Dall-E is cool but can we talk for a second about how great Tim is? Not only an artist as a designer but also a cool, humble and focused person. No wonder why he is on the team!
@@kangtheconqueror8359 How would you stop it? It will replace us all, and any countries which decide to stop development will simply be rendered obsolete by other countries which will continue. The only solution is to change our socioeconomic structure to allow for us all to benefit from the technology, rather than simply benefitting a very small subset of the entire species.
@@karenreddy Gather a large group of a resistance to sabotage and destroy all AI servers and technology. Which won’t happen so..just be depressed af and hate life that I’ll never reach my dreams because the industry I want to apart of will be replaced by machines and eventually it will destroy us.
I felt bad for Tim a few times because you can tell he was stressed about this whole situation. He essentially had to prove his worth to the team while up against a computer that spits out 10 images in 10 seconds. I hope if you ever do anything like this again to please give Tim some reassurance (or at least show the reassurance on camera) because this kind of broke my heart watching this guy put all the effort he had into this challenge. All-in-all, Tim, you did a phenomenal job, and the MKBHD team is lucky to have you! Your images were stunning, and your care and attention to detail were unparalleled. Keep up the good work!
You will probably like to know, that DALL•E can't be used for commercial purposes ( for now ), so visual artists won't lose their job ( FOR NOW )... EDIT: Actually, I think it can be used for commercial purposes now, so...
@@krtkks watch it again and take note of his body language - it says it all. He was extremely stressed the entire time, even stating the Tiger Atlantis challenge "broke" him. This technology almost directly challenges his position, his expertise, and his role with the team. Whether it was overtly stated or not, Tim felt threatened by DALL-E. Even if Marques never said he was going to lose his job, Tim was thinking that, and perception is reality.
As an artist myself, this whole video gives me anxiety. TIm did a phenomenal job, but I would be lying if I said my desire to root for him in this situation didn't come with a human bias toward other humans. Ultimately, this video proves how much time and human effort had to go into producing these images, vs the AI who (aside from all of the data and time tht has already gone into programming it) required relatively no time/thought. Things are going to change a lot soon, for all of us.
Imagine telling your grandchildren about the old job of the “graphic designer” and them being in disbelief that an actual person used to spend all that time and effort to make every detail on an image when the computer could just make any photos, and even videos, that you asked for instantly? I’m worried AI is going to disproportionately benefit the wealthy, and eliminate their need for a lot of The Poors.
Well, this is why humans use utensils. Not to make the work of an human, but to make repetitive or hard works feasable and much cheaper. Maybe we will have better covers for fantasy books 😂
yeah, design is soon going to go the way of the weaver, the sower, the telephone operator.. automated away by the ever-increasing power of technology. the only thing we can do is prepare for the storm ahead, and adapt.
@@Dejawolfs How do you plan for a storm that you know nothing about? How do you prepare for every job being automated within our lifetimes? It’s a scary future!
@@tyler.walker only thing i know is that you somehow pass into law a base salary for every person. the reality however is probably that all wealth will pass into the hands of an extreme few individuals.
Tim’s creativity should be more featured here. As a graphic designer myself I am blown away by his attention to details and composition. It’s very interesting and fun to see a fellow artist’s thought process while creating, I could never be that thorough with mine since a lot of concentration and creative juice is needed to just come up with the idea, much more put it on canvas. He’s a very talented artist! Would love to see more of Tim and his work!
@@squeebploozer not everything that requires a "design" or "decision making" is "graphic design". Matte painting, collage, photo manipulation, are not graphic design
I love how although Tim has mad skills when it comes to Photoshop, he still took a second to watch a tutorial. That just shows me you can't always be the best at every aspect of your job. It's okay to seek help sometimes.
I learned a while ago you can’t always know everything, but it’s good to know where to find it when you need it. Also just good to stay on top of new tools and workflows
Many apps are complex with thousands of features...you're not gonna know it all at all times, and apps are actively developed so staying up to date to such a degree on all apps you use is impossible. The key is to know what the capabilities of the app are, so you can research and learn the feature when you need it.
Any professional programmer will tell you, a good chunk of their job is just Googling for existing solutions/algorithms people have already made or guides on how to do something specific. It's the same to a certain extent for most sciencific/artistic jobs.
There is something comforting about seeing a professional graphic designer pull up a tutorial on how to do things. Shout out to Unmesh and PiXimperfect for delivering the best Photoshop tutorials on RUclips!
I was nervous for him! People looking over my shoulder while I'm working on a project and hearing "i like it yeah" or really any comment is just so uncalled for. I hate getting critiques while I'm in the middle of something
The only word I can use to describe this video is "fascinating", seeing an expert explain his thought process and seeing how much time it takes to make those edits really makes me appreciate the work even more. Good job Tim, I wouldn't replace you with an AI for sure
@@Ez-se2dl IA can help as a tool like in the nvidia canvas check it out but ask a computer to have creativity is something we us humans have so lets say a convination betwen is the key but ask a IA to make the whole work is like ask digytal artist to create whitout a pc youre askyn for the tools to do the whole work aint no balance on that at the end we make the IA we give the IA the comands is all about the porcents of participation aint no gift for lazy people level up ...🍄
@@murraybritton6729 Its definitely more than one year. Dall -E and all these AIs are first of all currently not available to public so not everyone can use these. Secondly, I have noticed that the more object specific you get, Dall-E struggles. Like if you gave it a proper tech object like mac studio or maybe iphone 13, it will struggle way more. Plus it struggles with human faces, like if they specifically need Marques in a picture it will struggle but a graphic designer knows won't. Like take the thumbnail from the Mac studio video on main channel with marques sitting on the mac studio. I don't think Dall-E will be able to create anything even close to it What Dall-E does is very impressive. It excels a lot in representing those everyday objects, like the goat or the tiger, astronaut etc considering the time it takes but its still perfect enough to replace a full blown graphic designer
@@AdityaGupta-om8ez these struggles are specifically because the data set Dalle-2 was trained on was curated to prevent copyright infringement, false identity creation, and adult imagery. It isn’t a limit of the AI or it’s technique, it’s an intentional limit placed by the designers so the public could use it without creating legal issues. It’s why Google’s Imagen AI isn’t going to have a public testing because they didn’t limit their data set the same way. The minute you allow the public to create realistic Ao generated images of politicians, corporate products, or real people doing things- you create severe legal issues. It will be working through those legal quandaries that slows the roll out of this to the public, not the state of the technology itself.
@@AdityaGupta-om8ez I think one of the big shortcomings too is a lack of revisions, you can't explain to it why the feathers don't look good, and how to fix them. There's a lot of things a human can just do more concisely, while this is kind of like image roulette.
Despite how hard I try to remain optimistic, it’s pretty stressful watching AI progress as a young 3D artist. I quite honestly can’t imagine life being worth living for me if the visual art industry is automated. It’s difficult not to get emotional about it because Im so passionate about my career and one day, all the work I put in/experience I have could mean essentially nothing. I always question “just why would you replace artists?” because it seems a bit evil on the surface. Artists typically love what they do and find fulfillment from their work. This isn’t hamburger flipping. Additionally, artists are already constantly shamed for all their quirks but only accepted because their skills prove useful to many businesses who rely on their creative talent. I cant imagine how poorly artists would be treated by the business class if their skills served no monetary value. I’d rather be f&@king dead than work an boring, awful job for decades just because society hasn’t figured out how to automate it yet. The only thing that keeps me optimistic is that I love what I do so much I shouldn’t really care if a robot can do it better than me. I do it because it brings me peace and calms my mind. It’s no secret, art is therapy for many artists. I also hope people love supporting other people and don’t care as much about support robots. Handmade clothing and jewelry still exists despite much of the field being automated so that’s somewhat promising. So hopefully when all digital art can be automated, our society will find a way to value and support digital artists. Idk. I’m always cautious to publicly comment on the topic because it’s so emotionally charged, I’m afraid I’ll change my mind tomorrow. Nonetheless, this is how I often feel about it when my anxiety goes through the roof. I hope my perspective serves some purpose other than to be mocked 😅 If so, I make a video about it to inspire digital artists to keep pursuing their dreams. The script is already written haha
Go for it bro, we will support humans rather than robots, when it comes to art, i need to feel that there is a human touch behind this, if it is done by a robot, it doesn't matter if it is 10x better, it won't mean anything to me.
This is how I choose to think about it too. Just because a robot can do it faster and nicer doesn't always mean it's better. Art is and always will be about the artist and the context. As they say, art is all about perspective. I think manes lanes puts it quite nicely up there, it doesn't mean anything if it wasn't made by a person.
very-bad-things-to-self went up dramatically in the rust belt when they automated all those manufacturing jobs. It's not going to end well for a number of artists. It's also not going to end well for the researchers. Putting your name in bold on a ground breaking paper that directly results in tens of thousands of people losing their livelihoods is how you make a *lot* of enemies. But you already know how this goes: no one's going to listen, we're all going to hate it, and then 20 years maybe we'll reconsider. Climate change, leaded gasoline, cfcs, asbestos, plastics, car centric suburbia, surveillance, Afghanistan, social media-- I mean we can just keep going.
this is just the greedy ones trying to take the job of everyone just because they can. these guys are pure evil and do not deserve anything good from society. I cant believe so many people on the comments are exited for the AI this is the dehumanization of creativity and art.
Deepmind created an Ai called gato, which uses the same technique as dalle, but it can do basic household chores and answer questions about its environment. Many more jobs could be taken.
Okay, this video has now put into perspective how insane Dall-E actually is, it’s easy to overlook how it’s not just “stitching” photos together, it’s actually manipulating and creating light and everything else that is so easy to overlook as an outsider.
Yeah this is what I saw as well. Even if Tim's arguably followed the prompt closer, all three of Dall E's (maybe somewhat less on the robot one tho) looked like complete photos while Tim's looked photoshopped.
I was impressed by DALL-E when I saw how many ready-made images Tim needed to complete the task. It wasn't just one human against one AI, it was a human team with access to nature, cameras and art collections against an AI alone.
@@HoriaCristescu DALL-E needed 650 million images to do its thing, that's the library it has committed into memory during its training as the weights of its processing nodes.
I just wanted to give Tim a hug....I could feel his anxiety and how this competition is unfair...I know it was just a playful time but it shows the importance to recognize the human value involved in a process and how sensitive and hard can be all this new context where living in... It's an one way path
That Tim dude seems like such a genuinely good human being also being very intelligent, creative and funny! What a wonderful life affirming experience to see him and watch him work. Thank you so much for making this.
