00:00 🖌️ The challenge in Midjourney is generating consistent characters across different scenes. 00:29 🚫 No need for Photoshop or Dream Booth model; the method is entirely within Midjourney. 00:43 🌟 Learn to seamlessly insert character "Elara" into any scene regardless of pose, lighting, or composition. 01:38 📝 Writing a simple description as a prompt is sufficient for generating characters. 02:58 ✂️ Use Very Region to erase and modify character parts for better integration. 04:07 🔄 Combine image and text prompts for precise character generation. 05:24 🌈 Consistency lies in key features like chin shape, eye color, and nose size. 06:15 🔄 The method can be applied to any character style or genre. 08:07 🖼️ Create a character design sheet for reference in different poses and styles. 09:32 🌌 Use image prompts to shift character into new settings or scenarios. 11:48 🧐 Fine-tune features like eyes, mouth, nose, and ears for consistency. 15:54 🛠️ Iteration and fine-tuning lead to a consistent and versatile character.
TIP: if you want photo realism character sheets instead of animated, writing a variation of this works: A series of photos of a young woman in various poses, showcasing the same person but in slightly different ways. The woman is [insert description of character]. Once you generate a few times, you'll find a very realistic depiction of what you want, scale and generate variations from there and boom, you have the necessary photos you need to either use as is or save as a model reference.
This is seriously fantastic!! As a Midjourney newbie, this has been incredibly helpful for me, I'm finally starting to be able to create some of my more esoteric / non-human characters more consistently. Thank you, Glibatree!
Glad you enjoyed! Yeah, it took me a bunch of trial and error- but I felt it was only a matter of time after Vary (Region) came out. The feature truly was a game changer.
The "no nonsense, get to the point, not asking for money" energy here has made an instant fan. Yes, everyone needs to make their money especially content creators taking time to help others, but those of us without the money....a profound thank you!
THANK YOU, this is the best walk through of the info I need for creating illustration for my graphic novels. You have just saved me a ton of hours struggling to get the right look. I really appreciate your info and will be checking out your other videos over the weekend. Keep up the great work.
This was a huge help, creating a consistent character has been a challenge for me and stumbling across this video really helped. Yes I liked subscribed and rang the bell so thanks again.
Interesting. I had done the first part by following another tutorial where I created a prompt with various angles. You have taken it to the next level. Thank you for your time and energy and a clear concise video without the usual RUclips junk, jokes or self-indulgence. I will have to try this method tomorrow!
Thanks! It's working pretty good at first attempts. I was able to switch it from a painterly style to portrait pretty easy using this guide. :) Now I'll try to see if I can keep it in the same cartoon style next. Appreciate your hard work!
Bro! Subscribed ! Saved my life! I was struggling to turn artisans into wood and clay bust sculptures ant this tutorial + Midjourney 6 beta = saved my ass...
At last, the tutorial I've been hoping for! Your insights are always game-changing, and your advice never disappoints, and this topic on consistent character development is spot-on. Your perspective on maintaining character consistency in Midjourney is as incredible as I expected Can't wait for the next one! Thanks again 😄👍
I am a beginner and was desperately trying to get the same character in other poses using seeds and images I created before but got stuck.This helped me a lot! Thank you so much!!!! :D
Waiit I literally screamed with excitement when you said to push the like button and the entire outside of the button lit up! HOW did you do that?? Futuristic asf loved it
The issue I have with vary region is that everytime I use it the results are horrible, most of the time eyes will have different size for no reason, other times one eye will look one way and the other eye the other, other times eyes will not look anything like the reference, when changing the face the same will happen, gigantic heads, weird shaped heads, etc same issue for most features, I tried describing the eyes only, mentioning the eyes only, adding the style from the original prompt, removing the style from the original prompt, etc. Maybe it's because I'm using it for 2d images but it's not been fun.
It took "no" time for me to slide my mouse over the "subscribe" button and smash it to smithereens. And I mean.....FAST.....quick.....REALLY speedy. The immediacy with which it was pressed is probably unprecedented, I'm sure. In other words.....thank you for this. You rock!!!
