INFUSE Themes into Characters & Scenes - Midjourney Character Design

[@max-imize]

Oh yes! It's that time of the week again! We are going to LEARN 🚀

[@TokenizedAI]

Means a lot hearing this from a fellow KZbinr ☺️🙏🏻

[@DigitalAscensionArt]

I have to go back through the whole series again. This series is 🔥!! Thank you for another great video Christian!

[@TokenizedAI]

In hindsight, I think I may have to create an alternate video for Part 1, since some of the stuff in there has been debunked.

[@DigitalAscensionArt]

@@TokenizedAI The landscape of AI generated art evolves so quickly that I have to watch KZbin daily for the changes. Of course content that you put out a week or 2 earlier is now somewhat obsolete. I think it’s a great time to get it. I think AI art will be very complex down the road because of all of its capabilities. You’re doing great work! Keep it up!! You’ve taught me a lot!

[@taunasonn-lemarbe7267]

I enjoy your videos a lot. As a traditional artist unfamiliar with anything digital, your series has helped me with MJ. Thank you

[@TokenizedAI]

That's really nice to hear :) And kudos to you for embracing this new technology in spite of being traditional artist.

[@shrvn110]

just stopping by to say we love you! sorry i dont comment on every single video but your videos are GOD level value! thank you.

[@TokenizedAI]

Thank you 😅😘

[@Great.AnotherChristian]

The community only benefits from everything you, my man! Keep it up!!!

[@TokenizedAI]

🤗 Thanks

[@DMDrew]

I think this will help with what I was talking about on part three in the comments I was adding too many theming elements for the setting and it was kind of messing up my character. I think now it’s because I wasn’t really understanding the number ratios. Solid video man

[@TokenizedAI]

Yeah, this method can really help cut down on unnecessary stuff in your prompt. Especially if you have a general idea of your settings but aren't too picky about the exact details.

[@escagedo]

I didn't get the weights to properly work until I saw this video. Thanks for your work and dedication, this is an amazing channel.

[@TokenizedAI]

Awesome 🙂

[@Cifero-d5m]

love your way to explain. Great content, please keep on :)

[@TokenizedAI]

More to come! 🙂

[@generatedbyaistories]

Very useful! Thank you for your examples! Seeds are so relevant and unknown...

[@TokenizedAI]

Glad it was helpful!

[@RocketShipTM101]

Bravo!! Great stuff

[@TokenizedAI]

Glad you liked it!

[@ATRcams]

Thanks for all of those!

[@TokenizedAI]

My pleasure :)

[@GrimmHollywood]

Hope this isn't the end of Carla, it was a great journey. Would love to see some multi-character stuff with her once V5 releases, and see what she can do in that new version too. Thanks for the MJ lessons, Christian 👊

[@TokenizedAI]

Probably not. But I need a break from her :D

[@GrimmHollywood]

@@TokenizedAI ahahaha fair enough! 👍

[@LittleFellaDynamics]

Yeah that is really useful for controlling the introduction of a theme into the image >> looking forward to playing around with this technique... perfect thank you 👌💯👀🎯😎🌟

[@TokenizedAI]

Have fun playing around with it!

[@shelleyplum2286]

ty so much for providing these great tutorials

[@TokenizedAI]

My pleasure!

[@ozmenlaps3038]

You are so awesome) thank you and keep up the good work)

[@TokenizedAI]

Thank you! Will do!

[@WilliamAlexanderVT]

Simply fantastic!

[@TokenizedAI]

Thank you! Cheers!

[@sig7813]

Great work! Thank you

[@TokenizedAI]

Glad you liked it!

[@jonathanfesmiresteampunkau6983]

Thanks for the series! I found it very helpful, including to my own Midjorney experiments.

[@TokenizedAI]

Glad it was helpful!

[@A20Digital]

Wonderful and informative series! will be my number one go to for training others on this process.

[@TokenizedAI]

Are you teaching prompting to people? 🙂

[@yveskeroack1612]

Love it! Keep up the good work!

[@TokenizedAI]

Thank you!!

[@Retro_AI]

Thank you, that's great content! 🔥

[@TokenizedAI]

Appreciate it!

[@shaunaoc]

Love this! Thank you!!

[@TokenizedAI]

So glad! :)

[@backstabba]

One thing I am curious to try is to find out how impactful is a word 'theme' here. I would naturally go with the same but leave such word out.

