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

DALL-E Chess in Jungle and Dunes

imgur.com

Post reply on HN
Print view View original

31–40 of 57 posts

Re: DALL-E Chess in Jungle and Dunes

#31

Post by legutierr »

jstanley wrote: post #22
legutierr wrote: post #20

Earlier quoted context omitted.

Consider, however, that the output of these systems may not be copyrightable. So, when you move human involvement up to a higher layer of abstraction, it’s possible that the economics of the whole effort will be fundamentally transformed. Meaning, if these systems displace human artists, copyright itself may cease to be a motivator of economic activity—removing a significant incentive for the production of new art. A…

Sorry, I'm not seeing the downsides. That all sounds like a big improvement. And regarding point 2: do you think human artists are as good as they are without already having seen lots of great artworks produced by others? Human artists don't create art from an empty vacuum of nothingness either.

You don’t see a downside to there being fewer artists creating art?

Art benefits humanity not only because we consume it, but also because we produce it.

Making art is part of what makes us human.

Re: DALL-E Chess in Jungle and Dunes

#33

Post by dimmuborgir »

Lots of weird artifacts which are very hard to fix.

The argument that users can now generate professional grade art by bypassing artists entirely feels so strange. I have access to Dall-E. To generate images without artifacts, you have to do one of these: a) Do a lot of cherry-picking which can be expensive. b) Prompt should be about an abstract concept which can "tolerate" any number of artifacts. c) Prompt should be about a common/generic concept that you have already seen a lot of times on the internet.

I think the biggest use case of Dall-E will be in removing creative block for artists.

Re: DALL-E Chess in Jungle and Dunes

#34

Post by EwanG »

fab1an wrote: post #15
la64710 wrote: post #7

Somehow I find the Dall-E and othe AI generated pictures revolting … is it the choice of colors or what I don’t know? It’s like looking at an art piece without a soul ..

DALL-E has that quality, yes - I did a project where I tried to recreate my own photographs with DALL-E, which shows both its potential and limitations right now: https://twitter.com/fabianstelzer/status/1551663900776595461 (obviously ignoring the potential to conjure things you cannot photograph...)

And I suspect this points out one of the more common future uses of the tool. Not replacing photographers per se, but certainly taking a large chunk out of the already shrinking Stock Image market. If instead of having to find a stock image that I can use as the base for my work (usually with at least SOME tweaking) I can just describe what I'm actually trying for as a final product...

I'm not going to be getting rid of my Sony A7riv anytime soon, but this certainly would discourage me from trying to increase my library at Getty

Re: DALL-E Chess in Jungle and Dunes

#35

Post by arecurrence »

kuboble wrote: post #19
quadcore wrote: post #8

I dont know how artists feel about DALL-E but as an amateur I feel bad. "This should be forbidden" bad. I guess the root of this feeling is the same as the one Copilot gives OSS programmers, it feels like theft and copyright enfringement. The pictures in this case uses techniques and colors scheme widely used by illustrator in the entertainment industries. Some of them are even above the average quality and that's sc…

For what is worth, dall-e is great for exploring but it's nowhere near to being able to deliver a particular image you might have in your mind. I wanted a very particular, well defined scene: - A pig and a donkey play poker at the poker table. - The pig is using a computer while playing and we can see the screen of the pig. - The pig must look like a pig - The donkey must look like a donkey - The cards and chips must…

I suspect this is where an API and additional cost reductions will move the needle even before we improve the models themselves (which seems to be coming at a rapid pace right now). I can see a scenario like this working well in the future:

1. Get close via prompt debugging to what you want (effectively where you are now)

2. Run an image generation pipeline that creates 10,000 images or an infinite stream

3. Run each image through an 'image to text' step for vector similarity filtering

4. Take images that have very similar 'image to text' similarity scores to the original prompt and present to the user.

Once we can run models of this quality locally, it can even be a job that runs overnight and you wake up in the morning to a set of results to look at.

Re: DALL-E Chess in Jungle and Dunes

#36

Post by aetherson »

dimmuborgir wrote: post #33

Lots of weird artifacts which are very hard to fix. The argument that users can now generate professional grade art by bypassing artists entirely feels so strange. I have access to Dall-E. To generate images without artifacts, you have to do one of these: a) Do a lot of cherry-picking which can be expensive. b) Prompt should be about an abstract concept which can "tolerate" any number of artifacts. c) Prompt should b…

I don't think that Dall-E as it currently exists is a big threat to professional artists.

It's not super hard to imagine a noticeably improved version of Dall-E being a serious threat to professional artists, though. It's a question of how hard it will be to make some linear improvements to Dall-E.

As a side note, I created an image in MidJourney from a prompt, which got a fairly pretty image that had some serious facial asymmetry problems, then uploaded the image to Dall-E and erased half the face, letting Dall-E fill it back in with a much more symmetrical look, and that kinda felt like the future. Using various AI models as tools for the things they do best.

Re: DALL-E Chess in Jungle and Dunes

#37

Post by corysama »

seestem wrote: post #27

Is DALL-E deterministic, like if I type the same phrase it will always generate the same images?

I’m general, most Deep Learning processes these days are non-deterministic because 1) they only care about statistical correctness 2) there are some speed advantages in ignoring the existence of race condition bugs if you don’t care about being deterministic

Re: DALL-E Chess in Jungle and Dunes

#38

Post by corysama »

legutierr wrote: post #31
jstanley wrote: post #22

Earlier quoted context omitted.

Sorry, I'm not seeing the downsides. That all sounds like a big improvement. And regarding point 2: do you think human artists are as good as they are without already having seen lots of great artworks produced by others? Human artists don't create art from an empty vacuum of nothingness either.

You don’t see a downside to there being fewer artists creating art? Art benefits humanity not only because we consume it, but also because we produce it. Making art is part of what makes us human.

I’ve used Midjourney for months now. Artists love it. It will lead to fewer people creating art the same way the cars led to less people traveling. It’s like having a pre-concept artist for for concept artist. Instant style boards to run by your client.

Re: DALL-E Chess in Jungle and Dunes

#39

Post by minimaxir »

fab1an wrote: post #4

Amazed we haven't seen more DALL-E / Midjourney on HN. Probably the most astonishing new tech I've used since booting up a computer in the 90s. Just generating images is barely doing the tools justice though - you can create entire mini movies with it, like SALT (a 70s sci-fi adventure happening on Twitter): https://twitter.com/SALT_VERSE/status/1536799731774537733 DALL-E's inpainting feature is incredibly powerful t…

I made a post on trying to do absurd-but-controlled food photography with DALL-E 2 (https://minimaxir.com/2022/07/food-photography-ai/ ) which did get upvotes on HN but apparently did not make the front page (so may have been flagged?)

Re: DALL-E Chess in Jungle and Dunes

#40

Post by sbergot »

desindol wrote: post #6
fab1an wrote: post #4

Amazed we haven't seen more DALL-E / Midjourney on HN. Probably the most astonishing new tech I've used since booting up a computer in the 90s. Just generating images is barely doing the tools justice though - you can create entire mini movies with it, like SALT (a 70s sci-fi adventure happening on Twitter): https://twitter.com/SALT_VERSE/status/1536799731774537733 DALL-E's inpainting feature is incredibly powerful t…

Because it’s a novelty abused for attention with clickbait popculture references. It’s not ready for primetime.

Some people are already using generated images to replace stock art in low budget zines.
Post reply on HN