I think they want you to use Midjourney generated faces because of something called latent spaces (internal vector representation of the faces which are kind of a meta representation). This allows Midjourney to create different angles of the face using the latent representation as a jumping off point leading to better accuracy. That’s my guess based on the little ML math I know.
Mariko! Haha. Someone's been watching 'SHOGUN', I reckon!😉😄Awesome as ever bro. Your tutorials are truly a (Katana) cut above the rest, in terms of detail and photographic insights. Thank you soooo much for all your hard word & shared knowledge. Hontoni arigatou gozaimazu, Jungle-san! 😄 Love from Wales, UK. x
You're the best. When I come here, I always get just what I need. Also, you explain everything so well, step by step and in detail. You can tell you take pride in what you present, and I truly appreciate it. Will always buy your guides!
The massive Midjourney V6 update for achieving consistent characters in seconds is fantastic. SmthOS enhances the experience by providing a user-friendly platform for leveraging these new features. #MidjourneyV6 #AIArt #smthOS
I love your videos and channel. I have really learned a lot. I truly appreciate all of the time and effort you put into these tutorials. I know they take a lot of time. I want you to know that we all appreciate the work!
Very useful. I still think the character is not stable enough. There is always some percent that is not equal. But maybe it will be better in the future since it's a very new option. It's more visible if u generate common characters. Like doctors or office employees. Specially European. I never made to keep the character stable through all different styles. Still I find this tutorial very helpful and also bought your guide. Looks great. Thank you. Maybe u could release a notion guide about stable character creation too. ❤
Now, the big breakthrough will be for the ‘consistent character’ not to give their face to everyone else in the scene. I note in your examples the character is largely alone in their universe, which is a pretty big limitation, don’t you think? That said, your video is really helpful. Thank you
Good video! I've seen another RUclips tutorial of this same content and the outside photos he used of himself worked very well in MJ. I've created good look-alike images of me in another ai generator but they do not work in MJ at all. I hope consistent characters keeps improving, but for now, I've given up on the idea of being a consistent character myself. 🙂
Actually, you CAN add extra unique characters to the same image using the pan button. You can also swap/modify a character using "vary region." Try it.
Thank you very much, you made My Day. The tutorial about "cref" is the best I have seen up today. I kindly ask to keep doing it. Do you plan to make a fullbody character in many poses as tutorial using ""cref"?🤩
Just love your videos, also can you make a video who to prompt to create an full ai movie and greater cinematic scenes, just like you described that girl's, Thanks man for this value, really appreciate it.
First of all, thanks for your tips and for sharing your knowledge and contributing so much to the community! I've been testing the new character consistency method, and it seems to work quite well, with possible improvements but for now it serves its purpose. Better than the old method of dividing the image into 4 splits. Now then... I wonder: would it be possible to use it or create something similar for objects? Could both be combined? I've run several tests, but I haven't achieved anything. It would be great, wouldn't it? Warm regards Cyberjungle!!
This feature doesn't seem to work well at all with men, especially over 30, like me. I guess it was trained on a lot more young beautiful women, for obvious reasons. And especially redheads and Asian women.
is there any midjourney pro, who is experienced with multiple consistent characters in different styles? i need illustrations for different books. different detailed scenes with 1-3 consistent characters each. i pay for the projects. on upwork and fiverr they dont have any skill.. just answer on my comment if you think you can help and earn money with it
I tried to do it your way, it did not work, it refers image denied, request cancelled due to image filters.... do you know what happened? I used a picture which I created on midjourney
does this work with product design concepts as well? I want to find a way to have midjourney create the same product but in different angles. Much appreciated
Wow, amazing doing 500 samples and all look rt ON! 🎉 I’m hoping MJ will be able to recreate our own or client image uploads soon as yes it currently makes us less attractive and bit odd. Wondering if there is a work around now.
Can I do this with my own face or does MJ not support this yet? Like if I were to use an image of myself or a person (Customer to put him in a setting for a prompt for say, business?) Thanks for the video.