[@TokenizedAI]

I do think it makes a slight difference. Especially if you use stronger weights.

[@TchakalGames]

YOURE THE BEST, hello from 🇧🇷

[@TokenizedAI]

🙏🤗

[@danjones7561]

You're truly the boss :)

[@TokenizedAI]

Not sure about that but thank you very much 😅

[@isek.a.i.]

Great series! Thanks so much!

[@TokenizedAI]

You're very welcome! And thank you!

[@manuelprsnl]

Very cool! I'll test these and tell you my results :)

[@TokenizedAI]

Please do!

[@thingschangemusic]

We created and animated a complete music video with images from midjourney : kzbin.info/www/bejne/nHm9YYqpabuNqs0 If you wanna know anything about this, dm us or ask in the comments of the video

[@TokenizedAI]

Awesome work :D

[@thingschangemusic]

@@TokenizedAI Thank you!

[@Ben42-Ben42]

Love it. I struggle with an easy promt, so I have to learn the magic

[@TokenizedAI]

Practice makes perfect.

[@TheTimedCarbGuy]

I think to be able to also do that I have to rewatch the whole series one video a day and also invest an hour of experimenting on each of the series part's in MJ :)

[@craigstriz]

Good lead on from the last video!

[@TokenizedAI]

Thanks!

[@GreyFoxRL]

I have really enjoyed watching and learning from your part 1 thru 6 series character design tutorials. Thank you so much for taking the time making these videos for us to learn Midjourney v4. Version 5 just recently came out and I am not having as good luck with the comic book characters looking as good as v4. The v5 is more toward realism and does do hands and eyes better but not as good as v4 in the comic book realm. If you do find a way to do the same or similar course in the comic book characters in v5, I would love to go through that series of videos as well since it does do hands and other things better than 4. Anyway, thank you again for all the hard work putting out these videos for us. :)

[@TokenizedAI]

v5's default look is realistic. That doesn't mean it can't do comic book styles. It just takes far more words to get the right style. v5 is currently in an advanced mode which actually gives you far more control. That makes it more difficult for beginners because v4 was actually doing all the heavy lifting for them.

[@GreyFoxRL]

@@TokenizedAI Thank you for the fast reply. So, do you think I should learn to characters first in v4 and then move to v5 later? Also, do you know if I create a character in v4 and get an image I like but has bad looking hands, can I take that image to v5 and maybe do a remix or something that v5 can help the hands look better?

[@TokenizedAI]

Different versions will always work differently. All I'm saying is that if you want pretty images now, then use v4. If you don't care about that and prefer to use the improvements in v5, then do that, but you'll have to learn a lot about prompting in order to get what you want.

[@GreyFoxRL]

@@TokenizedAI Thank you so much. You have been a big help.

[@Dazreil]

I'm pretty sure MJ blends the image prompt and the text prompts separately the then blends the results together.

[@TokenizedAI]

Yes, I agree. Did the video come across as if I was stating the opposite?

[@rethinkprompt]

I'm so happy that I found your channel.... Do you have a video on How not to mess up faces? or correct them?

[@TokenizedAI]

Whether faces get messed up or not, largely depends on the steps you took to get to the point there they do mess up. And in other cases, you can't really prevent it.

[@deep2mixer]

You can try negative prompts.

[@byzmusic7forever]

Hi Christian! I'm new to Midjourney and really enjoy your tutorials. I have learned a lot of useful stuff and I am very grateful to you for that. I'm still curious if you could tutorial explaining how I can create my own illustration style, not just characters and themes. Thank you very much.

[@TokenizedAI]

Can you elaborate what you mean by "own illustration style"? Midjourney can't really be trained like that. In the end, you control the style with your words. You just need to find the right combination.

[@byzmusic7forever]

@@TokenizedAI I was referring to a certain kind of drawing in all forms, that is, whether it is cartoons, or illustrations, or fantastic, or regardless of the applied theme or the character created, to keep a certain style of drawing. But you are right about Midjourney and thank you very much for your reply. I follow you with love.

[@FreibierDas]

I think the img prompt weight is a way lower then you expect atm. If you have a theme that is dominated by men, the theme negates the gender of your img + prompts easily. In my case 5 img prompts of a female + promps like @05:55. Got it fixed by simply prompting female+theme. It encountered this first when I tried to apply a dollar bill theme. and every dollar bill bears the portrait of a man :(

[@TokenizedAI]

Absolutley possible. I just had to make an ussumption for the purpose of illustrating how the weights don't always have the same impact. In the end it doesn't really matter much since we need to go through multiple variations anyway until we hit the right level.

