BUILD SCENES with Consistent Characters in Midjourney

[@chadlight5066]

I am a subscriber. Your series is the best on KZbin I've seen on this topic, which for many of us is critical to production development. Well done sir.

[@TokenizedAI]

I really appreciate your comment. Thank you 🙂

[@dmitryboldyrev7364]

Please make more videos about consistent illustration. Especially how to add multiple characters and fill the scene with background characters etc. Its really interesting.

[@chariots8x230]

12:53 I love that Midjourney added the option to change the aspect ratio, but it would be helpful if we could use our own custom backgrounds (based on reference images that we provide), and place our characters within those custom backgrounds. For projects where we need a lot of control over the scene that we create, it’s important to not only have custom characters, but also custom backgrounds, and maintain consistency of these elements. When combining our characters with a background, we may want to show the scene from different camera angles, and change things like the lighting conditions (or weather conditions), but it should still maintain the same consistent look even when these changes are made.

[@jipjanneke11]

Yay! Carla is back! I love her, please continue this series further 😊

[@chariots8x230]

Thanks for sharing! 🙏🏻 Videos about consistent custom characters & building scenes are always appreciated.

[@nicolasleblanc5722]

Your content is phenomenal. I love exploring the mysterious world of Midjourney with you. I'm glad I purchased your Master of Midjourney course. Keep going!

[@IronCladMedia]

Fabulous, Chris, just fabulous. You have a way of knitting out the best of what Midjourney can and cannot do. Thank you for your generosity!

[@TokenizedAI]

You are so welcome! 😊

[@hocPhatlasuhuongthucaonhat]

Your videos are really stand out from the rest. You are responsible for what you say. Awesome!

[@chariots8x230]

15:35 I’m very interested in combining multiple characters together in a scene, since this is usually important for creating a full scene. But I don’t want to add the same character twice in a scene because there’s usually no point in doing that. Instead of adding a random male character to the scene like you did earlier in the video, would you be able to add a custom male character instead? I’d like to know if it’s currently possible to add multiple of my custom characters into a scene. Also, I want to be able to create interactive poses with my characters. Instead of my characters being placed in different areas of the image, I may want them closely interacting with each other. Some examples where the characters might need to closely interact is if they’re engaging in combat sports like wrestling or martial arts, or if the characters are hugging each other (or if they’re holding hands). This is something I’m hoping will be possible soon, because it would be really helpful in creating scenes for comics and graphic novels.

[@blsemetan7232]

For the shoulder strap problem or clothing etc. I've been editing the head references in Photoshop... Photoshop all of them with the same shirt, jacket, color, etc... then upload them to Midjourney for reference.

[@TokenizedAI]

Yeah, you can do that. But in this case I specifically wanted to reuse the references from the previous video, without any changes.

[@blsemetan7232]

@@TokenizedAI Understood, keep up the great vids! 👍

[@EiranBailey-fl1zi]

have you tried using the --no shoulder straps parameter

[@agnesslovehealz]

Exactly yup

[@agnesslovehealz]

And can use a remove prompt I think too

[@KriGeta]

amazing tutorial but i have created my character by hand, is there any way to create consistence poses with that?

[@TokenizedAI]

Tricky with an external image

[@KriGeta]

@@TokenizedAI if possible, is there a way to make a similar character like 90% close to the original?

[@davidmichaelcomfort]

Really good video. I was able to create consistent characters using Vary Strong by using an image prompt of the original image and then weighting it by say 0.25 in the remix. I wrote up a Medium Post about it: "Generating Consistent Characters in Midjourney v5.2".

[@TokenizedAI]

The new Vary-Strong in combination with Remix is indeed quite a game changer.

[@Achievly]

Videos are excellent man, one of the few that really provides detailed repeatable steps/templates!

[@TokenizedAI]

I appreciate that! 😊

[@ReverendInsanityAudiobook]

This is filled with information that I did not know. Much appreciated. This is going to up my game so much!!!!