@@Exitof99 You think there wont be any diminishing returns? Its very hard to improve on what they already have, it wont be exponential increase thats for sure.
@@outlander234 It seems a lot of the results are based on the passes it f does at some stage, I forget the exact term, so presently, it's also that the current version probably could do better given more passes. But still, the present version is still having a hard time generating useable results. It does well in some areas, but there are defects that are easy to pick up on, weird fractal spots, missing/extra fingers, and so on. I think there is still a long way to go yet. If you've seen Stable Diffusion, it looks like that might surpass the noise-based method that Dall•E 2 uses.
tim was such a great guy, i felt bad for him so seriously that i got teary-eyed ,heart racing like crazy, am a graphic designer too i know how stressed and intimidating this is , having to compete with an A.I knowing how intgeligent those thing has gotten this days you really dont know what to expect, also thought of u losing ur job to a computer flashing right before your face is not a joke it not funny at all. no matter how confident you are with design, u remember am competing with a robot that might be better, i absolutely was in tims shoe at the moment, my heart pounding like crazy not knowing my fate yet, i definately just started having plan b for a moment, thinking if i lose my job what my next move. tim we smoked it bro........... hell yeah tim i wish to know u in person
No, thank you sir! Been a big fan of your channel for a long time and your content has been incredibly helpful over the years! Always finding so many creative workflows in your videos that have changed mine for the better. Your insight and knowledge are deeply appreciated! Much respect to you and your channel!
That is the current value of AI. Even at this point it can make art creation far more efficient, increase the supply of visual ideas and thus devalue it, even though a designer is still needed.
Yes, that is the best part of the current image generators, I wish these AI programmers focused on automating the idea generation phase as well as the "boring" manual parts of art. I like UV mapping and retopology, but if there was an AI that could give me instantly useable UVs and clean topology, I would be ecstatic. AI could empower small artists in many ways, coloring keyframes for animation, help with figuring out perspective, manipulating existing images in a non destructive workflow, etc. But the current use of AI art with prompt based image generation and cloning existing artist styles, is just taking the human part out of art, Wich is a massive oversight that the executives and corporations will use to save money, because why pay a human artist a liveable wage if they can just use an image generator to make 95% of the art and have a low payed artist just refine and do small touch-ups?
@@qaztim11 it's easy to automate prompt generation, but not ideas which fit a specific need, style, with countless variables requiring subtlety of judgment. Maybe someday, but we're still far. It would require several levels beyond what we have today, including project comprehension, by itself. AI currently isn't taking artists... It's taking craftsmen. People who practice a craft (which I have). But people with good ideas, directors, artists, will still remain competitive. In fact, maybe even more so, as they will have more freedom and ability to leverage their ideas with AI. If you are creative, even with as early as we are in the process, it's possible to develop some interesting stylistic approaches to imagery by combining known styles. Even humans draw from what we know. The art remains human, as it is guided by humans, but the craft will be AI driven.
@@karenreddy I would assume that in the future the AI image generation changes so that people can give better guidance of what they want rather than just brief description in words. But for now the big challenges still lies in improving the actual image creation process. How to make the user able to describe what he wants better is then also another thing.
Nothing can replace the creativity, originality and the personal touch of an artist, it is the authenticity of his/her piece of art that matters and not the works of a gadget or a machine and that is the big difference. Two thumbs-up for you Tim. 👍👍😊
For anyone interested, the painting is a depiction of the Swedish queen Kristina, Who held the crown for a time in Sweden during the 1600s. The painting is at the national gallery in Stockholm, and it could be said to be a "tour de force" of the royal spirit and legitimacy to rule, through allegories of the spirit; the blowing wind i the handkerchief, the water fountain in the background, and the sphere, representing the world.
I find it interesting how most people who watch MKBHD are generally interested in the advancement of tech, tech that has and inevitability will replace all human jobs, yet we all (myself included) are still rooting for Tim!! AI is coming from everyone though…
For some poor souls it'll come sooner than the others, like for Tim. If AI gets evolved to a certain point where it can easily does most of the jobs then I'm pretty sure the economic system will take a shift in a new direction/reworked to benefit the people thus people might not go broke but for now, Artist are about to go broke for the sake of " progress".
I for one welcome AI. My only fear is how we humans go about embracing it and how willing we are to change for a nicer but different world. Current systems that are deeply embedded into our culture like capitalism(which requires human labor to work or we'll be homeless because AIs have taken all the jobs) need to be wildly rethought or completely replaced. The biggest issues are the words "different" and "change" because of how so many people are religiously married to these systems and ideas which inturn will or could cause a lot of political strife as people are unwilling and or unable to imagine or embrace new systems and ideas. Best of luck to us wise apes. :)
@@poopoodemon7928 You wouldn't be embracing this AI if you were an artist. It's essentialy a slap in the face to creativity/originality. Imagine honing your skills all these years for someone to just take some free pictures and write an AI that ultimately replaces you.
@@MusicComet Skipping the whole "hurt ego" thing as that's something personal and requires self-help, we should differently re-structure our society so that being replaced wont result in a financial/materialistic problem.
@@MusicComet Trust me on this - AI will create more job roles than it eliminates. Us humans will just have to do what we do best, which is to adapt and use the AI to advance our abilities and speed.
The ones that adapt the best will still be needed to do the prompt, the edit, the design, the concepto. AI is just a tool. In the old days you were a carpenteer, now you operate the machine that makes the chairs.
I am only half way into this video and I wanted to say that Tim is reminding me of my daughter who works as a professional graphic artist for a production company. The stress I see her go through and the pressure she is under on a regular basis is palpable. My heart goes out to you, good sir. I am looking forward to the reveal at the end of this video.// edit - I just watched the reveal. What a relief. Great work, Tim, you killed it.
scary part is the time comparison. yes, Tim is master of his craft, but compare that results payment vs. resource cost then its starting to be pretty scary. one great development would be to ask DALLE for draft images and made final editing with human
@@fitybux4664 I don't think you understand. DALLE creates images that have never been created before. They are saying use the images DALLE gives you and make final edits.
This was a fun video highlighting how well a development team can work together. I'm impressed at the knowledge, skill, and level of expertise Tim showed in his creations. Dall-E can replace any entry level graphic artist or help someone like me that needs graphic design but it can't yet replace the intuitive artistry of a professional designer.
Tim definitely deserves a bonus for this. Working for a RUclipsr who probably makes millions while he makes a regular person wage and he gets to be basically threatened with a massively powerful AI and forced to “prove his worth” idk I’m never that person, but this felt tone deaf af and like watching a kid be bullied.
I agree, tone deaf is a good way to put it. No one in the video displayed much empathy for Tim’s experience/stress, not sensitive to the situation at all.
also, relaaaaax, human artists will still be used for lots of projects and movies/games etc. AI makes stuff that looks cool but it doesn't look genuine, it has lots of errors in the details AND it's not consistent in the least. You still need a skilled [insert job here] to control the AI and bring it to it's true potential.. For example, i see lots of pics out of MidJourney that i never managed to make it do, and those have a lot of specific photography/art key words in them...
imagine making a game for example, it would be pretty shit if every picture, every character is not consistent from a style or even facial features point of view with the rest. And that's what you get with these AI generators... It's pretty hit or miss
Hi Tim, you're AMAZING! You can explain what u did and adjust as it is needed and u provide a story, that provides so much more information about the build process. U bring in your own aesthetics and thoughts, a machine can not do (at least not now). Just keep doing it, that is what drives creativity.
The goat and the deer are super impressive on Tim’s part, awesome to see it come together as he worked on it The robot women one was mind blowing for Dall E tho, the computers were so realistic and they had space between them and they did completely cover the wall, making it seem as if it’s actually possible The way Tim did that part, while still cool, just wasn’t realistic, by taking a front on image and stitching them together, then giving it perspective, it makes it look really flat as you lose all depth for the PC screen Still very impressive though
Technological advancement and automation disproportionately benefits the already well off. Replacing jobs and with the jobs that still exist, pocketing all the extra profit. The average adjusted wage has not gone up much from 1960 despite huge leaps and bounds in efficiency. Normal people undeniably loose power with DALL-E.
Love the video the only thing I wish is if you would show a full screen image of the photos side-by-side instead of in the lower corner So we could see a bit more of the detail
''now the training arc for dalle2 begins'' or '' this is dalle1 and this is what is beyond dalle 1 it is called dalle2 and to go further beyond this is dalle3
As a fellow graphic designer , I feel your pain Tim , LOL. But you did AWESOME! my jaw dropped at your works , and I’m not just saying it to say it 😁 tho I will say all the stress and anxiety of feeling like you had to prove yourself was very relatable lmao. But I still feel like your work DEFINITELY had more feeling than the AI. maybe it’s from being in the profession for so many years , but I could tell it was crafted with care and fun imagination 💖
As a fellow AI designer , I feel your pain DALL-E , LOL. But you did AWESOME! my jaw dropped at your works , and I’m not just saying it to say it 😁 tho I will say all the stress and anxiety of feeling like you had to prove yourself was very relatable lmao. But I still feel like your work DEFINITELY had more feeling than Tim. maybe it’s from being in the profession for so many years , but I could tell it was crafted with care and fun imagination 💖
Photo manipulation is a completely different thing and many graphic designers don't do or like doing. We use the same tools but that's about it. Tim however did a great job here!
As a creative professional, I’m most interested in exploring how Dall-E can help give unique inspiration images or rough starting points that can launch my own creative work. I doubt I would ever rely on it to complete a finished piece but for components in the process it will be hugely valuable.
I think Tim is super good at his job and it was awesome to see him working. Kind of feels like a person who would be awesome to have as a friend. Also, if you had little more budget, Adobe stock for example, he would have had even better resources and more to choose from to do those really good compositions.
As a Graphic Designer with over 18 years of experience, I have catastrophized EVERY generational advancement of tools as "gonna put me out of a job!" Now, after playing with GPT-3, DALL-E 2, MidJourney, etc. I see these as amazing time-saving tools! In my experience, the first 80% of a design is the "easy" part, and the last 20% is the "hard" part. Because of AI, I can now spend 99% of my time on that last 20%!
When used by a professional that combines design, painting, editing and prompting in a masterful level it may become one big time-saving and productivity increasing tool. The amateurs just can play with it, pros will be the ones having the best from it and mostly the ones that adapt the best.
