As mentioned in the video, the channel has been hit with the dreaded RUclips "Invalid Traffic Bug"-- if you'd like to support the channel, please consider joining the Patreon! www.patreon.com/TheoreticallyMedia
The problem I see with all AI that attempts to do text is that, depending on the AI, it attempts to make all or most text all upper case 🤮. I haaaaate all upper case text for a variety of reasons. Until some AI fixes this terrible case bias, I'll stick with traditional text creation tools. The comic book model looks quite worthy of checking out though. 👍
@@eri7-11 did it yesterday! And over the weekend I’m just going to do the old “turn it on and turn it off” trick that some folks have reported (occasionally) works. But the whole thing has been super eye opening, and has really driven home the point that, for the health and longevity of the channel, I can’t rely on RUclips as a primary revenue source. Given the other options of Audience support via Patreon, or just taking on every sponsor that slides into my inbox? Well, that choice is pretty clear to me!
How would you go about getting a consistent images for your comic? When I asked it to draw a space craft, the AI altered the appearance of the space craft thought out the comic.
It can't. that's the flaw of AI. That's okay. Someday it'll be more consistent, but by the time it does, you could learn how to draw things yourself. AI steals from artists and creators and most of all and more importantly, also steels your potential. No one respects, because it is entirely built on theft.
That argument is still in play. And to be honest, that only applies to US copyright law. I believe Japan has already decided to let AI fly, And you can copyright if the work has been transformed by a human, but there aren’t any definitions of what that actually means. How do you quantify what transformation is? If I write dialogue balloons, is that 10% a change? 50%? Who decides what that means? Personally, I think the whole fight is doomed to fail, namely because Warhol’s Campbell Soup artwork has already been upheld. The fairness thing, I totally get. Look, I’m a proponent of UBI, but that’s a whole other discussion…
@@TheoreticallyMediawith that type of ideology you won't mind people using your "work" as a baseline for tweaks and edits. After all if it wasn't for AI most people wouldn't have the opportunity to profit from professional artists. If copyright does side with the AI, everything will be up for grabs
@@TheoreticallyMediaso you wouldn't mind me reuploading your videos with a "speech bubble added"? Or a watermark of my channel - would that be enough for you to consider it transformative?
@@TheoreticallyMedia - Is it? Whatever the US does, other countries will have to follow due to basic international agreements and treaties alone. You might want to re-check. Japan has not decided to just let 'AI fly', they've explicitly stated that the law still allows for copyright claims against AI. They've just made the distinction between generation vs. training, and that training is still ok for research and educational purposes, which is the same everywhere. Also the Warhol Foundation recently lost its case vs. Goldsmith, so no, just because Warhol's Campbell Soup case was upheld, isn't the blanket carte blanche as you think. So no, the 'fight' isn't remotely close to being 'doomed' to fail.
@@NovaaGrind Guys don't mind these talentless people, they always defend AI art because they cant do anything for themselves. It is true, if you follow that ideology so then everyone is going to copy each others to the end. Period.
what i find especially lacking in those ai tools is the consistancy to keep characters or places ... it would be good to save models and keep them to try to generate and narrow scenes from them
Yes I came here because I had an idea to see if anyone had something like mid journey but focusing on character, building and comics. I would like to create something where I can detail a character's appearance and personality generate those image. Then give that image a name. Let's say his name is Luke now when I talk about Luke the AI will recognize his appearance and put him in the scene Then you will always have Luke. Luke can write a motorcycle swim with dolphins or even eat a hot dog
You couldn' t be more right. I try alternatives: using artists on inputs, ages of art/paiting/comic strips, and so on, but I think I will always only be able to create concepts with consistancy, not a great work with a same line of art.
From what I'm seeing, some people try saving 4 to 8 best result pics and then using all of them in the prompt in future generations. Like maybe the 'average' of these is close. I might try this and also see how it works in MJ 6.
He has the code up as open source, so I suppose it could be. Let me know if you need a link! (He mentioned that you might be able to train it on some LORAs!)
@wholeness. AI is a scam, like bitcoin. Don't ever use this stuff. If you have a great idea, you'll just be handing it over for free. These generators cannot create comic books. They only replicate existing artwork. If you're happy with that, you cannot be an artist. In time, once the bottom drops out of this fad, genuine human artwork will double in value.
You always worry about how this could threaten people's jobs, but with some improvements it could be a useful tool for those of us who have ideas but not the artistic skill to properly visualize it. AI shouldn't be used to do all the work, but as simply another version of the brush and canvas guided by actual people and their ideas and imagination it's perfectly valid
Sure, skip hiring a human artist and rely on A.I., your final product will end up looking exactly like the product of the thousands of others who had your same brilliant idea and chose the same A.I. engine. Good luck with that when it comes to finding a way to stand out in what will soon become a very VERY crowded marketplace. Why not reach out to a real human fellow artist, share your idea, exchange thoughts, be friends, create a fantastic original piece of art, be sad together if right away doesn't get the recognition you both think it deserves and celebrate together when you reap the results of your collective work.
Here's my problem: the layout of a graphic novel is just as important as the content of each panel. You are showing us panels with cut off content. Instead, it should be a very hyper-stylized text-to-image service where you describe the story, the panel you need, an aspect ratio, and possibly some dialog for the super-lazy. Then we can take that content and do it right in Indesign. INSTEAD, this thing is trying to pump out four images at a time, clogging up the servers and attempting to do all the work at once instead of asking simple obvious questions like "Is this a two-page splash?" or "Is this part of a nine-panel page?" This, as created, is neither useful for web comics (wrong format) nor print comics (no structure).
I mean, yes and no. The thing with this is that it’s trying to make a comic page, but it has no instruction to create a narrative between the panels. So you end up with panels that look (stylistically) like they should match, but actually don’t. But, if you generate 4 pages of them, you’ll find that you can cherry pick various panels to create a narrative. It might not be the one you originally intended, but that’s sort of the fun of it. It’s why I bought up the Marvel Method. I’m sure Stan had to think on his feet for a few of those issues. (Then again, Kirby and Ditko were so good, he probably didn’t have to do too much extra work!)
And there's always a winery or somewhere dude give it a break okay just as still experimental stuff did you think the wright brothers would have gotten anywhere if they were so chickenshit and wearing girl panties they took a chance in a dare
Very interesting video- and the technology is amazing, but I do have a question- (an honest question about how you feel- this isn't meant as a way to upset anybody. I think AI will affect everything, not just art) -but as you promote the use of the tech for comic creation, what are your feelings about the fact that, at some point very soon, videos like this will also be created by AI? -replacing you and your channel, or at the very least, taking more of your traffic away and demonetizing you further? This is all uncharted territory, so Im not taking a stance here, Im just asking a question about how you feel about that possibility.
As far as personally replacing me? I don’t see it. The channel has largely been built on 3 things in my opinion: 1) obviously the tech itself, and covering the latest developments as they happen. I get beaten by Matt Wolfe on this all the time. I’m pretty sure that man never sleeps. 2) My personality and viewpoint, or…dad jokes, for lack of a better term? So, truth be told, I generally write out every video in full script form. It is very dry and very bland. But as I’m shooting, I’ll start improvising, and going on tangents or coming up with non sequiturs. AI just isn’t good at that. 3) I do a lot of hands on examples on the channel. That is one thing I kind of pride myself on and is something that I think separates me from a lot of the other channels in the space. I’m not calling anyone out or anything, but most channels will say: “hey; here’s a thing, isn’t it neat?” Whereas I tend to do things like: “Hey, here’s a thing, isn’t it neat? So I made a recreation of the famous scene from Titanic, only using the tool, I’ve turned Jack and Rose into rabbits.” Even just in this video: most would probably say: Here’s Comic Factory, this is what it does…” but I’ll take the time to create 2 pages with it, and in doing so also talk about how awesome Mobius is. So, no. I’m not worried about AI replacing me. There are too many elements that comprise “me” that any AI system just can’t factor. And that isn’t coming from a place of ego, I mean that about all of us. Sure, there will be a sea of NPC AIs delivering the news soon enough, but I tend to think that people like connecting with people, so when someone runs across a video I made; they’ll come back. …or they’ll roll their eyes and say: “ugh, this guy again?”
@@TheoreticallyMedia - I appreciate the thoughtful reply- the tools AI employs are mind blowing for being so new to the game, relatively speaking- and I hope that AI, indeed, helps people, but I fear that AI will start living "for" us instead of helping us live better for ourselves... I dont know if that makes sense, but its worth having meaningful discussions about, thanks:)
Big comic book fan breathlessly promoting AI that wants to put comic book artists out of business, while complaining that a computer algorithm is hurting his revenue. Am I missing something?
Haha, that's a solid take. Look, I don't think AI is going to put comic book artists out of business...in the US, the direct market and comic book publishers are doing that job on their own. If anything, I think this just allows artists to accomplish more, and not grind themselves into the ground to produce one page-- or gives someone who has always dreamed of doing a comic, a shot at making one without having to break the bank hiring an artist, inker, colorist, letterer, and then foot the bill for printing. They're tools. Artists can use them to make amazing stuff, non-artists can play with them to make...honestly, stuff that will probably be sub-par and no one will want to read anyhow.
As a colorist, I don't agree with this. The industry is only interested in making money, especially big companies. This is a huge blow to honest artists. soon there will be more "Ai Art" than real ones. and the algorithm will make it very difficult for people to find the REAL ARTIST and their art. Even though independent artists have great works to share...