The MJ discord space has an open source ethos of "share your prompts, help other's learn." You gave text of the prompt to create a character sheet within the first five minutes of this video. Your efficiency is much appreciated.
I think it would be cool (and alot easier) if one day, the tech advances to where you can just drag and drop an image of a character that's already in the style you want and you can just simply tell the ai to "have this character doing anime style leaps across rooftops" Click and done
ive beein watching videos about character consistency for hours. This is the first one that actually shows an extremely good process. Also I dont know what you meant with "this video is already long" I would have watched you for 90 minutes at this video / explanation quality. --Subscribe
update. I have midjourney now. My stuff doesnt look photorealistic at all no matter how often i write "Photo of" and "photorealistic picture" I also tried out different modes, raw mode, stylize and everything. It just looks fake as f. Dont know what he is doing differently at all. pretty shtty
I am new to this and using a PC desktop and can't figure out how to do a screenshot on my PC or get my characters into the prefer option set. Can you give me some more insight?
@Glibatree. Do me a favour. Keep making these videos. You just keep doing it as the trends change, and people will start to see you as an authority on the subject. Awesome content - even a year later.
liked the video because you do NOT get pushy about liking the video, and I did, like that vibe haha,... then I stayed cuz the content is great too. Thx for the help, navigating AI prompts is both so simple and so ridiculous at the same time for a new user lol
Hello, great video. around 4:10 you explain how to add a prefer option set into the vary region prompt box, however, when I add my prefer option set I receive an error regarding JSON formatting. Could this be a change in midjourney since you created your video 10 mos ago?
I can't thank you enough, finally great detailed advanced videos! I've been struggling with consistency juggling between /blends, using photoshop and back in, etc. This set up is worth it for the longer projects. Thanks again. What drives me a bit nuts is how MJ wants to constantly have the character look directly into "the camera", so 99% of images look like great rendered portraits. Getting consistent different angles would be amazing. Any tips?!
It's really important to reference the pose and angle separately and clearly in the prompt. My video about my Photorealistic Formula - as pretty practical insights about this
the only problem is the character overall isn't consistent. I hope midjourney some day will make so that they can make consistent characters. I tried to make myself with real pictures and just like the redhead in your video, it would make look-alikes with small but significant face trait differences :( btw if anyone ever wants a real person to make characters from youre welcome to make/use my photos for free, even earn money on it, just as long as you don't pretend to be me :)
Anytime I'm attempting to use a preferred option in a prompt it looks to be coming up as an unrecognised parameter specifically when referenced in the varied region section. Everywhere else it looks to work fine. Anyone else get that too?
Confused. I'm new to midjourney and I don't have Vary region. Just Vary subtle and vary strong, and it doesn't allow me to do what you demonstrate. Do I need to upgrade to a higher version of midjourney?
Thanks for the video. Honestly to me it's a bit confusing though. As if some of the intermediate steps have been skipped. If you could do this with a more step by step approach it would be super viral. Thanks in any case.
I'm having a little trouble. I'm able to set the preferred option, but once I try --the name, I get an error message saying that there is an unrecognized parameter : my preferred option name. Any idea what might be going wrong?
My issue with this process is I when I collect the most representative photos of my character, and try to set them with the /prefer option, the output photos look nothing like my input photos, and I have to start over. So it's not iterating forward, but reverting to something away from my input photos.
This is simply not working for me, the Midhourney Bot says that the custom option is now set - I give it the URLs from the images, a short description - then when I ask it to imagine the character name I used for the custom prefer option set - it generates seemingly random stuff, it's like - what I've done has has absolutely no effect. I'm so confused. I've watched, re-watched, taken notes and tried time and time again... really frustrating as this is an amazing video - I thought I understood and was following everything step by step...
