Apply a CONSISTENT Character STYLE - Midjourney Character Design

[@ArslanAkhmedov]

This series is like the most anticipated TV show on Netflix, totally awesome 10 / 10

[@TokenizedAI]

Hahaha, I don't know about that but I'm glad you guys are enjoying! :)

[@thebluriam]

Truth! I wake up in the morning hoping new episodes in this series are available.

[@variableprime]

Tots

[@cypherzeros]

Totally! I haven't binged this hard in years...LOL

[@paulkonye8380]

Incredible series, and I love your channel!!! I am a storyboard artist, and maintaining consistency of character over many boards is essential. The nature of Midjourney works against this as it is best working up spectacular singular imagery. But your tips and tricks with weights, seeds, and emphasis shows there is light at the end of a very long tunnel.

[@TokenizedAI]

Awesome, thank you!

[@DunceInAwhile]

Easily the best Midjourney prompting tutorial/tips video series I have seen! I have been looking for an in-depth guide to maintaining character consistency. Thank you, sir, for your work!

[@TokenizedAI]

Really appreciate the kind feedback! 🤗

[@WAAXUS]

@@TokenizedAI Hi Christian, what types of URL links of photos you mean? I pasted URL links from upscaled photos while doing tutorial with you, and i have error. Greetings

[@springbreakflorida]

Excellent example.! Prompting is a skill that is very valuable for creators. 😁

[@TokenizedAI]

Yuop, totally agree. So much stuff we can potentially do with it 🙂

[@areacode3816]

Same as another user I am using Midjourney to create a graphic novel based on a dystopian book I wrote last year. Two days ago I figured out how to create a scene using 3 complex reoccurring characters in a complex specific scene.. that looked pretty good. To my knowledge it is one of the first of it's kind. So I'm pretty proud of that! Crazy I used a lot of the tricks shared here, just each new thing seems to push midjourney to it's limits more. 2 characters not so bad...3 wow so complex and messy it's not practical to use unless a scene is simple and vital.

[@TokenizedAI]

Yeah, 3 characters is definitely extremely difficult. I agree that it's not practical in most cases, at least not the way it works right now. But it's fascinating to see how far we can push the limits 😁

[@areacode3816]

@@TokenizedAI just reloading that final image to see if midjourney could add some magic literally overwhelms the AI making it give low res or unrelated images back 🤣. I hope they find a solution for this.

[@andycification]

@areacode3816 I am working on a graphic novel to help promote my feature film. I have many scenes with 4 people in the scene. I would be so grateful if you could share what you have learned regarding putting 3 characters in a scene... Thanks!

[@r.davidsen]

Brilliant stuff! It's amazing what AI can do, but even more amazing that you've taken the time to understand how it works in detail. I haven't been this creative in many years. Thank you so much!

[@TokenizedAI]

Much appreciated! 😊 My objective is to show everyone that a lot more is possible than they might think. What they do with that knowledge is an entire different story though 🤣

[@r.davidsen]

@@TokenizedAI In that case, your objective is more than fullfilled, as you're doing a wonderful job! As for how people utilize their knowledge... Well, there are weirder people out there than all these strange, blurry images that the AIs churn out. 😆

[@tbgr]

OMG Multi prompts !! Thanks for your videos, I've been through the Carla Caruso series and it's amazing ! I can't wait to try to apply those methods. I've already shared your videos to friends because you're doing a really great job and higly detail what you do. Keep it going, I can't wait to see the next videos :)

[@TokenizedAI]

Have fun! 🙂

[@pedrogda6392]

Tus videos son realmente educativos y prácticos. Gracias por enseñarnos el camino.

[@eddiedixon1356]

Finally, this is what I have been looking for, you rock! Thank you for your time, ☺

[@TokenizedAI]

Happy to help!

[@robo3383]

Wow, what a great tutorial. I am blown away by what you are demonstrating is possible. Thank you for your work. I can't wait for #4 in the series.

[@TokenizedAI]

Some of this stuff is super challenging to figure out and takes me many hours of pulling my hair out. But once I find a solution, even if it's not perfect, it's super rewarding 😊

[@robo3383]

@@TokenizedAI Not trying to blow smoke but halfway through this video I wondered how many freaking hours it took you to produce it. You take some of the hard work out of the process for us but I don't doubt you work your butt off. Thanks again.

[@TokenizedAI]

@@robo3383 In this particular case it took me about a day to figure things out. But once you do, everything after that is considerably easier to reproduce.

[@GrahamThomson502]

This information in this video and the past videos are a game changer for using MidJourney. Awesome work!! :)

[@TokenizedAI]

Thank you so much! :)

[@ManoloMacchetta]

man, it s late in the night, i should sleep, but you are kiiling it with this tutorialz! brilliant!

