stable-diffusion / AI - PART TWO

[Dickminer]

Originally Posted by Johnny Cage

I have to say that finally getting controlnet posing to work consistently is a game changer for me.

e.g. this yoga kneeling type pose is one I would have never been able to properly create before.

See, these are the moments when I feel ambivalent about AI honeys.
On the one hand, I appreciate the extra hot content it provides.
On the other hand, saying that I'd love to bury my face in that butt doesn't have the same ring to it knowing this lady is not real.

Originally Posted by maxhitman

Make changes to the words if you see it should describe the images better.
DO NOT NAME THE Faces or the main subject person in the pictures !

Now this what gets me, the guy called Aitrepreneur never mentioned any of this when I looked months back and also has different settings to Olivio Sarikas, I wish people would know exactly what they were doing when releasing a video instead of trying to fudge it.

[maxhitman]

Originally Posted by Zammis

Now this what gets me, the guy called Aitrepreneur never mentioned any of this when I looked months back and also has different settings to Olivio Sarikas, I wish people would know exactly what they were doing when releasing a video instead of trying to fudge it.

Yes, I agree 100%
There are too many videos with alot of amateurs being posted.
In a way I understand that people want to help out with using this
Stable-Diffusion, but the FACT is, this thing is so new to everyone, that
no one is really a professional at using it.
I am not a professional either. I just discovered the best way to write a
great prompt which I can use to create the best images I want to see.

I have researched and read alot of information about this SD
and collected as much info as possible. It´s like trying to collect the pieces
of a puzzle to put them all together to create a complete picture.

One thing I did discover is, in that particular Olivio Sarikas video it is the
best video Tutorial.

this video...

http://youtube.com/watch?v=9MT1n97ITaE&t=759s

It has easy to follow instructions on how to install the Dreambooth program and how to make a LORA quickly.
I did exactly what he says and I have the Dreambooth installed and I am
making LORA very easy. Everything is working great. I have also seen the
Ai-Enterpreteur videos and they are good too, but sometimes they forget to tell certain important facts and lessons.

One thing I also discovered is that the Dreambooth program which makes LORA and other things, can train the A.I. on images which are in bigger sizes, but
the fact is, it works much better if you use the sizes of 512x512 ,
or the size of 768x768. If you place your images inside a square-box of these
sizes , the Ai will get better training on the pictures.
The images you use are extremelly important. They must be very clear
and have details.
When you reduce the size of a High-Resolution portrait photo to a size of
768x768 to fit into that square-box, it seems that the image may have lost
some details and quality, but the fact is, this A.I. can still "read" all the
tiny little details and information which is in that small photo.
The Ai will later create an artwork using that tiny image and add
alot more details to it too.

This Ai works with billions of tiny micro-sized pixels and then re-arranges them and places all of these micro-pixels together in a super-quick way
to create an image or artwork.
So, HOW does an image look like while its being created ?
Here is an experiment which you can also try doing at home.
Start your SD program and have it make for you a batch of 4 images.
Make sure its of a woman wearing lingerie standing in a living room.
This is what I did.
Then you look at your little (cmd.exe) little-black-pop-up window to see
the Ai progress as it creates the images. Just when it is about half way
(50%) of a completed image, the electricty shuts down.!!!
This surprised me.
All of my house lights were shut down! It was some type of electrical problem, which some times happens around here in my town when they
are doing some electrical-grid work, or when there is a heavy thunderstorm and rainny weather. Its very normal and it does not effect my computer.
So, when the lights come back on, I re-start my pc.
I go and take a look at the "outputs" folder to see if the Ai created any
pictures while it was at 50% at making them.
To my surprise, it did create "some thing". What it looked like was the most
bizarre looking woman creature I have ever seen.
It looked like a horror film !
The Ai was half-way into creating a really cool image, but since the pc
was shut down, it quickly stopped.
What I saw was a body which looked like a horrible clay figurine, with alot
of distorted muscle tissue and body parts. A really ugly Frankenstein looking monster.
This is how the Ai creates a human person as its collecting all those
tiny-micro pixles and putting them together to create an image,
which will later look like a real person.
You can also try this at home and see what your Ai creates.
Just shut down your PC exactly when the Ai is about 50% into creating an image. Re-start your pc and then go look into the output folder to see
what pictures it was creating.
You will be surprised how this Ai works !