I FIGURED IT OUT. Okay - so, before this video I had watched one where the guy said MidJourney doesn't care about capitalization - WELL. In this case, make sure you DO write in your characters name exactly as you typed it in when you first taught it. And "--character name", not "character name" Hopefully this might help someone else x
8:40 and.... i completly lost it and dont know what to do in this place, tryed to search on google but i copleatly do not understand what should i do, how to screenshot and post it back
when I add the --"preferred option set name" to the vary region prompt, I am getting an "unrecognized parameter error." Is this something new to version 6.0? I have followed everything to the tee and have created a nice parameter option set that will generate a pretty consistent character, but I can't seem to use it in the vary region.
I get the same and before a few days I tried it and it worked pretty fine. Idk if it is different on phone or pc. Bit it is very frustrating now that it doesn't recognize the parameters I set
Great job on your video! Heard of Stylar AI? It's a web tool for logo, 3D, and interior design, with image upload and auto prompt capabilities. Still evolving, but could enhance your creative process.
At 7:20, you mentioned being able to replace and update the images in your /prefer option. How do you do that? Do you just add a new URL image to the "value" field after typing the character's name in the "option" field, or do you have to paste all URL images again?
I just ran a test and this does not seem to work with photos that are not generated by MidJourney. I wanted to see if I could use this method to make some photosets of my boss, who is a millionaire who's NEVER in the office, but we use his photo on marketing material, so I thought I'd try generating some photos of him doing various things. In my tests, it works when using images generated from MidJourney, but it doesn't get his likeness when I use this method with photos of him. MidJourney generates a face that kinda looks like him, but it's the same face that MidJourney has generated before. The faces from this method aren't even close to the reference either. Pink hair, beard/no beard, sunglasses, etc. it's borderline random.
00:00 🖌️ The challenge in Midjourney is generating consistent characters across different scenes.
00:29 🚫 No need for Photoshop or Dream Booth model; the method is entirely within Midjourney.
00:43 🌟 Learn to seamlessly insert character "Elara" into any scene regardless of pose, lighting, or composition.
01:38 📝 Writing a simple description as a prompt is sufficient for generating characters.
02:58 ✂️ Use Very Region to erase and modify character parts for better integration.
04:07 🔄 Combine image and text prompts for precise character generation.
05:24 🌈 Consistency lies in key features like chin shape, eye color, and nose size.
06:15 🔄 The method can be applied to any character style or genre.
08:07 🖼️ Create a character design sheet for reference in different poses and styles.
09:32 🌌 Use image prompts to shift character into new settings or scenarios.
11:48 🧐 Fine-tune features like eyes, mouth, nose, and ears for consistency.
15:54 🛠️ Iteration and fine-tuning lead to a consistent and versatile character.
Pinned for easy access ♥️, thanks man
TIP: if you want photo realism character sheets instead of animated, writing a variation of this works: A series of photos of a young woman in various poses, showcasing the same person but in slightly different ways. The woman is [insert description of character].
Once you generate a few times, you'll find a very realistic depiction of what you want, scale and generate variations from there and boom, you have the necessary photos you need to either use as is or save as a model reference.
This is seriously fantastic!! As a Midjourney newbie, this has been incredibly helpful for me, I'm finally starting to be able to create some of my more esoteric / non-human characters more consistently. Thank you, Glibatree!
Thanks! Please keep up the good work!
Much appreciated! Thanks so much
Wow, thank you for the detailed instructions. I would never of figured this out on my own. Too cool!!
Glad you enjoyed! Yeah, it took me a bunch of trial and error- but I felt it was only a matter of time after Vary (Region) came out. The feature truly was a game changer.
The "no nonsense, get to the point, not asking for money" energy here has made an instant fan. Yes, everyone needs to make their money especially content creators taking time to help others, but those of us without the money....a profound thank you!
Oh that is very kind, thank you for watching and I hope it helps 🙏
your comment made me realize what he had done, and i did the same.
THANK YOU, this is the best walk through of the info I need for creating illustration for my graphic novels. You have just saved me a ton of hours struggling to get the right look. I really appreciate your info and will be checking out your other videos over the weekend. Keep up the great work.
just be aware that no image created this way is under copyright. The words you made, yes. The image no.
