち極めたゾ！

MODEL: sd-v1-5-fp16.ckpt + wd-v1-3-float16.ckpt (0.7:0.3)
Steps: 40, Sampler: Heun, CFG scale: 11
Seed: 3331985109, Face restoration: GFPGAN
Size: 1024x1408, Denoising strength: 0.4, First pass size: 0x0

0 1
2022-11-14

Gaon from the sampler bc I’m still not over the Alice in wonderland aesthetic

24 94
2022-11-14


anything_v3実験。ちょっと多めのstepでいくつかのサンプラーにgirlだけで出力。これがV3の女の子のベースなのかな。
しかし、安定度ぱねえな……

prompt:girl
negative prompt:
Steps: 47, Sampler: Euler a, CFG scale: 12

2 23
2022-11-14

どうぞ！
設定（Scale、Samplerなど）がわからなかったので、いろいろ入れてみました。

0 3
2022-11-13

ボツ絵の供養。

MODEL: v1-5-pruned-emaonly.ckpt
Steps: 100, Sampler: Euler, CFG scale: 11
Seed: 24805162, Face restoration: GFPGAN
Size: 1408x1024, Denoising strength: 0.6, First pass size: 704x512

2 6
2022-11-13


model:freckle_mix_0.85-dbmai_0.15（マージ）+kl-f8-anime2.vae
Sampler: DPM++ 2M Karras
手足Inpaint修正

18 148
2022-11-13

🛸 I now have a Carrd! Need to find me somewhere else? Here are literally all my links to all my socials, streaming channels and my Ko-fi/Fivver!! Here are some of my thumbnails as a little sampler🛸


🛸 https://t.co/nr08zxhO5z 🛸

2 1
2022-11-11

got official DPMSolver++ sampler working on Mac.
today, Cheng Lu added a trick to improve performance on <15 steps. probably k-diffusion doesn't have this yet.
dynthresh only works on pixel space; remains an unsolved problem on latents.
https://t.co/22rDBzDWiW

1 40
2022-11-06

with DPM-Solver++(2M) sampler, we get coherent images in 5 sampler steps!
and these aren't Heun steps (where n steps = 2n-1 model calls), this is just 5 model calls! less than 3.5 secs on Mac!
Katherine released this implementation yesterday in k-diffusion — great work as usual!

21 156
2022-11-06

"Automatic1111 added more samplers, so here's a creepy clown comparison"

0 17
2022-11-06

ScaleとSamplerの設定を定期的に試してみてね。同じプロンプトで結果が大きく変わることがあります。

( プロンプトはここです：https://t.co/txCW5oCsey )

　

0 3
2022-10-27

classifier-free guidance:
ask model to denoise gauss noise.
no condition: model predicts a salad.
shrine maiden condition: model predicts graffiti of faces.
CFG is "what makes shrine maiden different from salad", multiplied by your guidance scale.
repeat this every sampler step.

1 4
2022-10-26

WIP

ADAM SANDLERrr ---
Atom Sampler <3 @/Telta14

0 6
2022-10-23

Noxw ()'s new sampler album, NONE OTHER THAN, is now available on digital and data cdr!! link in the thread!!

10 10
2022-10-23

Samplerによって同じパラメータでも生成される画像はかなり変わるのね。それぞれk_euler_ancestral, k_euler, k_lms, ddim。Eulerってこれのこと？ 常微分方程式解いてるの？ まあ流石に昔すぎてもう忘れてるわ。https://t.co/w1FGgGxOAe

0 3
2022-10-22

I don't get how people get such clean results in anime style. All of mine look a bit weird. How much is it down to the model, sampler or prompt?

0 12
2022-10-21

Interesting experiment: After adjusting the sampler, step, and scale it gives us another style even use the same prompt.

0 7
2022-10-21