stable-diffusion / AI - PART TWO

[maxhitman]

Originally Posted by Pixeldelic

Hello guys! I'll begin today with SD and I've read both threads here, thank you very much! I have a silly question, is this the 1.5 version? http://github.com/AUTOMATIC1111/stable-diffusion-webui

I followed Scott Detweiler Instructions how to install the
Stable Diffusion version 1.5 and it worked very good for me.

See this video and follow his instructions tutorial. (LINK BELOW)

HOW TO INSTALL S.D.

Stable Diffusion 1.5 - Windows Installation Guide [Tutorial]
- Check the links below video for ALL the necessary downloads
you will need to install SD into your Home Computer.
- Below I also posted for you the file names you must download, so you will know.

Follow his VIDEO instructions, word for word, and you will have your
personal A.I. Stable-Diffusion working at home.
It should take you about 30 minutes, perhaps a little more, or less.
DO NOT do it fast, so you do not make mistakes.

http://youtube.com/watch?v=ycQJDJ-qNI8

You only need 4 things to install to get started with your personal A.I.
What ever you install later, it is your choice. If you followed the instructions
and wrote that word "git pull" - every time you start your S.D. 1.5 your
program will get automatic upgrades and updates and your menu will have
new settings and goodies in it.

Install Python from here
http://python.org/
Use Python 3.10.6! Any other version might cause issues.

the file should say... python-3.10.6-amd64
-----------------------------------------------------------------------

Install GIT from here
http://gitforwindows.org

the file should say... Git-2.39.1-64-bit
-----------------------------------------------------------------------

Install this version of the Stable Diffusion UI
http://github.com/AUTOMATIC1111/stable-diffusion-webui
Stable Diffusion v1-5 Model Card

PLEASE NOTE, THIS IS THE ONLY OFFICIAL PLACE TO DOWNLOAD THE CHECKPOINT!
Download the models directly from hugging face:
http://huggingface.co/runwayml/stable-diffusion-v1-5

the file should say... v1-5-pruned-emaonly.ckpt (4.27 GB)

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

After you install everything RE-START your computer and wait a
little bit (4 minutes) for the PC to read all your new files properly.
Then you start your New S.D. program.
I also suggest you use OPERA browser instead of the Google-Chrome browser
to place the new LINK to start your Stable Difussion 1.5 Menu

[maxhitman]

She was a sexy Queen

Elizabeth II
A.I. created fake - Stable Diffusion v.1.5
Lora model by grtmate
-
25+ year old face pictures version
-


-

-

-

-

-

-

[SynchroDub]

Margot Robbie

V1



V2

[SynchroDub]

Monica Bellucci (sort of)

[Talionic]

Originally Posted by maxhitman

Some people ask me,
how many celebrity faces are inside Stable Diffusion ? , and can it create them all?
It does create some of them VERY GOOD, while some other people you may need to
keep prompting your A.I. to create a few more images, until the A.I. starts to create
their faces very good. What I mean by this is, keep using the same prompt asking to
see a particular known face and create over 50+ images. Usually after the 40-50
images created the A.I. will begin making the faces very well.

So,
HOW MANY Famous Faces are there inside Stable-Diffusion ?

Here is the complete list...
Each Page has 100 names ... and there are 1,996 Celebrities to read all the names
In the pages you can also see How many images of each person are used for A.I. training.

http://laion-aesthetic.datasette.io/laion-aesthetic-6pls/celebrities

I've been following some of your advice in this thread and it's really taken my celeb nudes to another level. I used your prompts and trained my custom Gillian Anderson model over Realistic Vision and it came out great. Here's an example:

[maxhitman]

Originally Posted by Talionic

I've been following some of your advice in this thread and it's really taken my celeb nudes to another level. I used your prompts and trained my custom Gillian Anderson model over Realistic Vision and it came out great.

She looks wonderful and very sexy in those pictures !
Congratulations!
I loved to see that old series "X-Files" and even bought the complete episodes from season 1 and season 2 on DVD.
I have not done any A.I. pictures with her yet. She is still on my To-Do list.

Keep in mind that the Prompt I wrote is very helpful for creating
certain Photo-Realistic style pictures ONLY.

You can add or change some words in it, to Fit the person or the
background you would like to see.
But
try to keep the way the prompt is written because of the important
places certain words are placed in the prompt.
It gives the words more importance so your A.I. can create a better
looking realistic image.
Each section of the prompt-script describes in detail to the A.I. what to
create exactly what you want to see.
Each section is separated by a comma.