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

Hurray! last nights job at making a new Lora came out amazing!
It is the version 2 for the Armando Huerta colored art style which he
often did for his sexy "Bad Girl" PIN-UP art.

It is now posted for everyone to download at Civitai ,and to enjoy making
a diferent style of sexy Pin-up Art ... using your favorite celebrity faces
if you wish. Armando the artist was very famous for doing that.
Its very cool that people love this style of art.

Armando Huerta = Pin-Up Artist Series (NSFW) Color version

http://civitai.com/models/50226/armando-huerta-pin-up-artist-series-nsfw-color-version

-

-

Originally Posted by maxhitman

Yes, I agree 100%
There are too many videos with alot of amateurs being posted.
In a way I understand that people want to help out with using this
Stable-Diffusion, but the FACT is, this thing is so new to everyone, that
no one is really a professional at using it.
I am not a professional either. I just discovered the best way to write a
great prompt which I can use to create the best images I want to see.

I have researched and read alot of information about this SD
and collected as much info as possible. It´s like trying to collect the pieces
of a puzzle to put them all together to create a complete picture.

One thing I did discover is, in that particular Olivio Sarikas video it is the
best video Tutorial.

this video...

http://youtube.com/watch?v=9MT1n97ITaE&t=759s

It has easy to follow instructions on how to install the Dreambooth program and how to make a LORA quickly.
I did exactly what he says and I have the Dreambooth installed and I am
making LORA very easy. Everything is working great. I have also seen the
Ai-Enterpreteur videos and they are good too, but sometimes they forget to tell certain important facts and lessons.

One thing I also discovered is that the Dreambooth program which makes LORA and other things, can train the A.I. on images which are in bigger sizes, but
the fact is, it works much better if you use the sizes of 512x512 ,
or the size of 768x768. If you place your images inside a square-box of these
sizes , the Ai will get better training on the pictures.
The images you use are extremelly important. They must be very clear
and have details.
When you reduce the size of a High-Resolution portrait photo to a size of
768x768 to fit into that square-box, it seems that the image may have lost
some details and quality, but the fact is, this A.I. can still "read" all the
tiny little details and information which is in that small photo.
The Ai will later create an artwork using that tiny image and add
alot more details to it too.

This Ai works with billions of tiny micro-sized pixels and then re-arranges them and places all of these micro-pixels together in a super-quick way
to create an image or artwork.
So, HOW does an image look like while its being created ?
Here is an experiment which you can also try doing at home.
Start your SD program and have it make for you a batch of 4 images.
Make sure its of a woman wearing lingerie standing in a living room.
This is what I did.
Then you look at your little (cmd.exe) little-black-pop-up window to see
the Ai progress as it creates the images. Just when it is about half way
(50%) of a completed image, the electricty shuts down.!!!
This surprised me.
All of my house lights were shut down! It was some type of electrical problem, which some times happens around here in my town when they
are doing some electrical-grid work, or when there is a heavy thunderstorm and rainny weather. Its very normal and it does not effect my computer.
So, when the lights come back on, I re-start my pc.
I go and take a look at the "outputs" folder to see if the Ai created any
pictures while it was at 50% at making them.
To my surprise, it did create "some thing". What it looked like was the most
bizarre looking woman creature I have ever seen.
It looked like a horror film !
The Ai was half-way into creating a really cool image, but since the pc
was shut down, it quickly stopped.
What I saw was a body which looked like a horrible clay figurine, with alot
of distorted muscle tissue and body parts. A really ugly Frankenstein looking monster.
This is how the Ai creates a human person as its collecting all those
tiny-micro pixles and putting them together to create an image,
which will later look like a real person.
You can also try this at home and see what your Ai creates.
Just shut down your PC exactly when the Ai is about 50% into creating an image. Re-start your pc and then go look into the output folder to see
what pictures it was creating.
You will be surprised how this Ai works !


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

Hurray! last nights job at making a new Lora came out amazing!
It is the version 2 for the Armando Huerta colored art style which he
often did for his sexy "Bad Girl" PIN-UP art.

It is now posted for everyone to download at Civitai ,and to enjoy making
a diferent style of sexy Pin-up Art ... using your favorite celebrity faces
if you wish. Armando the artist was very famous for doing that.
Its very cool that people love this style of art.