This is actually the most valuable midjourney video I have seen! I gonna give it a try... A lot
Holy cow, you figured it out man. Thanks!!! The best method I’ve seen this far!
Awesome man! I hope it leads to some great results for you :)
Man, I had zero idea that "prefer" was a thing, and I've been messing with midjourney since v1. thanks for the info dude!
SAME i was looking for this option this whole time
thank you! this definitely will improve my image gen. honestly didn't know this was a possible technique! Thanks again for the help
Awesome! Good luck generating your characters, I hope it goes well!
This was a huge help, creating a consistent character has been a challenge for me and stumbling across this video really helped. Yes I liked subscribed and rang the bell so thanks again.
Interesting. I had done the first part by following another tutorial where I created a prompt with various angles. You have taken it to the next level. Thank you for your time and energy and a clear concise video without the usual RUclips junk, jokes or self-indulgence. I will have to try this method tomorrow!
Yes sir!!!! this is the whole reason I started the Discord journey, and I highly appreciate you spreading this knowledge can't wait to get started.
He has another video explaining how to use the prefer option set ❤
Thanks! It's working pretty good at first attempts. I was able to switch it from a painterly style to portrait pretty easy using this guide. :) Now I'll try to see if I can keep it in the same cartoon style next. Appreciate your hard work!
Amazing to hear! Glad you were able to get started with this so quickly. Curious for updates as you keep honing in on your character
Bro! Subscribed ! Saved my life! I was struggling to turn artisans into wood and clay bust sculptures ant this tutorial + Midjourney 6 beta = saved my ass...
Damn it. Straight to the point. Speaks fast and answers what I had searched. Thank you!
This is a great tutorial! Please keep sharing these insights with us so we can all get better with our creations!
Been searching for THIS video for a year. Thanks, excellent content.
Yes, finally a video showing multiple steps to a character that can be used over and over, thanks so much Glibatree.
Thank you so much for teaching this technique! It is easy and produces fantastic results.
At last, the tutorial I've been hoping for! Your insights are always game-changing, and your advice never disappoints, and this topic on consistent character development is spot-on. Your perspective on maintaining character consistency in Midjourney is as incredible as I expected Can't wait for the next one! Thanks again 😄👍
Wow, you warm my heart! I'm glad I didn't disappoint. Good luck with your characters!!!
a great tutorial! prefer option set is a powerful tool, and I have created some amazing characters with a consistent look! Thank you!
THANK YOU! I've been looking for a method to do this for months!
I am a beginner and was desperately trying to get the same character in other poses using seeds and images I created before but got stuck.This helped me a lot! Thank you so much!!!! :D
Love all this exposure, thanks so much!
Instant subscribe! Thank you for your insights. Going to watch this a few times until the info sticks
This is very helpful for someone who does have a lot of images of a character already but enjoys putting them in other scenes. Thank you!
Thanks for not doing click bait. People are already interested in this subject - no need for that. Subscribed and thumbs up - as requested.
Amazing! Thank you so much for sharing. All very clear and explained at a good pace. Love the characters you've created :)
I can barely wait to try this. Thank you so much!
Brilliant! If this works, it can really solve a problem for me. Thanks. Can't wait to try it
Definitely subscribing, this will help a lot for my channel as I need to keep the characters consistent with my story animations.
Incredible! Thanks for sharing this amazing method!
Waiit I literally screamed with excitement when you said to push the like button and the entire outside of the button lit up! HOW did you do that?? Futuristic asf loved it
The issue I have with vary region is that everytime I use it the results are horrible, most of the time eyes will have different size for no reason, other times one eye will look one way and the other eye the other, other times eyes will not look anything like the reference, when changing the face the same will happen, gigantic heads, weird shaped heads, etc same issue for most features, I tried describing the eyes only, mentioning the eyes only, adding the style from the original prompt, removing the style from the original prompt, etc. Maybe it's because I'm using it for 2d images but it's not been fun.
It took "no" time for me to slide my mouse over the "subscribe" button and smash it to smithereens. And I mean.....FAST.....quick.....REALLY speedy. The immediacy with which it was pressed is probably unprecedented, I'm sure. In other words.....thank you for this. You rock!!!
