Create FACIAL EXPRESSIONS for Characters - Midjourney Character Design
HTML-код
- Опубликовано: 8 июн 2024
- 📙 Midjourney COURSE mastersofmidjourney.com Beginners to Advanced
🚀 FREE Midjourney Cheat Sheet tokenizedhq.com/freebies/mj-c...
🔗 FREE Promptalot Extension promptalot.com/extension
🔗 FREE Supporting Material tokenizedhq.com/freebies/vide...
Folder: 2023-02-15 - Facial Expressions
🌐 Check out the full blog post:
tokenizedhq.com/midjourney-fa...
-
📰 AI Newsletter 👉 Your Inbox tokenizedhq.com/newsletters/ai
🐤 Follow me on Twitter / chrisheidorn
💼 Follow me on LinkedIn / christianheidorn
📸 Follow me on Instagram / christianheidorn
💬 Join the Tokenized AI Discord tokenizedhq.com/invite/discord/
🎬 Character Design Series 🎬
Playlist: • Create CONSISTENT Char...
Part 1: Create Consistent Characters • Create CONSISTENT Char...
Part 2: Place Characters in Action Scenes • PLACE a Character in A...
Part 3: Apply a Consistent Character Style • Apply a CONSISTENT Cha...
Part 4: Create Multi-Character Scenes • Create MULTIPLE Charac...
Part 5: Create Facial Expression for Characters • Create FACIAL EXPRESSI...
Part 6: Infuse Your Characters with Themes • INFUSE Themes into Cha...
📺 Recommended Related Videos & Playlists 📺
Monetizing AI: • Make Money with Midjou...
Activate GOD MODE in MJ: • Become a GOD in Midjou...
The SECRET Language of MJ: • The Surprising TRUTH a...
How to Use --SEED in MJ: • How to Use the --SEED ...
Ho to Use NEGATIVE Prompts in MJ: • How to Use Negative Pr...
-
Are you struggling with emotions and facial expressions in Midjourney?
Or perhaps you've been so focused on getting your character's outfit right in Midjourney that you completely forgot to give a personality?
Let's face it, life is bland without emotions and you'd be surprised to see the unbelievable range your characters can have.
In this tutorial, I'll show you how to inject passion and emotions into your character's facial expressions.
This video is Part 5 of a longer series about Midjourney character design. Part 6 isn't quite ready yet, but it will be released within the next couple of days.
So stay tuned.
-
⏰ Timestamps ⏰
00:00 Injecting Emotions into Characters
01:06 Creating our Character Concept
03:29 Blending the Character
05:39 Creating Emotional Facial Expressions
12:44 Creating Lively Portraits
14:22 Creating an Artistic Portrait
-
#midjourney #midjourneyai #midjourneyart #aiart - Наука
Your tutorials took my Midjourney image-creation to a whole new level. THANK YOU!!!!
You have no idea how happy this makes me :)
I love how you brought Carla to life.
I thought it was just me not wanting enough to learn that's why I kept closing videos halfway from almost every other midjourney creator on youtube. Now I realize it's them, and how their thoughts are scattered all over the place, which makes their vids difficult to watch. How you explain everything makes your videos so easy to watch and learn from, well thought out, and very easy to understand, always staying on point instead from jumping to 10,000 other things. I also love how you consider that your viewer might not understand something you talk about and tell them where you've already talked about it and where to find it. I love how doesn't appear to be from expert to newbie but from one learner to another.
"From one learner to another" sums it up quite well.
So much is changing right now and at such a fast pace. There is no such thing as an "expert" in any of these fields. Everyone is experimenting and learning. Some quicker than others, but everyone is learning.
By far you share the best content about midjourney and the best edit, well done
And here I am, thinking that I'm an absolute video editing beginner. Which I truly am. Editing is so painful tbh 😅
I'm new to Midjourney and your videos obviously took a lot of time and are so great. You also should be commended for releasing this information free. All I can say is WOW! Thank you so much!
I really see no point in charging people money for a tutorial to be honest. I'm working on a course myself, but courses have a very specific purpose. I hate it when people gatekeep certain things behind a paywall. It's silly because it's not like this information isn't available out there already. It's just that not many people go through the trouble of putting it into a nicely consumable format such as these videos. Either way, thank you so much for your feedback. Truly appreciated! :)
Thank you, I'm learning so much and really appreciate your calm and to the point delivery.
