|
Thy dinner done, Thy meal come, In belly as it is on plate [that's it that's the prompt]  
|
#
?
Jun 20, 2023 09:54
|
|
|
|
| # ? May 28, 2024 14:26 |
|
|
|
|
#
?
Jun 20, 2023 14:01
|
|
|
I like how it threw some Dali-style melting tableware in there. Just for the hell of it.
|
|
#
?
Jun 20, 2023 14:22
|
|
|
Aertuun posted:Gosh. I need to try out that model. The pictures you did were fantastic. Thank you! It's definitely my favorite model for illustration type images Sedgr posted:I like to do my drinking under the sea. These are a great mood! There's a joke about being drunk and wet here but I can't think of it. E: Oh I forgot to say I loved your Alice in Wonderland inspired images too! TheWorldsaStage fucked around with this message at 15:54 on Jun 20, 2023 |
|
#
?
Jun 20, 2023 14:51
|
|
I thought they were with Midjourney as well.
|
Can AI generate maps that aren't complete gibberish? Any kind of map, from a standard political map to Middle Earth to a treasure map.
|
|
#
?
Jun 20, 2023 15:31
|
|
|
There are specialized algorithmic map generators : Medieval city generator Planet generator Ect, Ect, but stable diffusion/midjourney is going spit out something that could be used with some work.
|
|
#
?
Jun 20, 2023 15:43
|
|
|
Count Roland posted:Can AI generate maps that aren't complete gibberish? Generally, yes. Stable Diffusion takes more work to get a clean result typically, Midjourney will generate gibberish text. Stable Diffusion: Prompt: A political map with states colored in Blue or Red based on their votes, solid colors with clean bold line work, vector art, digital art Style: Low Poly  Midjourney Map on aged parchment of a medieval region --ar 3:2 --upbeta --q 2 --s 750  Political map of a country, breaking down regions by color, digital art, vector graphics with bold lines and solid colors --ar 3:2 --upbeta --q 2 --s 750 
|
|
#
?
Jun 20, 2023 15:46
|
|
|
Wonder how long it'll be before MidJourney can throw in words related to (but not directly from) your prompt on things like maps. For example, this "digital map of the robotlands". It's pretty easy to imagine it eventually replacing the gibberish with words relevant to the prompt.  Will it be within another year? Recall it's been over a year since DALL-E 2 came out, now. So far there's the one on hugging face that can represent exact text, so I guess it'll just be a matter of plugging in additional text AI
|
|
#
?
Jun 20, 2023 15:53
|
|
|
Soulhunter posted:Midjourney Hah, the ocean on the left like every fantasy map ever.
|
|
#
?
Jun 20, 2023 16:19
|
|
|
scifi fantasy elevation map wrapped around a sphere, starlight background Negative prompt: font, captcha, finger, hand Steps: 20, Sampler: DDIM, CFG scale: 10, Seed: 1284387071, Size: 768x512, Model hash: 2336dbf342, Model: dreamshaper_631BakedVae, Version: v1.2.1  not what I expected but fun. I'm borrowing some concepts that have been done in the thread before on earlier models.It missed the mark on wrapping the map around the sphere (which I believe someone did using MJ and that worked wonderfully for a non Earth planet) I think it's the rest of what I tacked on to increase coherency because I had some fantasy worlds earlier I just like how the entire thing turned out. pixaal fucked around with this message at 16:26 on Jun 20, 2023 |
|
#
?
Jun 20, 2023 16:24
|
|
|
AARD VARKMAN posted:Wonder how long it'll be before MidJourney can throw in words related to (but not directly from) your prompt on things like maps. For example, this "digital map of the robotlands". It's pretty easy to imagine it eventually replacing the gibberish with words relevant to the prompt. I think it'll be a focus. The same things happened with hands where they got called out on it, and now hands are way better. Seems like every low hanging fruit imo Same thing with llm's not being able to spell, count, etc.
|
|
#
?
Jun 20, 2023 16:56
|
|
|
w00tmonger posted:I think it'll be a focus. The same things happened with hands where they got called out on it, and now hands are way better. Seems like every low hanging fruit imo yeah gpt4 ime has already fixed the spelling and what little counting i've done with it (not much)
|
|
#
?
Jun 20, 2023 16:59
|
|
|
well this is horrific, nicely done
|
|
#
?
Jun 20, 2023 17:12
|
|
|
Hi, I'd like to call in a request. Could I get Stankonia? By Outkast? But like a good example and a bad example, in terms of number of fingers. Anything with fingers=wrong and fingers=good, actually, if it doesn't want to make Stankonia. Edit: nm, no longer needed 
Mescal fucked around with this message at 01:52 on Jun 21, 2023 |
|
#
?
