Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

DALL·E now available in beta

openai.com

Post reply on HN
Print view View original

551–560 of 579 posts

Re: DALL·E now available in beta

#551

Post by tomgp »

scrollaway wrote: post #519
POPOSYS wrote: post #511

Earlier quoted context omitted.

Unfortunately the content of that project is a hostage of Facebook now - similar to ransomware gangsters they force you to do something to get the data, in this case you need to create an account and take part in that global surveillance network. I do not understand why people do that.

Hyperbole will get you nowhere good (Just ask RMS). How about wording your comment in a way that highlights why it’s a shame these pictures aren’t accessible for those without a Facebook account, and skip the whole “you’re murdering puppies” bit?

Being a creepy sex pest will get you nowhere good (Just ask RMS).

The hyperbole is good marketing for a certain audience.

Re: DALL·E now available in beta

#553

Post by MrBra »

andybak wrote: post #497
ZeWaka wrote: post #319

Earlier quoted context omitted.

I mean, waifu2x and similar waifuxx libraries are free and open-source, there's really no reason to pay for it if you're working on a desktop.

Things have moved on a considerable amount since waifu2x Try https://github.com/n00mkrad/cupscale

I've recently updated waifu2x and I've seen it now supports lots of algorithms for different use cases and contexts and it also supports other tasks like frame interpolation. So could you briefly explain in what is cupscale better than it?

Re: DALL·E now available in beta

#554

Post by MrBra »

jazzyjackson wrote: post #381
Sohcahtoa82 wrote: post #163

The name "OpenAI" to me implies being open-source. I have an RTX 3080 and will likely be buying a 4090 when it comes out. Will I ever be able to generate these images locally, rather than having to use a paid service? I've done it with DALL-E Mini, but the images from that don't hold a candle to what DALL-E 2 produces.

From what i've seen it's all about the VRAM if you've got 60GB available to your GPU then maybe you can get close I'm really curious if Apple's unified memory architecture is of benefit here, especially a few years from now if we can start getting 128/256GB of shared RAM on the SoC

I was reading on Reddit that you would only need 12 GB of VRAM for DALL-E 2 to run on a local machine https://www.reddit.com/r/dalle2/comments/w3sbt3/comment/ih01....

Re: DALL·E now available in beta

#555

Post by MrBra »

ronsor wrote: post #232
Sohcahtoa82 wrote: post #163

The name "OpenAI" to me implies being open-source. I have an RTX 3080 and will likely be buying a 4090 when it comes out. Will I ever be able to generate these images locally, rather than having to use a paid service? I've done it with DALL-E Mini, but the images from that don't hold a candle to what DALL-E 2 produces.

I'm not sure if any current or next-generation GPU even has enough power to run DALL-E 2 locally. Anyway, OpenAI is unlikely to release the model. The situation will like it is with GPT-3; however, it's also likely another team will attempt to duplicate OpenAI's work.

Thanks to the amazing @lucidrains there's already an open-source implementation of DALL-E 2: https://github.com/lucidrains/DALLE2-pytorch and a pretrained model for it should be released within this year.

The same person is also at work on an open-source implementation of Google's Imagen which should be even better (and faster) than DALLE-2: https://github.com/lucidrains/imagen-pytorch.

This is possible because the original research papers behind DALLE-2 and Imagen were both publicly released.

Re: DALL·E now available in beta

#556

Post by andybak »

MrBra wrote: post #553
andybak wrote: post #497

Earlier quoted context omitted.

Things have moved on a considerable amount since waifu2x Try https://github.com/n00mkrad/cupscale

I've recently updated waifu2x and I've seen it now supports lots of algorithms for different use cases and contexts and it also supports other tasks like frame interpolation. So could you briefly explain in what is cupscale better than it?