Thank you :)
It’s almost like you’re manifesting dream characters. People’s shapes and forms change in dreams a lot. There is still that consistent sense, though, that you are interacting with the same character when changes to their shape or form occur. You are almost manifesting dreams.
thanks a lot ! your mj content is the best i have found on the net and your cool style makes it even better !
Wow, thanks! 🙂
So in the end, the effort you put is almost equal to drawing an image by yourself.
So I guess there's no more reason for some people to claim that it's effortless and "easy"? 🤷🏻♂️
You've done it again, ol chum. This is more of what I need for my characters and I appreciate your hard work, demos and articulate tutorials. Keep rockin' and journeyin'.
Everyday I'm promptin' 😅
Amazing job
Endlich weiß ich, wie ich zu den Bildern meiner Vorstellung gelange! Vielen vielen Dank dafür!!
Nichts zu danken! :)
Brilliant!
Thanks for sharing! That's incredible.
Glad you enjoyed it!
THanks mate! Very interesting
My pleasure!
You have some of the best midjourney tutorial.videos, thank you so much
Wow, thanks!
That was indeed impressive!
Thanks :)
Hi Christian, thank you so much for the series. I find you videos super helpful, concise and to the point. They should be getting 10x views! About IMG prompt, would you say this is akin to *training*, using your own model / seed? Anyway, I just wanted to say thanks and may see you on Discord later!
No, not really. MJ doesn't "learn" from it. It basically just uses the concept and tries to blend it into the final image. It's not really different from using multiprompts and adding a new segment, other than that it's an image and obviously it will be visually more precise.
I love this
Cool 😎
About front view, change that to "looking at camera, facing camera, facing forward"
I love the last part when you stated that creating image on Midjourney was not that easy..
Because it's true, no matter what many of the haters believe.
It would be cool if we could combine character facial expressions will body poses, such as a full-body shot, or half-body shot. Facial expressions & body language need to be combined together in order to create results that tell a story.
EDIT: I see you explored this here 14:59
Christian. Facing the. camera is also handy. 👍Great videos.
Yes, sometimes. But I've found it not particularly reliable.
i saw literally all the video about AI images from you uahsuahsuahsu, youre the best
Wow, thanks!
Fix the eyes using Photoshop neural filters. I had one character who kept "looking" at the reader. In Photoshop, I was able to turn his eyes slightly to the west / left, which made it look far more realistic.
"--no strabismus" also does the trick here.
It is crazy how easy these emotions are made. So to keep the style the same and get a good version of the emotion, would you just reroll until you got a best fitting image?
Did you use PS to place her head with buns head on the other body?
Absolutely great material, I'm going try to build a character that looks like my granddaughter and make comic book stories with her in it (thanks to you I'll get there sooner).
Good luck! 🙂
Love the videos so far. 🎉 Just fyi, it still says coming soon for part 6 at the end of this video.
Thanks so much for the heads up. I forgot about that :)
I've installed MidJourney today to with the intention to create images that can tell a story (that is going to unfold I hope...) and your videos have contributed with this decision Christian. I've seen a lot of RUclips channels that quickly lean into the young girls / anime / manga themes and I like that your channel stays away from this. Great content and I've subscribed!
lol, I think it's because those people have certain fetishes 😆 Also, internet culture (for whatever reason) seems somewhat obsessed with anime.
I can appreciate some anime but it's not really my thing as an overall style.
@@TokenizedAI Exactly mt thoughts. The whole internet seems to be slowly flooded with these dynamics. But again, love your work!
Respect
4:25 It would make more sense for the image reference to have more weight than the text prompt when crafting the character. This is essential for creating consistent characters in different scenes. Maybe Midjourney will eventually fix this problem. 🤔
I'm sure they will.
The information in these video is just awesome. It’s amazing I’m working on creating my own characters and using these as concept art in my character modelling in Blender.
You must show us your character models when they're done! 😀
@@TokenizedAI well am still learning Blender, might take a while lol
They look great! The image prompts really help!
Yup, they do!
Man, you are incredible. Great stuff, please keep it up 🙏
Thank you! Will do!
Amazing work & such valuable insight into the process of development. All your videos have been so inspiring & useful. Thank you.
Thank you so much 🙏🏻🤗
Exciting and fun to watch! Love your organized, step by step presentation. Great series!!
Glad you enjoyed it!
The white eyes may come from the fact that a lot of Marvel heroes have white eyes due to their masks (Spiderman, Wolverine, …) or when they intensively use their powers (firing energy beams out of their eyes or having a high energy level as is, like Storm controlling the weather, Thor with lightning, Cyclops loosing his mask, …). One would have to change the comic book style to one not having heroes with supernatural powers.