[@TokenizedAI]

Great to hear 😁

[@giri.goyo_yt]

Ah. You continue to help my skills grow. Thank you so much. Keep rockin' it.

[@TokenizedAI]

Thanks, will do! 🙂

[@6daysoflight]

Really smart! Saves me years of lifetime👏👏👏 so remix is the key feature in this workflow.

[@TokenizedAI]

Thanks! And I also think this only scratches the surface 🙂

[@TheTimedCarbGuy]

Awesome. The recap at the end was also helpful, thanks :)

[@jimberry7865]

Top notch!!! Ultra clear and organized.

[@TokenizedAI]

Much appreciated!

[@InternetInitiatives]

I just purchased your course and I must say bro you are INSANE the work you have put in is seriously the best I have seen in a while appreciate you thank you for your efforts!!!

[@TokenizedAI]

Thank you so much. That was always my plan. Over-deliver on the promise 🙂 Was a ton of work though too 😆

[@AmAl-km7tc]

Hi, Could you tell me what I would get from purchasing your course versus just watching your series of youtube videos? Thanks

[@TokenizedAI]

The landing page is pretty clear on this.

[@Kratos30000]

You are amazing! Thank you so much for your hard work! ❤️ btw, appreciated the biographical video a lot 🙌

[@TokenizedAI]

Glad you enjoyed it! 🤗

[@ATLJB86]

Carla is back baby!

[@TokenizedAI]

🦾💃

[@livingandtraveltoday]

Chris… I just published my first children’s book because of you! THANK YOU. I spent hours watching and implementing what you were teaching. Just saw your course and as soon as I have some bucks this week, I’m all in. Thank you!

[@TokenizedAI]

That's fantastic news! 😁 Love to hear stories like this. I know you guys are out there, I just rarely get to see the stuff you make.

[@thesupportcrew1]

This is so amazing! Thank you very much Christian!

[@TokenizedAI]

My pleasure

[@maxwell-cole]

Definitely awesomeness for the detailed instructions 🙌🏽. Thank you for sharing this.

[@TokenizedAI]

You are so welcome!

[@timothypruitt4578]

Another great video. Thank you, Christian.

[@pc-bh2in]

Thanks for making this Christian. I have a question please. When using "pan" feature, is the entered prompt only applying to the newly created pan area, or the new area + the original area? I asked this because you didn't mention the girl at 11:30 in the prompt when using panning. I hope you understand what I'm asking. Thanks for creating this tutorial.

[@TokenizedAI]

Technically, with pan you should only be entering what's in the new area. But sometimes it also works with the original prompt.

[@pc-bh2in]

@@TokenizedAI I see, thanks for your reply Christian.

[@agnesslovehealz]

The reintroducing the character is cool can crop and then use canva as welll and crop thr characters and making scneses alone and place but yes its unreal the control and ability we have w prompts and consistency as well as quality is mind blowing

[@AnnIsabelle76]

As always, another great video. Thanks for sharing.

[@TokenizedAI]

You bet

[@mbcase]

Using red hair is a common cheat for "consistent" characters. Because there are fewer redhead face examples that Midjourney has sampled, and the hair distracts the mind from facial differences. I challenge anyone to create a consistent character that isn't a red haired woman. Or a pirate with an eyepatch, which is impossible! Try to create just one image of a pirate with an eyepatch!

[@TokenizedAI]

Looking for the hair in the soup. The eyepatch issue also has nothing to do with consistency. That's a coherence problem.

[@jamesmcdonald1724]

Very cool. I’m the beginning, where are you getting the links for the reference images? The discord links are super long normally

[@bostjandervaric454]

Always the best. Thank you very much 🙏

[@TokenizedAI]

So nice of you

[@BaysideGraphics]

Wondering if your methods would work on objects rather than characters. I am trying to created multiple views of non-characters to use in camera shots for A.I. video. Would love to know more! Thanks.