Sure, with the way they pay designers today, graphic designers are forced to entrepreneurship. They're gonna have receptionists doing designs with a.i.
@@Nalexandros847 You’re not going to have a job soon, ai will be able to do everything without need for a human artist. ai is a tool for artists only right now, only for a very short window of time until humans are no longer needed. Artists may end up being paid as prompt engineers until AGI arrives
As much as I was rooting for Tim, this is just Dall-E 2. Future Dall-Es can improve at a much faster rate and do more things than any human possibly can. It’s just a matter of time before AI comes for all jobs done on a computer. This includes accountants, financial analysts, even software engineers and developers.
I'm not too sure about software engineers though... But it seems like the lucky winners of this whole mess is musicians since AI can't create a decent music on par with music industry standards.
@@PK-ow1kj honestly listening to ai music would defeat the whole purpose of music. Plus artists make most of their money from live performances and the whole celebrity aspect of it, ai definitely can't copy that
@@PK-ow1kj As a software engineer I can totally see how AI will do a lot of the simplier tasks. The designer would simply state what they need at runtime and instantly the AI writes the program to make it do that. Code neatness and structure is a human need. The AI will just hack out something that works.
the unnerving reality is the fact that this was a contest at all shows we're not far from the point where the AI will win. Especially if we're taking into account time and cost... Sorry Tim =/
The reality is that for no real life non-twitter-commission paid artist ever do they regularly make things that dall-e can make. The way Dall-e works is it makes a profile for each word, then creates a sort of "platonic ideal" for that word out of existing images, and riffs off that, then combines them. For things like original character designs where there's not a billion labeled images to pull from, or anything where the producer wants any amount of control over the final product besides a few words, dall-e is only useful for quick idea-generating concept art type stuff. The real useful stuff here from a professional perspective are the openai tools that artists can use that help to minimize the more tedious parts of digital art without sacrificing creativity or control, like inpainting
@@acblook actually it is not making images from existing images. Coming from an AI machine learning engineering background, it's a bit more interesting than that :)...Yes it's been trained on real images but what it generates Is in fact completely original as in, never existed before on any platform. Its is not putting together parts from other pictures, it's genuinely mapping out each pixle based on the model. Look onto the math behind the models it's truly amazing
The point in the future is not to replace him but give him that as a tool. Imagine to say you want that bird feathers on that deer. It’s fun but too random to be useful. I am sure you could do that if the let you use it.
Exactly what I was thinking! It's the perfect tool for him to use as a Graphic Designer to get it to do the heavy lifting of brainstorming ideas, so he can focus his efforts more on the later steps of making the content, yielding even better results.
@@iPl4yeRTv Wait, so he'll tell the AI to make 100 unique, incredible images of a concept he wants to create, then he'll spend half a day making his own, singular, similar concept for... who? Himself? That's all well and good, but if he wants to make a living doing his art, who would ever pay him when an AI can do it 1000x faster with 1000x more options... for free? A lot of people in the comments just don't understand what we're seeing here. This changes everything.
@@ChrisCapel you dont understand what design is and what its meant to do, the AI tech simply gives good results on a purely visual scale ... but the designer's skills are not defined by their ability to make good looking visuals but rather their ability to solve problems and reason with the client as to what will actually help them ... AI is never gonna be able to do the critical thinking that a human does especially in context of specific problems that need to be solved Design is not just art and even artists do things with a lot of thought and intention behind their work, most graphic designers are already aware of AI tools like this and they mostly just see it as a tool that'll help them
@@paramshah98 Sure, the client will give the "designer" notes and the designer will punch those into the prompt window and the AI will refine the image based on the client and designer's thoughts. None of this changes the fact that the vast majority of designers will be out of a job.
I've been using MidJourney for a bit now and while it is amazing, it is still a little way off. Some of the pictures it produces are awesome but they rarely match what I'd interpret as the prompt people give it. I watched someone give the prompt 'the fear of the number 6 as a poster' and the first 12 attempts all had a number 8 on the poster with illegible writing underneath. I tried about twenty times to get a picture of 'a friendly robot petting a cat', a straightforward prompt for human. I got twenty pictures of a somewhat robotic cat's head. I think it will be a while before they 'take our jobs' but at the same time it is exponential so maybe next week.
My daughter is 14 and DALL-E is soul crushing for her aspirations. So bizarre to live in a world where AI possibilities are affecting your children's emotional state.
i wouldn't worry. If anything i would say the fact that your daughter will produce art that is really human makes it almost even more valuable than before! keep those aspirations alive! I like to do art sometimes and have been using these text-to-image for ideas and stuff. When i like something i transfer it to canvas to give it a human touch. All the AI does is cutdown on time to find inspiration and alot of the busywork of actual art-making. The value was always in the ideas and im sure your daughter will have some amazing ones of her own!
well people need to stop supporting it and giving it so much attention. Ignore and let the phase.. phase out. I think issue is people are using all these tools and boosting the hell out of it. Need to ban this stuff I feel. But from what i've seen AI becomes dull and boring after the wow phase...
@@MohamedHassan-rh9iu I think this SHOULD be the case, but the reality is that in this microwave "I-want-what-I-want-and-I-want-it-now" society, many people don't care how the "art" is derived. Sure, there will be some who place well-deserved value on actual art created by individuals, but the sad truth (I think) is that those art lovers will be the exception rather than the rule.
This is such a great video !! I just would love to have a comparison between real life graphic design tasks.. those are very nice pictures / ideas, but we rarely need to do such surrealistic images for corporate designs.
I think the fact that the judges knew it was a computer vs human skewed the results. It should've been simply text and what image best represented the text. No context
Dall-E is somewhat daunting for graphic designers, especially since it's able to produce content instantly. But I don't think it can replace graphic designers entirely Tim in his first prompt said it took him 3 hours to get to the point he did, which is reasonable considering he's working with no brief or a plan. Where I work, it's exactly the same for literally everything, I don't get told what to produce or achieve, the only words I ever hear are "Make It Pop" - make "what" pop? So imagine my boss' surprise when I'm 3 hours in and I only have a series of concepts and not a finished product. To people who don't work in this field or understand it, the concept of how long it takes to produce something with little to no information is alien to them. People like my boss don't deserve to work with real people, Dall-E is perfect for them. Then there is the outlier, which are people who are looking for something specific, but find every possible way to help the designer understand what they need.
The arguably more incredible thing about this experiment, though, is the fact that the AI renders these images in ten seconds or less, while it takes Tim several hours to render one.
The problem for designers come when the ai is good enough, and the choice becomes to use the very cheap ai instead of paying the better, but way more expensive human.
@@fredrikjohansson The problem with this is that due to copyright laws, Ai generated art cannot be copyrighted which means for designing it is commercially useless and this is arguably how it should be.
Watching this (and the original MKBHD Dalle video) after playing around with Dalle 3 for several days...Tim wouldn't stand a chance against Dalle 3. It's hard to believe that this was only a year and a half ago.
This is the same reason I came back here.. Dall-e3’s accuracy and fidelity is mind blowing. This technology has exploded and become SO advanced in just 18 months. If I was a graphic designer I’d be shaking in my boots rn lol
I saw this video when it came out more than a year ago, and I'm watching it again now. He would stand no chance today with the most recent models, more so if you hired a so called "prompt-engineer" and gave them both the same time. But today DALL•E 3 doesn't need well-thought prompts.
The scary thing is that these AI image makers are just getting started, and developing amazingly fast. Poor Tim might win today, but tomorrow might be a different story.
They are, and I would say that this wasn't even a fair comparison for the ai. He had a very long time to work, in that time how many versions would the ai make and then you'd have to take the best one from those images and compare it to the artist. Or, how long would it actually take for the ai to make an image that surpasses his image? 10 seconds to generate an image... how many images in just an hour? This is insane. Edit: We may not even know how to get the most out of the prompt, maybe there will be classes on it in the future.
Amazing what the AI can do but Tim’s pieces are just head and shoulders above them. The robot in front of the computers is insanely good. Stop worrying Tim, there’s no way they’ll replace you. In fact, they should give you a pay rise!
The thing about this is that this AI went from its first, laughable version tbh, to this in about two years. In two more years it could literally generate perfect, unique images. It also generates 10 pictures every 10 seconds, which means you could just refresh again and again until you find one that you like. And Google just made an EVEN BETTER version of Dall-e 2 called Imagen. That's how fast progress is really happening.
@@yahiiia9269 It actually wasn't even two years, it was just over one year between the initial Dall-e release and Dall-e 2. That's how insane this progress has been, I keep trying to tell people but most are unaware.
@@acaustik8763 It could literally bankrupt tons of industries and completely shift focus on a lot of subjects. The sheer amount of insanity that is about to unfold because of AI is intense and I hope it proves just how pretentious humanity thinks it is when a machine can do anything faster and better than any human can. Humanity needs to be taught a lesson when it comes to humility. I remember the artists saying that they wouldn't be the first ones to go, but here we are with insane AI models that can literally make words into pictures in SECONDS.
This comment will age badly. Tim's stuff is great, but an AI able to spit this stuff out in minutes, in any style you want and we're just at the very beginning of it all? Imagine how much more control and specificity you'll have with future versions. It's nice to see all the goodwill towards Tim in the comments, but the writing's on the wall. It's over for all but the tip top tier of illustrators, concepts artists and photographers. And I say this as a fellow artist. This will come for filmmakers too. Sad and scary and exciting, but ultimately sort of depressing.
Love this video, and great job explaining the process so thoroughly Tim!! For other viewers' perspective, I think designers generally agree that AI is a helpful tool. I use Gigapixel's AI upscaling to save time on enhancing photo quality, Photoshop's content-aware fill for photo corrections, and Pinterest's like-for-like photo analysis for finding the right reference photo. Dall E is great for composition iteration and for giving us references on how to interpret things stylistically. Each of these help save time once we adopt them into our workflow.
Dall-E (and Google’s DeepMind version) doesn’t “composite” multiple images in the way that Tim implies he is approaching this. Every image is a unique creation that has never existed before, yes, it is based on all of the images it has been trained on along with the styles of multiple artists but the artwork is original as I understand it.
how do you feel about the fact that those creators haven't been asked for permission and OpenAI now wants to sell licenses for this latent space that original artwork can be drawn from? They took 400 mio images, no one knows exactly which and from where, and theoretically it could've been fed even with Tim's own work without him knowing.