I've been trying to experiment with it, but no matter what scene I describe, such as a wide angle shot, riding in a winter sled being pulled by a beautiful horse --ar 16:9 ... no matter what weight I give to the character weight ( --cw ) MidJourney repeatedly creates a close up shot of the woman and creates an image that makes no sense, ie: with maybe a horse at some wonky angle, but definitely not a zoomed out shot of the character riding in the sleigh with the horse pulling the sleigh across a winter landscape. What am I doing wrong?
Yes unfortunately somestimes Midjourney does that (defaulting to close up shots when character reference in the prompt) You can try: 1- Use “full body shot” as keyword 2- Describe also your character’s shoes or footwear (so Midjourney can show that hence full body) 3- Use Style Raw. Because default Midjourney aesthetics have a lot of close up shots 4- Use an image prompt together with character reference. Ensure that your image prompt reference has full body shot and ensure image weight (iw = 3) means maxed out
@@cyberjungle ... yes, I asked for fully body shot, wide screen image, described her boots, described a horse pulling a winter sleigh, she is riding in the sleigh ... but I want a fantasy-realism 3-dimensional style with a bit of manga influence and I find that V 6.1 and especially RAW gives only realistic images, often with very 'crunchy' texture (pretending to be detail). I'm trying to create something artistic, not photo-realistic ... Version 5.2 used to make much more artistic images and would recognize art styles if I mentioned them ... Version 6.1, especially RAW seems to completely ignore requested styles ... but Version 5.2 doesn't work with character reference, it gives an error message and tells me to choose 6 or 6.1.
@@galefraney ah I see 👍 I completely agree about crunchy texture and poor realism of 6.1 I know in few days / weeks they will roll-out 6.2 with improvements they collected from 6.1 feedback. Fingers crossed for 6.2 and hopefully it will work better with abstract / illustrative art better and recognize styles. By the way, if mentioning a style name doesn’t work you can also use style reference (sref) Just a reminder / suggestion ✌️
@@cyberjungle ... I also uploaded a very specific style reference and it only seemed to take colours from it, nothing else looked similar. I used to be really excited and amazed when using MidJourney ... now I just get discouraged and "hide" pretty much every single image it outputs ... at least if I were able to permanently delete them so that I'd have the relief of removing the crunchy, poorly rendered, inaccurate images forever, it might not seem so bad to know they're not still cluttering up my Archive section ... in addition to wanting to know why image quality seems to have greatly diminished recently, I'd also like to know why MidJourney doesn't provide a proper "delete" button. I thought that maybe it was just me ... I've been a MidJourney subscriber since it first launched, so I look at the gallery to see what others are creating and I see that most of it is the same generic stuff ... MidJourney outputs fine images of certain topics, is OK with photographic faces and sexy women who like they're posing for selfies (LOL) ... or super heroes with exaggerated muscles, and cute pets ... but when it comes to something a little more creative, more often than not it's a crunchy poorly rendered mess. 😨
11:36 It doesn’t look like Midjourney can show the same character at different ages. 11:51 Also, I don’t think this character looks like a 5 year old. She appears to be around 12-13 in my opinion.
@@cyberjungle Not in reality, the message literally states that it is because the "image", and I already figured out, is has to do with the way you copy the URL, in addition to the fact that you get more accurate results IF first you create a value for that image through the "/prefer option set" function. Thanks anyway!
If it’s someone Midjourney knows (world knowledge wise) then you can mention person’s name in the prompt and make some artwork. However you can’t upload a celebrity face and make art from it (at the moment)
Hi Junglitos, here is link to the Notion Library for Midjourney V6 shown in the video:
cihanunur.gumroad.com/l/midjourneyV6-style-guide
I think they want you to use Midjourney generated faces because of something called latent spaces (internal vector representation of the faces which are kind of a meta representation). This allows Midjourney to create different angles of the face using the latent representation as a jumping off point leading to better accuracy. That’s my guess based on the little ML math I know.
the most clear and details explanation on cref video I ever watched, cheers!