[@TokenizedAI]

Largely depends on how uniquely distinguishable the object is and whether you have a base prompt that reliably produces similar images.

[@nathanreetz]

Good stuff, thank you for all of this info!

[@TokenizedAI]

You bet!

[@thewebstylist]

Awesome series brotha 🎉

[@TokenizedAI]

🕺🦾

[@olexiisokolov]

Hello. Will this work not with people, but with animals, aliens, monsters? cartoon characters

[@TokenizedAI]

Why wouldn't it? Only one way to find out :)

[@COLORINGBOOK888]

Thanks for the information it truly helps: I’m Corey nice to meet you

[@TokenizedAI]

Nice to meet you too, Corey 🙂

[@achewlim]

i have a question, if we use ref like that with white background, most of the results will return us white background too, how can we fix that? i tried remix with ref removed, still getting back the white bg

[@TokenizedAI]

You need to use "Vary (Strong)" and you obviously need to replace "white background" with some sort of environment. If you're using Remix with low variation mode activated, then it obviously won't work because that's an entirely different type of Remixing. I'm pretty sure I show these steps with quite a bit of detail 😉

[@thisearth1443]

I will men I will . Y really killing it 🎉

[@Best_SS]

Okay, I give up! Great content but I've watched a dozen tutorials including yours on how to change the expressions and still create consistency and my efforts have all ended in tragedy. What am I doing wrong?

[@JClaytonBrown]

I don't understand how to add multiple image references in the same prompt. Your example shows 8 image references, and then a prompt. How do you do that??

[@TokenizedAI]

Just put them next to eachother and leave a space. /imagine

[@AG_before]

Nicely done! Thank you. 👍

[@TokenizedAI]

Thank you!

[@chariots8x230]

19:14 It will be great when Midjourney finally decides to introduce inpainting 😊

[@TokenizedAI]

Soon

[@chariots8x230]

@@TokenizedAI Hopefully :)

[@0Spiritus0]

They'll need to. Adobe is becoming more of monster (in a good way) than ever before. Definitely want to see Midjourney reach greater heights.

[@agnesslovehealz]

Overall really great video thanks

[@filbertflubberfish9541]

where can i get a book of the midjourney language to program

[@TokenizedAI]

Amazon?

[@davidcummins8125]

I'm wondering whether there is a workflow using character sheets. I've played around with generating comic character sheets, e.g. 4x4 grid of the same character with different expressions or varying by some other feature. I find that the items in the grid are reasonably consistent. The shame is that you can't start with a single item from the grid and outpaint from that (easily anyway).

[@TokenizedAI]

Yeah, like a "crop" feature that you can then expand again. That would be cool.

[@davidcummins8125]

@TokenizedAI Exactly. It would be easy to express in a prompt too, e.g. --crop top%,left%,width%,height%.

[@0xFreeWill]

Is there any way to reference a scene instead of a character? --sref just copies the style. What if i want variations of the same scenery? Like my house for example, same house but maybe it looks older.

[@JClaytonBrown]

Also, how are your image references so small? Mine are enormous.

[@TokenizedAI]

I'm not sure I understand what you mean by "small". If you mean the shortened URL, those are created by Midjourney itself, once you've entered a command that contains an image prompt.

[@assortedmusings1281]

how do you get that "copy prompt" button?

[@TokenizedAI]

Search for the chrome extension "Promptalot Helper".

[@manfrombkk]

Yes, these new features open up another level/compartment of pandora’s box. And this is only the tip of the iceberg: V6 and beyond…….

[@marisanoelle6599]

If I want to put a second character i've createdin midjourney next to the first - how do I do that?

[@controlaltai]

Face swap discord bot from insight face is a simple solution for consistent characters. Appreciate your hard work on the videos though.

[@batidavital]

from the midjourney channel?