@@LarsRichterMedia Well isnt it like you looking through a picture album or old illustrations for inspiration? Dall E learns what looks good and how things relate to one another, and then create new stuff based on it. Just like humans do.
@@terogamer345 If you want to declare Dall E to be coscious, then yes. Any conscious entity no matter how their mechanism of perception happens to work woud have the right to perceive a public space. A mechanized visual harvester on the other hand that potentially copies your data and processes it into a latent space from which it draws original visuals we might want to look at a little different than what "humans do". At least for the moment would be my opinion.
@@LarsRichterMedia It's similar to how humans will view someone's art and then imitate but not directly copy and paste it. That's how people grow their capability to make art, being exposed to different styles. Also, is it not extremely transformative?
damnn,,as a designer i wanted tim to win this and thankfully he did. OMG, fuk this Ai. this was so stressful. good job tim. you represented all of us and made us proud. love you my boy.
I believe the best part would be to augment Tim with Dall-E for quick idea generation, and then make it by hand. While computer can definatelly create convincing images, it lacks human touch in sense of humour, semantics and general feeling of intent in the work.
Dall-E’s deer and robot pics were better to me, and they took 10 seconds - Tim smashed the Mona Lisa one though. I think anyone under the impression this kind of technology won’t take over is massively mistaken
So far man's advantage in such contests is creativity, unpredictability and the addition of something subjective. This is best seen in the image of the Robot and the Goat. Tim didn't just transform words into a picture, he added something from himself. Apart from the technical quality, that is why his graphics are better in this case. They offer something more than just the basis implied by the description. But the sad truth is that this too will be faked by algorithms sooner or later. It is just a matter of time. And it's a few years rather than a few decades. There's no point in fooling ourselves. In about 20 years all content, written, visual, audio, will be possible to create to some extent in an automated way. It will be faster and faster, more and more perfect and closer to what people do today. There is no escaping it. The AI cat is already out of the box.
Lets just be hopeful, remember the first days people were discovering electricity, one can only imagine the fear they had thinking about the future. But, since then we discovered so many things about electricity that we realise there is no need to fear it.
People call me crazy when I tell them we're only a few years, if not months, away from AI beginning to sweep across creative industries. Character, concept art, paintings, that's already a done deal. As people continue to use it, AI is trained to get better and more specific. So eventually logos, posters, motion graphics, entire CG scenes, clothing, interior design, architecture, music, etc. It will all be done by an AI in mere moments. Eventually, an AI will be trained on popular music, and it will be released into the wild, only then will a legal battle ensue because of that industries power and resistance to offer any music for AI to learn off of, but it will ultimately be too late for the artist, musicians, designers, and engineers. If we don't take a stand against it now we're finished. "Oh, but don't worry. The artist will still be there to touch it up." No. They won't. Once these things are trained enough and get fast enough, all it will need is a prompt. And it won't have to be a very complicated prompt to get what you want either, as the public trains them on complex prompts they slowly begin to understand better what you're looking for until it can be entirely dumbed down. It will be able to come up with its own prompts if it wanted, its own styles, it will be able to take into account physics and math to create structures and re-engineer many of our technical designs to be superior. Sure, it will be some sort of golden age for progress and innovation, but what will happen to the artist? What will happen to all those jobs? Because even if companies still hired any, they would only need VERY few. So, should I even pursue a future in art/design? Should I post my art online when I know silicon valley is gonna swoop in and add it to its AI without my permission? This is the end goal of every site we casually upload our content to. If u wanna get really crazy I'd say this is the breaking point for the Internet. The line between free use and personal property online will finally be treated as a serious issue. NFTs were an attempt to solve that, but ultimately were just a disgusting mess. The problem is once trainable AI gets into the hands of the public, there will be nothing to stop people from stealing anyone's content and training an AI to replicate their work and then call it their own.
There's no stopping this. AI art generation is the will of the companies who artists have sold themselves out to for the last 100 years or so. A large section of commercial work will be off-loaded to AI, but I think what is left over will be better human artists and real art, not souless concepts or animation made for unfair to minimal pay. Harder to find, less profitable for people, but still better, though Im unsure how long it would last in that state. People are so numb or bitter nowadays that I don't see anyone caring until it comes for their job. Really, the only thing that should be lobbied for is for image hosters to disallow image scraping for use in AI data sets unless given permission by uploaders. We already know how hard this is to prevent, hell it might even be too late. AI is a data parasite on the back of thinking humans. I don't know what goes in to collecting training data for this purpose, but if allowed to continue with this level of freedom, it really will destroy many careers because the only solution to preventing AI from developing is for humans to not produce at all. - Dystopian - There should also be special laws placed upon organisations that develop and use AI, such that they are known and must abide by some code of conduct. They are not normal businesses and shouldn't be allowed to operate as one when they wish to employ artificial intelligence instead of humans. I'm fairly sure work is being done to implement some control.
I don’t know why I feel triggered and upset after this video. 🤷🏻♂️ - this was mean spirited. I wanted to hug Tim. The software was scarily good. I can see graphic designers becoming a rare luxury.
Dall-E is cool but can we talk for a second about how great Tim is? Not only an artist as a designer but also a cool, humble and focused person. No wonder why he is on the team!
Not only that, dude also made sure to credit those Unsplash artists. Incredible.
Dall-E 3 will have a personality based off of 700 million cool, humble and focused people. Tim is still in trouble.
WORD.
@@Poolzclosed sad but true. AI will replace emotional partners in the future. Like who needs a partner expect to reproduce.
AI doesn't even need to be humble, or have it's own opinion, or have "rights" that otherwise s human would and should have.
Tim: “So what are we doing for a video today?”
Marques: “You are re-interviewing for your job.”
John Henry comes to mind...
🤣🤣🤣🤣🤣🤣
Fucking cruel. People’s jobs are nothing to make light of. AI needs to be stopped...but it won’t :/
@@kangtheconqueror8359 How would you stop it?
It will replace us all, and any countries which decide to stop development will simply be rendered obsolete by other countries which will continue.
The only solution is to change our socioeconomic structure to allow for us all to benefit from the technology, rather than simply benefitting a very small subset of the entire species.
@@karenreddy Gather a large group of a resistance to sabotage and destroy all AI servers and technology. Which won’t happen so..just be depressed af and hate life that I’ll never reach my dreams because the industry I want to apart of will be replaced by machines and eventually it will destroy us.
I felt bad for Tim a few times because you can tell he was stressed about this whole situation. He essentially had to prove his worth to the team while up against a computer that spits out 10 images in 10 seconds. I hope if you ever do anything like this again to please give Tim some reassurance (or at least show the reassurance on camera) because this kind of broke my heart watching this guy put all the effort he had into this challenge.
All-in-all, Tim, you did a phenomenal job, and the MKBHD team is lucky to have you! Your images were stunning, and your care and attention to detail were unparalleled. Keep up the good work!
All true ✅
You will probably like to know, that DALL•E can't be used for commercial purposes ( for now ), so visual artists won't lose their job ( FOR NOW )...
EDIT: Actually, I think it can be used for commercial purposes now, so...
Had the same feeling
I came here just to say this. You nailed it.
@@krtkks watch it again and take note of his body language - it says it all. He was extremely stressed the entire time, even stating the Tiger Atlantis challenge "broke" him. This technology almost directly challenges his position, his expertise, and his role with the team. Whether it was overtly stated or not, Tim felt threatened by DALL-E. Even if Marques never said he was going to lose his job, Tim was thinking that, and perception is reality.
As an artist myself, this whole video gives me anxiety. TIm did a phenomenal job, but I would be lying if I said my desire to root for him in this situation didn't come with a human bias toward other humans. Ultimately, this video proves how much time and human effort had to go into producing these images, vs the AI who (aside from all of the data and time tht has already gone into programming it) required relatively no time/thought.
Things are going to change a lot soon, for all of us.
Imagine telling your grandchildren about the old job of the “graphic designer” and them being in disbelief that an actual person used to spend all that time and effort to make every detail on an image when the computer could just make any photos, and even videos, that you asked for instantly?
I’m worried AI is going to disproportionately benefit the wealthy, and eliminate their need for a lot of The Poors.
Well, this is why humans use utensils. Not to make the work of an human, but to make repetitive or hard works feasable and much cheaper. Maybe we will have better covers for fantasy books 😂
yeah, design is soon going to go the way of the weaver, the sower, the telephone operator.. automated away by the ever-increasing power of technology. the only thing we can do is prepare for the storm ahead, and adapt.
@@Dejawolfs How do you plan for a storm that you know nothing about? How do you prepare for every job being automated within our lifetimes? It’s a scary future!
@@tyler.walker only thing i know is that you somehow pass into law a base salary for every person. the reality however is probably that all wealth will pass into the hands of an extreme few individuals.
Tim’s creativity should be more featured here. As a graphic designer myself I am blown away by his attention to details and composition. It’s very interesting and fun to see a fellow artist’s thought process while creating, I could never be that thorough with mine since a lot of concentration and creative juice is needed to just come up with the idea, much more put it on canvas. He’s a very talented artist! Would love to see more of Tim and his work!
I’ll be uploading a full timelapse video soon with all three prompts on my personal channel! Should hopefully shed some light on how I made everything
As a graphic designer myself, I found myself more interested in just watching Tim’s workflow. We all do it differently, so it’s fun to see.
and as a graphic designer you should know this wasn't graphic design but photo composition (artistic)
@@corgikun2579 it was a piece of visual art that Tim designed. this is also.... graphic design.
@@corgikun2579 photo manipulation*
@@corgikun2579 photo compositing is a form of graphic design…
@@squeebploozer not everything that requires a "design" or "decision making" is "graphic design". Matte painting, collage, photo manipulation, are not graphic design
I love how although Tim has mad skills when it comes to Photoshop, he still took a second to watch a tutorial. That just shows me you can't always be the best at every aspect of your job. It's okay to seek help sometimes.
I learned a while ago you can’t always know everything, but it’s good to know where to find it when you need it. Also just good to stay on top of new tools and workflows
Many apps are complex with thousands of features...you're not gonna know it all at all times, and apps are actively developed so staying up to date to such a degree on all apps you use is impossible.
The key is to know what the capabilities of the app are, so you can research and learn the feature when you need it.
Any professional programmer will tell you, a good chunk of their job is just Googling for existing solutions/algorithms people have already made or guides on how to do something specific. It's the same to a certain extent for most sciencific/artistic jobs.