@@RarebitFiends No. Because the final decision rests with audiences. If you think audiences are all highly discerning, 'cost is of no importance' types, then please let me know where these people are.
Not bad. I would love to see an AI tool that is able to accurately copy our original characters, and generate images with them by displaying them in a consistent way in each scene.
You can do something similar one image at a time with 'reference' controlnet in automatic1111. Not always 100% but pretty close. AS for generating panels... yeah more people need to focus on that concept along with this.
It's already sort of possible by "training" your a.i based on your own artwork. Granted you'll probably need to get familiar with some coding or at least command lines in something like Python.
Interesting, but where does this art come from? Is this scratch made by the software or it based on something preexisting? What are the copyright implications? How much futzing before it’s not just ai?
So, to be honest: in my super non legal opinion, sure. It’s part of a larger rant I’m brewing, there’s obviously a big fight brewing about copyright etc, but ultimately it doesn’t matter to you and I. Legal teams are going to duke it out, and I’m sure it’ll make its way to the highest courts. But- A) This is just US law. Japan and other countries have already decided that Ai art is completely fine. B) this is a legal issue. You’re a creator. Don’t let paperwork stand in the way of you making stuff. Think of early hip hop/rap. I mean, the EARLY stuff: those artists didn’t care that they were looping copyrighted records. They were just creating. Do that. Or…maybe become an old school rapper? I don’t know, how’s your flow?
@@mf-- that is 100% true. Although I think the likelihood is fairly low. Unless you were say, doing a Batman comic, which would be fanfiction anyhow. I suppose if you had a bland idea like “Harry Potter, but they’re Cats” someone could “copy” you, but I also tend to think that idea wouldn’t gain much traction. Quality storytelling will hold an audience, nothing more.
‘Armorica’ is the Roman name for the part of France we now generally refer to as Brittany. I think the ‘Armorican’ style may be a reference to the European comic ‘Asterix the Gaul’ which is situated in that particular part of France. I’m not sure though because this style in Comic Factory doesn’t match the style of the original comic very well. Anyway… great, informatieve work again Tim. Thank you!
Ahhhhh, there it is!! I couldn’t for the life of me connect Armorica to comics! Asterix is it! I kept thinking Nemo in Slumberland or Moore’s League of X as a style! I think with some Asterix-y prompts, that style might show through a bit more. I’ll say the models can be a bit finicky and picky about what it produces. Thank you!!!
I confirm! The style is a bit difficult to catch and we don't really see the difference I think, because I didn't want to be too explicit in the keywords. It would work better with a LoRA but currently the comic editor doesn't use one (the tool is more like an homage and its goal is to have fun, not to plagiarize living authors or copyright holders, so I'm trying to stay vague in the targeting of styles).
@@TheWatchernator So, I think if you give it more of an "Asterix prompt" -- basically, something more in line w/ that style, it'll produce better results. Give it a shot and let me know!
It CAN, if provided with LORAs, which the base model does not have. That said; it is open source, so perhaps someone will pick up the mantle and run with it?
It is kind of a “garage project” that is run by one person. I can imagine there’s been some traffic strain recently. Maybe a little duct tape and a kick will get it going again!
The comic generator is a cool idea but the consistency is a real problem when you're looking at 4 panels on one page and the characters look different on each panel. One person may be wearing a blue suit on one panel, and a white suit on the next. For me the best solution is to pair stable diffusion with loras to individually generate each panel, and using inpainting to redo any inconsistencies.
@@robertdouble559 yeah. just 2 more weeks. You have no idea how difficult it is to create a good comic page. these examples are just random pictured put in frames. not a comic page.
Agreed. I think it’s a good stepping stone if you think you might be getting the comic bug. And I think it’s a fun thing to play with if you actually know what you’re doing. As it stands: I think you can produce a cool mini comic with it, but nothing like a 800 page graphic novel. …but I guess that’s for right now. We’ll see in a few months where we are!
To be honest, right now, you don’t. There are some tricks you can pull- like archetype characters (the spaceman in this example), or prompting very basic character designs, say: Red Haired women in a black dress, and then use a face swapping tool afterwards as a finishing touch. I think all of that will be worked out in the next few months though.
@@TheoreticallyMedia sir I made my own manga by Bing ai but the problem is the girl'face is same but hair is not same so how can I change hair I mean is there any only website there where I can change some details or edit and I can make a character consistent by doing some changes
I mean, I’m always the guy who ignores ToS. I tend to think it’s a CYA on their end, more than prevention on your end. There isn’t really a quantitative method to tell if an image is AI generated. You can eyeball it, sure- but without seeing the source file, you’ll never really know. And then ultimately, there’s no way to tell which model you used as well. Will we see it in a Coke or Nike ad? Nope. But, for your own personal projects? I say whatever, go for it. If you ever create an empire that’s as big as Coke or Nike? Well, that’ll be a worthy problem to have!
@@TheoreticallyMedia Sure, AI does have its benefits, but it's clearly trained on other people's art, their style, and frankly, copyrighted work. So by all means, "go for it" but also mark somewhere that it's AI art and not your organic originally drawn/painted art. There has to be some ethical guidelines for this unprecidented flood of art AI platforms. The Renessaince masters are turning in their graves! Now, if people want to pass AI art as "their own" that's between them and their conscience, and how you stated "you'll never really know" if an image is AI generated. To those that sneak this into manifested reality, the good thing (from dedicated artist's perpectives') is that, sure, we will never really know, but also the AI art creators passing these "works" as their "art" will know one thing for sure- and that's that they can't draw and need someone else to do the work for them so they can just slap their name on it. Except, the "someone" is a program, most likely written by someone of the same caliber of short comings in one area. Draw your own stuff, people, or at least label that it's AI, like a decent human collaborating creator would do in the indicia or credits of any comic book.
@@TheoreticallyMedia EDUT: btw, it's nothing personal against you, because I truly enjoy your chennel, and have learned a lot. Mostly things others are doing, but you have a friendlym, straightforward way of explaining that gets to the point and is easy to follow. That aside, it's just a matter of time before there'll be more AI/fake art than real art. Most proper artists using Ai are blind to what's coming, and are just contributing to their own demise, just as much as the coders of these art AIs are contributing to AI out-competing artists. Prompt writing is an art, some might say. But prompt writers are not artists, and your whole "you'll never really know, and F it go for it" approach to this, is part of the same mentality every person saying some AI-generated image is their own "artwork," Also, "for your own personal projects" is a moot point if people are not disclosing they used AI; ipso facto, not made by you.
@@TheoreticallyMedia It's hardly CYA, it's that legally the generated work cannot be copyrighted. The AI creating it can never hold a copyright to transfer to whoever prompts it. Anyone trying to use it commercially is open to having their work stolen. Just like all these AIs are stealing from actual artists. Oh yeah, this style is 'Moebius'-like. I wonder how that happened? But yeah, people talk about AI as if it's not just outright theft.
As a writer and artists, these new comic generation sites are pretty interesting, but they are really in their infancy right now. Anyone who already has a story, and a vision is going to find them incredibly limiting to use. Inconsistent characters/models, fixed panel layouts, and the bazaar interpretations of prompts to name few issues. The tech is moving so quickly though. Hopefully these issues will be addressed soon. With the fixed styles, I do wonder how quickly the audience will get burned out from the lack difference from comic to comic.
Audiences quickly become jaded to new technology due to their overuse. Like CGI use in films. I suspect the temptation to produce cheap material quickly will cause enough people to reject ai produced artwork, as being cheap and artificial. A compromise between ai and artists will have to be reached.
You can! There’s a few different ways, this is the latest and the one I like, and although this one is about video, it works for still images as well: Your Face In Any Movie! Easy/Free Video Faceswapping Is Here! ruclips.net/video/Y8CMwl2Fv0g/видео.html
The comic book generator is so interesting. It's a shame you can't lock Midjourney to a character / style set so you can generate imagery with specific characters and outfits etc. I'd love to make a comic out of my feature screenplay just to sell it in.
MJ has a new inpainting feature too, so maybe with enough prompting we can get more of the same character (I know I did panels in the past, but it did lose the plot fast without an easy way to fix small things).
@@GRNKRBY I think that’s a pretty ideal use of the Inpainting feature, and one that I do want to dig into at some point! I almost like this platform because it’s uncontrolled. It almost feels like working with a stubborn artist that you have to work around!
As an artist, I feel sad with AI art... It might be a tool to help you with your art, but those of us who put specific style and much time practicing... I remember working hard to learn facial features and how my first ever portrait felt like an amazing milestone... Now anyone can just generate an image... Even if it doesn't replace real artists, evil company leaders are going to prefer "cheap and obtainable" if it cuts cost corners. It's already hard with regular competition to make a living as an artist. Now if you can't compete, regardless of putting your heart and soul into your work, it'll be even harder.