If you were to write the same selection of words but place them
in different areas of the prompt-script, you will get a totally different result
which you may not like. It will not have the same effect.

Recently an A.I. picture creator also discovered this way to write a good
prompt script and made a little book about their discoveries, which they
put up on Civitai.
The style that he wrote their prompt is very close to my style, except
they are leaving certain parts out, which in my prompts, I usually write
and include even more words in the script.

You do NOT need any extra LORA or T.I,. for your A.I. to make these extra things,
because the information is already inside your A.I. for it to make it into your image.
Things such as eye color, hair color and hair styles, breasts sizes, body sizes and types, all types of clothing, different lighting techniques for indoor photoshoots, outdoor lighting, Weather conditions and sun placements, seasons of the year, camera types and camera lens, and so much more.
You can just write it into the prompt and your A.I. will create it.

I will leave you this link for you to download the Book from Civitai...

Creating Photorealistic Images With AI: Using Stable Diffusion
by PromptGeek - Uploaded Jul 29, 2023

http://civitai.com/models/118213

It is a (75.04 MB) PDF book with about 180 pages - easy to read and has
alot of pictures as examples.

He even made a little video about it. So you can get an idea how
this works.

You Tube

You Tube

Important NOTES - in the video he describes how he uses some Lora or a
pre-trained model for making "Better eyes " and "Skin", but I suggest to test them first,
because they can have a bad effect on your image if you are creating a Famous
celebrity person and it may distort the faces a little. It is not a 100% full-proof.
These are the ones he mentions ...

http://civitai.com/user/Polyhedron_AI/models

He also uses some model for "Negative" prompts, often called "Negative Embeddings",
which I suggest for you to test them on certain images, but not use
them constantly.
These can also have a very negative effect on your final photo-Realistic
type of pictures created by the A.I. - They are not 100% full-proof.

I personally just use my own style of already written Negative Prompt script and use it on all my images.
It says Exactly what I do NOT want to see the A.I. to create.

The reason of some problems using these "negative embeddings" is
because some of these negative Models may contain information about certain effects you would like to see in your images,
but if the information is already inside these pre-trained Negative-models, your A.I. will not create them for you.

I have NO IDEA what these negative embeddings have inside them
which will be useful for me.

Imagine if the Negative-Model has written the word "Matte" in it, but you want your final image to have a "matte look" for a more artistic style
of photograph?
Your A.I. will not create what you want to see.

Imagine again if for some reason, you are using a LORA or a T.I. of a person.
But the images of that person has very shiny skin, almost "wet like"?
Everything is looking shiny and not good.
What to do?
You would write , "(realistic matte skin:1.1)," , or, "a Photo with a
Matte Texture" into your prompt.
But the A.I. refuses to obey your orders, because inside the negative embedding has the word "Matte" !

I have no idea what words the negative-Model has inside it, so I do not
use them.

[Talionic]

Originally Posted by maxhitman

I will leave you this link for you to download the Book from Civitai...

Creating Photorealistic Images With AI: Using Stable Diffusion
by PromptGeek - Uploaded Jul 29, 2023

http://civitai.com/models/118213

It is a PDF book with about 180 pages - easy to read and has
alot of pictures as examples.

They even made a little video about it. So you can get an idea how
this works.

http://youtube.com/watch?v=z-AoELaJfn0&t=309s

Thanks, I'll check it out

[maxhitman]

Originally Posted by Talionic

Thanks, I'll check it out

Yes, do download it and use it. You will like it.

The prompt is very similar to the way I write my version, but is more
simplified and easier.

My way is much more complicated for beginners.
Back in January I was already doing many tests and collecting certain
words which acted like "trigger words" for the A.I. to better understand
what to create.
I started to collect a big list of special words and testing them.
Testing their places in the prompt-script and what worked best for me.
So far, my book is over 300+ pages long and I keep discovering new
and cool things about this A.I.

One thing that really amazes me, is the amount of places and locations
the A.I. is able to create as backgrounds for my sexy female models.
All the information is already stored inside the A.I. "brain" and you do NOT
need any extra Lora or T.I.to do this.

Do some tests for your images... Write in Google Search
" Best vacation spots in the word "
or
" Best photography places in the world "

Pick out any website which has some nice looking places and locations.
Just copy the name of the place and put it as your "Background" in your prompt-script.
In the section before you write the location, write how the weather is like.
For example, if its a "Sunrise", or "Sunny", or Sunset, or foggy, or a
Thunderstorm Clouds with rain type of weather.