Cupscale supports multiple models tuned for different types of content (including Anime like waifu2x): https://upscale.wiki/wiki/Model_Database

Re: DALL·E now available in beta

#558

Post by ryanmercer »

bemmu wrote: post #58

I was supposed to be making a video game, but got a bit sidetracked when DALL·E came out and made this website on the side: http://dailywrong.com/ (yes I should get SSL). It's like The Onion, but all the articles are made with GPT-3 and DALL·E. I start with an interesting DALL·E image, then describe it to GPT-3 and ask it for an Onion-like article on the topic. The results are surprisingly good.

This is the most wonderful thing ever.

Re: DALL·E now available in beta

#559

Post by ryanmercer »

nsxwolf wrote: post #234
peteforde wrote: post #228

I have been having a blast with DALL-E, spending about an hour a day trying out wild combinations and cracking my friends up. I cannot imagine getting bored of it; it's like getting bored with visual stimulus, or art in general. In fact, I've been glad to have a 50/day limit, because it helps me contain my hyperfocus instincts. The information about new pricing is, to me as someone just enjoying making crazy imagines…

I'm already bored of it. When you have everything, you have nothing.

Same. I generated several thousand images and found it a chore, outside of the daily theme on the discord server, to try and even think of anything to query. It was also discouraging when sometimes you'd hit pure gold for 4-5 of the 6 images, then you'd be lucky to get 1 out of the 6 that was worth saving for several more queries. Now it's down to 4 images and... yeah...

I'm not going to try and profit from the images, I don't need them for any business uses or anything, so to me it was a fun for a while and now just something I'll largely put out of mind.

Re: DALL·E now available in beta

#560

Post by ryanmercer »

nwienert wrote: post #378
peteforde wrote: post #366

Earlier quoted context omitted.

I don't know how to say this without sounding like a jerk, even if I bend over backwards to preface that this isn't my intent: this statement says more about your creativity and curiosity than a ceiling on how entertaining DALL-E can be to someone who could keep multiple instances busy, like grandma playing nine bingo cards at once. Knowing that it will only get better - animation cannot be far behind - makes me feel…

Dall-e has novelty, but no intent, meaning, originality. Yes the author can be creative at generating prompts, but visually I haven’t seen it generate anything that feels artistically interesting. If you want pre-existing concepts in novel combinations then yes it works. It’s good at “in the style of” but there’s no “in a new style”. It has a house style too that tends to feel Reddit-like.

> I haven’t seen it generate anything that feels artistically interesting.

I'd disagree.

- One of the first queries I did made some interesting chairs, I would genuinely buy the first if it was sanely priced: https://www.ryanmercer.com/ryansthoughts/2022/6/17/dall-e-2-...

- One of the first H R Giger inspired works I did made some really interesting computers, I would (and may) hang some of these https://www.ryanmercer.com/ryansthoughts/2022/6/19/dall-e-2-...

- The first wood carving query I did generated solid gold of Vikings eating pizza, this is the kind of thing I'd see in a hole in the wall restaurant and absolutely love https://www.ryanmercer.com/ryansthoughts/2022/6/17/dall-e-2-...

- The first "painting" in this series I may very well print and hang in my office https://www.ryanmercer.com/ryansthoughts/2022/6/27/dall-e-2-...

- These H.R. Giger chairs 100% look like something I'd expect to see in a modern art museum https://www.ryanmercer.com/ryansthoughts/2022/6/24/dall-e-2-...

- If you want to create new characters for something, and are lacking inspiration, I think it could be extremely useful to artists. For example these variations of Don't Hug Me I'm Scared https://www.ryanmercer.com/ryansthoughts/2022/6/21/dall-e-2-...

I've got thousands of queries, and a LOT of them have generated things I genuinely see as having artistic value, I've probably got 200~ images that I would 100% hang/display in my home ('woodcarving' and 'stone carving' queries rarely disappointed me)

is it some unique form of art, no, but can it produce works I want to see in a medium or style that already exists to a level that it is believable as authentic human art, absolutely.

People like me, with zero artistic ability, are able to take part in creating visually pleasing works. I imagine artists would also find great value in it, being able to feed a few queries in with what they are thinking of creating to draw inspiration, or even putting their own work in and generating variations that may lead to inspiration for new works.

Post reply on HN