“Evil company leaders” are going to bone you as an artist, AI or not. Have you seen mid-low tier artist page rates for Marvel or DC? They’re below minimum wage. Don’t focus on getting a job drawing Batman. Spend your efforts to create something you own. Something that your audience recognizes as you. Build and cultivate that audience. That’s how you make it as an artist. (That does for every medium of art)
I’m not saying this is the original poster of this comment, but: having now done a bunch of these videos and interacted with a lot of people via comments, there is an underlying attitude with some naysayers I’ve come to recognize. There’s this defeatist “I’m not going to bother trying because of this technology” not “I’m quitting my profession because of this technology” it’s “I’m not going to try” Making a living in art is perhaps the hardest path you can choose. Some view AI as a barrier to entry, and it’s an easy scapegoat right now. But what they don’t realize is that there are 1000 other obstacles behind that. Again, not speaking about or to the original comment here…
I'm not going to waste time crying about AI. It's pointless. Every artist is aware of the 1000 obstacles related to making a living as an artist. I don't think there is another profession that typically has "starving" as a common descriptor. Thousands of talented people have been dissuaded from choosing art as a career over the centuries because of that reality. I know many of them. AI is another (exponentially higher) barrier to entry that logical, thoughtful, forward-thinking young people will contemplate when choosing what career will allow them to pay off their college debt and feed their families in the future. And by the way, they're coming for your job as well.
Don't get discouraged, AI users will never experience what you just mentioned, the marvelous feel of nailing a drawing, a creation, something you can proudly say, I drew this, with pencil and paper. You just say something really important with all this AI dumbness, it is just an image generator, it has always existed but now it is much more polished. Art is a journey, being a pro artist is hard, and now even harder, but with your creativity and passion, plus hard work you can pull it off. Best of luck and keep going!
You currently can! The results might be a bit wonky, though. There are a few videos on the channel where I go over those tools. Check out a recent one on a free video generator called Haiper! It was just a few videos back!
What I see is a bit repetitive to say the least. Almost every shot is straight on. They'll need to create some camera-angle commands to diversify the cinematography. But for writers and film makers, this is very useful.
Agreed. I know the dev is still playing with the project, but it is a “garage project” of one person, haha, so I think of this as more “proof of concept” But, agreed with your point! It is fun to play around with though!
You should just be able to get there? It might be updating right now. It is a one-person operation, so it does tend to go down when he is working on updates!
I have a BFA bachelor fine arts and I have no idea what the style that last was. If you have any idea which AI I can take a screenshot or a sample of a cartoon style I want and then turn it into video my video into that style that would be amazing thank you for your work.
Can you use your own character design to make comics with? That is consistent with the same characters and the same costumes. And every frame screen or was it just random like AI usually is?
Well, I’d argue that if you resequence the raw output, and add captions/dialogue balloons, that would constitute alteration. But, I’m not a lawyer and the courts are still fighting that one out. I’ll say, the transformative argument is going to be an interesting one, as Warhol’s Campbell Soup can art has already been upheld. So, we’ll see
@@TheoreticallyMedia you should read the new law on it issued by people who deal with copy write infringement its very simple and concise they issued it as a sentannce instead of a paragraph of law jargen.
It’d be great if comic factory the hands/fingers can drawn out the hands. They are either nothing,?- scribble or warped out nubs. I wonder why that is?
Yup, it is open source. And Julian has indicated you can even train Loras with it. That said, he also admits that his code is a “garage project” and lacks any proper documentation, so you’ll have to do some legwork on your own. I respect a messy workspace!
It’s possible. There might be some argument over copyright etc if you disclose that it is AI generated, but the truth is: that fight is ongoing and besides that, many counties have already OK’d data training…sooo. That said: what you’ll quickly find is that making it (which is still going to be challenging) isn’t the hard part. Getting people to read it, and/or buy it? That’s the tough part.
@@TheoreticallyMedia Good to know! Thanks. I think if the subject matter (story, characters, etc) grabs people's attention enough it might work, depends what it's about I guess, that's the fun part to work with. Let's see what happens!
I went to the comic factory site, it says you can use it once before registering....When I tried to sign up, it said "hugging face refuses to connect"....Ummmm okay.
Same. Try again in a bit, I know Julian was playing around with the UI. The button might have gotten lost in an update. Just remember: this is just a dude working in his garage, not a fully fledged company product. Haha. It is literally a garage band!
Boy I can't wait until AI is making all of our art and music and poetry and writing all our books. All this while we lug rocks from one pile to another for the corporate overlords that own the AI.
Okay, problem: None of the character models are consistent from panel to panel. Even in the late 90s and early 2000s with Spawn, there were fans that would write in why his shoulder pad had a different number of spikes from panel to panel. Here, the buggy is different from panel to panel as is 'Space Guy's' character model.
True-- I think if you were to expand this idea, you'd need to bring it in to another platform for inpainting. Which, actually just proves the point that none of this is "easy"
The ai program doesn't allow uploading of my own character modelsheets. So it doesn't place my characters in the scene. But what this can do is set up the layout, and then I can draw my characters within the layout. Saving time.
Yeah. It’s kind of what I like about it. It isn’t a “prompt and you have a magic page” it takes human work to make it happen. It’s an important aspect of these tools!
I will never believe in using AI to make manga or comics... That's something that should be done by a human so you actually see their artstyle and the likes
another problem is trying to keep the same characters in your comic with out the face changing to something diffrent how can you have a hero if you cant keep the same person
People like me people use this to make comics àre passionate ,them making something that they have wanted to make for a long time, even doing it in second means they will still cherish it I im doing a comic/manga by hand and I'm making a reversed one with this ai to compare it to my hand made comic for ideas and advice
That’s excellent to hear, and exactly what this kind of technology excels at! If you also do your own art, you might want to check out Prome as well: ruclips.net/video/P1-VpUFHoEo/видео.htmlsi=snUUe4WezzhV5Zd4
I mean. It’s a one person project, not a VC backed product with thousands of employees. Think of this as more of a research project that will eventually lead to the “thing”
Ha, so that is a 1980s Peavey Horizon 2. Made in the US, swamp ash (heavy!) I love it-- 24 fret super strat, with a weird system on the tone control: Kick the tone up to 10 and it is a single coil, bring it down to 7 and it's a humbucker. Oh, and it has a out of phase switch too. Just a weird and awesome guitar I got lucky with on Facebook Marketplace!
Yup, that was from a Reddit Thread that Julian posted. Lots of good technical info there if you've got the good brain. I have the middle teir one. Haha, I'll upgrade at some point...maybe more RAM? Here's the link: www.reddit.com/r/StableDiffusion/comments/163ikmd/wip_comic_factory_a_web_app_to_generate_comic/
Yeah, my current trick with all of these tools is to use very basic character designs: “a woman with red hair in a black dress” or that sort of thing. That will usually get you a relatively consistent character that you can modify with face swap tools, or Photoshop. That said; I think we’re only a few months away from seeing consistent characters in all these models.
@@TheoreticallyMedia If only there were an option to select characters, settings, objects, and design details on materials as defaults for a consistent narrative, not just the style. And a way to save those settings to carry over into later projects.
Terrific find, Tim! I like how you used Comic Factory, despite its limitations, to make something creative and coherent. Have you tried Comic Life 3 for laying out those images? I used it for a graphic novel. It does all things comic except making the images. So you have panels, text bubbles, sound effects, filters -- it's really handy if you have images but would like prebuilt pages and appropriate tools. You can even lay out the panels in whatever way you want. It's something like $10. It might come in handy once MJ actually shows us the storytelling tools! And so sorry to hear RUclips cut your income. As if they don't take most of the money anyway. I hope you get many Patreon supporters, you definitely deserve it!
I clicked on this video and then was listening to it in the background while working. Heard your voice and thought, wait. MetalJesus rocks has a second channel. I listened for another minute or so and thought, he sure does. I brought the video up to double check, but no, you are not MetalJesus. Your voice is sound exactly the same. Check it out if you haven't heard that already.
Technically you can’t. That’s an issue all of the AI generators work now. The workaround is to use archetypes. Like, in this: an Astronaut. Because the bot will mostly use the same idea of an Astronaut over and over. There are always nagging details that it’ll miss, but you can fix them via photoshop or image editing. Stuff like, James Bond, Woman in a black dress with red hair, etc. those are the (bland) characters that tend to result in very similar (though not exact) characters.
Oh, you'll love this little bit of Internet Lore: (Not mine, just an old Reddit post) "At a scifi convention hotel party with a Klingon theme I was talking with a gentleman who spoke Klingon. He said it came in handy when he was on a business trip to Japan and got lost when he got off of at the wrong subway stop. He tried speaking English and broken Japanese with other commuters but they were ignoring him. Then he saw a Japanese man with a shirt that had Klingon writing on it, thinking "Oh what the hell" he approached the Japanese man and uttered a Klingon phrase. To the gentleman's delight the Japanese man muttered a Klingon phrase back and they both had a lively conversation in Klingon since they both shared the language. The gentleman was given the right directions and found his way again."
Per their latest: that’s up to you. If you’re a design firm hired by Nike, and their policy is no, then no. If you’re a local pizza shop who is designing a new logo? Sure.
I see a lot of comments about the lack of consistency in AI-generated image sequences. How do you achieve the desired consistency to make your own comics? Very difficult - learn how to draw, master perspective and color theory. It will take years and years, with no guarantee that one day you will be able to even come close to the heels of masters like Hergé, Moebius, Bilal, Hugo Pratt, Will Eisner, Robert Crumb, and many others. Better start now!
The whole copyright thing is in flux. Currently you can’t output AI art and just copyright it. It has to be manipulated by a human being and then it can be. But there is no definition of what manipulation is, or to what degree. 10% manipulation? 50%? How do you quantify that with art? In the case of the “AI Marvel Method” that I detail in the video, I’d argue that is a significant amount of manipulation.
