~ Paint By Numbers ~
To say A.I. has a mind of it's own is a paradox. In one sense, the statement couldn't be more true... but on the other hand, without an intelligence to guide the intelligence there wouldn't be much intelligible to ascertain.
I am probably going to get lynched for this post once the singularity occurs and A.I. takes over. Shhhhh
As a curator for the Alien Art Hive Community i've come across numerous posts that feature the use of generative art. In fact, generative art is one of our featured topics. There are numerous methods for using Artificial Intelligence solely to generate art as well as augmenting one's own work.
I will share some of those options and methods for you all to check out and play around with... but to be honest, most all of it has already been covered in depth by other awesome community members. I don't really feel like i have a lot of new information to add or share but i know there will be some folks on my blog that are not familiar with our Art community and the posts that have delved deep into this subject. So, i hope they don't mind me tagging them and adding their links.
Essentially, many of the most common methods could be linked and described here along with explaining the technical jargon but i will summarize it with an analogy of my own. Feel free to correct me in the comments below if necessary because i know this is a gross oversimplification.
Generative Art is basically using numbers to paint based on prompts that the A.I. uses pitting it's left brain vs. it's right brain against each other. This conflict refines each iteration and run to try and best come to the conclusion it deems to be what you asked for. I find it to be really cool that order is drawn out from mathematical chaos. Sometimes, it still results in chaos though...
~ Skynet Dreams ~
All the ways i have experimented so far have been free options. I am sure paying to play would produce faster, more efficient, and better results... but free ain't bad either.
There are two website platforms we will start with that you can try out and play around with for free (or upgrade to a paid service) if you want to. You don't have to know how to code or understand the technical terms because the platforms do most of the work. You just select what you want and input your options and press a few buttons and boom... you have art! They are easy to use and fun to explore.
The first one is ArtBreeder which has a bunch of cool options to try out. You should go and look and see what all you could do there to have some fun!
I made that portrait in ArtBreeder from a portrait i drew in pencil on paper. You can see the initial source portrait here below. The A.I. removed the piercing but other than that i really loved all the refinements. A very realistic recreation of my artwork that i was quite pleased with. I will probably tweak some of the settings and play around with all the options now that the main image is uploaded in there. It's crazy all the tweaking you can do on ArtBreeder.
I also made a post a long time ago called Giant Slayer and Artificial Intelligence when i tried out ArtBreeder for the first time. I took a character i drew and cropped his head and had the program make a portrait of him. Here is that outcome.
The second website is called DeepDreamGenerator which also has a lot of really cool features you can try out for free too.
I put in one of my old portraits that i did back in high school with Crayola Crayons, some ink pens, and i think some highlighter to move the wax around. It's a bit abstract so i thought the A.I. would have some fun with it and i was not wrong!
Here is the initial image i gave the program to work with.
I am currently still running the image back through to get to a deeper A.I. dream state... but so far i am on pass 5 of 13 on Dream Level 5. If i can get to 13 passes then i will level up to Dream Level 6 and have created a Rare Deep Dream! I will share the final result for you to compare at the end of the post.
Some have used one program to create an output artwork and then run it through the other program to tweak the art further. The fun and experimentation is really only limited by your own creativity. These tools are just now in their infancy and this technology is burgeoning... so hop on now and enjoy the wild ride!
~ A Lot of Help from my Friends ~
My primary method of Augmenting my Art with Artificial Intelligence has been through Google Colab Notebooks. @kaliyuga has a nice post with a Library of AI Google Colab Notebooks. She also created a community called The Latent Space for A.I. generated art if you are interested in joining the fun over there.
I have to take my hat off (if i was wearing one) to @jotakrevs for all his hard work refining and bug fixing code for many of those Google Colab Notebooks. He and i spent a good deal of time troubleshooting the VQGAN+CLIP with video options notebook. For some reason that one really seems to be the buggiest. For all of Jotakrevs hard work (which i imagine isn't completely finished yet) i am going to make him a small beneficiary to some of my part of the rewards for this post. Thanks bro!
As we speak (and i type) i am running the VQGAN+CLIP with video options notebook in the background. Hopefully, no bugs or code errors will pop up and i will be able to share that video with you all at the end of this post. It's actually the first time to get a video in this fashion.
I would recommend reading all the information in that notebook before giving it a go... and it may be helpful to play around with the original notebook to get a bit more familiar with the process first. You can find the regular VQGAN+CLIP notebook to play around with at that link. You will have to translate the page to English unless you speak Spanish, but that is not so difficult.
Prior to this video option notebook method, i was having to run the normal gamut of just the images produced from the normal notebook and then splicing together the timelapse video results in an editor after the fact. I would do a run, save the final image, and then use that image as the starting point for the next run and so on.
Here is an example of the normal and more cumbersome method i was using. This example uses 7 runs of around 400 iterations per run all spliced and edited together. I also added some music for effect!
Yes, that's right... i guess the A.I. at Youtube realizes the A.I. output is NSFW. So, you have to be logged in to view the video. I will be adding the NSFW tag just to be safe here. I did file an appeal as this is an abstract output of art that has no real discernable issues that would truly justify the restriction in my humble opinion.
(Looks like my appeal was successful and the restriction has been lifted! yay)
I wanted to also take a little time to mention a few other people that have been very encouraging and supportive to one another as we all try to navigate our way through this stuff figuring it out together!
