Thank you very much for this graduate-level lesson on character consistency. I've never seen anything like it before, after searching for over a year on this topic. I've subscribed to your channel
It's pretty good. The only issue I've experienced is that it's really nice hard to get the character to look away from the camera, or to be active in the scene. When you start with a consistent portrait style, all the subsequent prompts tend to be portraits as well. So, not the best for storytelling, but otherwise pretty neat.
I've had the same issue, the character tends to default to staring into the camera. Had moderate success with using the camera as ref itself. Eg - back to camera, looking away from camera, etc
I am new at midjourney, I am wondering if it is possible to make 2 of my consistent character to interact in a scene. Or maybe you have a video for it? And I know you are showing midjourney alpha goodies, but it will be great if you are showing the old DIscord way :)
Thanks a lot :) do you have any suggestion how I could create multiple characters and create pictures with two, three or even four consistent characters?
Thank you for another excellent video! Question: I noticed that your GPT created prompts with multi-prompt parameters (i.e. using the ::). In the Midjourney documentation, it doesn’t list v6 as being compatible with that parameter. Is that just an oversight on Midjourney’s part, or has that multi-prompt parameter indeed been deprecated?
Version 6 definitely supports (or at least follows uses) multi-prompts, and i did a lot of experimenting - and controlling the prompts this exactly was the best way to get actual consistency. Otherwise little changes in scene, like being in a gym, would still change your character (ie, add muscles)
All the characters in my images have the same pose / facial expression on all six photos. No matter how often I rerun the prompt. Any idea what to do in that situation?
Typically yes. You definitely have a little bit of style transfer ability. But you'll want to use a low --cw value - if you your trying to adjust the style.
thank you for these videos as they are helping me to create character profiles for a book- I did not name my first character, and I created a second character, both of which need to interact. What is the best way to go about this? It is both a creative and organizational question- thanks in advance
Also, I'm creating these to plug into Pika, but the costumes change...I want the background and costume to remain the same, but her action would be different. How do I do that? Thank you!
Yes, on Discord, I would say that that would be the best approach to save the cref so it's easy to use over and over again. (You'll probably still appreciate the prompts from my gpt) Technically, the function is closer to using /suffix directly. But that isn't really practical if you ever want to generate more than one character. Good luck generating 😀
prima vista it seems like your work surface in midjourney will do the things a lot easier than I can do them on discord; did I miss something I should know now to switch back from discord to... yes, to what you are using? Btw: great work!
Yeah, I was getting a lot of these too. I chose this prompt format to escape the standard Midjourney look- I also think starting close-up and zooming out like this is good for that. Phrases like "looking to the left" will be much more likely to work if your generating the image from pretty close up
@@glibatree hey so what do you mean by this. Do you make a whole new prompt and divert from your GPT prompt? Im having the same issue... would love some help!
Hi, thanks a lot for your tutorial, I am testing your process in discord. You uploaded 5 images, I am in discord, how do I replicate this operation, is it cref??? or some other setting.
Yeah that's right, cref from the alpha site and discord are pretty similar, but you will need the image urls: th-cam.com/video/KwAwWjCRaCY/w-d-xo.htmlsi=RWuHhqvuB4KnQfhp&t=42
Yeah, haha if you noticed it a touch different, I am working on a new video that will also be using it and things sometimes need to change a bit when ChatGPT updates etc
I've been using midjourney a while, but haven't checked out other channels. 1 of your videos you invited people to check out your channel, but I don't see what it is or how to find it?
I think you mean my when I invited people to my Discord channel - which also includes some threads to generate Midjourney art. I always have my discord link in my description.
Ok. But how do I get the same character consistency between seasons? Without having to generate a new character. I want the same character from session to session!!
On my glt, you can simply paste in an image of your character, or a prompt you want to start with. Once you've created your references you can use them in midjourney during any session. Just grab an old prompt with a cref and use it again.