On the other hand, what I really absolutely do not like is when Midjourney uses the long hairs of the character instead of drawing tears (see 17:33). With the red hair it looks like the character is bleeding out of her eyes. 😱 I would not have an idea how to tell Midjourney not to do that.
And as far as the squinting of the eyes is concerned, that´s also totally weird…
Love your videos. Learning much of them. Thx 4 sharing.
Greetings from Switzerland 🇨🇭
Yep, I totally agree with everything you've said. MJ is amazing but it drives me nuts sometimes.
Wow wow wow
😁
I love your content, thank you sharing this with us.
My pleasure! 🙂
Great video, as always. But I don't understand how you are getting the waist up "cowboy shot" by prompting "wide angle full body image." The latter prompt gives me a head to toe image.
That's a very fair question. So the thing is, you always need to look at what your inputs are. In this particular case we had a reference image that was a full body shot. Then we had the first segment of the prompt that was also most likely going to generate a full body shot (although not always!) and the second segment was a concept that most of the time, would probably give us a portrait shot.
When Midjourney receives inputs with different angles, it will usually blend them into an "average". So for all intents and purposes, let's assume the image prompt gets 50% of the weight. The text prompt also get 50% of the weight, but since there are 2 segements with equal weights, each one get 25% of the total weight.
50% Image reference with full body
25% 1st segment with mostly full body concepts (but not always)
25% 2nd segment with almost exclusively portrait shots
All of that tends to blend into something that resembles a cowboy shot.
Would these also work if its not a closeup portrait shot , or even multiple person in the shot ?
Multi-character would be difficult or even impossible if they all should have different expressions. But for a wider angle shot it should work. Just bear in mind that MJ might not let you use a really wide shot because you're talk about specifics in the face. It really depends on the whole prompt.
How many image prompts do you think is the most effective for generating consistent pictures?
3-5 is usually enough, depending on what you're doing. Anything beyond that has diminishing returns.
wow!that's amazing!You are very patient and explain it in detail.
Thank you :)
Hello, could you activate the Spanish subtitles of this video? Thank you very much for your time!
Herzlichen Dank, super Infos. deutsche Untertitel wäre supercool :)
Das macht RUclips doch automatisch, wenn du sie aktivierst und Deutsch auswählst, oder?
Christian…best MJ series on characters ever. Thanks!
Thank you! 🤗
@ 15:12 how did you change the face??
You probably know this already; I've recently felt that if I situate description relating to camera angle shot light scene at the beginning of the prompt rather than the end, it seems more effective. perhaps you could test this out?
I'm pretty sure I already do this in my video. This is a fairly old video.
Camera angle, I agree. As for lighting, I disagree. Lighting should be further back.
@@TokenizedAI I tried to dive into cinematography in order to table all the terms and technical names for camera angles and lighting techniques but it was obscured by dozens of opinion pieces in the search results. If you're looking to experiment, perhaps this could be a direction? I'm not trying to give advice; your content is great and I'm jus sharing an idea I thought may have value. Thanks.
I love your videos! I have a feedback about the thumbnail for this one: because it's an extreme example of a laughing smile, it's actually a little bit too much and enters the uncanny valley, which might stop people a little bit from clicking on your video, I want people to click it so they can see it! Maybe choose something just a little bit less extreme
I actually run A/B tests on the thumbnails. The funny thing is that the stuff that makes people click usually has little to do with that. I'm genuinely surprised by the results sometimes 😅
@@TokenizedAI Haha, fair enough! Looking fwd to your thoughts on v5!
At 15:10 how did you replace the head?
With an image editor?
What you do is amazing .. when medjourney shows up ppl though ohh thats so eazy .. but you show that is so much need to learn to get the results you wish .. good job
Thank you! That's exactly what I'm trying to show people :)
Hey.. i click on the image and I do not see the envelope. I do see all the emos but there is not envelope.
Why?
Because you're right-clicking, which only shows your 5 most used reactions. You should be clicking on the "Addr reaction" button instead.
Thanks for your tests and work !
Pleasure! As always 🙂
Excellent, can’t wait until part 6
Part 6 is going to be fun 😁
i don't get the envelope icon...
great tutorial as usual, one question - @15:10 of the video, how did you replace the long hair head with the bun head? did you use seed or image reference, or prompts in midjourney? or other photo editing software? thanks!