I'm using it to pretty much pre-viz my novel. So characters looking the same thruout doesn't matter as much as perspectives, segments of the monster's spidery legs -- all sorts of great mood-setting stuff. @@TheoreticallyMedia
I'm actually more of a fan of Danka's earlier work. That Comic Factory looks awesome! I don't understand why every old school art image creator tool from Paint to Photoshop had combined a way to add text over your image, but I've yet to see an AI art generator have that feature. Seems like it would be a no brainer. Ideogram is still hit or miss with the text. You can type in lyrics to be rapped or sung by music generators, but I can't get a little 14-point Times New Roman into my comic's thought bubble? Still pretty amazing/cool stuff! Hope YT shares the wealth better.
You would figure that someone would have suss’d out a ChatGPT plugin to pop over an image. But hey, at least this keeps the lettering folks working! And man, that is such a lost art. I still can’t believe there was an era where they did that BY HAND!
@@TheoreticallyMediaI'm old enough that I trained in constructing lettering by hand. (Do they still do that these days?) I only remember enough of that experience to admire the talent and skill of professionally produced work.
I make comics because the process is fun and I love the medium, all the craft that goes into it. This shit right here, is not fun, and there is no love involved at all. I feel bad for all the folks who wants to make comics this way.
Right on. I mean, this is not stopping you from making comics at all. There are some who will use it to generate comics, sure-- but the medium will go on, and I think most pro artists will end up using a hybrid of their art and some AI generated stuff. I've spoken to a few pros at the Big 2 (no names, for obvious reasons) and they aren't worried about AI at all. In fact, they're learning it. Different stokes...
Not positive, but "Amorican" might be a play on Alan Moore's work through American publishers. The style shown reminds me of the detailed backgrounds in things like LXG and Top Ten.
That’s 100% what I was thinking. Plus, maybe a little Little Nemo in Slumberland, but I’m not sure what that particular style of art is classified as. Great minds, though!
My fear of AI is not what it can do now but in a few years. Once they figure out a way of connecting narrative, anatomy, light, mood etc... every single human archetype, premise or story will be made available on demand.
True. But in a few years all of that can be applied far beyond art. Honestly, as the guy that makes these videos, I keep saying: we need to have serious discussions about UBI. Or some sort of society safety net.
@@TheoreticallyMedia The individual doesn't exist in isolation. After family, career is the most common place that people find a sense of meaning and purpose in their lives. By being productive in their work, through career progression and through the status that income provides, people derive a huge amount of their sense of intrinsic value. Without avenues to experience a sense of meaning in their lives- in ways that are meaningful to them- people become psychologically sick. If AI replaces human labor on a broad enough scale to justify discussion about a UBI, then vast numbers of people will be deprived of one of the key methods by which people experience meaning. We will be living in a society much more sick than it already is. The idea of a UBI is often given as a hand-wave solution to the ramifications of AI, but that's usually all it is- an idea. Not a particularly well thought-out one as far as I have seen.
I think there needs to be a “100% Made by Humans” label (like “Made in America” and so on) that can be applied to art that is created without use of AI. That way those of us who want to support human artists can do so. I’m very suspicious of AI and don’t particularly like it, but we’re not going to get this genie back in the bottle.
wow, this going universal, like big bang theory, which happened about 15 bill years ago, now it's here, im ina state of confusion now, mayb going into the lv sphere will clear things up>
unfortanetly the tool is not useable for alot of stuff as it seems you cant get any kind of consistancy with the Charicter it might work for SCI FI where they can keep helmets on but thats about it
I need to study more Herge, to be honest. Always liked him, but never fully dove it. Lots of that era of artists as well, Wally Wood, etc. one day I’ll have time…one day.
RUclips gets the same ad revenue either way, so why not deliberately throttle the numbers content creators see, so they don't have to pay out as much? They can just call it a "glitch" and laugh all the way to the bank.
It was super frustrating-- I'll say, the issue resolved for all of us that were affected after about a month (slightly suspicious that all the channels suddenly were ok, all at the same time), and then we all got hit again around November for about 20 days. I've just learned to factor in that this is a possibility every few months.
@@TheoreticallyMedia This sort of thing is obviously against the law. And other social media companies are getting sued for fudging their numbers. But few content creators are going to stick their toes in that water, or they lose everything. You have to just roll with it, for now.
Thanks for the video and easy explanation about the website. Unfortunatelly it does not work too fine. After plotting a simple story I have found that it says "Generating a new story", and after some seconds, the panels remain blank. I could only make it work 1/10 times I tried to. Lot of potential, but still much to work to improve.
@@TheoreticallyMedia Ok. It could be the explanation of the problem. Yes, now it works really fine. A good tool indeed! I enjoy a lot the availability of different comic book styles. It would be great the possibility of a panel by panel description, and to select the number of panels for page. But the tool is really impressive.
As mentioned in the video, the channel has been hit with the dreaded RUclips "Invalid Traffic Bug"-- if you'd like to support the channel, please consider joining the Patreon! www.patreon.com/TheoreticallyMedia
👋
The problem I see with all AI that attempts to do text is that, depending on the AI, it attempts to make all or most text all upper case 🤮. I haaaaate all upper case text for a variety of reasons. Until some AI fixes this terrible case bias, I'll stick with traditional text creation tools.
The comic book model looks quite worthy of checking out though. 👍
Try to turn off the embed link option on all videos, you can do a mass changing of this. This helped some people.
@@eri7-11 did it yesterday! And over the weekend I’m just going to do the old “turn it on and turn it off” trick that some folks have reported (occasionally) works.
But the whole thing has been super eye opening, and has really driven home the point that, for the health and longevity of the channel, I can’t rely on RUclips as a primary revenue source.
Given the other options of Audience support via Patreon, or just taking on every sponsor that slides into my inbox? Well, that choice is pretty clear to me!
@@LouisGedo hi Louis!! Great to see you!!
How would you go about getting a consistent images for your comic? When I asked it to draw a space craft, the AI altered the appearance of the space craft thought out the comic.
I'd also like to know!
It can't. that's the flaw of AI. That's okay. Someday it'll be more consistent, but by the time it does, you could learn how to draw things yourself. AI steals from artists and creators and most of all and more importantly, also steels your potential. No one respects, because it is entirely built on theft.
Keep in mind everyone you can't copyright any of this due to the laws just passed.... which ISN'T fair to the models (people) it pulls from.
That argument is still in play. And to be honest, that only applies to US copyright law. I believe Japan has already decided to let AI fly,
And you can copyright if the work has been transformed by a human, but there aren’t any definitions of what that actually means. How do you quantify what transformation is? If I write dialogue balloons, is that 10% a change? 50%? Who decides what that means?
Personally, I think the whole fight is doomed to fail, namely because Warhol’s Campbell Soup artwork has already been upheld.
The fairness thing, I totally get. Look, I’m a proponent of UBI, but that’s a whole other discussion…
@@TheoreticallyMediawith that type of ideology you won't mind people using your "work" as a baseline for tweaks and edits. After all if it wasn't for AI most people wouldn't have the opportunity to profit from professional artists. If copyright does side with the AI, everything will be up for grabs
@@TheoreticallyMediaso you wouldn't mind me reuploading your videos with a "speech bubble added"? Or a watermark of my channel - would that be enough for you to consider it transformative?
@@TheoreticallyMedia - Is it? Whatever the US does, other countries will have to follow due to basic international agreements and treaties alone. You might want to re-check. Japan has not decided to just let 'AI fly', they've explicitly stated that the law still allows for copyright claims against AI. They've just made the distinction between generation vs. training, and that training is still ok for research and educational purposes, which is the same everywhere. Also the Warhol Foundation recently lost its case vs. Goldsmith, so no, just because Warhol's Campbell Soup case was upheld, isn't the blanket carte blanche as you think. So no, the 'fight' isn't remotely close to being 'doomed' to fail.
@@NovaaGrind Guys don't mind these talentless people, they always defend AI art because they cant do anything for themselves. It is true, if you follow that ideology so then everyone is going to copy each others to the end. Period.
what i find especially lacking in those ai tools is the consistancy to keep characters or places ... it would be good to save models and keep them to try to generate and narrow scenes from them
Really good point this looks pretty cool but I was wondering about the consistency weather for characters or even props like ships or buildings
@@FluxNomad678there's surely some people freelancing using ai. But idk what other tools they use.
Yes I came here because I had an idea to see if anyone had something like mid journey but focusing on character, building and comics. I would like to create something where I can detail a character's appearance and personality generate those image. Then give that image a name. Let's say his name is Luke now when I talk about Luke the AI will recognize his appearance and put him in the scene Then you will always have Luke. Luke can write a motorcycle swim with dolphins or even eat a hot dog
You couldn' t be more right. I try alternatives: using artists on inputs, ages of art/paiting/comic strips, and so on, but I think I will always only be able to create concepts with consistancy, not a great work with a same line of art.
From what I'm seeing, some people try saving 4 to 8 best result pics and then using all of them in the prompt in future generations. Like maybe the 'average' of these is close. I might try this and also see how it works in MJ 6.
Thanks
Thank you so much! Means a lot!!
Can this currently be run locally?
He has the code up as open source, so I suppose it could be. Let me know if you need a link! (He mentioned that you might be able to train it on some LORAs!)