[@areacode3816]

Crazy how independently we used several of the same steps to do 3 character scenes. Mine I must admit was much more complicated relying heavily on outside programs. Can you do a video on advanced book cover design. I'm struggling on it. Specifically a book cover using reoccurring characters that is easy to edit and rivals pro covers in stores.

[@TokenizedAI]

Honestly, the list of topics I need to cover is already too long 😬

[@areacode3816]

Thanks for the reply. No worries I totally understand. Odd thing for midjourney, it gets the concept of book covers it just too often puts the design on an angled small book (with shadows on it) essentially making it unusable. There are work arounds but often those come with new problems.

[@TokenizedAI]

Just don't reference the actual word "book cover" or use "aerial view". That might help.

[@personone6881]

yeah man nice! but how do you CREATE the THEME in the first place before tryna “INFUSE” it?

[@TokenizedAI]

You don't "create" a theme in this context. You reference one that already exists. It's about infusing well-defined aesthetics/topics.

[@SaidThaher]

Impressive.. as alwayes

[@TokenizedAI]

Thank you! Cheers!

[@kaempf2512]

Hallo Christian, ich versuche bisher ohne grossen Erfolg ein reales Produktfoto in eine surreale Szene zu implementieren. Konkret eine Abwasserpumpe U3K von Jung Pumpen in einem überfluteten Keller mit Regalen, etc. mit einer verzweifelten Person. Ich habe dafür ein Originalprodukt hochgeladen und die Szene beschrieben. Auch mit Wichtungen :: habe ich gespielt. Leider wird die Form der Pumpe jedes mal komplett verändert, auch die Grösse der Pumpe kann ich nicht wirklich beeinflussen. Hast Du hier einen guten Tipp?

[@FreibierDas]

da mj deine prompts erstmal in ein rauschen verwandelt und dann dein bild erstellt ist die chance dafür gegen 0, mMn unmöglich. die pumpe kann im erzeugten bild höchstens als platzhalter für eine spätere bearbeitung mit einem weiteren programm dienen.

[@TokenizedAI]

Wie @FreibierDas schon sagt, ist das nahezu unmöglich. Das würde voraussetzen, dass das KI Modell auch mit Bilder von der besagten Pumpe trainiert wurde und diese auch als solche ausgewiesen waren (inkl. Name des Modells der Pumpe). Für sowas ist das eifnach nicht gemacht. Du wirst nicht umhin kommen, mit Photoshop oder etwas ähnlichem zu arbeiten. Midjourney ist halt keine Bildbearbeitungs-Software.

[@ppovoa1]

amazing, wonder if this works with stable diffusion :)

[@TokenizedAI]

I suspect not in the same way.

[@nonfungibleai]

Is theme infusion the same as "additive prompting" I have been seeing the use of that terminology? Over all great series.

[@TokenizedAI]

No, not really. "Additive Prompting" is a term that is being pushed by Nick St. Pierre a lot. However, I believe Shane McGeehan was among the first to actually coin the term. Overall, the process is more basic that the word suggests. With additive prompting you basically start with an initial prompt of 1 or 2 keywords. And then you keeping adding additional elements one by one and always generate a new set of images with the new word to see what it looks like. It's effectively just an iterative process, rather than stuffing 12 different words in all at once and hoping for the best. Basic process: - editorial style photo, farm house - editorial style photo, farm house, rustic style - editorial style photo, farm house, rustic style, surrounded by lush green grass - editorial style photo, farm house, rustic style, surrounded by lush green grass, golden hour sunlight That's additive prompting. The major difference is that theme infusion requires the use of multiprompts and it's not mean to "build out" a very specific result. Think of it like using a filter on an image, except that the filter doesn't just change the color and look, but it also adds tons of objects into the image.

[@nonfungibleai]

@@TokenizedAI thank you for that comprehensive answer.

[@cazfood]

I will Google "theme infusion" and send them bloggers a notice to cite you 😆

[@TokenizedAI]

Haha 😅

[@whyfollowmenow]

Does changing the chaos to a low number help?

[@TokenizedAI]

Help with what exactly?

