Jump to content

AI Generative Art


sbloyd

Recommended Posts

On the Copilot (formerly Bing) Image Creator, adding comic book covered a lot of ground that I was formerly using line art, cel shaded and the like, freeing up space in the prompt for other things.

 

It turned a prompt of 

full body, style of artstation, phoenix, woman made of glowing fire, glowing bright orange skin, skin glowing with inner fire, bare upper arms, hair made out of fire, glowing red flaming eyes, wearing red tech bulky containment suit, shooting flames from her palms

 

image.png.ef6076c7cc83b2cc18d4b863b99db3ba.png

 

to

full body, comic book, style of artstation, phoenix, woman made of glowing fire, glowing bright orange skin, skin glowing with inner fire, bare upper arms, hair made out of fire, glowing red flaming eyes, wearing red tech bulky containment suit, shooting flames from her palms

 

image.png.d587378676daaa5fbbdf76f11161460b.png

  • Like 4

Horizon Twilight, The Chernobyl Effect, XLR Mk8, Dodgeball, and a host of other alts all hanging out on Everlasting.

Link to comment
Share on other sites

I need to mess around with LORA creation some more.  My experiment the other day was in the right direction but... flawed somehow.  It turned out a lot of very chunky lines when it generated images.

 

Meanwhile, I got Microsoft to turn out 

image.png.447b734827a582295bd85c8d968f873b.png

which is a lot closer to how I imagine Horizon to be.  

  • Like 2

Horizon Twilight, The Chernobyl Effect, XLR Mk8, Dodgeball, and a host of other alts all hanging out on Everlasting.

Link to comment
Share on other sites

I've stated it multiple times on the forums:  I'd love to see ethnic neighborhoods added to Paragon City: Little China, Little Italy, Little Greece, Czech Village, etc.  With their Hmong origins, I've felt the Tsoo were a natural for a Little Asia neighborhood.  So I attempted this look. 

 

For the sake of the A.I.'s feeble understanding, I made some compromises.  Instead of a "Little Asia" I went with "Little China".  Rather than waste limited text, I went with "ninja" for a general description.  As always, A.I. struggles with proper rendering of fingers and swords, and some of the rejects were hilarious. Even so, some of the finalists have these problems.  It was unavoidable.  Also curious, the system had a hard time figuring out what a ninja looked like who wasn't in black.  Some of the rejects look liked they formerly attended Ringling Brothers' Clown College.    I've yet to figure out how to get the figures to quit mugging for the camera and get into a real action shot.  Lastly, initial shots of the Little China were overrun by cars and pedestrians, so I placed limiters up front. Other than that, this wasn't so bad. 

 

Description: Low-population, 4 cars, street-level Little China district at night in New England big city, with male ninjas wearing a variety of dark colors with Asian tattoos on their chests, wrapped forearms and shins, holding a kitana and a tanto and wearing paddy hats and Asian demon masks, in action pose

(for the top left shot, I edited out the sword info and opted for "in martial arts action pose".

 

2024_03_07_Tsoo.thumb.jpg.1b0ab64e1e0359ffebb8a0b773f1928e.jpg

 

 

Edited by Techwright
Added description
  • Like 1
Link to comment
Share on other sites

So, still trying to find something apropos for my living garbage concept creature/character. Current prompts are: 3d CGI render of a large creature made of corpses and garbage.

I changed it to large amorphous creature and got...less than optimal results. Changed amorphous to shapeless and got that final one.

Detrius4.jpg

Detrius5.jpg

DetriusGood.jpg

  • Confused 1
Link to comment
Share on other sites

Exactly what I’ve been asking for: Midjourney now allows you to re-use the same character in different scenes. Won’t be long before all the other AI programs do, too.

 

 

Link to comment
Share on other sites

I decided to give vanilla Stable Diffusion XL a try. I was getting iffy results before, so I did a quick Google and it said SDXL prefers Ancestral Samplers. Normally I use DPM++ 2m (Karras), but for these I switched to Euler Ancestral.

Blue_Idoru__Petite_and_slender_Japanese_woman__Dark_blue_hair_in_twin__S60188947_St100_G11.3.jpeg

Blue_Idoru__Petite_and_slender_Japanese_woman__Dark_blue_hair_in_twin__S60188950_St75_G11.3.jpeg

Blue_Idoru__Petite_and_slender_Japanese_woman__Dark_blue_hair_in_twin__S60188949_St50_G11.3.jpeg

Blue_Idoru__Petite_and_slender_Japanese_woman__Dark_blue_hair_in_twin__S60188946_St50_G11.3.jpeg

Blue_Idoru__Petite_and_slender_Japanese_woman__Dark_blue_hair_in_twin__S2721988964_St50_G11.3.jpeg

Blue_Idoru__Petite_and_slender_Japanese_woman__Dark_blue_hair_in_twin__S2721988960_St50_G11.3.jpeg

  • Like 2
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...