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

I used DALL·E 2 to generate a logo

jacobmartins.com

Post reply on HN
Print view View original

371–380 of 386 posts

Re: I used DALL·E 2 to generate a logo

#371

Post by karpour »

capableweb wrote: post #317
karpour wrote: post #314

I find this completely unethical. You're basically exploiting every single artist whose art was used - without agreement - as training data for Dall-E. It would be different if all the training data was art that was explicitly licensed for this.

Isn't this similar to how every single piece of art is created? You look at bunch of propriety art, copy it to learn how to create it yourself (just for learning, not to sell those copies), eventually learn enough to start being able to create art from scratch and then start selling your unique art? Would watching a lot of animated movies in order to learn how to create good animated movies yourself be unethical as w…

No, it's not. Art created by humans is not just looking at other art and trying to copy it. It's a culmination of a whole life of experiences and the personality of the artist. Looking at other peoples art is inspiration and useful for learning technique, but only a very small part of the bigger picture.

Re: I used DALL·E 2 to generate a logo

#372

Post by karpour »

sabertoothed wrote: post #319
karpour wrote: post #314

I find this completely unethical. You're basically exploiting every single artist whose art was used - without agreement - as training data for Dall-E. It would be different if all the training data was art that was explicitly licensed for this.

I have very mixed feelings on this topic. I share your sentiment -- BUT: Human brains also use anything the human can see, feel, hear for training. And what you produce in terms of creative outcome is a result of your experiences. But you don't owe anyone anything for training your human brain -- even if you use your brain to sell paintings, music etc.

I think it's easy to - seeing the results - draw too many parallels between artificial neural networks and human brains. Art created by humans is very different. Dall-E gets fed tagged images and produces images that match tags. Art created by humans works on an entirely different level.

And I stand by my point, it should be artists who decide whether their work should be used as training data for networks that get commercialized. If your work is used as training data, it is essentially an integral part of a product that is being sold without consent. Does this sound ethical?

Re: I used DALL·E 2 to generate a logo

#373

Post by jherrmann »

cube2222 wrote: post #49
burlesona wrote: post #47

I’m fascinated by how much this is exactly like working with a human artist who doesn’t really understand the domain that you are wanting to represent with an image. Iterate, iterate, iterate. It seems like the most valuable thing this could do is get some of that early exploration out of the way faster and easier than a human can do it, get to two or three concepts that feel like they’re in the neighborhood, and the…

Agreed. At the end of the article I also described a bit how I would see the evolution of such a tool, and it looks like we're thinking very similarly. --- Though I think the real breakthrough will come when Dall-e gets 10-100x cheaper (and faster). I would then envision the following process of working with it (which is really just an optimization on top of what I’ve been doing now): 1. You write a phrase. 2. You ar…

No post body was provided.

Re: I used DALL·E 2 to generate a logo

#374

Post by Terretta »

cube2222 wrote: post #120
allenu wrote: post #114

Earlier quoted context omitted.

I was thinking something similar. The editing process is still a human one, and I agree that the one chosen was weaker than a lot of the intermediate choices. It's a matter of taste, obviously, but to me the red ball with a nondescript sketched square around it feels unfinished. The yellow cartoony logos look more finished and professional to me.

Appreciate the feedback! I'll keep it mind, as I might still end up choosing a different one. The chosen one is closer to my original vision, but you do have a point that the yellow ones look more polished.

Strongly agree with others here that you skipped better options.

Also, since time immemorial, databases are cylinders and data comes in cubes.

For logo purposes, these are both strong, while the second adds “personality”:

https://i.imgur.com/j6P4Oh4.jpg

https://i.imgur.com/kM23GZV.jpg

I really like the design breaking out of the strong circle, and your hard hat idea was great. That last one could have been your logo “as is”!

Though you could consider replacing the green cubes with cylinders, or simply hand add rubix cube lines to these green cubes to make them data cubes.

https://duckduckgo.com/?q=data+cube&t=ha&va=j&ia=images&iax=...

Thanks for sharing the process!

Re: I used DALL·E 2 to generate a logo

#375

Post by cube2222 »

rockbruno wrote: post #327
cube2222 wrote: post #2

Hey, author here, happy to answer any questions! The logo was created for OctoSQL[0] and in the article you can find a lot of sample phrase-image combinations, as it describes the whole path (generation, variation, editing) I went down. Let me know what you think! And btw. if you get access take a look at [1] before you start using it. A ton of useful bits and pieces for your phrases. TLDR: DALL·E 2 is really cool, t…

I also tried to make it generate an icon for a product and I managed to get it to show me interesting things, but never got to make it actually draw it as one. Do you remember which prompt resulted in this macOS-ish app shape? https://jacobmartins.com/images/dalle2/DALL%C2%B7E%202022-08...