Jun 20, 2023 18:25
|
|
|
TheWorldsaStage posted:These are a great mood! There's a joke about being drunk and wet here but I can't think of it. Thanks! Everyone's stuff is great. Good work ai art thread! 
|
|
#
?
Jun 21, 2023 00:50
|
|
|
the interplanetary rally championship               
|
|
#
?
Jun 21, 2023 01:07
|
|
|
Marvel made the intro to their new television show using AI. https://twitter.com/PrimeTimeFilms_/status/1671441123259105280
|
|
#
?
Jun 21, 2023 13:12
|
|
|
BoldFace posted:Marvel made the intro to their new television show using AI. I get the feeling this sort of thing is soon going to instantly date a work to the mid-2020s, just like really ugly CGI did for the late 90s-early 2000s, or vaporwave to the mid-2010s.
|
|
#
?
Jun 21, 2023 13:24
|
|
|
BoldFace posted:Marvel made the intro to their new television show using AI. Make really nice super detailed super resolution images with a tile diffusion method. Pan and rotate, add some simple effects in post to make it seem more alive. - that's about all I'd do to attempt to make a TV show intro not have pulsating latent blobs. Oh god this is going to look so bad with streaming compression too. I still don't think AI is production ready, for filling in your gaps for a small project sure. But I somehow doubt this has the excuse of shoestring budget.
|
|
#
?
Jun 21, 2023 13:35
|
|
|
Fuschia tude posted:I get the feeling this sort of thing is soon going to instantly date a work to the mid-2020s, just like really ugly CGI did for the late 90s-early 2000s, or vaporwave to the mid-2010s. Yes, that's what I'm also very cognizant and very careful of when incorporating ai into my professional work. For instance: this intro already looks dated because when it was made, it was probably in the time of midjourney v3 or pre-LoRA SD 1.5
|
|
#
?
Jun 21, 2023 14:04
|
|
|
I could see it being an interesting choice if you're aim is to constantly change the opening or something. Like generate a new opening for every episode would seem like a good use case. Certainly marks it as very of this time and place.
|
|
#
?
Jun 21, 2023 14:08
|
|
|
doing a new opening every single ep could actually be p cool, especially if it meshes with the content of the show. Like idk if you had some time travel poo poo and what they do in the show changes the timeline the intro changes too.
|
|
#
?
Jun 21, 2023 14:21
|
|
|
BoldFace posted:Marvel made the intro to their new television show using AI. wow so they've already cut out tons of artists with this oof.
|
|
#
?
Jun 21, 2023 14:50
|
|
|
kinda coming around on the intro, except it's not ugly enough. most people keep trying to make AI art look cool or pretty, not enough are embracing ugliness. Like I keep trying to make cool sci-fi stuff but all my best stuff has been horrifying. Here's some I guess   
Roman fucked around with this message at 18:45 on Jun 21, 2023 |
|
#
?
Jun 21, 2023 18:43
|
|
|
KinkyJohn posted:Yes, that's what I'm also very cognizant and very careful of when incorporating ai into my professional work. For instance: this intro already looks dated because when it was made, it was probably in the time of midjourney v3 or pre-LoRA SD 1.5 There was a Disturbed music video made in Midjourney v3. IIRC it released like, the day after v4 came out, it was incredibly obviously outdated before it released, because when it came out everyone was excitedly playing with v4. https://www.youtube.com/watch?v=fpUpVznI4Yc
|
|
#
?
Jun 21, 2023 18:56
|
|
|
Matrix green shadowy government agents, 4k, vibrant, Marvel Universe Negative prompt: blurry, pristine Steps: 20, Sampler: DDIM, CFG scale: 7, Seed: 451063792, Size: 768x768, Model hash: 2336dbf342, Model: dreamshaper_631BakedVae, Version: v1.2.1  You could get this to move with a bit of effort. And whatever process you use will look like poo poo in a year.
|
|
#
?
Jun 21, 2023 19:55
|
|
|
pixaal posted:And whatever process you use will look like poo poo in a year.
|
|
#
?
Jun 21, 2023 20:15
|
|
|
  
|
|
#
?
Jun 21, 2023 21:17
|
|
|
Driftwood boxcar hobo is rad AF
|
|
#
?
Jun 21, 2023 21:18
|
|
|
  
|
|
#
?
Jun 22, 2023 00:00
|
|
|
BoldFace posted:Marvel made the intro to their new television show using AI. wow this is embarrassingly bad (as a corporate production) Mercury_Storm fucked around with this message at 01:02 on Jun 22, 2023 |
|
#
?
