|
|
# ? Jun 30, 2022 15:23 |
|
|
# ? May 30, 2024 07:22 |
|
wa27 posted:
Thanks for trying The Majula ones are pretty good, it knows that there is a beach and some towers. But weird that Dalle-2 knows nothing about Dark Souls, when Dalle-mini knows about it. e: from earlier in the thread Siegward of Catarina enjoying the sunset
|
# ? Jun 30, 2022 15:45 |
|
Almost exactly what I envisioned. Glorious, thank you.
|
# ? Jun 30, 2022 15:46 |
|
axolotl farmer posted:The Majula ones are pretty good, it knows that there is a beach and some towers. But weird that Dalle-2 knows nothing about Dark Souls, when Dalle-mini knows about it. Apparently Dalle mini wasn't related at all to Dalle-2, and that's why they changed their name. Which, if that's the case, they should have sent a C&D months ago because everyone thinks it's related in some way.
|
# ? Jun 30, 2022 15:58 |
|
Is there an equivalent thread for GPT3?
|
# ? Jun 30, 2022 17:11 |
|
Trash Trick posted:Is there an equivalent thread for GPT3? A couple yeah. There's a general procgen in PYF and a specific GPT3 in GBs
|
# ? Jun 30, 2022 17:52 |
|
Those look delicious.
|
# ? Jun 30, 2022 21:50 |
|
"Sub-Zero eating a sub sandwich on a submarine"
|
# ? Jun 30, 2022 22:19 |
|
wa27 posted:"Sub-Zero eating a sub sandwich on a submarine" wow
|
# ? Jun 30, 2022 22:23 |
|
Everyone knows this is a better prompt: "Dub-Zero eating a dub sandwich on a dubmarine"
|
# ? Jun 30, 2022 22:58 |
|
Yoshi Jjang posted:"an orca wearing board shorts, illustration" Love the "thought" process here. Orcas have a tail, but shorts have two holes, just make two tails per hole. Orcas have one tail, so that goes on one of the short's holes. Just make up a leg for the other hole. gently caress it, just add two legs for the leg holes!
|
# ? Jul 1, 2022 00:05 |
|
Wifi Toilet posted:gently caress it, just add two legs for the leg holes! this is dope
|
# ? Jul 1, 2022 00:07 |
|
TIP posted:this is dope Thats some av material. gently caress it, mods, male that my avatar. Tia.
|
# ? Jul 1, 2022 01:39 |
|
Okay, so the paladin needs to learn where you hold a sword and some need some cleanup, but drat. I guess I'll never want for fun custom portraits for the old Infinity Engine/Pillars of Eternity type games ever again. To get this look I wrote a bunch of short descriptions: "A dwarven warrior with blonde hair and a big grin. He carries a holy warhammer." Then all of them were followed by "Award winning oil painting by Thomas Cole. HD extremely detailed." I found they turned out a bit better if I had them all carry a weapon or a magic staff or whatever, although it was more likely to cut off the top of their heads. I tried to get glowing eyes or horns coming out of the forehead, but couldn't do it. I also have to share my full set of gnomish thieves because they're AMAZING "An old man with a handlebar mustache. He is wearing leather armor and aviator goggles. He is laughing. Award winning oil painting by Thomas Cole" the last set added "He is carrying a crossbow." Fintilgin fucked around with this message at 03:44 on Jul 1, 2022 |
# ? Jul 1, 2022 03:42 |
|
edging into robotnik turf on some of those
|
# ? Jul 1, 2022 04:27 |
|
TIP posted:Chewbacca as a Dragonball Z character Midjourney's better with anime apparently, behold the Chewbaccas. Chewbacca's Bizarre Adventure - easily my favorite Evangelion versions insist on being weird as poo poo, as is appropriate get in the chewbacca, shinji Chewbacca Wing Zero Kakashi Chewbacca Gotta catch all the Chewbaccas
|
# ? Jul 1, 2022 05:16 |
|
The paladin holding the sword that way is actually pretty interesting. It adds a level of symbolism to it, reminds me of a tarot card.
