Message Turncoat in a DM to get moderator attention

Users Online(? lurkers):
7 / 17 posts
Posts: 33405
0 votes RE: ⚡ Stable Diffusion ⚡

The original Wombo models used as reference for prompts: 

Set 1:
Wombo:

Posted Image

SD:

Posted Image

Set 2: 
Wombo:

Posted Image

SD:

Posted Image
Set 3:
Wombo:

Posted Image

SD:
Posted Image


They actually look pretty similar, which has been the overall goal. 

Ę̵̚x̸͎̾i̴͚̽s̵̻͐t̷͐ͅe̷̯͠n̴̤̚t̵̻̅i̵͉̿a̴̮͊l̵͍̂ ̴̹̕D̵̤̀e̸͓͂t̵̢͂e̴͕̓c̸̗̄t̴̗̿ï̶̪v̷̲̍é̵͔
last edit on 2/19/2023 3:59:41 PM
Posts: 33405
0 votes RE: ⚡ Stable Diffusion ⚡

Lol okay, so what started as experiments with Ghost Rider and Negative Prompt experimentation got really fun. I wanted to try to create a base form to work with for building the guy in Set 3 from above, but instead had fun letting Stable Diffusion get creative. 

The first set was basically just 'Ghost Rider, Marvel Comics, Realistic' for Positive Prompts, then I added 'Skull, Skeleton, Undead, Necromancy, Cartoon, Comic book, Toon, Anime, Manga' to Negative Prompts: 

Posted Image

Posted Image

Posted Image

Posted Image


I then added 'Leather' to the Negative Prompts: 

Posted Image

Posted Image

Posted Image


I then added to Negative Prompting  'X-men Costume' and refined the earlier one into 'Leather Clothing'

Posted Image

Posted Image


This base form was awesome, but not what I was looking for. It had no torso anymore after ridding of anything resembling Leather and the X-Men! 

I added 'Red Skull' to the Positive Prompts to try to give it a sense of direction (with Skull still in Negative Prompts), and this is where it gets cool: 

Posted Image

Posted Image

Posted Image

Posted Image

Posted Image

Posted Image

Posted Image

That last one's just woah. 

Ę̵̚x̸͎̾i̴͚̽s̵̻͐t̷͐ͅe̷̯͠n̴̤̚t̵̻̅i̵͉̿a̴̮͊l̵͍̂ ̴̹̕D̵̤̀e̸͓͂t̵̢͂e̴͕̓c̸̗̄t̴̗̿ï̶̪v̷̲̍é̵͔
Posts: 33405
0 votes RE: ⚡ Stable Diffusion ⚡

To continue on attempting to create the man in Set 3, I changed the Positive Prompt to 'Portrait of Ghost Rider, Marvel Comics, Male Surfer Stoner, Norville Shaggy Rogers, Realistic', with the same Negative Prompts as before: 

Posted Image

Posted Image

Posted Image

Posted Image


The two ideas for even their bad renders rarely consolidated the imagery together, save for the last one, so I added 'Portrait of' to the beginning of it which isolated the image to a single figure: 

Posted Image

Posted Image

Posted Image

Posted Image

Posted Image

Posted Image

Posted Image


That last one had me so "woah" that I tried adding 'Rainbow Colors' to the beginning, but that did not go as planned: 

Posted Image

Posted Image

Posted Image

Posted Image

 I've since revised it to 'Rainbow Flames', I'll edit the results as I receive them. 

Edit: Okay this isn't working either: 

Posted Image

Posted Image

Posted Image

Posted Image

Next is to test it with Rainbow Flames at the end of the Prompt instead. 

Edit: Still not quite, but closer: 

Posted Image

Posted Image

Posted Image

Posted Image

Posted Image


Now to try the middle of the prompt. 

Ę̵̚x̸͎̾i̴͚̽s̵̻͐t̷͐ͅe̷̯͠n̴̤̚t̵̻̅i̵͉̿a̴̮͊l̵͍̂ ̴̹̕D̵̤̀e̸͓͂t̵̢͂e̴͕̓c̸̗̄t̴̗̿ï̶̪v̷̲̍é̵͔
last edit on 2/19/2023 7:17:40 PM
Posts: 33405
0 votes RE: ⚡ Stable Diffusion ⚡

Rainbows are not friendly with the fusion danced Shaggy + Ghost Rider, but once I got rid of Shaggy and allowed for skeletal stuff and leather again, with a few different rainbow prompts: 

Posted Image

Posted Image

Posted Image

Posted Image

Posted Image

Posted Image


I decided to Negative Prompt 'Luchador Mask' to get it to stop doing the Army of Two motif: 

Posted Image

Posted Image

Posted Image

Posted Image

Posted Image


Posted Image


This program is so fucking cool, Negative Prompting is so mmmm ♡

Ę̵̚x̸͎̾i̴͚̽s̵̻͐t̷͐ͅe̷̯͠n̴̤̚t̵̻̅i̵͉̿a̴̮͊l̵͍̂ ̴̹̕D̵̤̀e̸͓͂t̵̢͂e̴͕̓c̸̗̄t̴̗̿ï̶̪v̷̲̍é̵͔
Posts: 4564
1 votes RE: ⚡ Stable Diffusion ⚡

I was using InvokeAI to generate images. It won't run as fast as the web site TC linked does unless you happen to have a newer Nvidia card, but you have more controls. This video explains the program:

Posts: 784
0 votes RE: ⚡ Stable Diffusion ⚡

I was using InvokeAI to generate images. It won't run as fast as the web site TC linked does unless you happen to have a newer Nvidia card, but you have more controls. This video explains the program:

 That's very good for matte painting backgrounds, and fast too.

 

Posts: 33405
0 votes RE: ⚡ Stable Diffusion ⚡

Dezgo released a new, improved render style on their platform for normal people to use. 

Shit's improving quickly: 

Ę̵̚x̸͎̾i̴͚̽s̵̻͐t̷͐ͅe̷̯͠n̴̤̚t̵̻̅i̵͉̿a̴̮͊l̵͍̂ ̴̹̕D̵̤̀e̸͓͂t̵̢͂e̴͕̓c̸̗̄t̴̗̿ï̶̪v̷̲̍é̵͔
7 / 17 posts
This site contains NSFW material. To view and use this site, you must be 18+ years of age.