Mariko! Haha. Someone's been watching 'SHOGUN', I reckon!😉😄Awesome as ever bro. Your tutorials are truly a (Katana) cut above the rest, in terms of detail and photographic insights. Thank you soooo much for all your hard word & shared knowledge. Hontoni arigatou gozaimazu, Jungle-san! 😄 Love from Wales, UK. x
You're the best. When I come here, I always get just what I need. Also, you explain everything so well, step by step and in detail. You can tell you take pride in what you present, and I truly appreciate it. Will always buy your guides!
The massive Midjourney V6 update for achieving consistent characters in seconds is fantastic. SmthOS enhances the experience by providing a user-friendly platform for leveraging these new features. #MidjourneyV6 #AIArt #smthOS
I love your videos and channel. I have really learned a lot. I truly appreciate all of the time and effort you put into these tutorials. I know they take a lot of time. I want you to know that we all appreciate the work!
Thank you so much ✨
Very useful. I still think the character is not stable enough. There is always some percent that is not equal. But maybe it will be better in the future since it's a very new option. It's more visible if u generate common characters. Like doctors or office employees. Specially European. I never made to keep the character stable through all different styles. Still I find this tutorial very helpful and also bought your guide. Looks great. Thank you. Maybe u could release a notion guide about stable character creation too. ❤
Pretty sure I've never hit subscribe faster. Well done.
Excellent video on character consistency!! Impressive output!
Now, the big breakthrough will be for the ‘consistent character’ not to give their face to everyone else in the scene. I note in your examples the character is largely alone in their universe, which is a pretty big limitation, don’t you think? That said, your video is really helpful. Thank you
Best MJ YT Channel
👍👏amazing!! learned a lot...Thank you so much brother!🙏 Love from India😍
Greetings to India 🇮🇳 👋🏼
🙏🙏@@cyberjungle
Good video! I've seen another RUclips tutorial of this same content and the outside photos he used of himself worked very well in MJ. I've created good look-alike images of me in another ai generator but they do not work in MJ at all. I hope consistent characters keeps improving, but for now, I've given up on the idea of being a consistent character myself. 🙂
Very, very good 👌🏻🙏🏻
Excellent work! Thank you for this and the rest of your tutorials!
Actually, you CAN add extra unique characters to the same image using the pan button. You can also swap/modify a character using "vary region." Try it.
Thank you very much, you made My Day. The tutorial about "cref" is the best I have seen up today. I kindly ask to keep doing it. Do you plan to make a fullbody character in many poses as tutorial using ""cref"?🤩
Amazing! Will this be in the "Midjourney V6 - Photorealistic Cinematic AI Photography Style Guide"?
Yepp already there ✅
Best yt channel 👌🏼 Great work sir🙏🏻
Simply a fantastic Masterclass, many thanks !!!
I work on a project of making my own comic with Midjourney, your tips are really coming handy, thanks ;)
Great tutorial, Thanks! It's on Notion Page too?
Absolutely outstanding.👏👏👏
perfect video, thanks! Are the consistent characters prompts included in your guide?
Will be added in April 👍🏼✨
This was superb - thank you!
Wow Amazing Work, Great Video!
This is perfect video for Midjourney subscribers!!
Just love your videos, also can you make a video who to prompt to create an full ai movie and greater cinematic scenes, just like you described that girl's, Thanks man for this value, really appreciate it.
perfect and very structured, nice!! =D
thanks for the email, really needed that
Great content bro. Keep it up!
Well done! I'm curious how you could do the same using your own face, or is this where FaceSwap comes into play?
Yepp this is a job you can do with FaceSwap only (for now)
First of all, thanks for your tips and for sharing your knowledge and contributing so much to the community! I've been testing the new character consistency method, and it seems to work quite well, with possible improvements but for now it serves its purpose. Better than the old method of dividing the image into 4 splits. Now then... I wonder: would it be possible to use it or create something similar for objects? Could both be combined? I've run several tests, but I haven't achieved anything. It would be great, wouldn't it? Warm regards Cyberjungle!!
for a single character, it is doable for all the AI. I want an example with several characters (3 characters) in the same scene
Love this!