[@TokenizedAI]

Enjoy! Hope you got a good night sleep nonetheless :)

[@kubita-thecall8795]

Thanks Leadership for such tutorials. Watching n learning from Eswatini 🇸🇿. SWAZILAND

[@TokenizedAI]

Welcome!

[@kubita-thecall8795]

@@TokenizedAI do you know of any of these ai sites that work well on android phones? Just curious

[@bad_writer]

Love this series, this is the best tutorial on comic book techniques I've seen so far. Thanks for making this

[@TokenizedAI]

Thank you 🙏🏻

[@megholbrook8334]

Excellent series, I've had a lot of fun generating my own original characters and placing them in various scenarios (most recently injecting them into isometric scenes with the help of your 'scaling' tips). Looking forward to learning more!

[@TokenizedAI]

Glad you enjoy it!

[@DonStone808]

just found your channel and am amazed! Thanks so much for sharing all this info!

[@TokenizedAI]

You are so welcome! 🙂

[@basiccomponents]

This is really innovative stuff, I got the best results in a while, I finally found a style but couldn't get close the results I wanted, keeping in mind what you showed here I definitely feel I'm on the right track, thanks!

[@TokenizedAI]

Thanks! :)

[@finlaycowan3681]

Thank you so much for this. You've answered so many difficult questions and solved so many problems with this one video !

[@TokenizedAI]

You are so welcome!

[@DavidPeyrouty]

So so gooooood!!! I'm looking forward to the sequel, as I imagine many are!

[@TokenizedAI]

Working my butt off trying to get it out to you guys as soon as possible.

[@DavidPeyrouty]

@@TokenizedAI My intention was really not to pressure you I know how time consuming it is! The amount and speed at which you publish your videos is already perfect... Good luck, and thanks again for sharing all this knowledge with us!

[@TokenizedAI]

Oh don't worry, didn't feel pressured at all. Just stating that I'm just as eager to release it 😁

[@datano74]

Love it! I was waiting for you to make this one with the same expectation of waiting for a blockbuster and I loved it ❤ thank you and keep up the great work now can’t wait for the next ☝️ ❤

[@TokenizedAI]

Haha, I'm not sure about the "blockbuster" part, but thanks! :)

[@thebluriam]

This series is probably the most advanced and real-world applicable AI Tool breakdown on the internet right now. I say AI Tool and not specifically "Midjourney" because I've not seen so much thoughtfulness and diligence behind the exploration of prompts in any other channels, not just for Midjourney ones (and I watch a lot of AI prompting videos). Seriously, these videos are probably my biggest inspiration for wanting to start making my own helpful AI Prompt videos, but I'm having an extremely hard time getting the ball rolling. I think I just need to do it.

[@TokenizedAI]

You have no idea how humbling these comments are 🙈

[@thebluriam]

@@TokenizedAI Truly, I mean even word. So much so that I'm working on doing my first video right now. I joined your Discord server today too since I just learned of it. I'm sure we'll talk there too

[@kennymonsters]

Loved it Christian, very well done. I wanted to point out that if you used the final prompt of Carla in the coffee shop in an actual comic book or graphic novel, that this is a great example of an establishing scene. Tony Moore (The Walking Dead) did this perfectly in issue #1 "Days Gone Bye" pg #1. The first panel shows a panoramic image of Rick and Shane in a shootout on a highway behind a cop car to establish the scene. He than went on to closeups of the characters with supporting dialogue using 4 smaller panels, followed by another wide panel at the bottom with a panoramic image of Rick getting shot by the suspect they were chasing. Fantastic work my friend

[@TokenizedAI]

I'm actually not particularly proficient in the comic book world, so this comment of your is quite fascinating for me. Thanks for the tidbit.

[@silversurfer4195]

A masterclass in and of itself. Excellent!!

[@TokenizedAI]

Many thanks!

[@elex05]

I'm totally binging the whole series now.

[@TokenizedAI]

😅🤗

[@DonatienGnakliZ]

Bravo ! for this very intersting and insightful series, very well done, thank you so much

[@TokenizedAI]

Thanks, appreciate it :)

[@MasonIce1999]

I love this series. Thank you. I have a question. I notice when referencing a URL you place it right after the /imagine prompt so why not do that with the --Seed? I was wondering if doing that would eliminate the need to re-type the entire description. I am pretty new so I hope this is not a dumb question. Thanks

[@TokenizedAI]

The seed doesn't do what you think it does 🙂 It's just a random input variable.