@dbddv01 was the first person i saw that was using A.I. to tell a story with a theme. He used a theme to create a fictional city guide to a music city. You have to go check out his posts with that theme because they are part of what truly inspired me to go and play around with all this... and look where we are now! His last post was also incredible. Here is the link to his first post on the city Broozel and in there are links to the continuation of his story! Utimate City Guide of Broozel for the enchanted traveller lost in music. A Rare non-existing books.
@lavista was also clever enough to have a theme and use the A.I. as an illustrator for creative writing. I would highly recommend his blog and to check out all the old posts from the beginning! I think you will be more than entertained and inspired when you finish! Here is his latest post titled The Forest Chronicles- The Portals. The creative writing alone is worth the visit to go check out the content... but the augmentation of A.I. illustrations is the icing on the cake!
And last but not least... (sorry if i left anyone else off my list here) is @juecoree who i think had some wonderful success experimenting. His last post After the Apocalypse- AI Art and Experiment was really cool along with his post Inside the Dungeon- AI Art and Experiment.
Kaliyuga has probably been the most prolific of us all. Go check out the blog there linked in her name and see what there is to see. Quite the spectrum of content and topics artistically expressed!
Jotakrevs blog has some amazing art and commentary on the processes. I would highly recommend going and checking some of the content he has produced! His portraits turned out amazing.
~ A.I. for the Assist ~
I personally don't believe that Artificial Intelligence of any kind is going to replace the analog artist or traditional mediums and methods. In fact, i really think it helps encourage the creativity and artistic side in those that don't feel to be artists themselves. It seems to me that the assistance provided by machine learning and all the different digital methods to generate or augment art is quite useful and inspiring.
For artists that perhaps doubt their skills or techniques... these digital options can be used to bolster their creative efforts and possibly influence them to grow and develop the skills that are lacking by being able to see it firsthand. I know there is a real concern for automation and it's impact on many industries and many livelihoods.
I really don't feel that concern is merited in the realm of art. Not unlike digital photography though, many people who actually suck will be able to be the next Ansel Adams or Leonardo DaVinci though because of their dependence on technology to do all the work for them. So, i guess it's kind of a double-edged sword.
Personally, i don't mind if folks are heavily dependent on the technology to produce art so long as they add or contribute something to the process. Not everyone continues art when they are a kid and grows to develop the skills and techniques. I do believe that everyone is an artist and somewhere along the line they turn away from that part of themselves.
So, using A.I. for the assist or as your personal illustrator is a great way to utilize the modern tools at our fingertips. I would encourage everyone to get back in touch with that part of themselves and playing around with some of the links in this post might just be a way to get that ball rolling!
~ NFTs & Final Forms ~
A couple of weeks ago the NFT Showroom, built here on the Hive Blockchain crossed 10,000 NFTs sold by artists on the platform! It's a pretty amazing feat and milestone to surpass and i am honored to have a few of my NFTs in the lot of the first 10k sold. I have collected a few as well along the way.
For this incredible achievement the NFT Showroom is hosting a contest which you can read about here for the rules to participate. There is going to be a lot of Hive that will be given away in prizes for Artists and Collectors. As of writing this post, there are 11 days left to get your entries in.
The cool thing is you don't have to be an artist you can simply be a collector. I thought that was a cool way to reward those that appreciate the artworks by including them in the celebration contest.
So, with this in mind, i will be participating by minting an NFT from some of the art in this post. If you want to see what i have up for sale you can click here to visit my NFT ShowRoom Gallery.
I said earlier that i would share the video that was produced from the notebook with those options. I wasn't able to take advantage of the slow motion feature so i just edited it to the slowest speed and added some music! Hopefully, my appeal for this video will go through as fast as the first one and it will be available for all by the time this post drops!
I have been running that video feature notebook in the background for a second video and hopefully, it will go without a hitch. Time will tell... hahahaha I've been running it for about 6 hours so far.
Now that we are close to the end of this post... time did tell and the slo-mo feature didn't work for me once again so i improvised! Thanks to @clayboyn for helping me troubleshoot cropping MP4s coming through in the clutch! It kind of works out better since i was able to add music again. I'm still waiting for some of the editing features before adding that as a final touch here!
I did say that i would share the DeepDream Generator image here at the end like i said i would. Rare Deep Dreams are any dreams or artworks produced by the site which went deeper than level 6.
I made it to the 3rd pass out of 19 at level 7... well beyond the limits of normal sanity. I guess this artwork is like inception... a dream within a dream. Let me know if you see any Electric Sheep in there!
So, here is the first video produced from the A.I. notebook experiment. After much travail, i just decided to go with it and get it minted. This will be by Genesis A.I. minted NFT. I hope you all like it. If the first one is this cool i really look forward to experimenting and playing around for future projects.
To celebrate the NFT Showroom 10k event i have listed the last 7 of 10 minted Genesis A.I. artworks for auction. The auction will be for the next seven days so don't miss your chance to scoop them up for only 10 Hive a pop.
Here is the link to hop over there and check them out. If you want editions 1-3 it will be a bit more costly. I would go for the auction price if i were you.
https://nftshowroom.com/gallery/castleberry_ai-augmentation-creations_red-glow-ai-genesis-mint
So, with a long day of creating all that went into this post and more i am going to call it a night! Thanks for taking the time to stop by and check out what's been going on! I appreciate your precious time spent here with me! I hope you all have the best week ever! Until next time... Castleberry signing out!
Bob Ross