Shouts out to my man Sevan. Y’all should check his amazing A.I course 🔥. Wholeness as always 🙏❤️
@wholeness. AI is a scam, like bitcoin. Don't ever use this stuff. If you have a great idea, you'll just be handing it over for free. These generators cannot create comic books. They only replicate existing artwork. If you're happy with that, you cannot be an artist. In time, once the bottom drops out of this fad, genuine human artwork will double in value.
You always worry about how this could threaten people's jobs, but with some improvements it could be a useful tool for those of us who have ideas but not the artistic skill to properly visualize it. AI shouldn't be used to do all the work, but as simply another version of the brush and canvas guided by actual people and their ideas and imagination it's perfectly valid
Nailed it. That’s pretty much the mantra of this channel!
So just skip over the artist, which does eliminate the job of an actual living artist?
Sure, skip hiring a human artist and rely on A.I., your final product will end up looking exactly like the product of the thousands of others who had your same brilliant idea and chose the same A.I. engine. Good luck with that when it comes to finding a way to stand out in what will soon become a very VERY crowded marketplace.
Why not reach out to a real human fellow artist, share your idea, exchange thoughts, be friends, create a fantastic original piece of art, be sad together if right away doesn't get the recognition you both think it deserves and celebrate together when you reap the results of your collective work.
Here's my problem: the layout of a graphic novel is just as important as the content of each panel. You are showing us panels with cut off content. Instead, it should be a very hyper-stylized text-to-image service where you describe the story, the panel you need, an aspect ratio, and possibly some dialog for the super-lazy. Then we can take that content and do it right in Indesign. INSTEAD, this thing is trying to pump out four images at a time, clogging up the servers and attempting to do all the work at once instead of asking simple obvious questions like "Is this a two-page splash?" or "Is this part of a nine-panel page?" This, as created, is neither useful for web comics (wrong format) nor print comics (no structure).
I mean, yes and no. The thing with this is that it’s trying to make a comic page, but it has no instruction to create a narrative between the panels. So you end up with panels that look (stylistically) like they should match, but actually don’t.
But, if you generate 4 pages of them, you’ll find that you can cherry pick various panels to create a narrative. It might not be the one you originally intended, but that’s sort of the fun of it.
It’s why I bought up the Marvel Method. I’m sure Stan had to think on his feet for a few of those issues.
(Then again, Kirby and Ditko were so good, he probably didn’t have to do too much extra work!)
And there's always a winery or somewhere dude give it a break okay just as still experimental stuff did you think the wright brothers would have gotten anywhere if they were so chickenshit and wearing girl panties they took a chance in a dare
😮😮Thankyou for sharing MY Sentiments Exactly---Kobayashi, Pres. of GM&M
Very interesting video- and the technology is amazing, but I do have a question- (an honest question about how you feel- this isn't meant as a way to upset anybody. I think AI will affect everything, not just art) -but as you promote the use of the tech for comic creation, what are your feelings about the fact that, at some point very soon, videos like this will also be created by AI? -replacing you and your channel, or at the very least, taking more of your traffic away and demonetizing you further? This is all uncharted territory, so Im not taking a stance here, Im just asking a question about how you feel about that possibility.
As far as personally replacing me? I don’t see it. The channel has largely been built on 3 things in my opinion:
1) obviously the tech itself, and covering the latest developments as they happen. I get beaten by Matt Wolfe on this all the time. I’m pretty sure that man never sleeps.
2) My personality and viewpoint, or…dad jokes, for lack of a better term? So, truth be told, I generally write out every video in full script form. It is very dry and very bland. But as I’m shooting, I’ll start improvising, and going on tangents or coming up with non sequiturs. AI just isn’t good at that.
3) I do a lot of hands on examples on the channel. That is one thing I kind of pride myself on and is something that I think separates me from a lot of the other channels in the space. I’m not calling anyone out or anything, but most channels will say: “hey; here’s a thing, isn’t it neat?” Whereas I tend to do things like: “Hey, here’s a thing, isn’t it neat? So I made a recreation of the famous scene from Titanic, only using the tool, I’ve turned Jack and Rose into rabbits.”
Even just in this video: most would probably say: Here’s Comic Factory, this is what it does…” but I’ll take the time to create 2 pages with it, and in doing so also talk about how awesome Mobius is.
So, no. I’m not worried about AI replacing me. There are too many elements that comprise “me” that any AI system just can’t factor. And that isn’t coming from a place of ego, I mean that about all of us.
Sure, there will be a sea of NPC AIs delivering the news soon enough, but I tend to think that people like connecting with people, so when someone runs across a video I made; they’ll come back.
…or they’ll roll their eyes and say: “ugh, this guy again?”
@@TheoreticallyMedia - I appreciate the thoughtful reply- the tools AI employs are mind blowing for being so new to the game, relatively speaking- and I hope that AI, indeed, helps people, but I fear that AI will start living "for" us instead of helping us live better for ourselves... I dont know if that makes sense, but its worth having meaningful discussions about, thanks:)
Big comic book fan breathlessly promoting AI that wants to put comic book artists out of business, while complaining that a computer algorithm is hurting his revenue. Am I missing something?
Haha, that's a solid take. Look, I don't think AI is going to put comic book artists out of business...in the US, the direct market and comic book publishers are doing that job on their own. If anything, I think this just allows artists to accomplish more, and not grind themselves into the ground to produce one page-- or gives someone who has always dreamed of doing a comic, a shot at making one without having to break the bank hiring an artist, inker, colorist, letterer, and then foot the bill for printing.
They're tools. Artists can use them to make amazing stuff, non-artists can play with them to make...honestly, stuff that will probably be sub-par and no one will want to read anyhow.
As a colorist, I don't agree with this. The industry is only interested in making money, especially big companies. This is a huge blow to honest artists. soon there will be more "Ai Art" than real ones. and the algorithm will make it very difficult for people to find the REAL ARTIST and their art. Even though independent artists have great works to share...
@@susysannIf an artist can not be more creative than a computer, then perhaps art is not the right career for them to begin with.
Yeah, 1 has absolutely nothing to do with the other🙄
@@RarebitFiends
No. Because the final decision rests with audiences. If you think audiences are all highly discerning, 'cost is of no importance' types, then please let me know where these people are.
Not bad. I would love to see an AI tool that is able to accurately copy our original characters, and generate images with them by displaying them in a consistent way in each scene.
I think we’ll see something like that in the next three months. Fingers crossed here as well!
Some application of control net should be able to make that happen
Have a look at what Corridor Crew have been up to with "Rock, Paper, Scissors"
You can do something similar one image at a time with 'reference' controlnet in automatic1111. Not always 100% but pretty close. AS for generating panels... yeah more people need to focus on that concept along with this.
It's already sort of possible by "training" your a.i based on your own artwork. Granted you'll probably need to get familiar with some coding or at least command lines in something like Python.
Interesting, but where does this art come from? Is this scratch made by the software or it based on something preexisting? What are the copyright implications? How much futzing before it’s not just ai?
I think it scours the internet and basically Frankensteins the final image.
After ai comic generation can we get copyright and publish in webcomic, bilibili.. Etc?
So, to be honest: in my super non legal opinion, sure.
It’s part of a larger rant I’m brewing, there’s obviously a big fight brewing about copyright etc, but ultimately it doesn’t matter to you and I. Legal teams are going to duke it out, and I’m sure it’ll make its way to the highest courts. But-
A) This is just US law. Japan and other countries have already decided that Ai art is completely fine.
B) this is a legal issue. You’re a creator. Don’t let paperwork stand in the way of you making stuff. Think of early hip hop/rap. I mean, the EARLY stuff: those artists didn’t care that they were looping copyrighted records. They were just creating. Do that.
Or…maybe become an old school rapper? I don’t know, how’s your flow?
You just cannot complain if someone else generates the same thing.
@@mf-- that is 100% true. Although I think the likelihood is fairly low. Unless you were say, doing a Batman comic, which would be fanfiction anyhow.
I suppose if you had a bland idea like “Harry Potter, but they’re Cats” someone could “copy” you, but I also tend to think that idea wouldn’t gain much traction.
Quality storytelling will hold an audience, nothing more.
‘Armorica’ is the Roman name for the part of France we now generally refer to as Brittany. I think the ‘Armorican’ style may be a reference to the European comic ‘Asterix the Gaul’ which is situated in that particular part of France. I’m not sure though because this style in Comic Factory doesn’t match the style of the original comic very well. Anyway… great, informatieve work again Tim. Thank you!
Ahhhhh, there it is!! I couldn’t for the life of me connect Armorica to comics! Asterix is it! I kept thinking Nemo in Slumberland or Moore’s League of X as a style!
I think with some Asterix-y prompts, that style might show through a bit more. I’ll say the models can be a bit finicky and picky about what it produces.
Thank you!!!
I confirm! The style is a bit difficult to catch and we don't really see the difference I think, because I didn't want to be too explicit in the keywords. It would work better with a LoRA but currently the comic editor doesn't use one (the tool is more like an homage and its goal is to have fun, not to plagiarize living authors or copyright holders, so I'm trying to stay vague in the targeting of styles).
@@julianbilcke6922 thank you Julian! Both for the comment and Comic Factory!
@@TheWatchernator So, I think if you give it more of an "Asterix prompt" -- basically, something more in line w/ that style, it'll produce better results.
Give it a shot and let me know!
Reminds me of some of the "albums" from Europe Comics
Is it doing consistent characters?
It CAN, if provided with LORAs, which the base model does not have. That said; it is open source, so perhaps someone will pick up the mantle and run with it?
the CB generator must be seeing a lot of demand. I tried multiple times and never got a comic. It finally worked.