[@nicole-wk]

I never subscribe, but I subscribed tonight because your videos are always so helpful and informative! Waiting on pins and needles for you to teach me how to get multiple consistent characters to interact because I'm about to lose my damn mind. But honestly, I think after watching this tutorial, I understand multi-prompting so much better that I might get there. One thing I noticed as I have been developing and fine-tuning my OC is that if I upscale the image I like at least 3x, and heart react to the one I like best, it starts to get a bit more consistent. Each upscale has a totally different vibe a lot of the time! When they get hard to tell apart, I know I am really starting to dial it in and save the seed numbers of those renders. I've also done some really weird things like get an image with a pose I like off the internet and slap my character's face on it (poorly in Canva, lol) and use that as an image ref. to get better poses. Or I find an outfit I like on a mannequin or a model and erase her head before I use the image as a reference. Or I will take my ref image of my character and remove the background and put her on something different in a different aspect ratio or something in Canva and then use that as a new ref. when I am trying to get a different scene - but I like how you've used Midjourney to create those scenes. It's been a lot of trial and error, of course, but those are a couple of things I thought are worth sharing in case they are helpful to you or anyone else who comes along and reads this. :)

[@TokenizedAI]

The things you are trying to do are not weird at all. In fact, they make total sense to me. And yes, I use some of that in my next video's attempts. I say attempts because (as you will see once it's released) I'm not yet convinced that it's actually possible to place multiple characters in a scene. Or at least not the way we would like to.

[@dmitristefantsov8975]

once again, exactly what I needed! great job, thank you =)

[@TokenizedAI]

So glad I could help :)

[@lestutosmidjourney]

Hi Christian, thanks for this (quite technical!) video. I would like to know why you do the first three steps of character creation (full body, medium and portrait image) to mix them? Why don't you create the character directly since you only use one reference image (URL 1) in your video? Thanks 🙂

[@TokenizedAI]

Good question. I think it's mostly preference. I've found that when you use a very specific character concept, it's quite difficult to reproduce consistently. But when you blend multiple images with very slight differences, the resulting image feels like it's closer to all 3 source images. In other words, your blended character is more likely to be perceived (!) as similar throughout your work. But that's really just a personal touch in the process. I guess you don't really need it....like I say in the video.

[@DMDrew]

How does this video only have 400 likes? This ENTIRELY changed everything I make on MidJourney to be unrecognizably better than my previous work.

[@DMDrew]

I'm struggling to get faces to look as consistent as yours, though. I got my image blended from 3 and loved it. Then I tried to run this: " wide angle shot full body image, young elf boy, dark purple messy hair, only dark purple hair, wearing purple wizard robes with a gold star pattern and a hood, wearing purple pointy elf shoes, wearing white genie pants, in a medieval style, in front of a castle in a cave, purple lighting hue --ar 3:2" having added "in front of a castle in a cave, purple lighting hue" hoping he would be in a cave with a castle in it. I did indeed get the character in the setting I wanted, and the body and outfit are spot on. But his face, my god it looks like some distorted painful abomination every time 🤣😭 any tips for somewhat consistent faces?

[@DMDrew]

tied this as well " Panoramic view of a cavern with a purple hue::1 distant shot of a cavern with a purple hue to the lighting::1 wide angle shot full body image, young elf boy, dark purple messy hair, only dark purple hair, wearing purple wizard robes with a gold star pattern and a hood, wearing purple pointy elf shoes, wearing white genie pants, in a medieval style, in front of a castle in a cave, purple lighting hue:: 6 --ar 3:2" and the background keeps getting better as his face gets more and more deformed...

[@TokenizedAI]

Not everything that you can imagine in your mind will necessarily work in MJ. I think it's heavily limited by what it's been trained on. For example, it's dfficult beyond the limits of comprehension why it's nearly impossible to position and white rabbit next to a honey badger without MJ constantly blending both animals into some freakish imaginary creature. To this day, I have not managed to do this. With cats and dogs it's a bit easier because there are heaps of images out there that show cats and dogs side by side. Rabbits and honey badgers....not so much.

[@DMDrew]

@@TokenizedAI that’s super weird I’ll just be happy if I can put a character in a different scene and they don’t have a blank face 😂 That’s what I’m running into the most lol it’s very creepy like a horror movie

[@TheTimedCarbGuy]

Also, the aerial shot and panoramic shot combination is such a cool idea - def gonna use that for the scenes of my timedcarb comicbook story :D Do you have another idea for solving the problem of carla being too big in the coffee shop without PS/AD and without making the following results of her face seem out of a horror movie? :)

[@naturehikingchannel5336]

Really enjoying your tutoring on consistent characters and I actually get impatient waiting for your next episode 😊 I’m wondering if it’s possible to have consistency in an extremely simple design eg: a simple flowing sketch of a character with limited colours and thick lines etc or is this only possible with much more complex character in comic style? I’m experimenting in the meantime and await your next instalment. Very much enjoying seeing you come across things I run into as well and your solutions for them.