The MJ discord space has an open source ethos of "share your prompts, help other's learn." You gave text of the prompt to create a character sheet within the first five minutes of this video. Your efficiency is much appreciated.
Great video! What a rabbit hole this will be!
Thank you, Doctor Octavius, for this amazing advices
Wow, this is a huge help. Thank you for posting this.
Loved this tutorial! Can´t wait to try it out :) Thank you!!!
This is awesome! And your method of teaching deserves an A+++. Thank you.
Appreciate it, thanks so much!
Great video bro keep them coming!
Thanks!
I am definitely doing something wrong. I keep getting an error that my custom option is "removed". Help?!
I think it would be cool (and alot easier) if one day, the tech advances to where you can just drag and drop an image of a character that's already in the style you want and you can just simply tell the ai to "have this character doing anime style leaps across rooftops" Click and done
THANK YOU! Very helpful video. Will need to rewatch again for sure though lol
Thank you very much. This is a watershed video.
Great Vibes and awesome tutorial
ive beein watching videos about character consistency for hours. This is the first one that actually shows an extremely good process. Also I dont know what you meant with "this video is already long" I would have watched you for 90 minutes at this video / explanation quality. --Subscribe
update. I have midjourney now. My stuff doesnt look photorealistic at all no matter how often i write "Photo of" and "photorealistic picture" I also tried out different modes, raw mode, stylize and everything. It just looks fake as f. Dont know what he is doing differently at all. pretty shtty
This is sooooo amazing. Thank youuuu!.
Incredibly helpful. Thank you!
Liked your vibes. And the video! 😊👍🏽
My new Manga project will start in 3, 2, 1...
Wow! This is a whole new world.
Learned something new! Thank you.
Fantastic! Thank you for sharing , I appreciate it a lot !
Best video on this subject, period.
Great content, subscribed!!
Brilliant tutorials as always! Thank you ❤
Thank you! Cheers!
I am new to this and using a PC desktop and can't figure out how to do a screenshot on my PC or get my characters into the prefer option set. Can you give me some more insight?
Subscribed because of this. 🚀🙌🏽
I did not get how he put the urls back to midjurney
Wow, this technique is really impressive! It kind of reminds me of something I saw on CGDream
@Glibatree. Do me a favour. Keep making these videos. You just keep doing it as the trends change, and people will start to see you as an authority on the subject. Awesome content - even a year later.
Very kind of you thanks!
liked the video because you do NOT get pushy about liking the video, and I did, like that vibe haha,... then I stayed cuz the content is great too. Thx for the help, navigating AI prompts is both so simple and so ridiculous at the same time for a new user lol
subbed at the end, cuz I like the vibe and content. Excellent work. Thx again for the help, and some inspiration. Ill be watchin.
Hello, great video. around 4:10 you explain how to add a prefer option set into the vary region prompt box, however, when I add my prefer option set I receive an error regarding JSON formatting. Could this be a change in midjourney since you created your video 10 mos ago?
i never thought i would be getting midjourney lessons from Ray Romano
I can't thank you enough, finally great detailed advanced videos! I've been struggling with consistency juggling between /blends, using photoshop and back in, etc. This set up is worth it for the longer projects. Thanks again. What drives me a bit nuts is how MJ wants to constantly have the character look directly into "the camera", so 99% of images look like great rendered portraits. Getting consistent different angles would be amazing. Any tips?!
It's really important to reference the pose and angle separately and clearly in the prompt. My video about my Photorealistic Formula - as pretty practical insights about this
Thank you SO MUCH for this!
the only problem is the character overall isn't consistent. I hope midjourney some day will make so that they can make consistent characters. I tried to make myself with real pictures and just like the redhead in your video, it would make look-alikes with small but significant face trait differences :( btw if anyone ever wants a real person to make characters from youre welcome to make/use my photos for free, even earn money on it, just as long as you don't pretend to be me :)
Anytime I'm attempting to use a preferred option in a prompt it looks to be coming up as an unrecognised parameter specifically when referenced in the varied region section. Everywhere else it looks to work fine. Anyone else get that too?