Tham chiếu hình ảnh tải lên của MJ vẫn còn hạn chế. Ảnh tạo ra vẫn không thể giống nhiều . Ngoại trừ ảnh đó của MJ tạo ra .😢
Nice. For 2 characters select Vary Region and do one character, then the other
This feature doesn't seem to work well at all with men, especially over 30, like me. I guess it was trained on a lot more young beautiful women, for obvious reasons. And especially redheads and Asian women.
thankyou for making these videos
thanks so much for sharing
Thank you bro, you rock !
SUper!! Thank so much!!!
This is gold
Tks
Like always an extremely useful tutorial. How can I get to the Midjourney Alpha Website?
They’re slowly rolling out to everyone. Stay tuned
Thank you! 🙏🙏🙏
Hello, thank you for this amazing tutorial. Did you updated this information on your Gumroad midjourney v6 guide?
In progress 👍🏼
Thank you ,very useful!
When using --cref, how can I change the body orientation and the direction in which the character's eyes are looking?
You can use ‘she is looking at the camera’ or ‘focus in her eyes’ in your prompt
I love your videos!
I followed along to get consistent character and it wasnt even close.
Why not? What was your prompt?
thanks for the video...why I can't save it in a playlist...?
Can you make her into an anime or cartoon? Simply adding "cartoon" in the prompt doesn't have any effect.
Thanks for the video! just a question how can I use Atomic prompt? I don't get it...
the ultimate guide ! thank you so much !!
😍
is there any midjourney pro, who is experienced with multiple consistent characters in different styles? i need illustrations for different books. different detailed scenes with 1-3 consistent characters each. i pay for the projects. on upwork and fiverr they dont have any skill.. just answer on my comment if you think you can help and earn money with it
Great vid bro
RIP digital artists lol
Yup!
Once again very well conveyed... Many thanks for that. I would like to give them a few more likes, does anyone have a trick for me 😀🙃😇
I tried to do it your way, it did not work, it refers image denied, request cancelled due to image filters.... do you know what happened?
I used a picture which I created on midjourney
Your are the best 😊
does this work with product design concepts as well? I want to find a way to have midjourney create the same product but in different angles. Much appreciated
yes it works to help with replicating product features, design and details but it's not always 100% accurate
yes it does. use the same cref format
@@cyberjungle Thanks friend
@@hfycentral Thank you!
Could you help me? Im freaking out, im doing exactly the same but im not getting consistency.
Me either, character varies quite alot and even with CW set to 0 clothing is still influenced by base photo.
Wow, amazing doing 500 samples and all look rt ON! 🎉 I’m hoping MJ will be able to recreate our own or client image uploads soon as yes it currently makes us less attractive and bit odd. Wondering if there is a work around now.
Do you have this Notion template available for download/purchase?
another great video
what is the stylise perimeter you have used? with these?
That lady has soooo many different orange outfits.
actually emotion means, what type of emotion?
What about keeping the environment/scene consistent? Like several scenes take place in the same kitchen?
Not possible at the moment
Can I do this with my own face or does MJ not support this yet? Like if I were to use an image of myself or a person (Customer to put him in a setting for a prompt for say, business?) Thanks for the video.
Unfortunately it only works with faces generated by Midjourney
Anyone like me screenshotting these incredible examples? Good one.
excellent
I've been trying to experiment with it, but no matter what scene I describe, such as a wide angle shot, riding in a winter sled being pulled by a beautiful horse --ar 16:9 ... no matter what weight I give to the character weight ( --cw ) MidJourney repeatedly creates a close up shot of the woman and creates an image that makes no sense, ie: with maybe a horse at some wonky angle, but definitely not a zoomed out shot of the character riding in the sleigh with the horse pulling the sleigh across a winter landscape. What am I doing wrong?