Hey!

I didn't prompt anything specifically, it came after a line of variations from a definitely-not-icon-looking picture.

Though I'd try tags like "iOS icon".

Re: I used DALL·E 2 to generate a logo

#376

Post by HmOh6WbJ »

mikecx wrote: post #79

Not sure if this will be considered off topic, my apologies if so. The article says that octopi is the plural of octopus, but it's actually octopuses. Octopus is originally Greek, not Latin and thus does not get the Latin plural -i, but instead would get the Greek plural -odes. Since it ends in a way English can deal with, the commonly accepted usage is octopuses (English) over octopodes (Greek) with octopi being the…

Similarly: cyclops -> cyclopodes

Re: I used DALL·E 2 to generate a logo

#377

Post by monkeynotes »

pyinstallwoes wrote: post #370
monkeynotes wrote: post #169

When AI reaches the point where we can talk to a system like DALL.E in real time and work with it to solve a problem, it's game over. Art will become a commodity. Human art and ai art will be indistinguishable, "artists" will become as common as "photographers" since the inception of digital photography and social media. Movie and TV scripts will be iterative with a creative director and AI working together. Animatio…

I think the burden shifts towards being able to imagine and describe the fantasy. Novelty, and artistic creativity is still required. You can bring a horse to water, but you can’t force the horse to drink it. Many humans don’t use their imagination let alone have the eloquence to describe a search space that contains novelty.

How is this any different to working with a human artist? If I wanted the real world Salvador Dali to draw me a picture of a kebab being eaten by a badger I'd have to tell him that's what I want. I'd also need to educate baby Dali first, feed him all the art and information he can take so that he has a model of the world he's operating in. I'll need to supply Dali with context of prior art, educate him on styles, literature, language, and all the other things that shape a human mind.

As for the humans that don't use their imagination, maybe they never want to talk to an AI artist, just as many humans don't care about art at all. Millions of humans don't care about social news, and yet FaceBook algos pump out content for people all day long.

Re: I used DALL·E 2 to generate a logo

#378

Post by dormento »

DonHopkins wrote: post #203
VladimirGolovin wrote: post #135

Earlier quoted context omitted.

Kai's Power Tools.

https://news.ycombinator.com/item?id=27288454 Love him or hate him (and I do both), Kai was all about cultivating his adulating cult of personality and dazzling everyone with his totally unique breathtakingly beautiful bespoke UIs! How can you possibly begrudge him and his fans of that simple pleasure? ;) In the modest liner notes of one of the KPT CDROMS, Kai wrote a charming rambling story about how he was once pas…

This was some really interesting reading, thank you internet stranger :)

Re: I used DALL·E 2 to generate a logo

#379

Post by nodja »

spyder wrote: post #354
nodja wrote: post #237

Earlier quoted context omitted.

The model starts from 64x64 8bit RGB image of noise (random pixels) so technically 1 in 3_145_728 (64x64x256x3) but most will probably be very close to each other as the color difference won't be that much. The image is then further upsampled by two other models which will change some details, but shouldn't affect the general composition of an image.

Maybe I'm wrong, but with these diffusion models there is randomness in every sampling step too not just in the initialization and they can have 1000 steps to generate a single image.

Ah good point, this would introduce more variation if the initial noise is close, but if the initial noise is exactly the same it probably means it was initialized with the same seed and the rest of the generation will be the same since the random algorithms are deterministic.

Re: I used DALL·E 2 to generate a logo

#380

Post by pyinstallwoes »

monkeynotes wrote: post #377
pyinstallwoes wrote: post #370

Earlier quoted context omitted.

I think the burden shifts towards being able to imagine and describe the fantasy. Novelty, and artistic creativity is still required. You can bring a horse to water, but you can’t force the horse to drink it. Many humans don’t use their imagination let alone have the eloquence to describe a search space that contains novelty.

How is this any different to working with a human artist? If I wanted the real world Salvador Dali to draw me a picture of a kebab being eaten by a badger I'd have to tell him that's what I want. I'd also need to educate baby Dali first, feed him all the art and information he can take so that he has a model of the world he's operating in. I'll need to supply Dali with context of prior art, educate him on styles, lit…

Basically, yeah. https://knowyourmeme.com/photos/1331836
Post reply on HN