|
# ? Jul 1, 2022 12:43 |
|
Half-swording is a completely accurate thing from the late medieval period, one hand on the grip and the other halfway up the blade. All those pictures own, being able to generate quick unique art for a tabletop gaming session is awesome. I applied for DallE2 access but I'm not an artist or known in any way so I doubt I'll get in. Would someone be willing to generate a request for me? "Oil painting of a three-masted frigate under sail, the sun rising behind it and illuminating the sails"
|
# ? Jul 1, 2022 12:54 |
|
https://replicate.com/laion-ai/erlich + = Redonionking fucked around with this message at 14:01 on Jul 1, 2022 |
# ? Jul 1, 2022 13:43 |
|
Foot Lettuce sounds intriguing. Does it taste like feet? Or look like feet? Smell like feet? Or are feet somehow involved in its preparation, like stomping on grapes to make wine?
|
# ? Jul 1, 2022 14:23 |
|
Kylaer posted:Half-swording is a completely accurate thing from the late medieval period, one hand on the grip and the other halfway up the blade. All of these ended up pretty good:
|
# ? Jul 1, 2022 14:45 |
|
it's a schooner!
|
# ? Jul 1, 2022 16:06 |
|
the XKCD guy has access and also did something with ships https://twitter.com/xkcd/status/1532187680733134848 https://twitter.com/xkcd/status/1538350320300052480/photo/1
|
# ? Jul 1, 2022 16:18 |
|
wa27 posted:All of these ended up pretty good: Thank you! It got the concept pretty well.
|
# ? Jul 1, 2022 16:37 |
|
Someone with access to the good poo poo please enter 'a skeleton engulfed in flames playing a guitar solo to a sold out stadium, photo realistic' e: tia
|
# ? Jul 1, 2022 16:48 |
|
Also, if anyone has access to Dall-E 2, could you try the prompt "Landfill filled with funko pop figurines, award winning photography by Edward Burtynsky"? edit: Also you could try some AI prompts that neither midjourney nor dall-e mini nor disco diffusion has gotten yet: "a red panda with blue fur, wildlife photography" or "a blue apple cut in half" double edit: Thanks! ymgve fucked around with this message at 17:07 on Jul 1, 2022 |
# ? Jul 1, 2022 17:00 |
|
Rev. Melchisedech Howler posted:Someone with access to the good poo poo please enter 'a skeleton engulfed in flames playing a guitar solo to a sold out stadium, photo realistic' didn't really get the stadium part (or the photoreal part), but this one looks the best imo: ymgve posted:Also, if anyone has access to Dall-E 2, could you try the prompt "Landfill filled with funko pop figurines, award winning photography by Edward Burtynsky"? These all turned out kind of same-y
|
# ? Jul 1, 2022 17:06 |
|
from what I understand, 'photorealistic' is generally worse than calling it a photo, since everything labeled 'photorealistic' is non-photographic art. anyway midjourney Tunicate fucked around with this message at 17:46 on Jul 1, 2022 |
# ? Jul 1, 2022 17:24 |
|
Just a random thought - I wonder if you'd ever see Adobe license this technology and the results would be exported into separate layers in a .psd. I don't want or think human artists are going away, but I could see AI-assisted stuff being serious business.
|
# ? Jul 1, 2022 18:18 |
|
Does “award winning” actually affect the image? Does it add a level of tastefulness?
|
# ? Jul 1, 2022 18:44 |
|
External Organs posted:Just a random thought - I wonder if you'd ever see Adobe license this technology and the results would be exported into separate layers in a .psd. Adobe has been gradualky adding neural powered features to photoshop already, just in a more controlled way. I believe the new content aware fill is powered by that.
|
# ? Jul 1, 2022 18:55 |
|
^^^ The fill tool has been like that for years, yeah. There's now more stuff like automatic subject (and sky?) selection and so on as well. Snowy posted:Does “award winning” actually affect the image? Does it add a level of tastefulness? I can't speak specifically to Dall-e 2 or these pictures, but in general yes it can. The models are trained on huge datasets that are tagged with stuff like that, so "award winning" would be more associated with stuff people like. Depends if it's from deviantart or fetlife or whatever of course.