Ah sorry, I did that in Photoshop 😁
@@TokenizedAI thanks for the quick response, keep up the good work!
Please make a video on how to show two characters in one scene fighting each other....I want to make an action scene in which two or more characters are fighting each other but midjuorney is not able to do this when I try...each time it make scene without action or fighting... please make a video on action scene of multiple characters in one scene
Please check all my videos. I have covered multi-character scenes already.
Ya watched all of your videos but and I have tried with different different method of prompting but midjuorney is not able to make multi character scene ... specially when it comes to action scene....
Show one character in action is easy task but when you are trying this action scene with 2 or more characters then midjuorney is completely fu*****CK it up ....
Like you tried to make Carla curuso action scene it was looking easy task try to add another character who is fighting to Carla curuso...then you will understand my problem...and thnx for replying ❤️❤️❤️❤️love u bro
How did you get the URLs?
How do I find the URLs in the DMs?
Right-click -> Copy link?
@@TokenizedAI Did it, got the URL but when I put it in Midjourney it says it's a invalid link.
Maybe on the fear part, is it because the midjourney detects different situations that the body reacts to fear? For example in a fight like "Karate Kid"...there was fear but in certain parts the expression didn't seem like what we are used to. Could a prompt of fear and retreat work? I'll try here
Followed on Twitter already o/
I honestly don't know. I also couldn't think of a better description for the expression 🤷♂
Thank you for all your Midjourney videos! I’m new to Discord so my question may sound dumb but I followed your link to join your Discord server; however, after accepting invite, nothing happens. Am I supposed to get an invite link from you? If so, where is the link? Thanks again!
So strange....did you get any specific error? Cause the link should work 🤔
Apparently using my phone worked! It didn’t when I tried on my iPad for some reason. Thanks again!!!
OMG YOU are the best....I suspect that English is not your first language like for ex I think if you used the word horrified instead of horror you may get a better result.imma try this myself AFTER i get through watching the rest of your videos and liking all of them LBVVS🤣😂
Du bist mein Lieblingslehrer !
Das freut mich 😁
You are the best, man!
Just doing my best :)
Great video! Could you make a similar video about body language? Expressive full-body poses? Emotional postures? I work more in full body images than I do portrait and have trouble getting the right pose even with reference pictures.
I think the other parts of the series (especially Part 2) might be helpful in this case.
@@TokenizedAI Part 2? Is it out already or are you working on it?
@@paullenoue8173 You just commented under Part 5. So yeah, Part 2 was already released weeks ago 😉
@@TokenizedAI Got it. Must have missed it, watched them out of order because there's so many AI Art videos out it's hard to keep track. Will check it out.
Brilliant stuff, thx again. I am still struggling with multi-character action scene and with practically no progress at all...
Action in a multi -character scene is almost impossible at this point.
@@TokenizedAI agree. can't wait to use future versions, rumor has it v567 will come out every once or second months, that's exciting
i dont see an envelope icon?
Because it only shows you the most frequently used icons if you right-click. Press the "add reaction" button and search for envelope.
@@TokenizedAI hi i did that but i dont see anything in my DMs?
@@t33w33 I suggest contacting the MJ support in the official Discord.
Bro I want to sell AI Art if I sell it on Amazon or Fiverr how they will deliver my AI Art to the customers do you have any idea? How it works if you know please tell me thanks
Not really sure what you mean by "sell AI art", since that take many forms. Just create it, put it up for sale and hope for the best. This isn't a "get rich quick"-scheme ;)
Master Chef Heidorn is cooking!
Hahaha 🤣🧑🍳🥩🥦🍽️🔪
You ended up using all four images of your cover survey, good idea! 👍
Injecting emotions to a character seems an obvious thing to do in trying to tell a story but judging from the MidJourney community feed only, it is not a very common practice. I am glad this video addresses this and adds very clears explanations on how to do it, very well done!
Merci encore pour tout ce travail de recherche et de démonstrations, je l’apprécie beaucoup!
Im thinking if you use V3 and use --IW (Insert weight(Higher than .25 because i think thats the default)) you should be able to then add prompts. When you put smiling in at the end, since the images have no higher weight, they arent as important as smiling
Except that no one wants to use v3. They'd never get images like these.
Very informative but I don't understand what your doing to make it lock in on your character. why isn't it giving you a deferent red head with braided buns in every image? I jumped in here from a discord link, I'm off to watch the rest of your series where you may have covered this already.
