|
IShallRiseAgain posted:Just posting a few more goosebumps covers I generated Please never stop making these.
|
# ¿ Oct 16, 2022 20:20 |
|
|
# ¿ May 14, 2024 18:35 |
|
A couple of days ago I was having trouble getting to bed, thinking about the imminent collapse of our society and trying to be ever so slightly less fatalist and cynical about the future. I thought back to how close man was to the brink of extinction in the ice age, and decided to use Dreamily to generate the wording for a story I had the shape of in my head. It was surprisingly easy to direct the AI to mostly stick with what I had in mind, and definitely improve it in areas. I got inspiration from what it wrote, and used it to spark new ideas and move the story and emotions forward in ways I wouldn't have otherwise. It got off course plenty, and I had to write a few really prescriptive things in order to point it back in the right direction (mostly in the story the young man tells), but I really like how it came out, which was more or less how I had imagined. The prompt involved two cavemen walking through a snowy mountain pass.quote:"Do you remember the stories?" The young hunter whispers to his partner, who seems lost in thought as he stares into the cold, icy nothingness. The tense changes at one point, and there's probably a few minor goofs here and there, but I was impressed. I ran it through an automatic plagiarism filter and it picked up a decent number of sentences yanked from everything from random nonfiction YouTube videos to old half-forgotten pulp fiction. Once these tools reach full maturity and the copyright issues are cleared up, it's really going to be scary how fully they'll be embraced by authors. I like this story enough that I decided to adapt it into a comic book as an experiment: This also involved a ton of image generation for every panel in MidJourney, picking the handful that felt close to what I had imagined or pleasantly surprised me, then selectively erasing things and adding in new prompts in DALL-E and expanding the scenes. I also manually adjusted the colors here and there to match a bit more, but nothing is painted or added by hand. After a certain point, what I was able to do with the tool started to make the visuals feel very repetitive, so I stopped. I couldn't get it to do specific poses or compositions which would add a ton to the visual storytelling, and it looses a lot of emotion that's present in the prose because of it. I couldn't even get them to consistently hold spears or bows, or show emotions easily. Still, a fun experiment. feedmyleg fucked around with this message at 04:05 on Oct 24, 2022 |
# ¿ Oct 24, 2022 03:52 |
|
Are there currently any online services which let you do proper AI upscaling with newer methods? I can't run things off my computer's GPU and all the in-browser upscaling methods I've seen are extremely out of date.
|
# ¿ Oct 25, 2022 18:00 |
|
IShallRiseAgain posted:Training continues on. Doing gods work. Can you generate some characters that weren't in TAS, like Jason Todd and Red Hood and the Court of Owls and Professor Pyg and such?
|
# ¿ Oct 27, 2022 19:30 |
|
IShallRiseAgain posted:It's not aware of the other characters, but I was able to use txt2img and img2img to make this: Hell yeah. This is great. What about Marvel characters? Would love to see its take on Iron Man, Spider-Man, Wolverine, etc.
|
# ¿ Oct 27, 2022 20:17 |
|
Congratulations, you've reinvented Jared Leto
|
# ¿ Oct 28, 2022 14:36 |
|
As someone who hasn't been able to play around with too much cutting-edge stuff because I have a crappy computer and am too dumb to figure any of the more complicated stuff out, MidJourney v4 is blowing my goddamn mind. What's the best in style transfer these days? If I put some time into it, would I be able to reasonably transfer a portrait photograph into the style of, say, Robert McGinnis or Mort Künstler? Or is that sort of thing still difficult for it?
|
# ¿ Nov 7, 2022 02:32 |
|
Ah sorry, to clarify—I'm wondering about taking an existing photograph and trying to get an AI to output a repainted version of that in an artist's style, while looking reasonably like the original person. I don't know if there's any real difference in complexity of that versus prompt-based image generation.
|
# ¿ Nov 7, 2022 02:46 |
|
I wrote and produced an adventure radio drama with a friend a decade ago that I've always remembered fondly. I decided to use Midjourney v4 to create an image that I've had in my head since then, but am not a good enough artist to paint myself: the cover of the trashy pulp book that the radio drama was (metafictionally) adapted from: Extremely happy with how it turned out. Took a lot of working and reworking of images that it spit out, but it more or less looks like exactly what I had in my head. feedmyleg fucked around with this message at 04:51 on Nov 8, 2022 |
# ¿ Nov 8, 2022 02:00 |
|
Appreciate it, all. Of course it's a mix between the AI generation and a good amount of manual effort with color matching and outpainting and such, but it feels pretty amazing that someone with extremely limited illustration talent can do this sort of thing. To give a sense of how many iterations this thing went through before its final form, here's some rejected elements all living together in an alternative version: TIP posted:this is great That was, uh, intentional. You see, cheap paperbacks don't have the highest editorial standards. That's it. ...thanks!