It is kind of a “garage project” that is run by one person. I can imagine there’s been some traffic strain recently.
Maybe a little duct tape and a kick will get it going again!
The comic generator is a cool idea but the consistency is a real problem when you're looking at 4 panels on one page and the characters look different on each panel. One person may be wearing a blue suit on one panel, and a white suit on the next. For me the best solution is to pair stable diffusion with loras to individually generate each panel, and using inpainting to redo any inconsistencies.
Remember how shitty almost all AI art looked 12 months ago? Just wait.
@@robertdouble559 yeah. just 2 more weeks.
You have no idea how difficult it is to create a good comic page.
these examples are just random pictured put in frames. not a comic page.
@@robertdouble559they're literally describing how it's going to improve lol.
I think this is cool for the hobbyist who wants to make a comic but isn’t a great artist. The human element is still needed for it to work.
Agreed. I think it’s a good stepping stone if you think you might be getting the comic bug. And I think it’s a fun thing to play with if you actually know what you’re doing. As it stands: I think you can produce a cool mini comic with it, but nothing like a 800 page graphic novel.
…but I guess that’s for right now. We’ll see in a few months where we are!
AI. NEEDS TO BE GONE NOT ROLL OVER AND MAKE SPACE FOR THE DEVIL...
But how to keep characters same and story consistent?
To be honest, right now, you don’t. There are some tricks you can pull- like archetype characters (the spaceman in this example), or prompting very basic character designs, say: Red Haired women in a black dress, and then use a face swapping tool afterwards as a finishing touch.
I think all of that will be worked out in the next few months though.
@@TheoreticallyMedia sir I made my own manga by Bing ai but the problem is the girl'face is same but hair is not same so how can I change hair I mean is there any only website there where I can change some details or edit and I can make a character consistent by doing some changes
Ideogram is great but the biggest drawback is as for now you can't use images commercially as say in the terms of Service :(
I mean, I’m always the guy who ignores ToS. I tend to think it’s a CYA on their end, more than prevention on your end.
There isn’t really a quantitative method to tell if an image is AI generated. You can eyeball it, sure- but without seeing the source file, you’ll never really know. And then ultimately, there’s no way to tell which model you used as well.
Will we see it in a Coke or Nike ad? Nope. But, for your own personal projects? I say whatever, go for it. If you ever create an empire that’s as big as Coke or Nike? Well, that’ll be a worthy problem to have!
@@TheoreticallyMedia Sure, AI does have its benefits, but it's clearly trained on other people's art, their style, and frankly, copyrighted work. So by all means, "go for it" but also mark somewhere that it's AI art and not your organic originally drawn/painted art. There has to be some ethical guidelines for this unprecidented flood of art AI platforms. The Renessaince masters are turning in their graves! Now, if people want to pass AI art as "their own" that's between them and their conscience, and how you stated "you'll never really know" if an image is AI generated. To those that sneak this into manifested reality, the good thing (from dedicated artist's perpectives') is that, sure, we will never really know, but also the AI art creators passing these "works" as their "art" will know one thing for sure- and that's that they can't draw and need someone else to do the work for them so they can just slap their name on it. Except, the "someone" is a program, most likely written by someone of the same caliber of short comings in one area. Draw your own stuff, people, or at least label that it's AI, like a decent human collaborating creator would do in the indicia or credits of any comic book.
@@TheoreticallyMedia EDUT: btw, it's nothing personal against you, because I truly enjoy your chennel, and have learned a lot. Mostly things others are doing, but you have a friendlym, straightforward way of explaining that gets to the point and is easy to follow.
That aside, it's just a matter of time before there'll be more AI/fake art than real art. Most proper artists using Ai are blind to what's coming, and are just contributing to their own demise, just as much as the coders of these art AIs are contributing to AI out-competing artists. Prompt writing is an art, some might say. But prompt writers are not artists, and your whole "you'll never really know, and F it go for it" approach to this, is part of the same mentality every person saying some AI-generated image is their own "artwork," Also, "for your own personal projects" is a moot point if people are not disclosing they used AI; ipso facto, not made by you.
@@TheoreticallyMedia It's hardly CYA, it's that legally the generated work cannot be copyrighted. The AI creating it can never hold a copyright to transfer to whoever prompts it.
Anyone trying to use it commercially is open to having their work stolen. Just like all these AIs are stealing from actual artists.
Oh yeah, this style is 'Moebius'-like. I wonder how that happened?
But yeah, people talk about AI as if it's not just outright theft.
If you make a comic with these A.I. tools can you sell the comic?
Currently, I don’t think anything is stopping you. Getting people to BUY that comic? That’s the tough part!
Does it let you create custom characters?
Not yet-- but Julian has indicated that someone can take the code and train a LORA, and then: Yes-- you would have that!
Can I upload figures I invent myself? And use them as a basic figure?
There’s actually a better tool for that! Going to be covering that one in about a week! Stay tuned!
As a writer and artists, these new comic generation sites are pretty interesting, but they are really in their infancy right now. Anyone who already has a story, and a vision is going to find them incredibly limiting to use. Inconsistent characters/models, fixed panel layouts, and the bazaar interpretations of prompts to name few issues. The tech is moving so quickly though. Hopefully these issues will be addressed soon. With the fixed styles, I do wonder how quickly the audience will get burned out from the lack difference from comic to comic.
Audiences quickly become jaded to new technology due to their overuse. Like CGI use in films.
I suspect the temptation to produce cheap material quickly will cause enough people to reject ai produced artwork, as being cheap and artificial. A compromise between ai and artists will have to be reached.
😊@@echopeakbicycling85
Can I add my own face photo to generated images? Thank you
You can! There’s a few different ways, this is the latest and the one I like, and although this one is about video, it works for still images as well: Your Face In Any Movie! Easy/Free Video Faceswapping Is Here!
ruclips.net/video/Y8CMwl2Fv0g/видео.html
The comic book generator is so interesting. It's a shame you can't lock Midjourney to a character / style set so you can generate imagery with specific characters and outfits etc. I'd love to make a comic out of my feature screenplay just to sell it in.
MJ has indicated that they plan to included "Storytelling" features at some point-- so I think they're working on just that! Fingers crossed!
Oh crap that might be me out a job hahahaha @@TheoreticallyMedia
MJ has a new inpainting feature too, so maybe with enough prompting we can get more of the same character (I know I did panels in the past, but it did lose the plot fast without an easy way to fix small things).
@@GRNKRBY I think that’s a pretty ideal use of the Inpainting feature, and one that I do want to dig into at some point!
I almost like this platform because it’s uncontrolled. It almost feels like working with a stubborn artist that you have to work around!
Haha@@TheoreticallyMedia You said fingers!
As an artist, I feel sad with AI art...
It might be a tool to help you with your art, but those of us who put specific style and much time practicing...
I remember working hard to learn facial features and how my first ever portrait felt like an amazing milestone...
Now anyone can just generate an image...
Even if it doesn't replace real artists, evil company leaders are going to prefer "cheap and obtainable" if it cuts cost corners.
It's already hard with regular competition to make a living as an artist.
Now if you can't compete, regardless of putting your heart and soul into your work, it'll be even harder.
It can be discouraging, but I don't see how this would replace you in any fashion honestly.
“Evil company leaders” are going to bone you as an artist, AI or not. Have you seen mid-low tier artist page rates for Marvel or DC? They’re below minimum wage.
Don’t focus on getting a job drawing Batman. Spend your efforts to create something you own. Something that your audience recognizes as you. Build and cultivate that audience. That’s how you make it as an artist.
(That does for every medium of art)
I’m not saying this is the original poster of this comment, but: having now done a bunch of these videos and interacted with a lot of people via comments, there is an underlying attitude with some naysayers I’ve come to recognize.
There’s this defeatist “I’m not going to bother trying because of this technology” not “I’m quitting my profession because of this technology” it’s “I’m not going to try”
Making a living in art is perhaps the hardest path you can choose. Some view AI as a barrier to entry, and it’s an easy scapegoat right now. But what they don’t realize is that there are 1000 other obstacles behind that.
Again, not speaking about or to the original comment here…
I'm not going to waste time crying about AI. It's pointless. Every artist is aware of the 1000 obstacles related to making a living as an artist. I don't think there is another profession that typically has "starving" as a common descriptor. Thousands of talented people have been dissuaded from choosing art as a career over the centuries because of that reality. I know many of them. AI is another (exponentially higher) barrier to entry that logical, thoughtful, forward-thinking young people will contemplate when choosing what career will allow them to pay off their college debt and feed their families in the future. And by the way, they're coming for your job as well.
Don't get discouraged, AI users will never experience what you just mentioned, the marvelous feel of nailing a drawing, a creation, something you can proudly say, I drew this, with pencil and paper. You just say something really important with all this AI dumbness, it is just an image generator, it has always existed but now it is much more polished.
Art is a journey, being a pro artist is hard, and now even harder, but with your creativity and passion, plus hard work you can pull it off.
Best of luck and keep going!
The idea is interesting...
if there was a way to train the generator .
I think I understand how to use it.
Technically, you could use the code to train a LORA, but that wasn’t the intended use case. But- it could be done.
Is there a link to Comic-Factory?
