Topic: Bitcoin AI Generated Images (Read 617 times)

Wow, so is this site the origin of the photographs used by altcoins' teams when launching their tokens? That always looked shady to me, but I never imagined those could be AI generated photos of fake people.

I'm testing it right now and the results really pass the same professionalism impression from those images posted in ANN threads. I could say many of the results presented are actually from real CEOs, marketing and Blockchain specialists.

Appearances can be really deceiving.

This is definitely true! The quality, the lights, the atmosphere and the overall aesthetic is just simply sublime with these computer-generated AI images. Imagine, since these were created in just one website, the potential of businesses using this as their promotional pictures/videos would definitely entice people to spark their curiosity on checking out cryptocurrencies.

There is one downside, however, that these computer-generated AI images extend not only to Bitcoin but also to real-life images as well. I remember that this website https://this-person-does-not-exist.com/en generates thousands of new faces that are currently being used by some ANN threads on their team member page.

Everyone can make a sandwich, yet the sandwich industry is alive and well.

I think there's already a market for AI generated images. If not, there will soon be. And the cool thing is that it can be much faster than the iterative process we currently have with artists.

You could just post a topic and receive images that match it, then you select whichever you want to pay for. Done.

No, I'm not selling it. I just create the images for fun and to illustrate posts about Bitcoin, somehow it works like a BTC advertisement. Some people already thought about earning money selling those generated images and launching NFT collections, but I'm not sure if others would have interest on it, since they can acquire the AI and start creating images by themselves. Copyright might be another issue, since we don't own those images as our properties, as far as I know. Anyone browsing on Discord MJ's channels can get an image you created for themselves and use it as they wish.

Behold, one of my AI-generated masterpieces. The level of precision, accuracy and efficiency demonstrated by the AI system is nothing short of remarkable.

Those two things are very different.

A private key is a specific, unique series of symbols, from a vast set of combinations.

A generated image from text, although impressive, is way simpler to obtain that the private key, in terms of probabilities.

Think about it, given a single image, you could generate millions of images that look basically the same to the human eye but are represented by different raw numbers (each pixel is usually encoded by 3 eight-bit numbers). JPG compression is based on this concept, they can represent an image with another image that looks the same to a human, using way less information.

Then, you have the semantics. "A picture of a cat" has virtually an infinite number of outputs. You can have a close-up, another farther away, different colors or types of cats, etc. The interpretation of the prompt will basically generate an infinite amount of valid outputs.

And so on...

On the other hand, a private key is unique and even if it's one bit off, it simply won't work at all.

Sometimes I just give up when I see the AI isn't understanding what I mean. But I still try changing one word or another on the prompt before. I've never gone too far like 25 renders. I usually go until 4-5 attempts, although many of the results in OP were achieved in one try.

Here it always render pretty fast. I think I've never waited more than 3 minutes. It worths mentioning there is something called "fast" mode. By acquiring the subscription you have 15 hours of renders on this mode. After that, your prompts join a kind of row, where you have to wait other people's renders (in fast mode) to complete first, until you finish yours.

You can ask any doubts in Discord MJ server itself. There is a specific channel for this kind of doubt and people there are very helpful.

Yes, as you can see my Neanderthal on the last image has 6 fingers. Still realistic, but MJ doesn't know counting fingers with precision.

Indeed, a subscription is a must. Paid 30$ on the second tier for a month.

Added to the personal experience of @Kryptowerk, I also wrote mine. I hope it clears your doubts out, @kryptqnick.

I have read and even seen videos of people on Youtube comparing the implementation of artificial intelligence to art to the invention of the camera, long time ago.
Their point of view is that cameras obviously did not replace artist but rather made artists not to focus too much on painting things in the real world, since people started to take pictures of politicians, cities, urban spaces, etc.

Allegedly, this pushed artist to explore new styles and draw more fantastical and surrealistic things which are only to be found within the imagination.
I am not sure how comparable was that invention of the camera to the artificial intelligence.

What do you think, do you believe it is a fair point what those people say?

I would say that the technology have slowly killed the living of many people. Usually, those who are not related online/digital but in exchange of it, it is that many opportunities have also opened up but only to those who have a knowledge on online/digital.

Maybe those traditional artist can try upgrading and try doing digital arts because this can be the future. If not then they are free to find a new career. I just do not say that the pictures shared by the OP are all nice. This wasn't the first time I saw this but there is a one thread here last time who upload almost the same pictures. It seems that we are now being invaded by the AI but I still prefer those good old drawings tbh.