[@mahasalbi1086]

I have theme that I like, but no artist is entitled to it, and this theme has no name. Is it possible to create theme based on an image I have? Thanks for the response, it will be so much helpful.

[@TokenizedAI]

Partially. It depends on what's in the image that you're using as a reference. The problem is that not only the style but also the actual concept of the image is used and blended with your prompt. It's usually significant enough to alter the actual image you want to create. It's usually better to find a combination of words that leads to the desired style, rather than using an image reference.

[@sceltadelcamper]

Perfect as usual. Why don't you create a serie about creating a 2d character sprite sheet for games with walk cycle, idle cycle, ecc...?

[@TokenizedAI]

If you knew how many topics I still need to cover 🙈

[@sceltadelcamper]

@@TokenizedAI I can imagine, but when you will have time it could be a good new direction! I can't find anywhere how long my prompts last in my basic account. Do you know if they are deleted after a fixed period?

[@TokenizedAI]

@@sceltadelcamper I don't think your prompts or images are ever deleted. Quite the opposite, they are probably used to further train the models.

[@Michael-sl2mm]

Sometimes I have very good image but something is wrong. Damaged eye or 6 fingers. Do you know how to fix it? I use Photoshop

[@TokenizedAI]

You can try to fix it by using DALL-E's inpainting feature, but right now, there's no simple fix.

[@paoloferri8680]

This series is amazing! I am watching it multiple times and I replicated / tweaked some techniques and I am writing it on my blog, I wrote you also an email, I have created an amazing character, also using the "white background" and "concept art character" expressions cheers!

[@TokenizedAI]

I'm traveling and visiting my parents right now, so I may not be able to respond to emails until early next week.

[@Xeronimo74]

Interesting. But is the word 'theme' really necessary? Doesn't seem to mata real difference?

[@TokenizedAI]

I haven't tested this much. But I'm pretty sure it does make a difference in certain cases. For example, "jungle tribe" and "jungle tribe theme" may give you different results because the former is more specific to an actual subject.

[@nightai]

Change speed to 1.25. Thank me later.

[@TokenizedAI]

😂 True. Everyone at their own speed I guess 😁

[@nightai]

@@TokenizedAI Time is money. Anything for a more relaxed pizza feasting 😅

[@Ǿœ0œǾ]

Too bad Midjourney ain't free to use.

[@TokenizedAI]

Why would it be? It's a high-end product and GPUs aren't cheap.

[@andrewcampion8200]

I asked ChatGPT to write you a script for a cartoon. Can you make Chapter 7 an actual story where you put this all together??? please! INT. MILITARY BASE - DAY Carla, a beautiful red-haired woman, wearing skin-tight khaki pants, boots, a brown jacket, and a tight white t-shirt, is standing in front of a tank. She looks serious as she inspects the tank. CARLA: (to herself) "This tank is going to be the key to our mission." Suddenly, her radio crackles to life. RADIO: "Carla, we need you in the jungle. There's been a disturbance." Carla jumps into the tank and speeds off towards the jungle. EXT. JUNGLE - DAY Carla emerges from the tank in the middle of the jungle. She looks around, trying to figure out what the disturbance is. CARLA: (to herself) "What could be causing such a disturbance in the jungle?" She spots a group of monkeys throwing bananas at each other. She can't help but smile at the sight of the playful animals. Her joy quickly turns to sadness as she realizes that this is not the disturbance she was sent to investigate. CARLA: (to herself) "I was hoping for something more serious." Suddenly, she hears the sound of gunfire in the distance. She rushes towards the sound and finds a group of rebels attacking a nearby coffee shop. Carla's anger boils over as she sees the destruction and chaos. CARLA: (to herself) "How dare they attack innocent civilians and destroy property!" Carla quickly takes action, using her tank to drive off the rebels and protect the coffee shop. INT. COFFEE SHOP - DAY Carla sits inside the coffee shop, sipping on a cup of coffee. She looks out the window at her tank parked outside, feeling a sense of accomplishment. CARLA: (to herself) CARLA: (to a fellow soldier) "I never knew a tank could be the solution to all my problems." The camera zooms out as Carla continues to enjoy her coffee, leaving the audience with a sense of satisfaction and accomplishment.

[@TokenizedAI]

ROFL! 😂 Listen, I'm sure I'll do that at some point, but I need to take a bit of a break from Carla. You know, like most relationships need every now and then. 😆