|
# ? Jul 1, 2022 19:19 |
|
Tunicate posted:Adobe has been gradualky adding neural powered features to photoshop already, just in a more controlled way. I believe the new content aware fill is powered by that. Oh interesting! I haven't messed around with Photoshop probably in 10-15 years, cool to learn.
|
# ? Jul 1, 2022 19:21 |
|
External Organs posted:Just a random thought - I wonder if you'd ever see Adobe license this technology and the results would be exported into separate layers in a .psd. Wouldn't work without a complete retooling of how the AI creates images Right now it's all a "package deal" - the AI generates the full picture all at once, there are no layers to separate
|
# ? Jul 1, 2022 19:21 |
|
ymgve posted:Wouldn't work without a complete retooling of how the AI creates images Dall-E can directly manipulate results within an-already-generated image though, or even edit an image you upload yourself. Layers might not be unimportant anymore if the AI has the context for the alteration. https://openai.com/dall-e-2/ Specifically the deal with the flamingo bit, you can do this right now with Dall-E proper.
|
# ? Jul 1, 2022 20:20 |
|
Another thing that's funny to me is that Dall-e reminds me a lot of this notorious holodeck scene from star trek voyager (starts at 0:50). We all thought it was absurd but look at us now. Clowns, all of us!
|
# ? Jul 1, 2022 20:30 |
|
External Organs posted:Another thing that's funny to me is that Dall-e reminds me a lot of this notorious holodeck scene from star trek voyager (starts at 0:50). We all thought it was absurd but look at us now. Clowns, all of us! so the computer was just making up some random bullshit, and it happened to be right
|
# ? Jul 1, 2022 20:42 |
|
ymgve posted:Wouldn't work without a complete retooling of how the AI creates images Load out of focus ladybug. Ask Dall-E to show a ladybug in focus. Get the same ladybug now in focus. https://twitter.com/thenickdude1/status/1538033909081944065 Rutibex posted:so the computer was just making up some random bullshit, and it happened to be right you can consider it as bullshit that makes the most sense to include out of the billions of similar patterns out there. EVIL Gibson fucked around with this message at 20:57 on Jul 1, 2022 |
# ? Jul 1, 2022 20:44 |
|
Rutibex posted:so the computer was just making up some random bullshit, and it happened to be right We've been been closer to bladerunner_enhance.mov
|
# ? Jul 1, 2022 20:46 |
|
|
# ? May 30, 2024 07:22 |
|
Jesus I hadn't messed with Dall-E's image manipulation thing before, this is the real power. I snapped a picture looking out to my back yard. You can then use a paintbrush to erase a portion of the image, then Dall-E will attempt to put whatever it is you ask it to via text prompt into just that area. I chose a complex area with intersecting lines of the concrete and puddle since it rained earlier. Looked like this before generation: Results: "a perfectly smooth chrome sphere" "a translucent cube made of acrylic" "a pile of sandwiches" The sphere one it attempted to keep the puddle, sort of, but just ignored the line in the concrete. For the cube one the internal reflections don't seem to match too much, but does a pretty good job reflecting a lot of what is supposed to be around it, while deciding that the straight angled line is now a curve. The pile of sandwiches is a bit too sunny imho, but it did add a bit of shadow that's too sharp vs. the surrounding area. I then applied the paintbrush to the entire right side of the picture, then asked it to generate "photograph of a backyard after a light rain" and these are the results: This is essentially the content-aware function in Photoshop but on the extreme end of wholly-building entire portions of the photo with text context. It isn't perfect of course, but for a single 10~ second blot of time and it gives you multiple results is pretty impressive. Keeps the wet concrete most of the time, keeps the angle of things correct for where they photo was taken, lighting is correct. I then erased the bottom portion of the concrete and asked it for "deep dark hole to the center of the earth" Asked it for "a high performance sports bike"
|
# ? Jul 1, 2022 21:23 |