Grildon Tundy
Member
Yo yo yo, check out my new profile pic: brought to you by putting Demons' Souls prompts into Stable Diffusion
Last edited:
-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
Get Stable diffusion locally on your PC (RTX card needed) no restrictions and no censorship.
- Thread starter Tumle
- Start date
Makoto-Yuki
Gold Member
it was the resolution! putting it back down makes it run again. i was just putting in the same resolutions as found on the site. i think it was like 512x768. 512x512 works fine.Might be a dumb question, but did you start getting an error after changing some of your settings? I recall that upping the resolution or steps can cause memory issues.
Edit:
thanks, that helpsSounds like it doesn’t clear the memory on your graphicscard after your prompts..
Oh and try copying there prompts about details and settings, but insert your scene prompts.
Also try the seed number they used and any other settings they have changed
Grildon Tundy
Member
Good stuff, man! If anyone wants to see how there's still some human artistry needed when working with this program, all they need to do is compare your pics to mine lol
I edited your reply down to my personal favorites.
1) I really want to know desert drunk guy's story
2) Same with poor dock hanger-outers
3) Hell king looks realistic in a way that makes it even more metal
4) Heavenly
5) Purgatory-y
6) Looks like a legit still shot from a high-budget movie
Last edited:
Pakoe
Member
Lmao, thanks!
Hah, first one is actually based on a friend of mine that looks like a viking and can chug away beer like it's water. So I wanted to create something based on him.
It took a few renders to get good ones. I used the following prompt:
realistic detailed image of an viking alone in the desert next to giant skeleton with horns surrounded by beer bottles in the style of francis bacon, surreal, norman rockwell and james jean, greg hildebrandt, and mark brooks, triadic color scheme, by greg rutkowski, open ceiling, highly detailed, painted by francis bacon, painted by james gilleard, surrealism, airbrush, ilya kuvshinov, wlop, stanley artgerm
No funny story for the second one, wanted to see how it would render architecture based around a dock.
a harbor with a concrete dock and a storage facility, summer season, very hot, dry desert, architecture, a realistic digital painting by greg rutkowski and james gurney, trending on artstation, highly detailed
Tumle
Member
Nice !it was the resolution! putting it back down makes it run again. i was just putting in the same resolutions as found on the site. i think it was like 512x768. 512x512 works fine.
thanks, that helps
you can use an upscaler ai, if you want the picture to be lagrer
Last edited:
Makoto-Yuki
Gold Member
i have gigapixel ai for making them largerNice !
you can aksets ude an upscaler ai, if you want the picture to be lagrer
Tumle
Member
trying my avatar in a new style
highly detailed, d & d, fantasy, highly detailed, digital painting, trending on artstation, concept art, sharp focus, illustration, global illumination, ray tracing, realistic shaded, art by artgerm and greg rutkowski and fuji choko and viktoria gavrilenko and hoang lap
was my ekstra prompts
highly detailed, d & d, fantasy, highly detailed, digital painting, trending on artstation, concept art, sharp focus, illustration, global illumination, ray tracing, realistic shaded, art by artgerm and greg rutkowski and fuji choko and viktoria gavrilenko and hoang lap
was my ekstra prompts
Last edited:
Tumle
Member
looks really great.Well I was trying what the AI would think Morgoth or Sauron from Lord of the rings should look like and it went quite well
I think Morgoth below is a nice start
Sauron not bad either
what ekstra prompts did you use?
also tried Morgoth
Last edited:
Tumle
Member
found a new program with Img to img rendering.
painted this in ms paint:
this is one of the results >)
sorry for bumping..
the program can be found here:
nmkd.itch.io
painted this in ms paint:
this is one of the results >)
sorry for bumping..
the program can be found here:
NMKD Stable Diffusion GUI - AI Image Generator by N00MKRAD
Generate AI images on your own GPU for free
nmkd.itch.io
Last edited:
HoodWinked
Member
Crazy the rate at things are changing seem to be exponential.
Last edited:
Dirk Benedict
Member
trying my avatar in a new style
highly detailed, d & d, fantasy, highly detailed, digital painting, trending on artstation, concept art, sharp focus, illustration, global illumination, ray tracing, realistic shaded, art by artgerm and greg rutkowski and fuji choko and viktoria gavrilenko and hoang lap
was my ekstra prompts
Unda Da Spooky-ass, motha fuckin' sea!
thefool
Member
Art has become so diluted it almost lost its meaning.
We're watching the first step of mass dissemination of craft (some would call it democratization). Artisanship moves away from manual labour to a (expanded) relation we have with machines and their ability to display what we intend.
In a way its akin to what the slide ruler and the pascaline did to mathematics, from counting with stones to tools that democratized arithmetic, eventually leading to computing technology.
We're watching the first step of mass dissemination of craft (some would call it democratization). Artisanship moves away from manual labour to a (expanded) relation we have with machines and their ability to display what we intend.
In a way its akin to what the slide ruler and the pascaline did to mathematics, from counting with stones to tools that democratized arithmetic, eventually leading to computing technology.
Croatoan
They/Them A-10 Warthog
what happens if you feed it a 3d render from Daz Studio or something?found a new program with Img to img rendering.
painted this in ms paint:
this is one of the results >)
sorry for bumping..
the program can be found here:
[/URL]
Tumle
Member
The more details the better I think.. there is a slider where you can choose how much influence it should take from the picture to incorporate in the end result.
I started looking into other builds via GitHub, and managed to get one with all the bells and whistles with both in painting(masking of an area and only changing that area), out painting (continuing an image at its boarders) and up scaling.