Should be in the description!
can you try bring an old comic book to AI motion or video? that would be cool seeing the old comic books into video
You currently can! The results might be a bit wonky, though. There are a few videos on the channel where I go over those tools. Check out a recent one on a free video generator called Haiper! It was just a few videos back!
send me the link?@@TheoreticallyMedia
At 7:03…what printer button? Your video doesn’t show the thing you are talking about
Yeah. My screen recording software didn’t pick that up. It should be at the bottom right.
Can we ceap the characters we use?
Is “ceap” a typo for keep? If so, sure!
What I see is a bit repetitive to say the least. Almost every shot is straight on. They'll need to create some camera-angle commands to diversify the cinematography. But for writers and film makers, this is very useful.
Agreed. I know the dev is still playing with the project, but it is a “garage project” of one person, haha, so I think of this as more “proof of concept”
But, agreed with your point! It is fun to play around with though!
I don't know how to use comic factory it shows CPU updating like that please do a short video of how to open it
You should just be able to get there? It might be updating right now. It is a one-person operation, so it does tend to go down when he is working on updates!
I have a BFA bachelor fine arts and I have no idea what the style that last was. If you have any idea which AI I can take a screenshot or a sample of a cartoon style I want and then turn it into video my video into that style that would be amazing thank you for your work.
What about the face of protagonist?
Can you use your own character design to make comics with? That is consistent with the same characters and the same costumes. And every frame screen or was it just random like AI usually is?
that would be the greatest if you could get ai to duplicate your own style
@@kennycooper294 and people
Excelent review. The APP still have bugs and it is not possible to repeat the character.
I do think that’s coming to all these various platforms in a few months. Someone is going to crack that open and then they’ll all follow!
rember to redraw it if you do anything since any picture that is ai generated can not be copy righted in the usa until it has been altered by creator.
Well, I’d argue that if you resequence the raw output, and add captions/dialogue balloons, that would constitute alteration.
But, I’m not a lawyer and the courts are still fighting that one out.
I’ll say, the transformative argument is going to be an interesting one, as Warhol’s Campbell Soup can art has already been upheld. So, we’ll see
@@TheoreticallyMedia you should read the new law on it issued by people who deal with copy write infringement its very simple and concise they issued it as a sentannce instead of a paragraph of law jargen.
It’d be great if comic factory the hands/fingers can drawn out the hands. They are either nothing,?- scribble or warped out nubs. I wonder why that is?
Wow. Pretty cool. Marvel Method indeed. Was thinking the same thing.
I love using the marvel method, it’s such a unique way of working that can lead to some pretty unexpected results.
@TheoreticallyMedia it can be a really effective positive feedback loop
Can I run the comic factory locally?
Yup, it is open source. And Julian has indicated you can even train Loras with it.
That said, he also admits that his code is a “garage project” and lacks any proper documentation, so you’ll have to do some legwork on your own.
I respect a messy workspace!
Comic Factory not working. it says internal server error
might have caught hugging face at a bad time, try now: huggingface.co/spaces/jbilcke-hf/ai-comic-factory
I've been wanting to do this with the intent of selling a fully realized comic book online. Is that possible?
It’s possible. There might be some argument over copyright etc if you disclose that it is AI generated, but the truth is: that fight is ongoing and besides that, many counties have already OK’d data training…sooo.
That said: what you’ll quickly find is that making it (which is still going to be challenging) isn’t the hard part. Getting people to read it, and/or buy it? That’s the tough part.
@@TheoreticallyMedia Good to know! Thanks. I think if the subject matter (story, characters, etc) grabs people's attention enough it might work, depends what it's about I guess, that's the fun part to work with. Let's see what happens!
I can't get it to keep the same car on the screen in each image. So you get a blue car which is then a red different car.
Holy shit, I'm speechless. Been working for years on my first long form comic and this is truly disheartening.
Nah, man. You still have your comic. Nothing takes away from that. Nothing. Take it from a guy who has published 2 comics!
@@TheoreticallyMedia What the?!
@@giri.goyo_yt which part?
The difference is you've grown as an artist and a person while making that comic... AI teaches you nothing, it's as bad for your soul as Twitter is.
@@TheoreticallyMedia That you published 2 comics! Where are these comics?
I went to the comic factory site, it says you can use it once before registering....When I tried to sign up, it said "hugging face refuses to connect"....Ummmm okay.
Can I import my photos into any of these?
How do you save these prompt results...when I try to click save or print at the bottom, nothing happens. Using macbook pro.
Same. Try again in a bit, I know Julian was playing around with the UI. The button might have gotten lost in an update. Just remember: this is just a dude working in his garage, not a fully fledged company product. Haha. It is literally a garage band!
@@TheoreticallyMedia ok thanks for the reply!
Boy I can't wait until AI is making all of our art and music and poetry and writing all our books. All this while we lug rocks from one pile to another for the corporate overlords that own the AI.
how did you keep your characters consistent looking, i keep getting different looking characters....
Let us know when Comic Factory allows you two remix separate panels, then I'll get it.
Well, it’s free…so you can mess with it now. But yeah, that would be a great function!
What does the prompt look like for a page like that?
Okay, problem: None of the character models are consistent from panel to panel. Even in the late 90s and early 2000s with Spawn, there were fans that would write in why his shoulder pad had a different number of spikes from panel to panel. Here, the buggy is different from panel to panel as is 'Space Guy's' character model.
True-- I think if you were to expand this idea, you'd need to bring it in to another platform for inpainting. Which, actually just proves the point that none of this is "easy"
can u please tell how am i going to save it and make more pages from it
Problems: Story is ready but server is busy. Panels not displaying.
The ai program doesn't allow uploading of my own character modelsheets.
So it doesn't place my characters in the scene.
But what this can do is set up the layout, and then I can draw my characters within the layout. Saving time.
Yeah. It’s kind of what I like about it. It isn’t a “prompt and you have a magic page” it takes human work to make it happen. It’s an important aspect of these tools!
I will never believe in using AI to make manga or comics... That's something that should be done by a human so you actually see their artstyle and the likes
another problem is trying to keep the same characters in your comic with out the face changing to something diffrent how can you have a hero if you cant keep the same person
How to access comic factory or Hugging face?
are there any apps in the market that can take my images and convert them to comic book panels?
People like me people use this to make comics àre passionate ,them making something that they have wanted to make for a long time, even doing it in second means they will still cherish it
I im doing a comic/manga by hand and I'm making a reversed one with this ai to compare it to my hand made comic for ideas and advice
That’s excellent to hear, and exactly what this kind of technology excels at!
If you also do your own art, you might want to check out Prome as well: ruclips.net/video/P1-VpUFHoEo/видео.htmlsi=snUUe4WezzhV5Zd4
Does anyone know, If it is free for commercial use? Thanks and greetings from Hamburg Germany.
Comic Factory tries to load your input, but then gives up and you're left with blank panels. Fail.
I mean. It’s a one person project, not a VC backed product with thousands of employees. Think of this as more of a research project that will eventually lead to the “thing”
What is that natural finish guitar?
Ha, so that is a 1980s Peavey Horizon 2. Made in the US, swamp ash (heavy!) I love it-- 24 fret super strat, with a weird system on the tone control: Kick the tone up to 10 and it is a single coil, bring it down to 7 and it's a humbucker. Oh, and it has a out of phase switch too.
Just a weird and awesome guitar I got lucky with on Facebook Marketplace!
Cool, maybe we can hear you play ot some time too.
@@zakblackhawk204 funny enough, you hear it all the time! All the background music in my video is written/recorded/performed by me!
Wow refreshingly good content, liked and subscribed
Where can I find the document that you share at 3:17 with the “workflow and tech breakdown”?
Yup, that was from a Reddit Thread that Julian posted. Lots of good technical info there if you've got the good brain. I have the middle teir one. Haha, I'll upgrade at some point...maybe more RAM? Here's the link: www.reddit.com/r/StableDiffusion/comments/163ikmd/wip_comic_factory_a_web_app_to_generate_comic/
@@TheoreticallyMedia thank you very much! It seems that this is all done by prompting, not any specialized models, right?
where do you go to edit the dialog in ai comic factory?
Can i post the comics as storyteling on youtube? Is there any potential IP infringing especially hugging face?
The hard part seems to be getting the exact same character/ship replicated in different poses and situations.
Yeah, my current trick with all of these tools is to use very basic character designs: “a woman with red hair in a black dress” or that sort of thing.
That will usually get you a relatively consistent character that you can modify with face swap tools, or Photoshop.
That said; I think we’re only a few months away from seeing consistent characters in all these models.
Also the most important part.
@@ronamick3352 Indeed. I think we're only a few months away from that being a standard feature with all of these image generators though.
@@TheoreticallyMedia If only there were an option to select characters, settings, objects, and design details on materials as defaults for a consistent narrative, not just the style. And a way to save those settings to carry over into later projects.
Terrific find, Tim! I like how you used Comic Factory, despite its limitations, to make something creative and coherent. Have you tried Comic Life 3 for laying out those images? I used it for a graphic novel. It does all things comic except making the images. So you have panels, text bubbles, sound effects, filters -- it's really handy if you have images but would like prebuilt pages and appropriate tools. You can even lay out the panels in whatever way you want. It's something like $10. It might come in handy once MJ actually shows us the storytelling tools!
And so sorry to hear RUclips cut your income. As if they don't take most of the money anyway. I hope you get many Patreon supporters, you definitely deserve it!
I haven't checked it out, but I will now!