Armando Huerta = Pin-Up Artist Series (NSFW) Color version

http://civitai.com/models/50226/armando-huerta-pin-up-artist-series-nsfw-color-version

-

-

Very arty indeed, I will say that while I can create a spot on Lora I find the whole exercise too time consuming and not all of the results are bang on, in actual fact if you were to pick the subject of say Sandra Bullock and render that you will get some fab results, but if you try it in a Lora they more often that not come out crappy so it kind of defeats the object.

I think that soon enough a different easier way will come about where the AI trains on one image where you can just drag and drop and click one button and all will be done, almost like a 3D scan

[BoInk2]

You can create those "horror pictures" by clicking on "Interrupt" during generation.

Backgrounds are also very good

[maxhitman]

Guess who?

Legend claims that the famous "Margarita Cocktail" drink was named after her
real name, when she was dancing under her real name in a Tijuana, Mexico nightclub.

-

-

She also knocked out two front teeth off of the main actor during their
fight scene during the filming. It was her first major dramatic role in
a Hollywood movie !

-

[Dickminer]

Did you make a Rita Hayworth Lora? Impressive! Most of the folks I have seen make celebrity actress AI art rather neglect golden and silver age era Hollywood.

Originally Posted by maxhitman

Guess who?

Legend claims that the famous "Margarita Cocktail" drink was named after her
real name, when she was dancing under her real name in a Tijuana, Mexico nightclub

Only last night did I watch Discovering Rita Hayworth, spooky



parameters

Rita Hayworth in colour

Negative prompt: Canvas Frame, Cartoon, 3D, Ugly Eyes, Deformed Iris, Deformed Pupils, Fused Lips, Fused Teeth, Low Detailed Skin, Semi-Realistic, CGI, Render, Sketch, Cartoon, Drawing, Anime, Text, Close Up, Cropped, Out Of Frame, Worst Quality, Low Quality, Jpeg Artefacts, Ugly, Duplicate, Long Neck, Doll, Child, Ugly, Morbid, Extra Fingers, Poorly Drawn Hands, Mutation, Blurry, Extra Limbs, Gross Proportions, Missing Arms, Mutated Hands, Long Neck, Duplicate, Mutilated, Mutilated Hands, Poorly Drawn Face, Deformed, Bad Anatomy, Cloned Face, Malformed Limbs, Missing Legs, Too Many Fingers, Closed Eyes, Small Eyes, Monochrome, Child
Steps: 28, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 2501039356, Face restoration: CodeFormer, Size: 512x768, Model hash: e6415c4892, Model: realisticVisionV20_v20

postprocessing

Postprocess upscale by: 2, Postprocess upscaler: 4x-UltraSharp

extras

Postprocess upscale by: 2, Postprocess upscaler: 4x-UltraSharp

Oh and if you want to see your image info to share with others just load an image into the PNG Info Tab and you will see how it was all made/created.

[maxhitman]

Originally Posted by Dickminer

Did you make a Rita Hayworth Lora? Impressive! Most of the folks I have seen make celebrity actress AI art rather neglect golden and silver age era Hollywood.

Many Thanks.

Yes, do I love the classic film stars, because I also watch alot of
classic films too. I admire those classic stars from Hollywood.
They were VERY Sexy when they were much younger.
They are Great for doing amazing Photoshoots or Artistic-Nude photos.
Also great to use for a fantasy style artwork.

Sometimes I have to colorize the black & white photos using Photoshop
and my special tecniques, because not many color photos exist of them.
But the final process of training the A.I. comes out really cool.
The A.I. is very much capable of doing their faces in a very realistic
manner, including their many face expressions and bodies.
These classic movie stars will get a new life in the A.I. world !

I have an idea to create ALOT of cool classic film stars, both female and
male stars, for us to use in this Stable-Diffusion A.i.
I also have alot of other cool surprises which I will be posting soon.

It would be cool to see them in a picture or a film scene next to
todays film stars.
Like, a romantic film scene with Cary Grant and Angelina Jolie.
Or maybe Marilyn Monroe with Brad Pitt !
Or maybe,
a Gladiator fight scene with Kirk Douglas and Russell Crowe !
The ideas are infinite and would be FUN to create!

Some "checkpoints" such as Realistic-Vision , does contain
some information and data with these classic stars. Some times it
can create their faces a little bit, but it is not 100% exactly like them.
They are still missing alot of details.
So, that is the reason I am also helping out making these LORA and
training the A.i. with their faces and face expressions,
so everyone can use them in their art and photos.