Yes unfortunately somestimes Midjourney does that (defaulting to close up shots when character reference in the prompt) You can try:
1- Use “full body shot” as keyword
2- Describe also your character’s shoes or footwear (so Midjourney can show that hence full body)
3- Use Style Raw. Because default Midjourney aesthetics have a lot of close up shots
4- Use an image prompt together with character reference. Ensure that your image prompt reference has full body shot and ensure image weight (iw = 3) means maxed out
@@cyberjungle ... yes, I asked for fully body shot, wide screen image, described her boots, described a horse pulling a winter sleigh, she is riding in the sleigh ... but I want a fantasy-realism 3-dimensional style with a bit of manga influence and I find that V 6.1 and especially RAW gives only realistic images, often with very 'crunchy' texture (pretending to be detail). I'm trying to create something artistic, not photo-realistic ... Version 5.2 used to make much more artistic images and would recognize art styles if I mentioned them ... Version 6.1, especially RAW seems to completely ignore requested styles ... but Version 5.2 doesn't work with character reference, it gives an error message and tells me to choose 6 or 6.1.
@@galefraney ah I see 👍 I completely agree about crunchy texture and poor realism of 6.1
I know in few days / weeks they will roll-out 6.2 with improvements they collected from 6.1 feedback. Fingers crossed for 6.2 and hopefully it will work better with abstract / illustrative art better and recognize styles.
By the way, if mentioning a style name doesn’t work you can also use style reference (sref) Just a reminder / suggestion ✌️
@@cyberjungle ... I also uploaded a very specific style reference and it only seemed to take colours from it, nothing else looked similar. I used to be really excited and amazed when using MidJourney ... now I just get discouraged and "hide" pretty much every single image it outputs ... at least if I were able to permanently delete them so that I'd have the relief of removing the crunchy, poorly rendered, inaccurate images forever, it might not seem so bad to know they're not still cluttering up my Archive section ... in addition to wanting to know why image quality seems to have greatly diminished recently, I'd also like to know why MidJourney doesn't provide a proper "delete" button.
I thought that maybe it was just me ... I've been a MidJourney subscriber since it first launched, so I look at the gallery to see what others are creating and I see that most of it is the same generic stuff ... MidJourney outputs fine images of certain topics, is OK with photographic faces and sexy women who like they're posing for selfies (LOL) ... or super heroes with exaggerated muscles, and cute pets ... but when it comes to something a little more creative, more often than not it's a crunchy poorly rendered mess. 😨
can anyone help - see when i'm putting in the URL, do i put in the full thing and with (?
Put full URL without (
11:36 It doesn’t look like Midjourney can show the same character at different ages.
11:51 Also, I don’t think this character looks like a 5 year old. She appears to be around 12-13 in my opinion.
Hi, the follow message appears to me:
Image denied
Request cancelled due to image filters.
what it could be?, thanks for you video.
Probably one of the keywords you used is against their moderation policy
@@cyberjungle Not in reality, the message literally states that it is because the "image", and I already figured out, is has to do with the way you copy the URL, in addition to the fact that you get more accurate results IF first you create a value for that image through the "/prefer option set" function.
Thanks anyway!
Just wondering why mostly AI work is about asian culture.
Hey man can we use two face in one shot with a consistent background like example office room , or a hospital room . That would be helpfull man if yes
so we cannot use someones (famous) face to make an artistic image in midjourney?
If it’s someone Midjourney knows (world knowledge wise) then you can mention person’s name in the prompt and make some artwork. However you can’t upload a celebrity face and make art from it (at the moment)
@@cyberjungle thanks for your reply. I appreciate your educational videos bro 🙏
After all these, are the images and videos created using ai monetizable in youtube?
Yes
@@cyberjungle but there is some new guideline that it cannot be? Are you sure?
nice:)
Why can't I use the new alpha site? It doesn't give me choice to do this.
They’re gradually rolling out to all users. Stay tuned 👍🏼
@@cyberjungle if I pay for a subscription can i use it?
@@Steger13 it’s not about paying for a subscription. They’re letting people in based on total number of images they generated (starting from highest)
@@cyberjungle ah ok ok, thanks . I have to wait than.
What to do with them ..nothing, these imageries are useless
Why can’t I get anything done with the character?? 😭😭😭😭who knows what could affect this? who can help?🫠
another great video