Yeah, the GPT I made for this video is still a time saver to this day, especially if I need to create new reference images so I can start generating a new character
Definitely worth a regenerate if you haven't tried that already. I spend the most time on getting the right grid collage. You can also vary region to regenerate just parts of the collage which can help. But, don't fret too much. You can always just try using them as references and see how consistent the output is. Good luck :D
Appreciate the thoughts, thanks and glad you liked the video itself! I am experimenting with some recording software and working out the kinks, so all feedback is helpful
If you want to really to use your own image, probably the best way is going to be pulling it and your character that midjourney generates into a photo editor. Photoshop out whatever you use. If you want Midjourney to create an image /inspired/ by your own- a Midjourney image prompt might be sufficient
Thanks my man! Music choice can be a little polarizing, but this was a track I used to use a bunch when I live streamed more on this channel - felt like a fun called back
Ok.. newbie here. Why does your Midjourney seem like it's in it's own room? Not sure where you're actually at in Midjourney lol. Ohhhhh it's Alpha. Why can't I get access? Ok found it (need 100 images). I really just posted a train of thought stream and answered my question in real time LOL.
One windows: Windows Key + Shift + S opens the snipping tool so you can take a screenshot of any portion of your screen. You can also go into your windows search bar, and type "Snipping Tool" and the built in app for it will show up.
@@glibatreethank you I was in the wrong Midjourney. Do you know if it can create images with 2 people at once or would I have to create them separately and add them to a scene
@@cynthiarobinson1087 CREF doesn't directly support it... but with inpainting you can generate the first person, then inpaint a different part of the image and change the character within the cref so it generates that character too and adds it to the image.
You're OP for this one. Much gratitude.
So glad you're happy! Thanks for watching!! 😎
Thank you very much for this graduate-level lesson on character consistency. I've never seen anything like it before, after searching for over a year on this topic. I've subscribed to your channel
Thanks so much
this is awesome, thanks for making this gpt!
Your stuff is always awesome. I just recommended this to all my students who are learning CC with MJ.
You're a legend. Thanks for much for sharing this.
It's pretty good. The only issue I've experienced is that it's really nice hard to get the character to look away from the camera, or to be active in the scene. When you start with a consistent portrait style, all the subsequent prompts tend to be portraits as well. So, not the best for storytelling, but otherwise pretty neat.
Have you figured this out yet? If so please tell... Thanks
I've had the same issue, the character tends to default to staring into the camera. Had moderate success with using the camera as ref itself. Eg - back to camera, looking away from camera, etc
@@Lucidity59 try multiple angles shot in ur prompt/u can write side angle,shot from the back...
I am new at midjourney, I am wondering if it is possible to make 2 of my consistent character to interact in a scene. Or maybe you have a video for it? And I know you are showing midjourney alpha goodies, but it will be great if you are showing the old DIscord way :)
great info. Will try it out later this week.
say hi to Hannah made me smile 👋
Thanks for Glibatree Consistent Character Assistant!
Your very welcome!
Very much appreciate this breakdown. TY
Glibatree, this is good!! thank you
Amazing! props to you my man!
Wow fantastic, thanks for sharing !
Great Tutorial, i was wondering how you setup your Chat GTP, is ther a Tutorial to create a own customized Architecture behind a own Promt ?
Thank you very much Glibatree
This is amazing! Thank you! 🙏
wow thank u, so useful!
Amazing tutorial! Really great tips, thanks. ;)
Hi, this is fantastic. Do you recommend using Midjouney Alpha or Beta?
Thanks a lot :) do you have any suggestion how I could create multiple characters and create pictures with two, three or even four consistent characters?
Thank you for another excellent video! Question: I noticed that your GPT created prompts with multi-prompt parameters (i.e. using the ::). In the Midjourney documentation, it doesn’t list v6 as being compatible with that parameter. Is that just an oversight on Midjourney’s part, or has that multi-prompt parameter indeed been deprecated?
Version 6 definitely supports (or at least follows uses) multi-prompts, and i did a lot of experimenting - and controlling the prompts this exactly was the best way to get actual consistency.
Otherwise little changes in scene, like being in a gym, would still change your character (ie, add muscles)
@@glibatree Fantastic! I really appreciate the follow-up. I’ll reincorporate the multi-prompts into my workflow, and I can’t wait to use your GPT!
@@trevorama Good luck!