I can only say what I experienced about 3 months ago. Things are changing very quickly, so take it with a grain of "probably already different".
- As you cannot directly influence the composition of these images (at least with MJ) but only indirectly by tweeking some parameters and adjusting your prompt, it usually took between 5 - 25 tries (renders) to come close to what I had in mind. Sometimes it also seemed impossible to really get there
- Rendertimes depend a lot on activity, servers were often overwhelmed so you would wait between maybe 15 seconds to 10 minutes just for one image
- it's a lot of trial and error but also needs some knowledge on advanced parameters to get to the desired result. Back then there were already some decent guides and a good discussion forum, so you should be able to find a lot
- imho postwork (in PS, or Gimp etc) is necessary almost always if you want to use these images in a professional context
- don't expect to get too much done with the free version, 25 renders are gone in no-time
- have fun

I just want to say that op's images are amazing. My  Internet connection is very weak right now, but I think I'll try playing with AI later. I tried it before with some online services where you could also give prompts and choose styles, but the images, while sort of beautiful, were often too bizarre, usually because of objects not being fully there, faces and bodied generally being muffled or overstretched, etc. Yours are much better. I'd love to hear more about these results. Like, how many tries does it take to make a good image, or how many minutes/hours it took to get all the images in the original post? I see you've mentioned a free trial of 25 images, but also that you've purchased a subscription.

I have to admit that AI is really growing rapidly, I haven't thought about it being able to actually decrypt bitcoin's private key, but it is really replacing humans in many areas. It is both worrisome and gratifying at the same time to see this development. The more technology develops, the more benefits it will bring to people, but at the same time, it will also take away a lot of people's jobs. As for whether AI can figure out bitcoin's private key, it's a question that should be taken care of, anything is possible.

As a hobbyist artist, I must say I have got mixed feelings about artificial intelligence and art.
On one hand, I believe it is cool how the technology has advanced to the point a program can understand the concept behind an artistic piece, this was science fiction not long ago. On the other hand, I am afraid this could negatively impact my fellow professional artists, who already struggle to make a living out their passion...

Anyways, cool pictures. OP.

Well maybe this can be possible if AI has the brain of a quantum computer  but we are totally doomed since they might predict everything  but before that maybe bitcoin cryptography change too.

That is what people say about the risk of quantum computers for bitcoin.

I believe all hackers would have to do is to discover the 12 secret words, and their respective sequence, from 2048 available words.

However, it must be a very powerful and fast program to find the correct solution in an acceptable time period, as there is an insane number of different possible combinations. At some point AIs might become a threat on this matter, but I guess it's too far from happening yet.

Well lets see how long it will take for an AI to take on the bitcoin cryptography. If it is possible to find private keys, then yes we actually will be doomed!

If you would actually read anything in this thread you'd notice most was soley done by AI indeed, some other pictures were slightly post-edited.
There is such thing as an obviously AI-generated-image anymore. Some are, yes, but in the right hands I bet you would not be able to tell which is which.
AI is not getting smart, it's still very dumb actually. Just read some discussions about chatgpt. That being said, these "AI-apps" are great tools for producers of any kind and are already taken off a ton of workload from real humans.

Like it or not AI is getting smarter day by day, maybe we are gonna doomed by the AI like some movies said But it is good pictures to can be wallpaper on a phone or computer, are those really generated 100% by AI

they already are. peole use gpt-3 for article content from 2022 already, as for quality all depend on setup and seting you can honestly creat high quality article wit hai.
as for ai art, in the first palce what you mean orignality? concept can be created by you when prompt the generated image,

Very good, well done!

---

Updated OP with more AI generated Bitcoin images. I didn't resist and purchased the monthly subscription to continue doing more experiments while trying to improve the final quality of the images.

I hope you like the new images!

thanks, and sorry to disappoint you. But I'm not the girl on the image. It was a totally random AI generated anime face. if you want to see me then here's my own portrait made using midjourney AI.

I would say it's 85% match to my actual face. And the AI did a great job turning my ugly face to this. The facial features are there but slightly changed in order to match the prompt I guess.

Using center makes it good looking IMO. Also are you the girl in the images?

These are what came out from my try for this in AI. Could do much more but I am lazy so can't think of what else to feed the AI to get more good results.
These two are created by Nightcafe AI. I have also tried midjourney which is very good, but you need premium subscription in order to use that unlimited times. But I have tried some with my free trial to create a portrait from my own image, and it was bang on.
As AI is progressing so fast and are getting capable of doing these crazy stuffs, I wonder what it could do to the market if someone use it in wrong way.

