Ironhide's AI art gallery - feel free to exhibit your own art

Ironhide wrote:Spoilered for size, there's nothing really NSFW.

A cute one to finish

How do feel about things like this providing tools for you to prompt or create art with your abilities? Apologies if this is a personal question I was just thinking about it.

My mum for instance never paints or draws with physical media any more but she's been making incredible scenes and worlds and architecture in virtual words for well over a decade now. She can simply move around a little bit. She's in remission but has been frail for most of my life.

I checked that lexica site and saw this which is funny

"Designed by teenage engineering. futuristic international gameboy console. highly detailed digital art masterpiece, . detailed render. sleek. slim. minimalism. stunning design. designed by dieter rams."

The model obviously has no concept of a useful number of input methods or, well, even what they are :lol:

Green Gecko wrote:
Ironhide wrote:Spoilered for size, there's nothing really NSFW.

A cute one to finish

How do feel about things like this providing tools for you to prompt or create art with your abilities? Apologies if this is a personal question I was just thinking about it.

I checked that lexica site and saw this which is funny

"Designed by teenage engineering. futuristic international gameboy console. highly detailed digital art masterpiece, . detailed render. sleek. slim. minimalism. stunning design. designed by dieter rams."

The model obviously has no concept of a useful number of input methods or, well, even what they are

Putting disability aside, as someone with an interest in art but limited artistic ability (I used to draw a lot and was into origami during my teens but wasn't particularly good at either) I've found tools like this are a useful means of creative expression even if all I'm doing is inputting a few text prompts and changing a few numerical settings here and there until I get the desired output.

I've previously used things like Terragen (3D landscape editor) and POVRay to generate art and despite being a total novice with both pieces of software I managed to obtain satisfactory results (for myself rather than something I would necessarily want to share or exhibit).

And yes, some times the ai model produces some unintended hilarious results (and accidental porn :lol:

)

Oh yeah I remember Terragen. Made some hills and a sunset over water with it.

Green Gecko + Strawber =

Green Gecko wrote:Oh yeah I remember Terragen. Made some hills and a sunset over water with it.

I got quite into messing with the atmospheric settings to get some nice lighting effects, looking at the screenshots I took it looks terrible but back in 2008 I thought it looked good :lol:

site23 wrote:If you're interested in generating anime style images exclusively, Ironhide, you probably want to install one of the anime focused derivatives of Stable Diffusion rather than using the base model. (You can get very interesting results out of the base model but it's a jack of all trades.)

Have downloaded a couple of more anime leaning models (checkpoints) recently (after doing some research into producing better images) and am currently looking into the smaller LoRA models to fine tune the results (there's a lot of character specific ones but that's a rabbit hole I'm slightly afraid to go down :lol:

).

I've also recently switched back to an Nvidia GPU as I got a little fed up with images taking 20+ minutes to generate, now its around 50 seconds.

More pics incoming.

Vermilion wrote:Green Gecko + Strawber =

lolwut

I like how it also gets gecko hands wrong.

(And they kind of do have hands, what with being able to grip onto stuff.)

It seems to have merged Phelsuma with some kind of crested gecko, and it also looks like its got confused between eyes and ears.

Been playing around with LoRAs. Lovely stuff. Feel like a bit of a fraud doing this and I hope that doesn't change, but it is fun to play around with. Also it's kind of making me interested in getting back into drawing and painting.
What's a LoRA? Additional models for Stable Diffusion that help push the image more in a specific stylistic direction. Like an artist's personal style. Understandably scary stuff. I can see avenues for this tech to be helpful to artists themselves and not end up too dystopian, but who knows what the future will hold. Silver lining for now is that this is all fairly surface level, just a fancy coat of paint that doesn't actually hold together that well. It appears to at a glance, but for something actually well constructed and specific you still — blessedly — need a human touch.

Anyway. This image (and all the rest) is made with the aid of the Disco Elysium LoRA:

These latter three are all made with the help of ControlNet, which analyzes images of people and spits out a 3D pose (body, face and hands) to help the image generation along. The first two are from the same (!) image that I took of a friend years ago (the first one is much less changed than the second).

They're really good OSC, I definitely think stylised stuff works a lot better than realistic looking images as the flaws and errors stand out a lot less.

twitter.com/weirddalle/status/1651690840530591745

Took a break from anime stuff this time

More incoming.

Random question: what is the correct term for the bandage type chest wrap things you see depictions of female samurai wearing (trying to generate a pic of Erza from Fairy Tail wearing her samurai type armour).

Tried various ways of wording it but every attempt just has bare tits instead.

Ironhide wrote:Tried various ways of wording it but every attempt just has bare tits instead.

Ironhide wrote:Random question: what is the correct term for the bandage type chest wrap things you see depictions of female samurai wearing (trying to generate a pic of Erza from Fairy Tail wearing her samurai type armour).

Tried various ways of wording it but every attempt just has bare tits instead.

Ironhide wrote:Random question: what is the correct term for the bandage type chest wrap things you see depictions of female samurai wearing (trying to generate a pic of Erza from Fairy Tail wearing her samurai type armour).

The keyword you want is probably sarashi.

Think I might need some kind of traditional Japanese clothing LoRA for that.

"Jacob Rees Moggy"

"Ironhide wants a samurai woman"

Vermilion wrote:"Jacob Rees Moggy"

I'm stealing that. 8-)

Trying a different version of Stable Diffusion (InvokeAI) as the one I'm currently using (Easy Diffusion) is a little bit janky and doesn't have as many features.

I wrote:Trying a different version of Stable Diffusion (InvokeAI) as the one I'm currently using (Easy Diffusion) is a little bit janky and doesn't have as many features.

That didn't go well, couldn't figure out how to add different models/loras so abandoned it.

Tried Automatic1111 instead and got it working well (after much faffing).

Ironhide's AI art gallery - feel free to exhibit your own art

Who is online

Affiliates