[@TokenizedAI]

Awe shucks...that "I actually get impatient waiting for your next episode" is so sweet 😁

[@huibert_devet]

@@TokenizedAI I have the same question. A simple design with limited colours and thick lines to make a cartoon for my blog. Is this possible with MJ? I want to give MJ examples of my own homemade cartoons and make scenes of them with MJ 😎. It would be grateful if you have a answer for me 🙂

[@shaunaoc]

Love this series! Thank you !

[@TokenizedAI]

You are so welcome! 🙂

[@LaorejaLab]

Amazing Job Christian ! greetings from Chile

[@TokenizedAI]

Thank you! :)

[@princessmw]

Hi Chris, I watched this when you first uploaded it; my apologies, I thought I smashed the like button then🙈. I love all your videos.

[@mozimiga]

Thank you so much for your video series! I am getting so much out of them! I am working on a graphic novel and have been doing the artwork myself using Daz Studio and redrawing the scenes myself in Procreate to make them "cartoon style". I am hoping to integrate AI into this workflow, so I am LOVING this series. THANK YOU!!!!

[@TokenizedAI]

Awesome. Really glad that so many of you are getting value out of these :)

[@vegandilemma1129]

Thank you Christian! This is very helpful...

[@TokenizedAI]

My pleasure!

[@awesomecoyote5534]

Great video for works arounds to get more consistent character images and various scenes. Hopefully AI image generators start making features to get more consistent characters and more complex scenes easily. I think that's the only things really holding back users from making more detailed comics or even animations, inconsistent characters and static, centered scenes.

[@TokenizedAI]

Yeah, I agree. However, as long as you're willing to use a combination of tools and don't necessarily insist on doing everything in MJ, you can already to a lot today. MJ + SD/DALL-E + Photoshop will get you pretty far.

[@ClubCreative23]

This series is AMAZING! I really needed this right now and the examples are invaluable. I'm working with dogs, so my prompts are different, but I'm using yours as a guide. Thanks SO much for this. PLEASE: I would love to know how to put both my dogs into one still, posed together. I've tried the two prompts, but I can't get them to be where I want them! If you are doing a lesson on this I would be VERY happy! Thanks, Christian - from Oz.

[@TokenizedAI]

Using animals as characters is both easier and more difficult at the same time, but for different reasons.

[@paris.granville]

Great video again. I appreciate your resassuring attitude. So clear! I have wondered about combining the background with a separate character in something like powerpoint. That way I could ask midjourney for a character sheet in different poses. But it seems like it would be too low res.

[@GabrielSantosStandardCombo]

Do you think it would be possible to achieve a similar result, but Carla has her back to the camera? Would we start all the way from the beginning and develop concepts of the character from the back? What about concepts of the character from different possible angles?

[@TokenizedAI]

Sure, but you'll have to create new concept art for her in different poses and from different angles. That's actually what you need to start doing once you want to create more elaborate scenes. It just doesn't work unless you do extra work.

[@StyleViewStudio]

I love you tutorials! They’re detailed, brilliant, and very specific… Also, very artistic. But on a humorous side, may I say as a woman who is also very active that it is very difficult to run around when you have such a big chest and big hips… Most most action heroes like Angelina Jolie in tomb raider she’s slender and muscular so just keep that in mind for future fun tutorials…❤

[@TokenizedAI]

lol, yeah. This is mostly due to version 4 of MJ which creates this super curvy female figures almost by default as soon as you add "marvel comic" to the prompt. v5 doesn't do this as much. Thank goodness.

[@mysterious_monolith_]

Seriously awesome!

[@bgreynolds_loess]

This is superb work. Thank you so much.

[@TokenizedAI]

So nice of you 🙂

[@nosferaturr]

Not so bad. You are inventive for sure. :)

[@TokenizedAI]

Thanks 😁

[@DeanWilson]

Awesome series of creating a character in Midjourney. I put it to the test and I got my farmer character perfect in almost every shot. However, when I added farmers "kids (2) boy and girl" to the scene, they looked like aliens. Any suggestions on how to fix this? Thanks.

[@TokenizedAI]

Kind of depends on what your entire prompt and any potential image prompts look like. I think you should watch Part 4 of the Series. You'll probably want to start learning how to do multiprompts.

[@DeanWilson]

@@TokenizedAI Thanks. yes, giving it a shot tonight.