 It is that imperfection "perhaps" that makes it art... for some.

 Those amounts that you mention remind us that above all we are in a business that seeks profit rather than creating works of art, unfortunately for those who think like this, art cannot be massive.

Oh! and fortunately for society this is still the case.

You need to spend 24$-30$ bucks monthly to use the AI without limits, and have a decent length of spare time to try different combinations until reaching the results you are looking for. And the results aren't totally smooth. There are sitll many imperfections someone would need to fix manually. As you can see the bull I've posted in OP has an extra frontal undeveloped leg.

Probably so. There are AIs modifying and creating songs as well, although the quality is low and sometimes bizarre, morbid.

I'm impressed with some results in many ways. I can't say they are original, but sometimes the AI recovers concepts which are amazing, but were forgotten for a long time already in history. Things that people nowadays couldn't re-create or reproduce with the same accuracy if it wasn't for the AI. Especially if you want to portray past decades and centuries on your images.

That's your opinion and I disagree.
- There are many original artworks that combine things in a way never seen before.
- AI is just a tool, it will never replace traditional art, but can be utilized for inspiration, as a basis for further modification etc.
- It can be used to copy existing art and create endless variations, but it can also be used to create something completely new. A good analogy is sampling in music.
- Sure, there is a lot of ugly, unoriginal, clichee, copy-cat AI images. A lot is an understatement. Definitely more than in traditional art but, like I said before, not exclusively.


I want to add that it depends on the type of AI-generator that you use.
Also, I would advise to pick some of the examples above and not combine them all, just adding everything doesn't necessarily produce better results.

These AI "art" generators make the work of low quality crypto news sites really simple, now they don't have to hire artist to draw a physical bitcoin for 1000th time, they can have AI do it in seconds for free. Maybe some of these news sites would even use AI to write articles, I'm sure the quality won't suffer because it's all nonsense either way.

The quality of AI art is impressive, but still it's nothing more than copying of existing human art, it lacks originality, it lacks concepts. It's only good for creating boring and forgettable illustrations for cheap products.

Try doing this and see what happens:


Found this hint on Youtube and that is what helped me to generate some of the images in OP. You can also find hints at Midjourney's Discord - chat - prompt chat or community forums - prompt faqs.

Great images!
I tried myself to generate some and it was a disaster. I guess you have to learn some tricks.

Thanks. Like I said it's been almost a quarter of a year that I created these using Midjourney. So don't remember the exact prompts. The BTC logo and some details were later edited in PS.

I remember using something like "most epic Kubrick movie scene" for the first two. Also setting the aspect ratio to a cinematic format like 21:9 or similar (or 215:100 I think?) helps for the AI to immediatelly go for a more movie-like look.
I think for the 3rd render with Guy Fawkes face I added "award winning photography" but in a painters style.

It was quite some fun to combine over-the-top-elements, sometimes even contradicting each other. Will probably try and see how far Midjourney has evolved since then, in the upcoming weeks.

Awesome bitcoin pumpkin world! I liked you managed to create your images seen from distance. They bring much more elements, details and composition to the scene. It truly looks like a HQ's story scene and just by looking that we can imagine there is a whole story going on that world.

Could you share which prompt keyword you use to cause that effect?

As @Maus0728 repaired the difference between months ago to now, I believe we are being helpful training the AI and guiding it through its learnship process. It's still a beta test and the AI is learning in real time to become smarter, more accurate with results and probably will be able to execute more challenging tasks futurely, as technology never stops evolving.

In some time, once it achieves the goals I mentioned above, there will be better use for it, without any doubts.

For now it's nice to put our imagination to work, materializing images and situations we thought could exist only in our minds.

Nice to see these images. Only the thing I never understand about AI generated images is, why are they being generated in the first place using AI? I mean AI should be used for something better. May be to make blockchain even high throughput and putting bitcoin into right direction.

I am not against these Cool images but I’m thinking it’s time for the AI’s to be put into good use.

I mean even fellow user like Mause can make such images with their hands!! 

Some nice AI-renders from OP here. It's amazing how well things like cinematic lighting works when done by AI.
I played around with that stuff about 3 months ago. Personally I prefere imaginative art more than realism.

