Promptography is the art of making photographs with a sequence of texts by utilizing an artificial intelligence as a medium. This is how photographer alpha7 came into being, very good at generating realistic portraits. you can make also others composition too

Recomendations:

Sampler method:

DPM++ SDE Karras or DPM++ 2M Karras

Steps:

20 - 50

CFG scale:

4.5 - 10

6.5 is recommended

Resolution:

Square format: 512 x 512

Portrait format: 512 x 768

Hires. Fix:

Scale: 1.5 - 2

Denoising strength: 0.35

Upscaler: 4x_NMKD-Superscale_178000_G

Tips for a better Results:

Textual inversion:

PA7_Portrait-HL: half length portrait

PA7_Portrait-MS

PA7_Portrait-FL

PA7_Portrait-CU

PA7_Portrait-ECU

PA7_Portrait-MCU

PA7_UnRealistic-Neg_v2

PA7_Photo

PA7_Photo-BW

PA7_Human-Likeness

PA7_Food-Photo

Extensions:

adetailer

model keyword

next-diffusion

tile VAE

ultimate upscale SD

Recomendations for performance improvement:

Go to settings-optimization

Token merging ratio to 0.5

If you have a GPU with low VRAM, use for:

2gb-4gb vram:

--lowdvram command line arguments

6gb-8gb vram:

--medvram on command line arguments

(edit webui-user.bat on Windows).

Position prompt:

avoid this:

"masterpiece, photorealistic, 4k, 8k, super realistic, realism" etc. it doesn't make any significant improvement

start your prompt with type of photography like candid photography, street photography, etc. or use PA7 textual inversion

Negative prompt:

PA7_UnrRealistic-Neg_v2

Big thanks to

SD 1.5

Photon

Cyber realistic

RealisticFusion

Juggernaut

Realistic Vision

epiCRealism

Analog Madness

and all of you who are making stable di

ffusion the best one.

for more exemples go to Facebook page

Artofdimona.

I hope you enjoy this checkpoint and

Let's PA7 give to you the masterpiece you want