|
# ¿ Nov 8, 2022 05:24 |
|
mobby_6kl posted:This is cool too! Did you get it to make this halftone effect or was that photoshoped later? Halftone and grain were Photoshop, as was some color correction. I also had to generate the gun separately and outpaint it into her hand. But I had to strategically put a background behind it then erase it later because Dall E really doesn't like guns and outpaints them into other things. To me this feels like the ideal way to use AI to make art—generate elements, use AI to adjust them to your needs, then combine with an artistic eye to create something that is AI generated, but filled with specific artistic intent This whole process left me yearning for being able to easily give text prompts to adjust the output. Certain things you can outpaint easily, like changing a collared shirt to a turtleneck, but things like perspective or lighting or pose would be incredible to adjust with text prompts and would take this from a toy to a true tool. Being able to reroll something with an additional text prompt would be ideal. Also, Midjourney is so much better than Dall E that it's a bummer to have to outpaint in a lesser program. MJ desperately needs outpainting. It's also a huge pain in the rear end that Dalle E outpainting can't take the whole image into account if you want to work with higher resolution images. feedmyleg fucked around with this message at 13:54 on Nov 8, 2022 |
# ¿ Nov 8, 2022 13:35 |
|
Another of my old radio dramas come to life... This time in the form of a magazine that adapts radio dramas into short stories. Depicted is the thrilling climax where former boxing champion Jonnie Kidd is kidnapped by a notorious gangster with a grudge, then flown above the grand canyon and forced to face a choice between being shot or jumping to his doom.
|
# ¿ Nov 8, 2022 20:53 |
|
Is there any web-based tool (with no complex setup) which allows me to rough in shapes and get finished images? I know the tech exists if you properly know what you're doing with this stuff, but I'm just a dummy who uses MJ on Discord. I tried loading a rough sketch of outlines into DALL-E but it just ignored it.
|
# ¿ Nov 15, 2022 21:58 |
|
pixaal posted:The Huggingface https://huggingface.co/spaces/anzorq/finetuned_diffusion is working right now, this breaks for hours at a time, but go to the img2img tab and it should do what you want, change the model from Arcane it's not very good you probably want it set to Mid Journey to start the others don't seem as well trained. This is perfect, thank you so much! I found some old sketches from a lost media project I want to try and generate full paintings of. My first few tests went... okay. I'll toy with it. Snowy posted:Any ideas on how I can get better, more consistent results? I’m getting lots of images that are way off the mark and cherry picked the ones above. Have you tried "animation model sheets"? This is what "animation model sheets of a cute hamster, multiple angles, chibi, in the style of ghibli" got me on MJ, first try:
|
# ¿ Nov 15, 2022 23:20 |
|
Post what you get, I'm curious what ends up resonating!
|
# ¿ Nov 16, 2022 00:16 |
|
deep dish peat moss posted:i made a post or 4 about doing this in mj v4 Amazing. This is exactly what I was hoping for, you're a mensch!
|
# ¿ Nov 16, 2022 14:28 |
|
How the heck is Remix not a default-on option on MJ? This is a total game-changer in inching this toward a proper tool. e: Also, I think the novelty of MJ4 is finally wearing off for me. I feel like I've played around with it a lot and found the limits of what it is and isn't good at, which just makes it so frustrating that the version of this that is actually going to do the things I want it to do is still a year or so away. Right now I can get some great results for what I want, but it takes a LOT of fiddling and experimenting and failure. Here's hoping I'm wrong and a lot of these developments are leapfrogged in the next few months. feedmyleg fucked around with this message at 16:48 on Nov 16, 2022 |
# ¿ Nov 16, 2022 16:45 |
|
I still have to redo hands in DALL-E like 70% of the time, but yeah I did notice a quality bump.
|
# ¿ Nov 17, 2022 13:20 |
|
Why does MidJourney make such bad dinosaurs It can make a decent T-Rex, but that's about it. And even then it does wacky things with the arms. Trying to make a triceratops just makes totally freakish monsters, even if you feed it seed images.