[@oliverli9630]

6:44 the passers by on V1.... lol one stands on the hand rail, the other stands in the gap

[@TokenizedAI]

Hahaha, good catch. I didn't even notice that :P

[@artcgix]

I've been following the series, and you have done an awesome job! Keep it up, she is till not standing in line though 😆AI:0 - Illustrators:1

[@TokenizedAI]

Hahaha, thanks. Yeah, there clearly still are a lot of limitations and it's pretty frustrating when I'm preparing a video that is meant to show how to solve a particular problem. But it's super interesting to see how far we can push the limits with what we have 😁

[@Brian-ej9ew]

Love the content and learning a lot of useful info. I was wondering how you can combine a scene with 2 central characters you have created instead of just 1 character like Cara.

[@TokenizedAI]

I'm working on a video about this, but I'm afraid it's not looking too good 😬

[@ionceexisted]

Hey Christian, I binged on your videos for the comic book character series. I am loving it so far and I find them so easy to follow. Just a question, for the URLs, do you need to add separators (eg: Spaces, commas, or ) between the URLs for the generated images?

[@TokenizedAI]

Just ad spaces between the URLs. In the video, when I mention it's just meant as a placeholder.

[@JoeCastellon]

I'm learning so much from you! Sadly, I only have like 5 images left on my MidJourney trial and at the moment, I am unable to justify the subscription 😞

[@TokenizedAI]

If you check the MJ Discord, I believe they have a subscription reward for the top image raters. But I'm not sure whether they're still doing that.

[@JoeCastellon]

@@TokenizedAI thank you. It's worth a shot!

[@_The_Wu_]

Would continuing to use the characters name help in the long run?. I noticed it isn't necessary with the seed prompts, but what about with the new scene images along with the URL, just to reinstate it for the 'training' purpose?.

[@TokenizedAI]

Nope, the names are irrelevant at this stage.

[@arknight8794]

Unfassbar gute Videosserie! Danke für die detaillierte Erklärung und die ganze Arbeit die du dir dafür machst. Darf man fragen für welchen Zweck du Bilder aus midjourney nutzt ?

[@TokenizedAI]

Vielen Dank 😊 Derzeit primär für die Cover-Bilder meiner Blog-Posts 😂 Allerdings beabsichtigte ich, über die reinen Lehrinhalte hinaus, tatsächlich auch zielgerichtete Grafik-Arbeiten für kleine Betriebe anzubieten. Also quasi die Dienstleistung, die klassischerweise eine Agentur übernehmen würde, mit Fotoshootings etc. um Werbebilder zu erstellen. Da gibt es enorme Einsparpotentiale bzw. die Möglichkeit, die Qualität der Werbematerialien zu verbessern. Und im lokalen Bereich zählt erstmal Vertrauen und persönlicher Kontakt. Gerade bei den Kleinen.

[@arknight8794]

@@TokenizedAI das klingt super und ein guter Freund und ich hatten eine ähnliche Idee. Falls du Interesse hast in der Richtung etwas Brainstorming zu betrieben, sag einfach bescheid.

[@TokenizedAI]

@@arknight8794 Würde ich gerne. Allerdings hab ich derzeit einfach zu viele Dinge um die Ohren 😅

[@arknight8794]

Alles gut eilt ja nicht 😂 darf man dich einfach auf Twitter anschreiben ?

[@TokenizedAI]

@@arknight8794 Email ist ehrlich gesagt erstmal besser (siehe About Page). Ich versuche Messenger-Dienste möglichst auf einen extrem engen Kreis zu reduzieren, da mich zu viele laufende Chat-Benachrichtigungen eher stressen.

[@1979chyeo]

Hi, what does the means? Need to upload the image to the web and add the link?

[@TokenizedAI]

Yep, those represent image prompts.

[@pierreg2780]

Why don't you use "Carla Carusso" anymore in your prompt?

[@TokenizedAI]

Because it's not actually necessary. Especially not when the character is shown from up close. I made started removing her name in Part 2 of the series. Have a look.

[@Felaura]

@@TokenizedAI This was also my question 😃 In part one, setting up the name and solidifying her character design took a lot of time and images and emoji reactions, but I'm curious about why it doesn't matter? If you hadn't done all that, would you still be getting these consistent results now?

[@andycification]

Great video, thank you! On the very last image of the girl in the coffee shop, you used an image (url) but I wasn't sure which image that was. Was it from the photoshop image where you resized the girl to make her smaller in the scene? Thanks

[@TokenizedAI]

Sorry if that wasn't clear. Yeah, I used the image that I edited in Photoshop, or in my case, Affinity Designer/Photo.