Right now I’m trying to get textual Inversion to work, so I can feed it my own pictures to learn from.. but it’s a little hard when you are green to the ecosystem of machine learning
Something like this:
Last edited:
Mr1999
Gold Member
I had not used this for weeks and after the recent git pull I see that I'm now able to get much higher res 1920x1080 off of my 3070TI and with --medvram --opt-split-attention I no longer run out of VRAM. Also highres.fix is new to me, it helps a lot when rendering at 1920x1088 now, beforre you would get quadruplets, 10 hands. This thing is awesome. Every single AI niche software I've tried it felt almost magical. DLSS for me was also magical in that it feels like you can get something for nothing but can be explained away with math. Here are my crappy creations, what I find satisfying is how random it is especially when trying to create killer robots.
some i ran last night. have not bothered to go through them. def need to add some negatives though.
some i ran last night. have not bothered to go through them. def need to add some negatives though.
Last edited:
Croatoan
They/Them A-10 Warthog
The more details the better I think.. there is a slider where you can choose how much influence it should take from the picture to incorporate in the end result.
I started looking into other builds via GitHub, and managed to get one with all the bells and whistles with both in painting(masking of an area and only changing that area), out painting (continuing an image at its boarders) and up scaling.
Right now I’m trying to get textual Inversion to work, so I can feed it my own pictures to learn from.. but it’s a little hard when you are green to the ecosystem of machine learning
Something like this:
What program are you using that takes images as input?
Tumle
Member
https://github.com/AUTOMATIC1111/stable-diffusion-webui
Its pretty Easy to set up, you just have to download some Extra programs.
But just follow the installation guide
Makoto-Yuki
Gold Member
i'm not getting anything as good as any of these.
i don't know if it's bugged but it's a pain in the ass i can't run multiple samples. i need to do each one at a time. if i could set a prompt and let it run x samples to then go through then that'd be better.
i don't know if it's bugged but it's a pain in the ass i can't run multiple samples. i need to do each one at a time. if i could set a prompt and let it run x samples to then go through then that'd be better.
Tumle
Member
What are you using?
Also keep the weights between 7 and 12
And for prompts copy the prompts from lexica and insert your own subject, until you get more comfortable with it.
But interested to know what version of stable diffusion you are using and what settings
Last edited:
Pakoe
Member
Some more results. Seems like v-scale sharpens the result more (if thats the correct word).
If i want results looking like old paintings, i stick to v-scale 7.50. Anything higher than that and it goes towards realism.
I also picked up digital art again on my ipad with the Procreate app using stable diffusion art as reference.
Which is super efficient since it save so much time trying to come up with composition, color schemes etc.
If i want results looking like old paintings, i stick to v-scale 7.50. Anything higher than that and it goes towards realism.
I also picked up digital art again on my ipad with the Procreate app using stable diffusion art as reference.
Which is super efficient since it save so much time trying to come up with composition, color schemes etc.
Tumle
Member
Its just opening up artistic possibilities for people who lack the skills and will further help artist speed up there process
Also I got textual inversion to work and trained my version to Recognise me!
Me as a GTA V character
Me as Woddy from Toy Story
Mr1999
Gold Member
Its just opening up artistic possibilities for people who lack the skills and will further help artist speed up there process
Also I got textual inversion to work and trained my version to Recognise me!
Me as a GTA V character
Me as Woddy from Toy Story
Nice. This is what im currently doing on colab, but I see its become exhausted with users, got kicked a couple of times and pretty slow gpu, decided to pay the $10 to upgrade to colab pro, no more disconnects and cut my time in half.
Last edited:
Tumle
Member
Im using runpod.io and renting an rtx 5000 for 0.30 cent an hour.. but maybe i should go with the colab plan..
Could also buy a rtx3090.. or wait for the 4000 series
Last edited:
Wildebeest
Member
Saw that some nerds were upscaling Fallout art, so I thought I would step into the needlessly controversial world of Guybrush Threepwood concept art.
Bonus: I used the banned technique and just made him look like Bowie.
Bonus: I used the banned technique and just made him look like Bowie.
Last edited:
GeekyDad
Member
Anyone interested in working with me to make animations with this software to accompany my music videos? I'm still uncomfortable trying to dive into this, still working on getting my DAT (Roland VS-880ex) recorder back up and running and relearning its workings. Maybe just listen to the tunes, and using your own imagination to create animations to pair with it?
Please and thank you.
Please and thank you.
Tumle
Member
Sounds like a great project.. I wish I could help.. but haven’t looked much into making animations with it..
L3E2
Member
This actually sounds cool, I'd love to learn how to make animations with this but the only problem is I don't know where to go for that as I only know how to do just photos. Is there an actual program (ideally free) that I can run on my PC that you or anyone here know of?
GeekyDad
Member
Tumle
Member
L3E2
Member
Coooool, let me look into this and I’ll get back to you when I’m more knowledgeable!
Jinzo Prime
Member
Teach me your ways, master!Been playing with stable diffusion a lot lately and getting much better results. Its really a great way to create wallpapers for your pc.
Sybrix
Member
At what point do you tell yourself "Those breasts are too big?"
If a woman needed two golf carts to move her breasts around would you say their too big?
Last edited:
Ironbunny
Member
I think we are getting there.
Well those would certainly be rather inconvenient.
Haemi
Member
What are the promts to get these backgrounds? Something with bokeh i guess
Ironbunny
Member
You can pretty much define the lens, focal length and aperture used and descripe what kind of bokeh you want. Also descriping lens reflection like anamorphic lens flares help. Background out of focus with lots of light sources.
Last edited:
Jinzo Prime
Member
It would be on huggingface.co if it's out. Haven't checked in a while.
Ironbunny
Member
2.1 seems to be the newest there. Using 1.5 myself.
Makoto-Yuki
Gold Member
I think this is the latest
stabilityai/stable-diffusion-2-1 at main
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
huggingface.co