Why didn't you use seed numbers instead of url?
Because it's a different process and the seed theory has been debunked already. I recommend you watch the entire series.
@@TokenizedAI debunked already? So the seed is of no use?
@@reycaribe Depends on the context. Seeds are of course of use. If you are referring to Part 1 of my series about "training" MJ, then I recommend reading the Public Service Announcement I added to the description.
There are better ways to achieve consistency, which I more or less start showing from Part 5 onwards. I also say this in the video I believe.
Thank you @tokenizedai because with your information,
you make the AI world of image generation a little more friendly for all of us who like it.
Just try to create value for everyone 😀
I wonder if this prompting works on animal faces, need to test it 🤔
Good question! Let us know what you find out. 😀
It’s v4?)
Yes.
You are absolutely brilliant
☺️🙈
Excellent stuff
Thank you! Cheers! 🙂
Very good my brother!!!
Thank you!
Great videos... just an fyi, your link to your discord is expired. Trying to find your server.
I don't get it. I set it to never expire and new people are joining all the time 🤔
@@TokenizedAI I don't know. I've tried several times and can't join. And then trying to search for your server I can't find it either
Hi! I enjoy the content of your video a lot.
I wanted to point out that I discovered you can only use up to 5 reference images.
I think you're confusing this with the /blend command, which is limited to 5 files.
@@TokenizedAII didn't mean /blend.
I entered a prompt without initiating /imagine first, which made a normal massage.
That message only showed me 5 images.
/imagine also seemed to totally ignore the last (9th) image i used.
Might be coincidence?
What you're describing (the first part) has nothing to do with Midjourney. That's probably just Discord limiting how many images get embedded in the message.
And if you're using 9 image references, then each one of those image only contributes 11% to the image prompt. If you also added a text prompt (which alone gets a weighting of 50% in v4, with the other 50% going to the image prompts) then you've diluted the impact of each one of those 9 images to less than 6%.
So of course it's going to be barely noticeable in your final image. But that doesn't mean that it's not taken into account.
@@TokenizedAI gotya! So in a way you can manipulate your image weights :)
How did you get that envelope icon?
Hit the reaction button, then search for envelope.
Amazing -- it seems like midjourney really learns from the prompts, doesn't it?
I'm not sure. What part of the video makes you think that? Everything in the video just because of the image references and text prompts.
@@TokenizedAI yeah I forgot about the references... having to carry those along all the time is not convenient, but you can get around most of the typing or cut-n-paste with /prefer to preset some part of the prompt
However I think the better approach is what ChatGPT does, it remembers what you said earlier in a conversation and also offers multiple conversations so you can talk about totally separate topics and don't get everything mixed up in one conversation.
What do you think, wouldn't that be better so you can work on projects long term with MidJourney?
Sure, but that would mean that you expect Midjourney to build an entire ChatGPT clone just to complement its primarily product. I think building a good AI art generator is already difficult enough. I think it's unrealistic at this point. Bear in mind none of this was even possible a year ago. We tend to normalize this stuff way too quickly and then start to become picky, ignoring the fact that this is already a major advancement as it is.
Mega cool 😎 danke!
🙏🏻
@Christian Heidorn: try --no strabismus;
Works surprisingly well :P
@@TokenizedAI As I said, I thought of it as a joke but...
I am in love with Carla 😄
🤣🤣🤣
Your discord invite doesn't work?
So many people keep saying this and yet, new people are joing all the time 🤔 What's the error you get?
No idea what the glitch was, but I'm in now 😊
I just joined via the link, no problemo.
18:34seconds again huh??
Pure coincidence that it was so close to the previous one.
midjourney could just add an emotion and poses command ...
I think you have an overly simplistic of how this works 😅
All the emotions of a real woman ;)
🤣
Why why why 😅😅😂😂😂😂😂
Seriously though....why! 😂
hi!!
Hey! 🤗
Those weird eye crossing anomalies are a pain
Yup
OK, I see it now. Time to get serious.
😁
Your workflow doesn't really work anymore. "Duplicate images detected" blocks you from using multiple similar images.
Sounds more like you used multiple identical images and not just similar ones.
It still works on my end.
@@TokenizedAI they're images based off of the same seed, but they're definitely not identical.
but if it still works for you then, keep on keeping on I guess.
Do any of the images look extremely similar (including perspective)?
@@TokenizedAI i managed to get it to work by introducing more variety into the backgrounds.