I really like the step-by-step progression of your tutorials. …I wonder if it would stop the colour bleed if you remove the white background in the Carla image reference

[@TokenizedAI]

It does, but it creates a new problem. You get tons of cropped images instead.

@@TokenizedAI Ofcource 🤦🏼‍♂️ Good to know. Thanks

[@SiraEkabutPlus]

Really good tutorial! Thx!

[@TokenizedAI]

Thanks! 🙏🏻

[@flaviomorofurian5271]

Would be possible to have more than one character interacting on a scene? Is it possible to set each character with their own seeds?

[@TokenizedAI]

You can't apply more than one seed. Applying a consistent style to a multi character scene would require something similar to what I did at the very end. I'm pretty sure it's impossible to do purely via a text prompt.

[@caiosaulnier6574]

Christian, congratulations for your amazing work, we can see how much effot you're putting here at this project. But i think that unnecessary that every time you change a little bit the prompt, you read it all again. I think that only communicate what has been changed and put a 1.5seconds print of it in the video would be ok, the video would be more dynamic and you would saving your voice HAHA (sry fot the bad english)

[@TokenizedAI]

Yeah, it's become a habit to repeat the whole prompt when recording. Simply because it's sometimes difficult for me to immediately see what's changed 🤣 I will try to focus more on just what's changed from now on. I agree that's a bit repetitive in some of the cases.

[@caiosaulnier6574]

HAHA, it's all right, the content of this serie still amazing, congratulations! And thank you for the answer :D

[@helmutroll4773]

is it true that it is easier to first start creating a full body shot before defining the close up variations? because I have started to make good variations of the face (frontal, left, right) but I have a really a hard time to get afterwards the full body shot, which is resembling to the close up faces by adding the urls. The closest I get is the "cowboyshot" but I just can't get the legs showing fully.

[@TokenizedAI]

For full body shots you need to create new fully body concepts first and then use those as image references.

[@Wiyose]

Thank you so much chris.

[@TokenizedAI]

You bet!

[@RequestforQuest]

Great tutorials! 👏 Do you think it's possible to develop such approach for a character, which already been created in a messy way - with lots of remixing and such? Or is it better to start from scratch?

[@TokenizedAI]

Hard to say. It really depends on how picky you are with the results. I'd probably start over if you don't like the current images.

[@RequestforQuest]

@@TokenizedAI thanks for the advise! Some time ago I've created a bunch of characters for an RPG campaign. But not with just text prompts, with images, and constant remixing of results. Like, Ok, I'll take this one and this and look what if I mix em. Now I have my ideal faces but it looks like their prompt history is too messed up to work with consistency ))

[@jblade057]

How would we specify to generate Karla without tattoos as each one have different tattoos? Also, how would you fix the strange faces of the coffeeshop customers? Creating negative prompts? Great tutorial video as always!

[@TokenizedAI]

I would probably just retouch the primary Carla imega reference and remove any tattoos she may have and then use that use image instead. As for the other customers in the shop, since they aren't protagonists, I would consider them a negligible detail. Let's face it, traditional comic books also have declining levels of details on most secondary/tertiary characters.

[@jblade057]

@@TokenizedAI thanks for your response!

[@landwolfcto]

Amazing series, thank you!

[@TokenizedAI]

Thank YOU!

[@alonsoperezlopezsilva2566]

Hello Chris I have a doubt on how to create a character out of the result from a prompt that was not meant to be a character, I liked the outcome so much that I wanted to turn it into a character.

[@TokenizedAI]

I'm not sure I understand your question.

[@RailinX]

That’s amazing!!

[@TokenizedAI]

Thanks! Appreciate it :)

[@vishalkhombare]

Lovely!!

[@TokenizedAI]

Thank you! 😊

[@lolskylar]

Hey Christian / Others. I replicated this procedure with a stylized non-human bipedal character design without success. Is it possible that you could address some of the challenges with doing non-standard human faces? Thanks!

[@TokenizedAI]

I'm afraid I didn't even fully understand from your message what you're struggling with. I get the non-human part, but not what you mean by "bipedal".

[@lolskylar]

@@TokenizedAI bipedal just meaning humanoid like characters. Not just stylized humans.

[@aurorajones8481]

Great video. My problem is integrating two characters into one scene. And I'm working with animals. So my other characters come out twisted and odd like its never heard of a rabbit squirrel or bird! Seriously i get some non animal like outputs. Aye...

[@TokenizedAI]

Yeah, the struggle is real 😂

[@tracktaylor2783]

question: does the urls have to be in inside < > ? or do u just copy them them in and seperate them with a single press on the space key? Great tutorial

[@TokenizedAI]

I think MJ automatically removes those if you add them. I just add the < > to clarify that they are variables and not actually part of the text prompt.