Here is an example from one of my prompts, you can test...

a Realistic Full Body Portrait Photo of actress ANGELINA JOLIE with a long Hair Style, Beautiful, Detailed face, Perfect Eyes, (Eye Makeup:1.1), (highly detailed skin:1.1), (perfect breasts:1.1), (perfect nipples:1.2), Perfect Slim body, NSFW, Nude, Pussy, Modelshoot style, Professional Photography, Soft Lighting, Photorealistic, stunning Sunset, Foggy, standing in a grassy plains of ICELAND background, blurred background, volumetric fog, RAW, analog style, sharp focus, 8k, high res, DSLR, high quality, Fujifilm XT3, film grain, award winning, masterpiece,


Then include one of these words as your background .... (as examples)

in a Fields of tulips in Keukenhof background,
in a Lavender fields in Provence France background,
in a Lavender fields in Fuji Shibazakura Japan background,
in a Shibazakura flower fields background,
in a Miracle Gardens in Dubai background,
in a Sunflower fields in Tuscany background,
in a Poppy fields in Zamora Spain background,
in a Canola fields in China background,
in a Oxbow Bend, Grand Teton National Park Wyoming background,
in a Denali National Park Alaska background,
in a Central Tower Of Paine Chile background,
in a Torres del Paine National Park Chile background,
in a Glacier National Park Montana background,
in a Chincoteague National Wildlife Refuge Virginia background,
in a Baja Peninsula Mexico background,
in a Yosemite Valley California background,
in a Blackwater National Wildlife Refuge Maryland background,
in a Canadian Rockies background,
in a Grand Canyon National Park Arizona background,
in a San Juan Mountains Colorado background,
in a North Cascades National Park Washington State background,
in a Cuyahoga Valley National Park Ohio background,
in a The Highlands Of West Virginia background,
in a Capitol Reef National Park Utah background,
in a Acadia National Park Maine background,
in a Lake Mead Recreation Area Nevada background,
in a Lake Tahoe California background,
in a Mono Lake California background,
in a Bandhavgarh National Park India background,
in a Big Bend National Park Texas background,
in a Yellowstone National Park Wyoming background,
in a Playa del Carmen, Mexico background,
in a Machu Picchu Peru background,
in a Pyramids of Giza Egypt background,
in a Salar de Uyuni Salt Flats Bolivia background,
in a Antelope Canyon Arizona background,
in a Mauna Kea Hawaii background,
in a Lake Louise Canada background,
in a Alps of Switzerland background,
in a Swiss Alps background,
in a Swiss mountains background,
in a Amalfi Coast in Italy background,
in a Everglades National Park of Florida background,
in a Taj Mahal India background,
in a amib-Naukluft National Park Namibia background,
in a Milford Sound New Zealand background,
in a Serengeti National Park Tanzania background,
in a Chefchaouen Morocco background,
in a Rice Terraces of Yuanyuang China background,
in a Cinque Terre Italy background,
in a Granada Spain background,
in a Arashiyama Bamboo Forest Japan background,
in a Tuscan Countryside Italy background,
in a Santorini Greece background,
in a Venice Italy background,
in a Iguazu Falls in Argentina Brazil background,
in a Santorini Greece background,
in a Tekapo New Zealand background,
in a Angkor Wat Cambodia background,
in a Petra Jordan background,
in a Tuscany Italy background,
in a Bromo Tengger Semeru National Park Indonesia background,
in a Marina Bay Singapore background,
in a Cherry blossoms park Japan background,
in a Valensole Plateau France background,
in a Bagan Myanmar background,
in a Stonehenge UK background,
in a Thailand exotic beach background,
in a Ribeira da Janela Stacks Madeira Portugal,
in a Fog Forest Fanal, Madeira Portugal,

[Johnny Cage]

Another dumb little thing I did to amuse myself, asked SD for young and attractive versions of the current and some previous first ladies.

Jill Biden



Michelle Obama



Laura Bush



Hillary Clinton

[Johnny Cage]

Bonus one, not a FLOTUS but Nancy Pelosi:



Would have been so funny if this one came out looking like Ali Landry.

If you aren't familiar, an early internet urban legend was people posting a picture of Ali Landry (the Doritos girl) on forums with the claim that it was actually young Nancy Pelosi. The picture in question had been doctored a bit to look older than it was, but still was hilarious how many people actually fell for it. :P

And since we are in congress already, everyone's favorite nonagenarian senator, Dianne Feinstein.