Here are some results (inspired by Halloween themed stuff):

Finally showing your true face 🤣 this is some sick fantasy you have about bitcoin. I wonder what other dark fantasies you have🤔. How could you fantasize about a 14 year old getting fucked by a bull? Good thing the bot knew what to do, lol.


Here some more.


Would like to know if bitcoin has a pussy for a bull to fuck?
I just want to know, why a bull has to rape an underage?🤣
@franky1, just kidding man.

 

they are very christian with results too, i was not banned, but putting in "bull having sex with a bitcoin" just displayed 2 cows standing next to each other

I think even the group of 4 images counts towards the 25 free limit, because I didn't render too many images separately and reached that limit anyway.

Yes, maybe you could try through a private chat with the AI's bot. I tried it today and prompt command worked, but I had already hit the limit, so no deal.

You are going to be banned if you write forbidden words. The AI is very strict against NSFW content and very sensible at same time. Some words aren't even improper, but they will be considered as such. I was banned for several minutes just because I used the words "belly button".

I know that, the one I posted is actually a variant, though I don' know if using all U images count towards your 25 free limit or not? If it doesn't then I could go through all of them to see the best one, note this was my first time on discord using this AI.
It has a feature I really dislike, it shows everybody's images in a scrolling page which is being updated with a few images every second, these kind of settings could be changed somehow but I didn't go through them all. Also knowing all or at least most of the functional commands would be helpful but that requires you to spend hours reading their guides and tutorials.

I also have a NSFW idea but am not sure if the bot accepts to render them or not.

Following is considered NSFW, reader discretion is advised.

Something like "a young supermodel blonde with big booty and tities wearing bikini, surrounded on an alley by 5 big mean green trolls with evil smiles on their faces, the girl looks really scared but the trolls only want to give her bitcoins because they're her onlyfans."

Or "kid batman with his parents get jumped after a movie night on a dark alley, suddenly the thief looks at Martha's pearly melons and points the gun at Bruce asking Martha to pull her pants down and bends over, he rapes Martha infront of Bruce and his father and then kills them both, then he gets nervous and tries to flee the crime scene, but a few bitcoin falls off his pockets, which later turns out Joker had touched them leaving finger print on them, hence the fued between Joker and Batman." This won't look good on a single image, maybe using GIF is better for this.

Lol that was a whole new scenario for a Batman movie.🤣

Great results! You can render each image separately by clicking the U1, U2, U3, U4 respective button below the images, after they are generated by the AI. So you can also discard the ones you didn't like or which aren't accurate.

The possibilities are infinite. The limit goes until your imagination and creativity can reach. It's just a matter of time until they introduce convincing movements, like short movies to this feature.

The AI images or say text are really improving these days and some of them like this one are providing really cool images with full creativity.I have also seen good number of threads on forum describing these AI related images or say generating logos or images for bitcoin and forum with good results.The technology has improved and advanced to high levels and in coming years we might see they writing up good articles also with complete knowledge.The AI is new era of technology.

The app uses so much  resources, lol the size of this thing is huge, enjoy these bitcoins, hoard them collect them while you can, well this is AI's era now.
 

I didn't know about this previous thread. I enjoyed those pictures as well! Midjourney is great! It's indeed under constant improvement and development.

Unfortunatelly I already hit my free trial limit of 25 generated images. Now they require a subscription in order to continue using the service. 30$ monthly to have access to unlimited number of generated images.

I think it worths, especially for people working on this area. If I had some spare money right now I would purchase it!

Observing the BTC symbol generated from this thread[1], it looks like after a few months that A.I is constantly learning, the logo in this post is pretty much perfect compared from the previous thread. No distorted shapes, colors or any deformation is presented.

[1] https://bitcointalksearch.org/topic/m.60725349

those are some pretty good AI renders.
previous generation AI images badly cropped,layered and merged pictures together where you can see the blurred layering and such

AI has come long way in last couple years.
im impressed by the quality of these results

Yesterday I stumbled across youtube videos displaying AI generated images creating new versions for what could be games and movies of our modern days, if they were made in the 1980's dark fantasy style. I thought it fascinating and found they were created at a Discord channel called Midjourney, where anyone can have access and create the images they want, through the AI, by writting simple words as prompt command.

Then I saw someone creating a Bitcoin related image, and started trying as well. You have to be creative with the words to reach the most accurate and interesting results possible.

I would like to ask you to try as well and share your results on this thread. I will be posting mine as well.

This is what I've done so far with the AI:

---