[@KycokOv4arku]

@@batidavital another dc bot, could work with midjourney images)

[@TokenizedAI]

Yeah, I see that more as complementary. Specifically for those who are incredibly picky about facial details.

[@controlaltai]

@@TokenizedAI I am acquainted with one of the devs who created the technology. I have made a video tutorial on it. It has some limitations but works on realistic photos as well as anime style / cartoon characters. If you see the tutorial you will get how it works and why it works and it's basically the only way to get very accurate results in different scenes.

[@GillBrooks]

Thank you :)

[@TokenizedAI]

You're welcome!

[@MichaelFlynn0]

incredible stuff.

[@srikantdhondi]

How to do the same in Stable diffusion?

[@TokenizedAI]

Outpainting, I guess?

[@agnesslovehealz]

And if the straps annoy anyone can remove in other programs

[@TokenizedAI]

True. But many people struggle with this.

[@lgrindrod3474]

Thank you!!!❤❤❤❤🎉🫂

[@TokenizedAI]

😁

[@manfrombkk]

Be interesting to see what would happen if one tried to duplicate your process, except using the current beta of the web version of Adobe Firefly

[@Author_SoftwareDesigner]

Use this incomvination with lasso and cutting and pasting characters in photoshop as well as photoshop's generative fill.

[@TokenizedAI]

Probably would be similar, except that the image quality might not be as good as MJ.

[@manfrombkk]

@@TokenizedAI agreed

[@diegognoatto8706]

twitching my brain in MJ i had an idea for your next video if you can crack the code to do this, i couldnt, i tried to get the same image made in daylight "it was a website background, so i thought would be cool to sticth the image when toggles dark-mode" the idea i reemagine the same image but in night time. i let the chalenge here :D

[@TokenizedAI]

Oh boy, I think that would be pretty tough to be honest

[@derhitpula]

Have you meanwhile tried vary custom feature to get two or more Consistent Characters?

[@TokenizedAI]

I think you mean Vary Region?

[@derhitpula]

@@TokenizedAI Yes, correct. Any Success?

[@TokenizedAI]

Midjourney's inpainting is messed up and doesn't do what it should in at least 30% of the time. It does certain things really well and completely fails at many others. It needs a lot of work, to be honest. But yes, face swapping is something it's relatively good at.

[@mikhah4828]

When I clicked pan to, the image is automatically panned without allowing me to edit. I can't add modification to the prompt. Help

[@TokenizedAI]

That's because you didn't activate Remix Mode.

[@sachindatt4045]

I am all ears!

[@TheGurthang99]

I say this with the deepest respect and only wanting to help you. In the beginning of the video, you show Carla in poses and pictures, but those do not happen later in the video. This is confusing for me. I was expecting to see Carla in a tavern playing a guitar being produced by a prompt.

[@TokenizedAI]

The video would have ended up being 45 minutes long if I showed every single example. Many wouldn't watch the video then. The approach is always the same. The only thing that changes is the image that you want to create. Start with the head, pan the body (and choose an activity/pose if you must), custom zoom out a bit, pan the environment, custom zoom out the full scene. No need to repeat the same thing multiple times.

[@reycaribe]

Maybe the word European is illiciting the shoulder straps. Think Lederhosen.

[@TokenizedAI]

lol, I never thought of it that way but I think you might be on to something 😆

[@alperklc7782]

Good day. I wish you would put your 'christian heidorn midjourney' course on the udemy platform so that we can buy it from there. In countries like ours, 200 dollars is a huge amount for us. It is way beyond our purchasing power. We would be grateful if there was a discounted version for us. We would also like to benefit from the experience and knowledge of artist mentors like you. Best regards.

[@zitrandy]

WAY TOO fast! Confusing.

[@zitrandy]

To fast, talking too fast. Very confusing. Makes me feel like giving up

[@erikfadiman]

Thanks Chris! This is so helpful!

[@TokenizedAI]

My pleasure!