It just shows me that Graphic Designers are becoming useless.
@@jsph_gstv hahaha. Yep, we’re a form a planned obsolescence
There is something comforting about seeing a professional graphic designer pull up a tutorial on how to do things.
Shout out to Unmesh and PiXimperfect for delivering the best Photoshop tutorials on RUclips!
we keep learning until we die!
Phlearn and PiXimperfect are some of the best around! I watch them all the time
shoutout texture labs
PiXimperfect created Photoshop, nothing shakes my belief on that.
Like 50% of a professional artist/developer/technician's job is copying/looking up stuff.
My favorite part as a designer was the visible frustration when people were looking over his shoulder early on.
I was nervous for him! People looking over my shoulder while I'm working on a project and hearing "i like it yeah" or really any comment is just so uncalled for. I hate getting critiques while I'm in the middle of something
@@JakeWolven especially since sometimes (often times) the final product looks nothing like what it looks like halfway
Ikr?
I thought I’m the only designer with this feeling
I can't work with people looking over my shoulder. I'll stop and wait for them to leave.
Best part, as a designer I found it hilarious and accurate that he pulled up a PixImperfect video at 12:18 to help him. 😂
Unmesh is GOATED, we all need him :)
He forreal is the GOAT
Yeah 😂😂 Umesh
I love Unmesh and his work flow he's my go to mentor
The only word I can use to describe this video is "fascinating", seeing an expert explain his thought process and seeing how much time it takes to make those edits really makes me appreciate the work even more.
Good job Tim, I wouldn't replace you with an AI for sure
At least for now
But you will.
@@Ez-se2dl IA is never gonna be able to create good art is something only conciusness can create at least at a good level
@@ernestoduenas2466 Let's see what happens in three years.
@@Ez-se2dl IA can help as a tool like in the nvidia canvas check it out but ask a computer to have creativity is something we us humans have so lets say a convination betwen is the key but ask a IA to make the whole work is like ask digytal artist to create whitout a pc youre askyn for the tools to do the whole work aint no balance on that at the end we make the IA we give the IA the comands is all about the porcents of participation aint no gift for lazy people level up ...🍄
Most funny IMO: In a studio full of big monitors and wall-mounted displays, a group of people stand around an iPad to watch pictures.
thinking the same lol
simple human being
But… it is to make it more private.. i guess
Makes it feel informal and less like a PowerPoint presentation. Like showing your friend a video on your phone.
Same. But also. They have a studio with multiple macs and monitors, and i do not. Tee hee. Fun all around tho !!
dude, im happy for poor Tim. I can tell he is so anxious and actually did this like his job is on the line.
I mean, cause it is…maybe not right this second but give it a year
@@murraybritton6729 Its definitely more than one year. Dall -E and all these AIs are first of all currently not available to public so not everyone can use these.
Secondly, I have noticed that the more object specific you get, Dall-E struggles. Like if you gave it a proper tech object like mac studio or maybe iphone 13, it will struggle way more.
Plus it struggles with human faces, like if they specifically need Marques in a picture it will struggle but a graphic designer knows won't.
Like take the thumbnail from the Mac studio video on main channel with marques sitting on the mac studio. I don't think Dall-E will be able to create anything even close to it
What Dall-E does is very impressive. It excels a lot in representing those everyday objects, like the goat or the tiger, astronaut etc considering the time it takes but its still perfect enough to replace a full blown graphic designer
@@AdityaGupta-om8ez these struggles are specifically because the data set Dalle-2 was trained on was curated to prevent copyright infringement, false identity creation, and adult imagery. It isn’t a limit of the AI or it’s technique, it’s an intentional limit placed by the designers so the public could use it without creating legal issues.
It’s why Google’s Imagen AI isn’t going to have a public testing because they didn’t limit their data set the same way. The minute you allow the public to create realistic Ao generated images of politicians, corporate products, or real people doing things- you create severe legal issues.
It will be working through those legal quandaries that slows the roll out of this to the public, not the state of the technology itself.
@@murraybritton6729 Deepfake can do this though
@@AdityaGupta-om8ez I think one of the big shortcomings too is a lack of revisions, you can't explain to it why the feathers don't look good, and how to fix them. There's a lot of things a human can just do more concisely, while this is kind of like image roulette.
Tim's eyes are like someone who saw the future and didn't like what they saw
Despite how hard I try to remain optimistic, it’s pretty stressful watching AI progress as a young 3D artist. I quite honestly can’t imagine life being worth living for me if the visual art industry is automated. It’s difficult not to get emotional about it because Im so passionate about my career and one day, all the work I put in/experience I have could mean essentially nothing.
I always question “just why would you replace artists?” because it seems a bit evil on the surface. Artists typically love what they do and find fulfillment from their work. This isn’t hamburger flipping.
Additionally, artists are already constantly shamed for all their quirks but only accepted because their skills prove useful to many businesses who rely on their creative talent. I cant imagine how poorly artists would be treated by the business class if their skills served no monetary value.
I’d rather be f&@king dead than work an boring, awful job for decades just because society hasn’t figured out how to automate it yet. The only thing that keeps me optimistic is that I love what I do so much I shouldn’t really care if a robot can do it better than me. I do it because it brings me peace and calms my mind. It’s no secret, art is therapy for many artists.
I also hope people love supporting other people and don’t care as much about support robots. Handmade clothing and jewelry still exists despite much of the field being automated so that’s somewhat promising. So hopefully when all digital art can be automated, our society will find a way to value and support digital artists.
Idk. I’m always cautious to publicly comment on the topic because it’s so emotionally charged, I’m afraid I’ll change my mind tomorrow. Nonetheless, this is how I often feel about it when my anxiety goes through the roof. I hope my perspective serves some purpose other than to be mocked 😅
If so, I make a video about it to inspire digital artists to keep pursuing their dreams. The script is already written haha
Go for it bro, we will support humans rather than robots, when it comes to art, i need to feel that there is a human touch behind this, if it is done by a robot, it doesn't matter if it is 10x better, it won't mean anything to me.
This is how I choose to think about it too. Just because a robot can do it faster and nicer doesn't always mean it's better. Art is and always will be about the artist and the context. As they say, art is all about perspective.
I think manes lanes puts it quite nicely up there, it doesn't mean anything if it wasn't made by a person.
very-bad-things-to-self went up dramatically in the rust belt when they automated all those manufacturing jobs. It's not going to end well for a number of artists. It's also not going to end well for the researchers. Putting your name in bold on a ground breaking paper that directly results in tens of thousands of people losing their livelihoods is how you make a *lot* of enemies.
But you already know how this goes: no one's going to listen, we're all going to hate it, and then 20 years maybe we'll reconsider.
Climate change, leaded gasoline, cfcs, asbestos, plastics, car centric suburbia, surveillance, Afghanistan, social media-- I mean we can just keep going.
@@maneslanes7412 thanks bro. I really hope so. I almost can’t believe this is real right now lol
this is just the greedy ones trying to take the job of everyone just because they can. these guys are pure evil and do not deserve anything good from society. I cant believe so many people on the comments are exited for the AI this is the dehumanization of creativity and art.
The Dall-e is scary, fascinating, useful, cool and unavoidable.
That is a _perfect_ summary, couldn’t have said it better myself
@@thebishopchess because you don't know rhyming 😄, just joking 😆
Deepmind created an Ai called gato, which uses the same technique as dalle, but it can do basic household chores and answer questions about its environment. Many more jobs could be taken.
And inevitable.
Damn well said
Okay, this video has now put into perspective how insane Dall-E actually is, it’s easy to overlook how it’s not just “stitching” photos together, it’s actually manipulating and creating light and everything else that is so easy to overlook as an outsider.
Yeah this is what I saw as well. Even if Tim's arguably followed the prompt closer, all three of Dall E's (maybe somewhat less on the robot one tho) looked like complete photos while Tim's looked photoshopped.
It's not sticking photos together, it creates images that have never been seen before from words.
I was impressed by DALL-E when I saw how many ready-made images Tim needed to complete the task. It wasn't just one human against one AI, it was a human team with access to nature, cameras and art collections against an AI alone.
@@HoriaCristescu DALL-E needed 650 million images to do its thing, that's the library it has committed into memory during its training as the weights of its processing nodes.
@@tylisirn Can't wait to see what it can eventually do with billions or trillions.
That was fascinating to watch! I'd love to see more :)
I just wanted to give Tim a hug....I could feel his anxiety and how this competition is unfair...I know it was just a playful time but it shows the importance to recognize the human value involved in a process and how sensitive and hard can be all this new context where living in... It's an one way path
That Tim dude seems like such a genuinely good human being also being very intelligent, creative and funny! What a wonderful life affirming experience to see him and watch him work. Thank you so much for making this.
Damn. Thank you 🙏🏻
@@mcmahontim YOU'RE A ROCKSTAR, TIM!!!!!
Can this be a series? I could watch a season of this.
I think this would literally kill Tim
the battle of the intelligences
I would love that. Maybe different designers vs Tim too
How about Team vs AI? Like testing each member's skills against a version of AI out there.
Definitely!
I just want to thank Tim for saving us graphic designers. My palms started to sweat when the critique came about.
Saving for now. This is just Dall•E 2, wait until Dall•E 3 next year. For reference look at Dall•E 1 and how poorly it does compared with 2.
@@Exitof99 You think there wont be any diminishing returns? Its very hard to improve on what they already have, it wont be exponential increase thats for sure.
@@outlander234 It seems a lot of the results are based on the passes it f
does at some stage, I forget the exact term, so presently, it's also that the current version probably could do better given more passes. But still, the present version is still having a hard time generating useable results. It does well in some areas, but there are defects that are easy to pick up on, weird fractal spots, missing/extra fingers, and so on.
I think there is still a long way to go yet. If you've seen Stable Diffusion, it looks like that might surpass the noise-based method that Dall•E 2 uses.
This takes Man vs Machine to the next level
tim was such a great guy, i felt bad for him so seriously that i got teary-eyed ,heart racing like crazy, am a graphic designer too i know how stressed and intimidating this is , having to compete with an A.I knowing how intgeligent those thing has gotten this days you really dont know what to expect, also thought of u losing ur job to a computer flashing right before your face is not a joke it not funny at all. no matter how confident you are with design, u remember am competing with a robot that might be better, i absolutely was in tims shoe at the moment, my heart pounding like crazy not knowing my fate yet, i definately just started having plan b for a moment, thinking if i lose my job what my next move. tim we smoked it bro........... hell yeah tim i wish to know u in person
12:16 Thank you, sir!