[@tracktaylor2783]

Thank you very Much for quick response.. it makes sence yes.. your videos are very well explained.. im getting quite good concistency now

[@thenext9537]

MJ is like a love hate relationship. I love that it does some *CRAZY* things I didn't ask for, but also hate it because it does some *CRAZY* things I didn't ask for.

[@TokenizedAI]

Haha, so true

[@robjdrum]

When you finally find something that you want, is that when you assign a seed to it? Or do you need to have the seed set before you land on something you like? I'm not understanding exactly how to use the seed.

[@robjdrum]

Got It! Watched your SEED video :) Thank you!

[@wallaoduser]

Thank you for your content. Very useful! Did you try creating sprite-sheet? Consistent character, multiple poses/expressions, to create frames for animations. like for a walk-cycle, idle.. etc

[@TokenizedAI]

I have an entire series on character design. This is just 1 part.

[@wallaoduser]

@@TokenizedAI awesome. where can i find it?

[@TokenizedAI]

On this channel. Not that difficult to find 😅

[@YassineWa7dLiKayn]

You're saving lives

[@TokenizedAI]

Not sure about that, but thanks 😂

[@shoptimelessboutique5858]

WOW WOW WOW

[@donnell9238]

Nice video and very great work, OG, but I'm a bit confused with the url part do we insert the url of the images there?

[@donnell9238]

nvm I got it man, you are the man lol

[@TokenizedAI]

Cool :)

[@hakosho2003]

Can you help me in creating a character performing blows (punches and kicks)? I'm having a lot of trouble with this

[@TokenizedAI]

Difficult to say without knowing the overall scene. I recommend joining our Discord community: discord.gg/dGhqN7u5 I think that's easier than me trying to help you solve something via the comment section ;)

[@hakosho2003]

@@TokenizedAI Alright I joined your Discord. Do I communicate on the "promp-talk" channel with you?

[@TokenizedAI]

@@hakosho2003 You can communicate in any channel you like. I recommend reaching out to others in the Discord as well, since I'm obviously not always there and that's the whole point of having a place for everyone to share and help eachother. For example, I'm not at home right now and probably won't be able to respond until much later anyway.

[@JoeDitzel]

Wow! Good stuff.

[@TokenizedAI]

Thanks!

[@Wuuzzaaful]

You just defined "over prompting" :). JK love it!

[@TokenizedAI]

Hahaha 🤣 I do feel like I'm overdosing on prompts.

[@tylergorzney8499]

What video editing software do you use for making your great videos?

[@TokenizedAI]

I use VN Editor, which is currently enough for me to get all the basics right. However, I must admit that it's not great if you need to fix audio issues.

[@rodrigo8269]

great video!

[@TokenizedAI]

Thanks!

[@davediamond]

Whoah. This is insane. Thanks heaps champ. I think I hhave a challenge for you. Based on this video. I think you’ll appreciate it!

[@TokenizedAI]

I've got so many challenges ahead of me, I can barely count them all :P

[@_ENOVA]

what will happen if I use a png of a character with no background (transparent) and try to merge it with some background? I mean - will MJ do it correctly and - if so - why didn't u do that? U obviously wanted to get rid of the white background, so why woudn't just do it in Ps?

[@TokenizedAI]

I didn't want to get rid of the white background. I wanted to show everyone what happens, when your image reference has a white background. Sure, you can remove it. But even if it's transparent, the lack of color is also information that is translated. You'll see a lot of cut of areas. Also because the dimensions are completely different.

[@reidosnoobs81]

im almost giving up on midjourney, i put the same prompt and it give me results that or not even close from marvel or any other mentioned here. It seems that it does not want to obey me.

[@TokenizedAI]

This is a massively old video. It doesn't even use the same version. Prompting works very differently in v6.

[@J4ME5_]

seed seems to never give me the same results

[@TokenizedAI]

--seed only give you the exact same results if your prompt is also 100% the same (including parameters).

[@chrismachabee3128]

Totally scintallating

[@TokenizedAI]

Thanks 😁

[@tolstoy_was_right]

why did you stop to use the name of the created Character? Is it became obsolete?

[@TokenizedAI]

Because for what I was trying to do, it wasn't important anymore. The more abstract the character, the less obvious the inconsistency becomes.

[@Nytellem]

It really refused to use the braided bun hair lol

[@TokenizedAI]

Yeah, I dunno. Maybe it just doesn't like that hairstyle 😆

[@VideoNash]

thanks

[@sysadmin9396]

Subbbbbbed

[@TokenizedAI]

🤗

[@MotiviqueStudio]

But if you're planning to write an actual comic book and NOT go outside of Midjourney... I mean... good luck?