No he doesn't. He's promoting plagiarism software and stealing from artists... He deserves nothing
@@animtonni - buggy whips
Thanks for all the great info! I love your channel!
Thank you so much! Me too!! Haha
I ❤your videos! Keep up the good work 👏
Can someone PLEASE tell me what the art style is called at 1:00 ?
Why does the text bubbles it adds are not in english?
I clicked on this video and then was listening to it in the background while working. Heard your voice and thought, wait. MetalJesus rocks has a second channel. I listened for another minute or so and thought, he sure does. I brought the video up to double check, but no, you are not MetalJesus. Your voice is sound exactly the same. Check it out if you haven't heard that already.
The link stopped working, and my avatar is from fiver before you assume, thanks in advance for any updates or new tools like it,
Regarding Comic Factory: How would You make AI maintain design of character between prompts?
Technically you can’t. That’s an issue all of the AI generators work now. The workaround is to use archetypes. Like, in this: an Astronaut. Because the bot will mostly use the same idea of an Astronaut over and over. There are always nagging details that it’ll miss, but you can fix them via photoshop or image editing.
Stuff like, James Bond, Woman in a black dress with red hair, etc. those are the (bland) characters that tend to result in very similar (though not exact) characters.
Awesome. I see you are a fluent klingon speaker (9:16) XD
Oh, you'll love this little bit of Internet Lore: (Not mine, just an old Reddit post)
"At a scifi convention hotel party with a Klingon theme I was talking with a gentleman who spoke Klingon. He said it came in handy when he was on a business trip to Japan and got lost when he got off of at the wrong subway stop.
He tried speaking English and broken Japanese with other commuters but they were ignoring him.
Then he saw a Japanese man with a shirt that had Klingon writing on it, thinking "Oh what the hell" he approached the Japanese man and uttered a Klingon phrase. To the gentleman's delight the Japanese man muttered a Klingon phrase back and they both had a lively conversation in Klingon since they both shared the language. The gentleman was given the right directions and found his way again."
Is Ideogram designs usable for commercial use?
Per their latest: that’s up to you. If you’re a design firm hired by Nike, and their policy is no, then no.
If you’re a local pizza shop who is designing a new logo? Sure.
I see a lot of comments about the lack of consistency in AI-generated image sequences. How do you achieve the desired consistency to make your own comics? Very difficult - learn how to draw, master perspective and color theory. It will take years and years, with no guarantee that one day you will be able to even come close to the heels of masters like Hergé, Moebius, Bilal, Hugo Pratt, Will Eisner, Robert Crumb, and many others. Better start now!
I would love to use this but there is no copyright. Until that's resolved i don't see pros using this.
The whole copyright thing is in flux. Currently you can’t output AI art and just copyright it. It has to be manipulated by a human being and then it can be.
But there is no definition of what manipulation is, or to what degree. 10% manipulation? 50%? How do you quantify that with art?
In the case of the “AI Marvel Method” that I detail in the video, I’d argue that is a significant amount of manipulation.
@@TheoreticallyMedia do we have any clarity on using AI to help write stories? How much has to come from human?
what ever happened to this new AI Comic generator it was incredible when it opened up now the output is mediocre, why would they do that? any ideas
Pirate Software recently worked with the community to 'fix' one of the traffic bugs on youtube,
This looks amazing, as a author of a graphic novel myself this could be an interesting tool to use that could be used to brainstorming new ideas.
I think this is a perfect use case!
I'm using it to pretty much pre-viz my novel. So characters looking the same thruout doesn't matter as much as perspectives, segments of the monster's spidery legs -- all sorts of great mood-setting stuff.
@@TheoreticallyMedia
I'm actually more of a fan of Danka's earlier work.
That Comic Factory looks awesome! I don't understand why every old school art image creator tool from Paint to Photoshop had combined a way to add text over your image, but I've yet to see an AI art generator have that feature. Seems like it would be a no brainer. Ideogram is still hit or miss with the text. You can type in lyrics to be rapped or sung by music generators, but I can't get a little 14-point Times New Roman into my comic's thought bubble?
Still pretty amazing/cool stuff! Hope YT shares the wealth better.
You would figure that someone would have suss’d out a ChatGPT plugin to pop over an image. But hey, at least this keeps the lettering folks working!
And man, that is such a lost art. I still can’t believe there was an era where they did that BY HAND!
Dan Clowes probably still does his by hand @@TheoreticallyMedia
@@TheoreticallyMediaI'm old enough that I trained in constructing lettering by hand. (Do they still do that these days?)
I only remember enough of that experience to admire the talent and skill of professionally produced work.
Did these artists concent to their art being trained in these models? These are direct tips of specific artists...so.
They absolutely did not.
Ideogram is very distorted and could not get the comic book creator to work.
Comic Factory might be getting hammered right now. I’d say to give it a shot in a few days, or in off hours when the load is lessened a bit.
I make comics because the process is fun and I love the medium, all the craft that goes into it.
This shit right here, is not fun, and there is no love involved at all. I feel bad for all the folks who wants to make comics this way.
Right on. I mean, this is not stopping you from making comics at all. There are some who will use it to generate comics, sure-- but the medium will go on, and I think most pro artists will end up using a hybrid of their art and some AI generated stuff.
I've spoken to a few pros at the Big 2 (no names, for obvious reasons) and they aren't worried about AI at all. In fact, they're learning it.
Different stokes...
Not positive, but "Amorican" might be a play on Alan Moore's work through American publishers. The style shown reminds me of the detailed backgrounds in things like LXG and Top Ten.
That’s 100% what I was thinking. Plus, maybe a little Little Nemo in Slumberland, but I’m not sure what that particular style of art is classified as. Great minds, though!
My fear of AI is not what it can do now but in a few years. Once they figure out a way of connecting narrative, anatomy, light, mood etc... every single human archetype, premise or story will be made available on demand.
True. But in a few years all of that can be applied far beyond art. Honestly, as the guy that makes these videos, I keep saying: we need to have serious discussions about UBI. Or some sort of society safety net.
@@TheoreticallyMedia A UBI isn't going to give anyone a sense of meaning and purpose in their redundant lives.
@@Pneumanon Correct. That's not what its supposed to do. Meaning and Purpose is up to the individual.
@@TheoreticallyMedia The individual doesn't exist in isolation. After family, career is the most common place that people find a sense of meaning and purpose in their lives. By being productive in their work, through career progression and through the status that income provides, people derive a huge amount of their sense of intrinsic value. Without avenues to experience a sense of meaning in their lives- in ways that are meaningful to them- people become psychologically sick.
If AI replaces human labor on a broad enough scale to justify discussion about a UBI, then vast numbers of people will be deprived of one of the key methods by which people experience meaning. We will be living in a society much more sick than it already is.
The idea of a UBI is often given as a hand-wave solution to the ramifications of AI, but that's usually all it is- an idea. Not a particularly well thought-out one as far as I have seen.
I think there needs to be a “100% Made by Humans” label (like “Made in America” and so on) that can be applied to art that is created without use of AI. That way those of us who want to support human artists can do so. I’m very suspicious of AI and don’t particularly like it, but we’re not going to get this genie back in the bottle.
I loved messing around with Comic Factory. I'm not sure why but after an hour is crapped out but it has some mad potential.
Yeah, it’s a one man operation, so it kind of has a bit of a duct tape feel. But, when it works, it really works!
wow, this going universal, like big bang theory, which happened about 15 bill years ago, now it's here, im ina state of confusion now, mayb going into the lv sphere will clear things up>
unfortanetly the tool is not useable for alot of stuff as it seems you cant get any kind of consistancy with the Charicter it might work for SCI FI where they can keep helmets on but thats about it
I've got a solution for that over in this video: ruclips.net/video/v_FXC0iq1Sk/видео.html
Working up a full comic tutorial using that soon...
That looks nice and all. And the sound of free is enticing until it's not free lol
How is Franco-Belgian not Hergé? Moebius was inspired by Hergé by the way...
I need to study more Herge, to be honest. Always liked him, but never fully dove it. Lots of that era of artists as well, Wally Wood, etc.
one day I’ll have time…one day.
Wish me luck boys hopefully I can officially release my own " maze comics "
This is kind of horrifying for those who painstakingly craft comics.
don't worry there are a lot of manual labor jobs open
RUclips gets the same ad revenue either way, so why not deliberately throttle the numbers content creators see, so they don't have to pay out as much? They can just call it a "glitch" and laugh all the way to the bank.
It was super frustrating-- I'll say, the issue resolved for all of us that were affected after about a month (slightly suspicious that all the channels suddenly were ok, all at the same time), and then we all got hit again around November for about 20 days.
I've just learned to factor in that this is a possibility every few months.
@@TheoreticallyMedia This sort of thing is obviously against the law. And other social media companies are getting sued for fudging their numbers. But few content creators are going to stick their toes in that water, or they lose everything. You have to just roll with it, for now.
Thanks for the video and easy explanation about the website. Unfortunatelly it does not work too fine. After plotting a simple story I have found that it says "Generating a new story", and after some seconds, the panels remain blank. I could only make it work 1/10 times I tried to. Lot of potential, but still much to work to improve.
You might want to try now. I know the dev was updating it-- might be working better at this point!
@@TheoreticallyMedia Ok. It could be the explanation of the problem. Yes, now it works really fine. A good tool indeed! I enjoy a lot the availability of different comic book styles. It would be great the possibility of a panel by panel description, and to select the number of panels for page. But the tool is really impressive.