No, thank you sir! Been a big fan of your channel for a long time and your content has been incredibly helpful over the years! Always finding so many creative workflows in your videos that have changed mine for the better. Your insight and knowledge are deeply appreciated! Much respect to you and your channel!
Tim is a really great Guy, he used to be a bit shy, but now he’s a great dude on camera with the team. MORE TIM!
Safe to say Marques has one of the best teams anyone can ask for!
I’d disagree, a producer watching the designer work after he expressed discomfort is a hostile work environment.
@@aeonjoey3d You missed the point, but carry on.
caz of 1. good work culture 2. good pay. 3. he's kind of the coolest YT star to work with.
Funny you say that I only saw guys on the team...
@@aeonjoey3d they are friends
I'd love to see one where Tim starts with a couple of the Dall-E photos and see where he can take them to improve on them.
That is the current value of AI.
Even at this point it can make art creation far more efficient, increase the supply of visual ideas and thus devalue it, even though a designer is still needed.
Yes, that is the best part of the current image generators, I wish these AI programmers focused on automating the idea generation phase as well as the "boring" manual parts of art.
I like UV mapping and retopology, but if there was an AI that could give me instantly useable UVs and clean topology, I would be ecstatic.
AI could empower small artists in many ways, coloring keyframes for animation, help with figuring out perspective, manipulating existing images in a non destructive workflow, etc.
But the current use of AI art with prompt based image generation and cloning existing artist styles, is just taking the human part out of art, Wich is a massive oversight that the executives and corporations will use to save money, because why pay a human artist a liveable wage if they can just use an image generator to make 95% of the art and have a low payed artist just refine and do small touch-ups?
@@qaztim11 it's easy to automate prompt generation, but not ideas which fit a specific need, style, with countless variables requiring subtlety of judgment.
Maybe someday, but we're still far. It would require several levels beyond what we have today, including project comprehension, by itself.
AI currently isn't taking artists... It's taking craftsmen. People who practice a craft (which I have). But people with good ideas, directors, artists, will still remain competitive. In fact, maybe even more so, as they will have more freedom and ability to leverage their ideas with AI.
If you are creative, even with as early as we are in the process, it's possible to develop some interesting stylistic approaches to imagery by combining known styles. Even humans draw from what we know. The art remains human, as it is guided by humans, but the craft will be AI driven.
@@karenreddy found the best comment
@@karenreddy I would assume that in the future the AI image generation changes so that people can give better guidance of what they want rather than just brief description in words. But for now the big challenges still lies in improving the actual image creation process. How to make the user able to describe what he wants better is then also another thing.
Nothing can replace the creativity, originality and the personal touch of an artist, it is the authenticity of his/her piece of art that matters and not the works of a gadget or a machine and that is the big difference. Two thumbs-up for you Tim. 👍👍😊
It’s not creativity, it’s specificity. An ai works off a database of art. You won’t find something very specific in said database to work off of.
For anyone interested, the painting is a depiction of the Swedish queen Kristina, Who held the crown for a time in Sweden during the 1600s. The painting is at the national gallery in Stockholm, and it could be said to be a "tour de force" of the royal spirit and legitimacy to rule, through allegories of the spirit; the blowing wind i the handkerchief, the water fountain in the background, and the sphere, representing the world.
I find it interesting how most people who watch MKBHD are generally interested in the advancement of tech, tech that has and inevitability will replace all human jobs, yet we all (myself included) are still rooting for Tim!!
AI is coming from everyone though…
For some poor souls it'll come sooner than the others, like for Tim.
If AI gets evolved to a certain point where it can easily does most of the jobs then I'm pretty sure the economic system will take a shift in a new direction/reworked to benefit the people thus people might not go broke but for now, Artist are about to go broke for the sake of " progress".
I for one welcome AI. My only fear is how we humans go about embracing it and how willing we are to change for a nicer but different world. Current systems that are deeply embedded into our culture like capitalism(which requires human labor to work or we'll be homeless because AIs have taken all the jobs) need to be wildly rethought or completely replaced. The biggest issues are the words "different" and "change" because of how so many people are religiously married to these systems and ideas which inturn will or could cause a lot of political strife as people are unwilling and or unable to imagine or embrace new systems and ideas. Best of luck to us wise apes. :)
@@poopoodemon7928 You wouldn't be embracing this AI if you were an artist. It's essentialy a slap in the face to creativity/originality. Imagine honing your skills all these years for someone to just take some free pictures and write an AI that ultimately replaces you.
@@MusicComet Skipping the whole "hurt ego" thing as that's something personal and requires self-help, we should differently re-structure our society so that being replaced wont result in a financial/materialistic problem.
@@MusicComet Trust me on this - AI will create more job roles than it eliminates. Us humans will just have to do what we do best, which is to adapt and use the AI to advance our abilities and speed.
The fact that you even challenge a human at this point means a lot. The future for graphic designers is cloudy
But what if humans use the AI and better the photo using their skills.
@@Hamza-xl2sd AI will replace an expensive Designer soon
The ones that adapt the best will still be needed to do the prompt, the edit, the design, the concepto. AI is just a tool. In the old days you were a carpenteer, now you operate the machine that makes the chairs.
I kind of feel for the men. Would be intimidating to see an AI make something in 10 seconds that takes you hours to make.
I am only half way into this video and I wanted to say that Tim is reminding me of my daughter who works as a professional graphic artist for a production company. The stress I see her go through and the pressure she is under on a regular basis is palpable. My heart goes out to you, good sir. I am looking forward to the reveal at the end of this video.// edit - I just watched the reveal. What a relief. Great work, Tim, you killed it.
13:32 Tim's face is the best
Shout out to Tim and his amazing talent
This was a great comparison makes me appreciate professional graphic designers like Tim even more.
scary part is the time comparison.
yes, Tim is master of his craft, but compare that results payment vs. resource cost then its starting to be pretty scary.
one great development would be to ask DALLE for draft images and made final editing with human
What do you mean "draft images"? Do you know how this works?
@@fitybux4664 I don't think you understand. DALLE creates images that have never been created before. They are saying use the images DALLE gives you and make final edits.
@@jsph_gstv weird. They should have just said that...
@@zerodrakengard it looks more like Fity Bux is saying, that DALLE doesn't do draft images
As a designer/illustrator I agree, I think DALLE could save a ton of time in the drafting/layout stage if it got smart enough
This was a fun video highlighting how well a development team can work together. I'm impressed at the knowledge, skill, and level of expertise Tim showed in his creations. Dall-E can replace any entry level graphic artist or help someone like me that needs graphic design but it can't yet replace the intuitive artistry of a professional designer.
Tim definitely deserves a bonus for this. Working for a RUclipsr who probably makes millions while he makes a regular person wage and he gets to be basically threatened with a massively powerful AI and forced to “prove his worth” idk I’m never that person, but this felt tone deaf af and like watching a kid be bullied.
I agree, tone deaf is a good way to put it. No one in the video displayed much empathy for Tim’s experience/stress, not sensitive to the situation at all.
@@tompotter8748 yeah I just felt bad for him the entire video, I can’t imagine someone feeling like this was a good video premise
it's kind of like The Hunger Games hahaaaa
also, relaaaaax, human artists will still be used for lots of projects and movies/games etc. AI makes stuff that looks cool but it doesn't look genuine, it has lots of errors in the details AND it's not consistent in the least. You still need a skilled [insert job here] to control the AI and bring it to it's true potential.. For example, i see lots of pics out of MidJourney that i never managed to make it do, and those have a lot of specific photography/art key words in them...
imagine making a game for example, it would be pretty shit if every picture, every character is not consistent from a style or even facial features point of view with the rest. And that's what you get with these AI generators... It's pretty hit or miss
With Tim you not only get the best photos but you also get the endearing qualities and personality of Tim.
DALLE2 tho...
''dalle 2, faster and way cheaper than your average graphics designer become a member now'' sounds like a very funny ad for it
Hi Tim, you're AMAZING! You can explain what u did and adjust as it is needed and u provide a story, that provides so much more information about the build process. U bring in your own aesthetics and thoughts, a machine can not do (at least not now). Just keep doing it, that is what drives creativity.
Tim’s interpretation of the second prompt was so cool, chilling, and detailed. Nonetheless impressive work from DALL-E!
The goat and the deer are super impressive on Tim’s part, awesome to see it come together as he worked on it
The robot women one was mind blowing for Dall E tho, the computers were so realistic and they had space between them and they did completely cover the wall, making it seem as if it’s actually possible
The way Tim did that part, while still cool, just wasn’t realistic, by taking a front on image and stitching them together, then giving it perspective, it makes it look really flat as you lose all depth for the PC screen
Still very impressive though
Technological advancement and automation disproportionately benefits the already well off. Replacing jobs and with the jobs that still exist, pocketing all the extra profit. The average adjusted wage has not gone up much from 1960 despite huge leaps and bounds in efficiency. Normal people undeniably loose power with DALL-E.
Exactly, this will just concentrate more money in fewer hands.
would really like to see some statics for that
Love the video the only thing I wish is if you would show a full screen image of the photos side-by-side instead of in the lower corner So we could see a bit more of the detail
Well done Tim! I was rooting for you the whole way. You’ve done the human race proud.
''now the training arc for dalle2 begins'' or '' this is dalle1 and this is what is beyond dalle 1 it is called dalle2 and to go further beyond this is dalle3
As a fellow graphic designer , I feel your pain Tim , LOL. But you did AWESOME! my jaw dropped at your works , and I’m not just saying it to say it 😁 tho I will say all the stress and anxiety of feeling like you had to prove yourself was very relatable lmao. But I still feel like your work DEFINITELY had more feeling than the AI. maybe it’s from being in the profession for so many years , but I could tell it was crafted with care and fun imagination 💖
As a fellow AI designer , I feel your pain DALL-E , LOL. But you did AWESOME! my jaw dropped at your works , and I’m not just saying it to say it 😁 tho I will say all the stress and anxiety of feeling like you had to prove yourself was very relatable lmao. But I still feel like your work DEFINITELY had more feeling than Tim. maybe it’s from being in the profession for so many years , but I could tell it was crafted with care and fun imagination 💖
This was scary. I don’t know what the future holds, but Tim is a legend.