Jun 22, 2023 00:59
|
|
|
There's a nonzero chance that it's intentional just to cash in on the AI frenzy to get people talking about the show. We're talking about it, so it worked.
|
|
#
?
Jun 22, 2023 04:29
|
|
|
It's likely that it was made a while ago with older tech. Progress is measured in months with this stuff. I have seen people posting that they thought it looked good. The *~horrified reactions~* aren't universal. Roman fucked around with this message at 04:42 on Jun 22, 2023 |
|
#
?
Jun 22, 2023 04:37
|
|
|
Roman posted:It's also likely that it was made a while ago with older tech. Progress is measured in months with this stuff. Weeks, it seems like. Meanwhile the lead time on TV production seems barely any shorter than film, these days.
|
|
#
?
Jun 22, 2023 04:42
|
|
|
I've found my new passion. Prompting the AI for an image from a nonsense judgement call and then asking it what is in the image with clip interrogation.  "The absolute best thing in the world!" No negative prompting or anything gives results like:  a hand holding a broken out tv in the desert with a desert road in the background and a desert landscape, Aquirax Uno, anamorphic lens, a jigsaw puzzle, environmental art  a silver box with a bunch of silver objects inside of it on a table top with a chair in the background, Eduardo Paolozzi, finely detailed, a computer rendering, assemblage  a wooden box with a lid filled with cans and cans of liquid and a metal container with a lid, Eden Box, dieselpunk, concept art, serial art  a man riding a horse in the water on a trailer in the water with a man on it and a horse in the water behind him, Avgust Černigoj, book cover, a poster, action painting  a person holding a container of food and a bag of chips in it's hands, with a bag of chips in the other hand, Anne Said, product photo, a stock photo, verdadism  a woman is smiling while looking through a tube of fire with the caption the atoists's 'billes twealit, Auseklis Ozols, promotional image, a poster, interactive art  a cat with a sweater on sitting on a bed with a quote written below it that says, the abuubest be est be best about the to world, albert aublet, cute and funny, a photo, art & language  a stack of food sitting on top of a sandy beach next to the ocean with a caption that reads, atown dixie, Camille Bouvagne, product photo, a pastel, superflat  a little girl sitting at a table with a cake on it and a caption that reads, be be able best bo ti be, Else Alfelt, cute and funny, a pastel, les nabis  a person holding a bottle of alcohol in a box on a table with a window in the background and a hand holding a bottle of alcohol, Altichiero, product photo, a 3D render, aestheticism  a person holding a camera with a lens in it's hand and a lens in the back of the camera, Beeple, anamorphic lens, a raytraced image, photorealism  a little girl that is sitting down with a book in her hand and a book in her mouth and a book in her hand, Charlie Bowater, cgstudio, a photorealistic painting, photorealism Some takeaways. Stable Diffusion for all its power in understanding the connection between images and words...can't loving read. It also thinks some of the absolute best things are food and booze, boxes of junk and electronics, cameras, and happy children. Maybe the occasional man on a horse, or hot wet woman. And a kitty. I can't decide whether its a goon or thats some real meaning of life type stuff right there. But it made me either way.
|
|
#
?
Jun 22, 2023 05:02
|
|
|
 AI's idea of a Gary Larson cartoon: 
|
|
#
?
Jun 22, 2023 07:22
|
|
|
Mercury_Storm posted:AI's idea of a Gary Larson cartoon: I have no idea why this is so funny but I love it.
|
|
#
?
Jun 22, 2023 09:28
|
|
|
I cracked up at work yesterday going through some previous pages in this thread. I couldn't tell you why it was so funny but I rarely laugh that hard.
|
|
#
?
Jun 22, 2023 15:15
|
|
|
Sedgr posted:
I'm pretty sure this thing talks to the main character in a David Cronenberg movie and also has metaphorical sexual organs.
|
|
#
?
Jun 22, 2023 15:36
|
|
|
|
| # ? May 28, 2024 14:26 |
|
|
Flexclip, which I sometimes use to add subtitles to gifs easily to post dumb one off memes in discord. Anyway they have a (bad) AI feature in beta right now. It has text to video and it's more text to slide show the text to image AI produces the same quality only just one image. https://www.flexclip.com/ The AI stuff is under "tools" if you go in from the editor because you are already logged in. If you aren't logged in the homepage should be proud to be offering AI and direct you there. some of the frames on this bull one have some movement so I guess it's a touch more but not much. (most notably the moon scene the clouds move) https://i.imgur.com/8eS1GQZ.mp4 only got 2 scenes on this one https://i.imgur.com/oOV8l3i.mp4 pixaal fucked around with this message at 15:58 on Jun 22, 2023 |
|
#
?
Jun 22, 2023 15:39
|
|