[@TokenizedAI]

If you really want to write a comic book, then why would you let the fact that you may have to learn some basic image editing skills stop you? MJ is just another tool like everything else. It's not a full-blown comic book generator. 😉

[@the_ironman]

Except it doesn't work... The composition of the image prompt is too strong. MJ doesn't get the concept of using only the character

[@TokenizedAI]

Sounds like you didn't watch past the first 5-6 minutes. Cause I clearly demonstrate that it can be done. Is it easy? Is it perfect? No to both, but I clearly stated that in the video multiple times. If you thought that you could just mash together an image with a text prompt with barely any effort and get impeccable results, then I'm afraid you're going to have to manage your own expectations 🤷‍♂

[@the_ironman]

@@TokenizedAI I appreciate your efforts. The method you're describing is the best I could come up with as well, but it is not feasible for a graphic novel to put that much effort into every image. For the novel I'm making I'm thinking to first get char images from all different angles and distances and hopefully that kind of works. But I don't have enough time to do that right now.

[@TokenizedAI]

@@the_ironman Everyone who plans on creating a graphic novel needs to take the limitations of MJ into account. Right now, everyone's assuming that anything is possible with MJ and that's why many are aiming way to high. All of this might be possible in the future but not right now. In my opinion, the best way forward is to look at what IS possible right now and then plan your graphic novel around that. Accept that certain scenes cannot be rendered directly in MJ. So you can either limit your storyboard to 1-character snapshots, simple 2-character scenes, or try to blend stuff within Photoshop. Define your boundary conditions and then work off of that. Anything else will just frustrate you.

[@RubykonCubes3668]

Artstyle is bound to be inconsistent because it evolves along with your tastes and skills. It even comes down to minute details. Anyone who draws or paints knows this. But alas, how is it comparable to a process as dissociating as GENERATING images? It is not. Honestly, i don't even understand why ai art exists at all. This genre feels like a weird loophole for people who know how to write but rely too much on the alleged myth that, online, images are more appealing to browse. Either way, i don't get it. Interesting video though.

[@TokenizedAI]

AI art exists because some researches were curious whether it could be done. Just like most other research fields. Unlike what many people seem to believe, I do not subscribe to the fatalist assumption that AI is going to take away all our jobs. Quite the contrary. AI is going to make us incredibly more productive in so many ways that aren't even imaginable yet. I know that isn't something that necessarily everyone wants, but it's bound to happen. For example, so many people out there are creative in many ways. That doesn't necessarily mean that they are able to draw or paint. But perhaps they are really good writers or just like to create stuff. In the past, many of these people may not have been able to create a fully fledged illustrated book because they couldn't afford an illustrator. This allows them to do it all by themselves and exactly how they want it. You can also switch it the other way around, with an illustrator using an AI text generator. Again, these are just basic examples. There are many more. And some people are incredibly creative but they're much better at seeing the bigger picture and putting all the "building blocks" together into a finished product. The classic jack of all trades, but master of none. Despite the derogatory use of the phrase, these are talented people. I don't expect everyone to understand the appeal of it. Especially because not everyone necessarily has any interest in doing anything but the specific craft they learned. But AI is going to be a massive creative enabler for many individuals around the world.

[@robjdrum]

Midjourney v5 sure has been a disappointment for me using these technics. The characters is generates all seem to be funky lookin.

[@TokenizedAI]

Please watch my video where I share my initial impressions on v5. It's not even the final release. It's not even beta. It's alpha. But it's actually grear, because it forces everyone to learn how to write better prompts.

[@logitoonz2246]

deep condolences for " Real Art ". F**k AI art

[@TokenizedAI]

Let us know if you ever need a hug 🤗

[@logitoonz2246]

@@TokenizedAI LOL i am drawing now, it is already for hours for making an art. But AI did only for a minute or les LOL

[@TokenizedAI]

I recommend you find ways to incorporate AI in your workflow rather than trying to fight a quixotic battle against something that's not going to go away.

[@logitoonz2246]

@@TokenizedAI

[@TigerWoodsLibido]

For me, all of this is easy enough until it requires photoshop…

[@TokenizedAI]

Learn and do hard things. That's where the value is. If it's easy, then what's the point? 😉

[@benzyn]

Brilliant man...just brilliant 👏🤍

[@TokenizedAI]

Thanks! :)

[@carmillasmith7743]

Is their a way to just give your character a name so you won't need to include the details everytime?

[@TokenizedAI]

It seems that you can now use "/prefer option set" to also include text prompt portions.