Photo manipulation is a completely different thing and many graphic designers don't do or like doing. We use the same tools but that's about it.
Tim however did a great job here!
As a creative professional, I’m most interested in exploring how Dall-E can help give unique inspiration images or rough starting points that can launch my own creative work. I doubt I would ever rely on it to complete a finished piece but for components in the process it will be hugely valuable.
Until what it gives you is so good you don't need to do anything
@@gueton5200 yeah and his employer thinks that they know longer need him
Great and entertaining video Tim did such I great job. I would of been freaking out being in his shoes lol
Dude, this is scary, i mean in a few years it can literally replace all the graphic designers
millions of peoples financial security, gone in less than 2 years i predict :(
@@cappypyramsaudpate5535 yup singers,dancers,musicians, heck even movie makers, and salespersons. 🤷
In less than a year.
at photo manipulation probably. But thats a very small aspect of graphic design.
This is why capitalism needs to fundamentally change. Less work should be a cause for celebration, not misery
8:25 That's such a graphic designer thing to say.
in the 2nd prompt, in tim's picture, the shadow is insane 🥶🥶
I think Tim is super good at his job and it was awesome to see him working. Kind of feels like a person who would be awesome to have as a friend. Also, if you had little more budget, Adobe stock for example, he would have had even better resources and more to choose from to do those really good compositions.
How dope is that!?
Tim is such a pro!
I love the atmosphere around THE STUDIO 🔥💯❤️
the robot woman guarding the computers one by DALL-E was actually so fucking impressive holy shit
As a Graphic Designer with over 18 years of experience, I have catastrophized EVERY generational advancement of tools as "gonna put me out of a job!" Now, after playing with GPT-3, DALL-E 2, MidJourney, etc. I see these as amazing time-saving tools! In my experience, the first 80% of a design is the "easy" part, and the last 20% is the "hard" part. Because of AI, I can now spend 99% of my time on that last 20%!
When used by a professional that combines design, painting, editing and prompting in a masterful level it may become one big time-saving and productivity increasing tool. The amateurs just can play with it, pros will be the ones having the best from it and mostly the ones that adapt the best.
I like your view..... crushed it
Sure, with the way they pay designers today, graphic designers are forced to entrepreneurship. They're gonna have receptionists doing designs with a.i.
@@Nalexandros847 You’re not going to have a job soon, ai will be able to do everything without need for a human artist. ai is a tool for artists only right now, only for a very short window of time until humans are no longer needed. Artists may end up being paid as prompt engineers until AGI arrives
Tim blew it out of the park!! His attention to detail was amazing!!
As much as I was rooting for Tim, this is just Dall-E 2. Future Dall-Es can improve at a much faster rate and do more things than any human possibly can. It’s just a matter of time before AI comes for all jobs done on a computer. This includes accountants, financial analysts, even software engineers and developers.
I'm not too sure about software engineers though... But it seems like the lucky winners of this whole mess is musicians since AI can't create a decent music on par with music industry standards.
@@PK-ow1kj i give it 2 years before the entire music industry will be threatened by AI. I dont underestimate the technological sinuglarity
@@PK-ow1kj honestly listening to ai music would defeat the whole purpose of music. Plus artists make most of their money from live performances and the whole celebrity aspect of it, ai definitely can't copy that
@@cappypyramsaudpate5535 ^
@@PK-ow1kj As a software engineer I can totally see how AI will do a lot of the simplier tasks. The designer would simply state what they need at runtime and instantly the AI writes the program to make it do that. Code neatness and structure is a human need. The AI will just hack out something that works.
the unnerving reality is the fact that this was a contest at all shows we're not far from the point where the AI will win. Especially if we're taking into account time and cost... Sorry Tim =/
I think the near future is the artist throwing prompts at the ai for like 30 minutes and then tweaking it.
@@satibel For a job? Maybe. For a hobby? Never.
The reality is that for no real life non-twitter-commission paid artist ever do they regularly make things that dall-e can make. The way Dall-e works is it makes a profile for each word, then creates a sort of "platonic ideal" for that word out of existing images, and riffs off that, then combines them. For things like original character designs where there's not a billion labeled images to pull from, or anything where the producer wants any amount of control over the final product besides a few words, dall-e is only useful for quick idea-generating concept art type stuff.
The real useful stuff here from a professional perspective are the openai tools that artists can use that help to minimize the more tedious parts of digital art without sacrificing creativity or control, like inpainting
Yeah good luck trying to put a copyright on it though. Companies can’t replace humans with AI because AI work can’t be copyrighted
@@acblook actually it is not making images from existing images. Coming from an AI machine learning engineering background, it's a bit more interesting than that :)...Yes it's been trained on real images but what it generates Is in fact completely original as in, never existed before on any platform. Its is not putting together parts from other pictures, it's genuinely mapping out each pixle based on the model. Look onto the math behind the models it's truly amazing
Love the quick Piximperfect cameo! Part of every single one of my design projects, gotta be honest. Such a cool challenge, love the results.
The Monalisa GoaT with Ipad that was amazing TIM!!. Also credits to AI as well as it was a difficult prompt to understand.
The point in the future is not to replace him but give him that as a tool.
Imagine to say you want that bird feathers on that deer.
It’s fun but too random to be useful.
I am sure you could do that if the let you use it.
Exactly what I was thinking!
It's the perfect tool for him to use as a Graphic Designer to get it to do the heavy lifting of brainstorming ideas, so he can focus his efforts more on the later steps of making the content, yielding even better results.
@@iPl4yeRTv Wait, so he'll tell the AI to make 100 unique, incredible images of a concept he wants to create, then he'll spend half a day making his own, singular, similar concept for... who? Himself? That's all well and good, but if he wants to make a living doing his art, who would ever pay him when an AI can do it 1000x faster with 1000x more options... for free? A lot of people in the comments just don't understand what we're seeing here. This changes everything.
@@ChrisCapel you dont understand what design is and what its meant to do, the AI tech simply gives good results on a purely visual scale ... but the designer's skills are not defined by their ability to make good looking visuals but rather their ability to solve problems and reason with the client as to what will actually help them ... AI is never gonna be able to do the critical thinking that a human does especially in context of specific problems that need to be solved
Design is not just art and even artists do things with a lot of thought and intention behind their work, most graphic designers are already aware of AI tools like this and they mostly just see it as a tool that'll help them
This is the reassuring comment I was desperate to find 😅
@@paramshah98 Sure, the client will give the "designer" notes and the designer will punch those into the prompt window and the AI will refine the image based on the client and designer's thoughts. None of this changes the fact that the vast majority of designers will be out of a job.
From Tim to Tim… this was amazing. Great job!
Tim not only treated this as if he had his job on the line, but all of us graphic designers/artists/photographers. We won't go down easy!
can we just appreciate what a great designer Tim is. dude nailed them all so perfectly.
I've been using MidJourney for a bit now and while it is amazing, it is still a little way off.
Some of the pictures it produces are awesome but they rarely match what I'd interpret as the prompt people give it.
I watched someone give the prompt 'the fear of the number 6 as a poster' and the first 12 attempts all had a number 8 on the poster with illegible writing underneath.
I tried about twenty times to get a picture of 'a friendly robot petting a cat', a straightforward prompt for human.
I got twenty pictures of a somewhat robotic cat's head.
I think it will be a while before they 'take our jobs' but at the same time it is exponential so maybe next week.
you just need to increase your poesy
Tim did an insane job! Good work!
Immediately jumped on the waitlist for this AI. It did insane work, especially with the robot woman. WOW.
Appreciations to Tim. I mean, he had represented the entire human race there.. right!!
My daughter is 14 and DALL-E is soul crushing for her aspirations. So bizarre to live in a world where AI possibilities are affecting your children's emotional state.
i wouldn't worry. If anything i would say the fact that your daughter will produce art that is really human makes it almost even more valuable than before! keep those aspirations alive! I like to do art sometimes and have been using these text-to-image for ideas and stuff. When i like something i transfer it to canvas to give it a human touch. All the AI does is cutdown on time to find inspiration and alot of the busywork of actual art-making. The value was always in the ideas and im sure your daughter will have some amazing ones of her own!
Not only children. It's also crushing my mental and emotional state as an adult.
well people need to stop supporting it and giving it so much attention. Ignore and let the phase.. phase out. I think issue is people are using all these tools and boosting the hell out of it. Need to ban this stuff I feel. But from what i've seen AI becomes dull and boring after the wow phase...
HYHAHSAHASAHHAAHHAHAHA
@@MohamedHassan-rh9iu I think this SHOULD be the case, but the reality is that in this microwave "I-want-what-I-want-and-I-want-it-now" society, many people don't care how the "art" is derived. Sure, there will be some who place well-deserved value on actual art created by individuals, but the sad truth (I think) is that those art lovers will be the exception rather than the rule.
This is such a great video !! I just would love to have a comparison between real life graphic design tasks.. those are very nice pictures / ideas, but we rarely need to do such surrealistic images for corporate designs.
I think the fact that the judges knew it was a computer vs human skewed the results. It should've been simply text and what image best represented the text. No context
Exactly.
Dall-E is somewhat daunting for graphic designers, especially since it's able to produce content instantly. But I don't think it can replace graphic designers entirely
Tim in his first prompt said it took him 3 hours to get to the point he did, which is reasonable considering he's working with no brief or a plan.
Where I work, it's exactly the same for literally everything, I don't get told what to produce or achieve, the only words I ever hear are "Make It Pop" - make "what" pop? So imagine my boss' surprise when I'm 3 hours in and I only have a series of concepts and not a finished product.
To people who don't work in this field or understand it, the concept of how long it takes to produce something with little to no information is alien to them. People like my boss don't deserve to work with real people, Dall-E is perfect for them.
Then there is the outlier, which are people who are looking for something specific, but find every possible way to help the designer understand what they need.
He did an incredible job and I think he nailed all 3 photos better than the AI
The arguably more incredible thing about this experiment, though, is the fact that the AI renders these images in ten seconds or less, while it takes Tim several hours to render one.
I liked the AI downy dear way more than Tim's dear. But the mona Lisa one from Tim was crazy good.
For now, considering the ai is barely finalised, lol.
The problem for designers come when the ai is good enough, and the choice becomes to use the very cheap ai instead of paying the better, but way more expensive human.
@@fredrikjohansson The problem with this is that due to copyright laws, Ai generated art cannot be copyrighted which means for designing it is commercially useless and this is arguably how it should be.