|
# ¿ Nov 21, 2022 08:46 |
|
Kharmakazy posted:I interrogated an image of a triceratops in SD and it thinks it's a trilops, which is a monster from a video game. I eventually tricked it by making a rhino then inpainting triceratops features little by little. So far I've been unable to replicate my success with a stegosaurus. Too unique, I think.
|
# ¿ Nov 21, 2022 21:35 |
|
Behold, the closest thing I can get to a stegosaurus after dozens of variations:
|
# ¿ Nov 22, 2022 10:52 |
|
That sucks. The way I maintain continuity with characters across multiple prompts is through putting in celebrities who "play" the characters. Otherwise you can't really generate multiple types of images of the same person easily, in my experience.
|
# ¿ Nov 24, 2022 18:01 |
|
I'm very interested in an image upscaler, would love some details from someone who's used it about what it's good at and what it's not. Does it really just focus on portraits and photos of subjects, or is it good at art and environments as well?
|
# ¿ Nov 24, 2022 18:18 |
|
Everyone post a duck
|
# ¿ Nov 28, 2022 01:18 |
|
Excellent ducks, all. What's the prompt? It looks so much like Steve Purcell's art but I didn't get similar results when I tried him before.
|
# ¿ Nov 28, 2022 11:27 |
|
"The cops said it was an accident. But me? I knew there was fowl play..."
|
# ¿ Nov 28, 2022 15:49 |
|
TheWorldsaStage posted:The purpose is horrifying but that design is sleek I mean, outside of the pods this is nothing new. The Soviets went nuts for these things in the late 60s and early 70s: ektachrome 35mm photograph of a 1970s Soviet kitchen device which dispenses liquid cheese, cup full of hot liquid cheese, product shot, promotional photo, product design, sleek design, Soviet 1960s design
|
# ¿ Nov 30, 2022 21:50 |
|
Cross-posting this since it's very relevant to the thread. I'm doing an illustrated Choose Your Own Adventure version of Jurassic Park in the Let's Play forum, a sort of side-quel that takes place the same day as the first movie. The whole thing is powered by (mostly) AI imagery, and hopefully will be a good bit of fun: Jurassic Park: Edge of Chaos Planning to incorporate AI text prompts, too, but we'll see how chaotic it gets.
|
# ¿ Dec 2, 2022 20:53 |
|
No. The dinosaurs will be very bad. Some of them will work okay, but there will be a lot of extra limbs.
|
# ¿ Dec 2, 2022 20:59 |
|
Yeah. I'm pretty hesitant at where this tech is going, but it's exciting being here right now.
|
# ¿ Dec 3, 2022 19:50 |
|
FunkyAl posted:Ok I will bite: explain art. Done.
|
# ¿ Dec 5, 2022 17:23 |
|
I've had a ton of failures throughout the day
|
# ¿ Dec 6, 2022 20:25 |
|
Rutibex posted:yeah I think what the developers are going for is a chatbot that won't lie and make up facts when you ask it for genuine factual information. If the chat bot understands that you understand its making things up than its ok Yeah, I was trying to get it to cast a fictional TV series, and it kept telling me it couldn't because it doesn't have knowledge of the future, or because such a series doesn't exist. I ultimately had to ask for suggestions for who should be cast before it finally caved and spit out some actors. It actually did a terrific job coming up with a cast at that point.
|
# ¿ Dec 7, 2022 12:55 |
|
"Make it sew!"
|
# ¿ Dec 7, 2022 21:00 |
|
Best I could get with image prompts: I find the trick is to use at least 2 images. Popoto posted:I tried making some pixel arts assets with midjourney. This poo poo is insane. There’s still a lot of touch ups to do but it’s gonna save me so many hours.
|
# ¿ Dec 7, 2022 21:54 |
|
Cabbages and Kings posted:I would encourage people to think about pixel waste and try to minimize it uh huh
|
# ¿ Dec 7, 2022 23:30 |
|
Yeah can you make him a bit... I don't know, beefier? Buffer? Chonkier? Swoller?
|
# ¿ Dec 8, 2022 00:16 |
|
Tried to do it with Data and Riker but the results felt too pornographic. That being said, this one is worth sharing:
|
# ¿ Dec 8, 2022 06:00 |
|
Yes but what if it is accurate and the computer has figured out how to predict tornados
|
# ¿ Dec 9, 2022 15:03 |
|
|
# ¿ May 14, 2024 18:35 |
|
Fuzz posted:Someone needs to train an AI based on the character art in the HADES games. Too easy.
|
# ¿ Dec 10, 2022 17:45 |