Thankyou, what a great help. Could you perhaps make a video about multiple characters? :)
All the characters in my images have the same pose / facial expression on all six photos. No matter how often I rerun the prompt. Any idea what to do in that situation?
Thank you for this great video. I enjoy watching your tutorials. 🤗
Appreciate it bro!!
Awesome explanation! Reminds me a bit of CGDream ai tool
Question are you on discord on your computer? Because if not I didn’t know mid journey had a website
Yes you can use the site now. I believe you might have needed to create a certain amount in discord first.
Seems like we need to create the character in the style we would like to keep them in later. Any thoughts on that?
Typically yes. You definitely have a little bit of style transfer ability. But you'll want to use a low --cw value - if you your trying to adjust the style.
thank you for these videos as they are helping me to create character profiles for a book- I did not name my first character, and I created a second character, both of which need to interact. What is the best way to go about this? It is both a creative and organizational question-
thanks in advance
Also, I'm creating these to plug into Pika, but the costumes change...I want the background and costume to remain the same, but her action would be different. How do I do that? Thank you!
I LOVE your GPT to create prompts for character consistency in MJ! I'm struggling to maintain ref body shape/size. How can i get better accuracy?
Thak you for such a fine job
LOVE it ~~~!!!
Very good explanation, nice music, too. Curious, which country are you from? So easy to understand your English...
Thanks, glad you enjoyed! I'm from the United States, in the northeast.
This is great. Thanks a bunch. I have a question though, how would this work with 2-3 characters in the same image?
so cool! thank you so much!
could you show how to create consistant multiple charachters in the same scene? And how to use the new editor tool in MIdjourney.
Gerat work !!
Hello, great tutorial. I want to get my characters in action, specifically sneaking but can't. Any tips? Thanks!
How do you get into the alpha? I can only get into the discord. There is no place to pay for the client.
Congratulations!
Also, is it possible to create a character based on real pictures and place this character in various different contexts?
Great stuff. If done in Discord, is this similar to using cref with "preferred option set" option name (where I use five reference images)?
Yes, on Discord, I would say that that would be the best approach to save the cref so it's easy to use over and over again. (You'll probably still appreciate the prompts from my gpt)
Technically, the function is closer to using /suffix directly. But that isn't really practical if you ever want to generate more than one character.
Good luck generating 😀
prima vista it seems like your work surface in midjourney will do the things a lot easier than I can do them on discord; did I miss something I should know now to switch back from discord to... yes, to what you are using? Btw: great work!
damn, I missed the update on midjourney 6 it seems, lol
I tell Midjourney to create the 6 photo collage and it makes all kind of strange things but I never have the 6 ID photos like you have
invaluable . thanks
Good tip.
Glad you found it useful
So could you actually create an entire comic book with a consistent ai character using this?
Thanks!
Thank you i like to use your combination with chat gpt ✌️
Glad you enjoy it, and you are very welcome!
When I try stuff like this, the characters tend to stare into the camera, they never look at other things in the scene.
Yeah, I was getting a lot of these too. I chose this prompt format to escape the standard Midjourney look-
I also think starting close-up and zooming out like this is good for that. Phrases like "looking to the left" will be much more likely to work if your generating the image from pretty close up
@@glibatree Which prompt format?
@@jesse8068 The one my GPT generates for closeup portraits
@@glibatree understood, thanks!
@@glibatree hey so what do you mean by this. Do you make a whole new prompt and divert from your GPT prompt? Im having the same issue... would love some help!
Can you explain why you are using chatgpt for the prompts? Why don’t you just prompt mj
Hi, thanks a lot for your tutorial, I am testing your process in discord.
You uploaded 5 images, I am in discord, how do I replicate this operation, is it cref??? or some other setting.
Yeah that's right, cref from the alpha site and discord are pretty similar, but you will need the image urls:
th-cam.com/video/KwAwWjCRaCY/w-d-xo.htmlsi=RWuHhqvuB4KnQfhp&t=42
thank you!!!
hallo i loved your video. But i have 1 question, how to make six panel like in the video via midjourney? thx GBU
What GPU did you use?
say hi to hannah sounded so human
could i possibly do this with another image generator ?
I am enjoying your GPT. Are you updating it?