Watching this (and the original MKBHD Dalle video) after playing around with Dalle 3 for several days...Tim wouldn't stand a chance against Dalle 3. It's hard to believe that this was only a year and a half ago.
This is the same reason I came back here.. Dall-e3’s accuracy and fidelity is mind blowing. This technology has exploded and become SO advanced in just 18 months. If I was a graphic designer I’d be shaking in my boots rn lol
Forreal tho, my body isn’t ready 😅
I saw this video when it came out more than a year ago, and I'm watching it again now. He would stand no chance today with the most recent models, more so if you hired a so called "prompt-engineer" and gave them both the same time. But today DALL•E 3 doesn't need well-thought prompts.
The scary thing is that these AI image makers are just getting started, and developing amazingly fast. Poor Tim might win today, but tomorrow might be a different story.
The best thing to do right now is learn how to harness these new capabilities
Did you have to call tim poor?😅
Get outta here robot!
@@chananjamajiji5412 the perspective of losing your job to a soulless machine is kinda dreadful
They are, and I would say that this wasn't even a fair comparison for the ai. He had a very long time to work, in that time how many versions would the ai make and then you'd have to take the best one from those images and compare it to the artist. Or, how long would it actually take for the ai to make an image that surpasses his image? 10 seconds to generate an image... how many images in just an hour? This is insane. Edit: We may not even know how to get the most out of the prompt, maybe there will be classes on it in the future.
Well done Tim! Skynet has some more learning to do before it takes over.
Amazing what the AI can do but Tim’s pieces are just head and shoulders above them. The robot in front of the computers is insanely good. Stop worrying Tim, there’s no way they’ll replace you. In fact, they should give you a pay rise!
The thing about this is that this AI went from its first, laughable version tbh, to this in about two years. In two more years it could literally generate perfect, unique images. It also generates 10 pictures every 10 seconds, which means you could just refresh again and again until you find one that you like.
And Google just made an EVEN BETTER version of Dall-e 2 called Imagen. That's how fast progress is really happening.
@@yahiiia9269 there are editing and painting features in AI tools that will raise the bar even further, why stop at first drafts
@@yahiiia9269 It actually wasn't even two years, it was just over one year between the initial Dall-e release and Dall-e 2. That's how insane this progress has been, I keep trying to tell people but most are unaware.
@@acaustik8763 It could literally bankrupt tons of industries and completely shift focus on a lot of subjects. The sheer amount of insanity that is about to unfold because of AI is intense and I hope it proves just how pretentious humanity thinks it is when a machine can do anything faster and better than any human can.
Humanity needs to be taught a lesson when it comes to humility. I remember the artists saying that they wouldn't be the first ones to go, but here we are with insane AI models that can literally make words into pictures in SECONDS.
This comment will age badly. Tim's stuff is great, but an AI able to spit this stuff out in minutes, in any style you want and we're just at the very beginning of it all? Imagine how much more control and specificity you'll have with future versions. It's nice to see all the goodwill towards Tim in the comments, but the writing's on the wall. It's over for all but the tip top tier of illustrators, concepts artists and photographers. And I say this as a fellow artist. This will come for filmmakers too. Sad and scary and exciting, but ultimately sort of depressing.
Love this video, and great job explaining the process so thoroughly Tim!!
For other viewers' perspective, I think designers generally agree that AI is a helpful tool. I use Gigapixel's AI upscaling to save time on enhancing photo quality, Photoshop's content-aware fill for photo corrections, and Pinterest's like-for-like photo analysis for finding the right reference photo. Dall E is great for composition iteration and for giving us references on how to interpret things stylistically. Each of these help save time once we adopt them into our workflow.
TIM your work is MILES ahead. AI may have skill but Tim has taste.
Great job Tim
Definitely an indispensable member of MKBHD.
I love how visibly nervous Tim is about his job security
Showing my process to the world was very much on my mind and I really wanted to show up for it haha 😅
Dall-E (and Google’s DeepMind version) doesn’t “composite” multiple images in the way that Tim implies he is approaching this. Every image is a unique creation that has never existed before, yes, it is based on all of the images it has been trained on along with the styles of multiple artists but the artwork is original as I understand it.
You are correct.
how do you feel about the fact that those creators haven't been asked for permission and OpenAI now wants to sell licenses for this latent space that original artwork can be drawn from? They took 400 mio images, no one knows exactly which and from where, and theoretically it could've been fed even with Tim's own work without him knowing.
@@LarsRichterMedia Well isnt it like you looking through a picture album or old illustrations for inspiration? Dall E learns what looks good and how things relate to one another, and then create new stuff based on it. Just like humans do.
@@terogamer345 If you want to declare Dall E to be coscious, then yes. Any conscious entity no matter how their mechanism of perception happens to work woud have the right to perceive a public space. A mechanized visual harvester on the other hand that potentially copies your data and processes it into a latent space from which it draws original visuals we might want to look at a little different than what "humans do". At least for the moment would be my opinion.
@@LarsRichterMedia It's similar to how humans will view someone's art and then imitate but not directly copy and paste it. That's how people grow their capability to make art, being exposed to different styles. Also, is it not extremely transformative?
Tim is more scary than DALL-E 2. Imagine how Tim 2 would be?
damnn,,as a designer i wanted tim to win this and thankfully he did. OMG, fuk this Ai. this was so stressful. good job tim. you represented all of us and made us proud. love you my boy.
I believe the best part would be to augment Tim with Dall-E for quick idea generation, and then make it by hand. While computer can definatelly create convincing images, it lacks human touch in sense of humour, semantics and general feeling of intent in the work.
Yeah it lacks those things…..FOR NOW!
Dall-E’s deer and robot pics were better to me, and they took 10 seconds - Tim smashed the Mona Lisa one though. I think anyone under the impression this kind of technology won’t take over is massively mistaken
Also Dall-E would give 100s of options per prompt, in seconds. The Team chose the prompts so maybe this is rigged lol
@@tbone84828 if tim has access to it, he could just send prompts and cherry pick the ones that look good. And maybe composite them.
@@satibel exactly my dude.
@@tbone84828 it makes 10 pictures in about 20 seconds. They probably picked the best out of one or two runs.
So far man's advantage in such contests is creativity, unpredictability and the addition of something subjective. This is best seen in the image of the Robot and the Goat. Tim didn't just transform words into a picture, he added something from himself. Apart from the technical quality, that is why his graphics are better in this case. They offer something more than just the basis implied by the description.
But the sad truth is that this too will be faked by algorithms sooner or later. It is just a matter of time. And it's a few years rather than a few decades.
There's no point in fooling ourselves. In about 20 years all content, written, visual, audio, will be possible to create to some extent in an automated way. It will be faster and faster, more and more perfect and closer to what people do today.
There is no escaping it. The AI cat is already out of the box.
Lets just be hopeful, remember the first days people were discovering electricity, one can only imagine the fear they had thinking about the future. But, since then we discovered so many things about electricity that we realise there is no need to fear it.
dale with a creepy smile: clock is ticking tim tim tim tim tim and end is near.
Corridor? But in all seriousness, his goat drawing was incredible
Imagine ai model which can make live videos in real time based on viewers mood and thoughts, amazing and terrifying at the same time.
It'll happen within two years I think
People call me crazy when I tell them we're only a few years, if not months, away from AI beginning to sweep across creative industries. Character, concept art, paintings, that's already a done deal. As people continue to use it, AI is trained to get better and more specific. So eventually logos, posters, motion graphics, entire CG scenes, clothing, interior design, architecture, music, etc. It will all be done by an AI in mere moments. Eventually, an AI will be trained on popular music, and it will be released into the wild, only then will a legal battle ensue because of that industries power and resistance to offer any music for AI to learn off of, but it will ultimately be too late for the artist, musicians, designers, and engineers. If we don't take a stand against it now we're finished.
"Oh, but don't worry. The artist will still be there to touch it up." No. They won't. Once these things are trained enough and get fast enough, all it will need is a prompt. And it won't have to be a very complicated prompt to get what you want either, as the public trains them on complex prompts they slowly begin to understand better what you're looking for until it can be entirely dumbed down. It will be able to come up with its own prompts if it wanted, its own styles, it will be able to take into account physics and math to create structures and re-engineer many of our technical designs to be superior. Sure, it will be some sort of golden age for progress and innovation, but what will happen to the artist? What will happen to all those jobs? Because even if companies still hired any, they would only need VERY few.
So, should I even pursue a future in art/design? Should I post my art online when I know silicon valley is gonna swoop in and add it to its AI without my permission? This is the end goal of every site we casually upload our content to. If u wanna get really crazy I'd say this is the breaking point for the Internet. The line between free use and personal property online will finally be treated as a serious issue. NFTs were an attempt to solve that, but ultimately were just a disgusting mess. The problem is once trainable AI gets into the hands of the public, there will be nothing to stop people from stealing anyone's content and training an AI to replicate their work and then call it their own.
There's no stopping this. AI art generation is the will of the companies who artists have sold themselves out to for the last 100 years or so. A large section of commercial work will be off-loaded to AI, but I think what is left over will be better human artists and real art, not souless concepts or animation made for unfair to minimal pay. Harder to find, less profitable for people, but still better, though Im unsure how long it would last in that state. People are so numb or bitter nowadays that I don't see anyone caring until it comes for their job.
Really, the only thing that should be lobbied for is for image hosters to disallow image scraping for use in AI data sets unless given permission by uploaders. We already know how hard this is to prevent, hell it might even be too late. AI is a data parasite on the back of thinking humans. I don't know what goes in to collecting training data for this purpose, but if allowed to continue with this level of freedom, it really will destroy many careers because the only solution to preventing AI from developing is for humans to not produce at all.
- Dystopian -
There should also be special laws placed upon organisations that develop and use AI, such that they are known and must abide by some code of conduct. They are not normal businesses and shouldn't be allowed to operate as one when they wish to employ artificial intelligence instead of humans. I'm fairly sure work is being done to implement some control.
I don’t know why I feel triggered and upset after this video. 🤷🏻♂️ - this was mean spirited. I wanted to hug Tim. The software was scarily good. I can see graphic designers becoming a rare luxury.
Best YT vid I have watched all year.
Now if you had somebody with the same time budget as Tim, tweaking the prompts it feed dall-e I think we would have a clear winner
Might as well keep Tim then. He's a beast!
exactly what i was thinking