Nice stuff Gilba
Thank you thank you, my esteemed developer friend
@@glibatree aw shucks.
Every time I watch one of your videos, I walk away more informed
Glad you feel that way, it's very much the goal! This was also a really fun one to work on
Confused. I'm new to midjourney and I don't have Vary region. Just Vary subtle and vary strong, and it doesn't allow me to do what you demonstrate. Do I need to upgrade to a higher version of midjourney?
Thanks for the video. Honestly to me it's a bit confusing though. As if some of the intermediate steps have been skipped. If you could do this with a more step by step approach it would be super viral. Thanks in any case.
Great tutorial. Thanks!!
I'm having a little trouble. I'm able to set the preferred option, but once I try --the name, I get an error message saying that there is an unrecognized parameter : my preferred option name. Any idea what might be going wrong?
🔥🔥Awesome, thx for sharing 😎👍
My issue with this process is I when I collect the most representative photos of my character, and try to set them with the /prefer option, the output photos look nothing like my input photos, and I have to start over. So it's not iterating forward, but reverting to something away from my input photos.
that's happening to me too.
This tutorial is amazing!
Excellent guide!
How do you get the image URLs? Someone please explain this part for me :)
unrecognized parameter(s): elara (or whatever)
I got an syntax error: unexpected token…not valid JSON after the first step. Does this still work?
Suuper good video explanation, thank you so much! You kind of remind me of Ted Mosby :D
This is simply not working for me, the Midhourney Bot says that the custom option is now set - I give it the URLs from the images, a short description - then when I ask it to imagine the character name I used for the custom prefer option set - it generates seemingly random stuff, it's like - what I've done has has absolutely no effect. I'm so confused. I've watched, re-watched, taken notes and tried time and time again... really frustrating as this is an amazing video - I thought I understood and was following everything step by step...
I FIGURED IT OUT. Okay - so, before this video I had watched one where the guy said MidJourney doesn't care about capitalization - WELL. In this case, make sure you DO write in your characters name exactly as you typed it in when you first taught it. And "--character name", not "character name"
Hopefully this might help someone else x
i have this same problem. i'm putting in the same name exactly as when i did the prefer option set but it's still coming up weird stuff
very helpful. thank you!
8:40 and.... i completly lost it and dont know what to do in this place, tryed to search on google but i copleatly do not understand what should i do, how to screenshot and post it back
when I add the --"preferred option set name" to the vary region prompt, I am getting an "unrecognized parameter error." Is this something new to version 6.0? I have followed everything to the tee and have created a nice parameter option set that will generate a pretty consistent character, but I can't seem to use it in the vary region.
I get the same and before a few days I tried it and it worked pretty fine. Idk if it is different on phone or pc. Bit it is very frustrating now that it doesn't recognize the parameters I set
Glad I’m subscribed to you
Great job on your video! Heard of Stylar AI? It's a web tool for logo, 3D, and interior design, with image upload and auto prompt capabilities. Still evolving, but could enhance your creative process.
Great technique works great thanks
How do you add 2 different characters in 1 shot
Thank you 1000 million times ^^
Love your vid ❤
great stuff! always
Thank you kindly!
At 7:20, you mentioned being able to replace and update the images in your /prefer option. How do you do that? Do you just add a new URL image to the "value" field after typing the character's name in the "option" field, or do you have to paste all URL images again?
I just ran a test and this does not seem to work with photos that are not generated by MidJourney.
I wanted to see if I could use this method to make some photosets of my boss, who is a millionaire who's NEVER in the office, but we use his photo on marketing material, so I thought I'd try generating some photos of him doing various things.
In my tests, it works when using images generated from MidJourney, but it doesn't get his likeness when I use this method with photos of him.
MidJourney generates a face that kinda looks like him, but it's the same face that MidJourney has generated before.
The faces from this method aren't even close to the reference either. Pink hair, beard/no beard, sunglasses, etc. it's borderline random.