Yeah, haha if you noticed it a touch different, I am working on a new video that will also be using it and things sometimes need to change a bit when ChatGPT updates etc
can this work with cartoon characters?
Multiple characters tips on MJ Alpha?
I've been using midjourney a while, but haven't checked out other channels. 1 of your videos you invited people to check out your channel, but I don't see what it is or how to find it?
I think you mean my when I invited people to my Discord channel - which also includes some threads to generate Midjourney art. I always have my discord link in my description.
Looks like we have to have alpha status to join
recently just started using aigc tool. thanks for sharing. do you know any free tool? what about Stylar?
Great tips. Please turn that awful background music down or get rid of it.
😂lol this crack me up but I do agree it’s like sir I’ll pick my own music while you teach me thanks but yeah lol
is there any way i can do this without gpt subscription?
Great video. Love the GPT you created. Please ditch the background music. Its competing with your vocals.
Ok. But how do I get the same character consistency between seasons? Without having to generate a new character. I want the same character from session to session!!
On my glt, you can simply paste in an image of your character, or a prompt you want to start with.
Once you've created your references you can use them in midjourney during any session. Just grab an old prompt with a cref and use it again.
Is this achievable if you begin with an existing image of a character?
You can certainly invent characters this way as well
How do you get that panel for midjourney?
is it still relevant with the character reference parameter?
Yeah, the GPT I made for this video is still a time saver to this day, especially if I need to create new reference images so I can start generating a new character
I'm not getting different angles or even expressions from your prompt. They're all just staring straight ahead.
Definitely worth a regenerate if you haven't tried that already. I spend the most time on getting the right grid collage. You can also vary region to regenerate just parts of the collage which can help.
But, don't fret too much. You can always just try using them as references and see how consistent the output is. Good luck :D
The video is very good, but i feels like the BGM is a bit strange, and the sound is loud.That's just my thoughts, no offense.
Appreciate the thoughts, thanks and glad you liked the video itself! I am experimenting with some recording software and working out the kinks, so all feedback is helpful
Shoutout to the whales singing in the background 🐋
This must be what they use for Max Joe Steel
I will never understand people who don't use dark mode
I hate dark mode. Makes me depressed😅
Good news! You don’t have to understand!
How can we add our own captured image as the background?
If you want to really to use your own image, probably the best way is going to be pulling it and your character that midjourney generates into a photo editor. Photoshop out whatever you use.
If you want Midjourney to create an image /inspired/ by your own- a Midjourney image prompt might be sufficient
I liked your choice of background music.
Thanks my man! Music choice can be a little polarizing, but this was a track I used to use a bunch when I live streamed more on this channel - felt like a fun called back
Ok.. newbie here. Why does your Midjourney seem like it's in it's own room? Not sure where you're actually at in Midjourney lol. Ohhhhh it's Alpha. Why can't I get access? Ok found it (need 100 images). I really just posted a train of thought stream and answered my question in real time LOL.
Hahaha, what a journey this comment was
Amazing!
What snippet tool? Or can I snippet anywhere
One windows: Windows Key + Shift + S opens the snipping tool so you can take a screenshot of any portion of your screen.
You can also go into your windows search bar, and type "Snipping Tool" and the built in app for it will show up.
@@glibatreethank you I was in the wrong Midjourney. Do you know if it can create images with 2 people at once or would I have to create them separately and add them to a scene
@@cynthiarobinson1087 CREF doesn't directly support it... but with inpainting you can generate the first person, then inpaint a different part of the image and change the character within the cref so it generates that character too and adds it to the image.
@@glibatree you are so helpful! It is awesome if you to actually take the time to answer questions
Glibatree is not writing code, its creating one image. help.
I thought midjourney is working in discord, this looks different
bro… wtf is that awful BG music… 😂
MJ alpha isnt avaible to us all yet
🙄 it’s always something. I can’t get into alpha
background music is such distraction. remove it
Wow. Are you a paid subscriber who buys merch? Geez
@@bobhs1605 He is right, its very annoying
@@bobhs1605 yeah his right mate
Jesus Christ loves you all
Lose the massage music mate. The video is almost unwatchable due to this.
I couldnt stay with this awful bg music🤮