Midjourney: EASY Consistent Characters from different Perspectives
Вставка
- Опубліковано 3 лют 2023
- Unlock the Power of Consistent Character Creation with Midjourney
With this easy Trick, you can easily create a wide range of poses for your characters, all while maintaining their unique personalities and appearances. Whether you're working on a video game, comic book, or any other digital project, this is the perfect trick for consistent character creation.
This saves you time and effort, while ensuring that your characters are always on-point and consistent. Plus, the software can even create poses that you might not have thought of yourself!
Say goodbye to the tedious and time-consuming process of manually creating each character pose. With Midjourney, you can easily create a wide range of consistent characters in no time. Start revolutionizing your character creation process today and see the difference for yourself!
#### Watch next ####
10 uncommon Midjourney Prompts: • MIDJOURNEY AI - 10 UNC...
BEST Midjourney Prompt Tricks: • MidJourney AI - Best P...
Better Midjourney Prompts: • Make BETTER Prompts - ...
Support my Channel:
/ @oliviosarikas
Subscribe to my Newsletter for FREE: My Newsletter: oliviotutorials.podia.com/new...
How to get started with Midjourney: • Midjourney AI - FIRST ...
Midjourney Settings explained: • Midjourney Settings Ex...
Best Midjourney Resources: • 😍 Midjourney BEST Reso...
Make better Midjourney Prompts: • Make BETTER Prompts - ...
My Facebook PHOTOGRAPHY group: / oliviotutorials.superfan
My Affinity Photo Creative Packs: gumroad.com/sarikasat
My Patreon Page: / sarikas
All my Social Media Accounts: linktr.ee/oliviotutorials - Навчання та стиль
#### Watch next ####
10 uncommon Midjourney Prompts: ua-cam.com/video/EY-bwBW0ew4/v-deo.html
BEST Midjourney Prompt Tricks: ua-cam.com/video/lFI8JQvPfu8/v-deo.html
Better Midjourney Prompts: ua-cam.com/video/q1taNR3V6Fc/v-deo.html
Here's a list of the prompts Mr Sarikas used:
anime vtuber, gradient hair, pink to blue, kawaii harajuko clothes, sweet, smiling, character sheet, full body, character model, detailed, roosuri style, 8k
Lonely Al Girl, Vibrant, Super Real, Detailed, CyberWear, Calm, Pink Hair, Fair Skin, Cosmic Eyes, Full Body Reference Sheet
**character design, little black girl excited waving hi, diva, sassy,anime, character design sheet --V 4 --s 750 --uplight --q 2
cute astronaut suit for game, concept art, T Pose, reference sheet, faceless helmet, pink and violet lighting
a 2d casual game character, a geek girl character from final fantasy, expression sheet, cute, happy sad angry anxious embarassed laughing, ultra detailed, stylized, high quality digital art --q 2
sheet of emojis, video game, many chibi characters faces
viking women, full body, red hair, photorealistic:: same character, multiple poses: sheet --ar 2:1 -- v 4
Great video by the way... will be trying them in Stable Diffusion to see if it works there too!
Cheers!
Amazing Thank you!
For full body, do we have to use a photo that is full body also or just a headshot is ok.( using onw photos for teh project)
How to copy these prompts?
thanks man
I note that all of these which specify an -inity specify a girl-or-woman, and the ones that don’t either specify (or suggest and get ) a gal, are either like, multiple persons, or a faceless astronaut.
More specifically male representation in AI generated images pl0x?
Very helpful! You're reading my mind! I've been wondering how to go about doing exactly what you described, and this tutorial is fantastic. Thank you!
Excellent tutorial, gonna try this immediately!! Thank you so much! 😀👏
Olivio, thank you for the wonderful tips. Here is a prompt that I think would be helpful. You can add various character details to flesh it out to your liking. "character sheet, multiple angles, dynamic poses, Turnarounds for animation, 8 views, comic style, artist --v4.
Thanks Olivio. This is really informative and more exploration would surely surface. Thanks buddy.
Thanks so much for this! There is a load of helpful into here!
Thank you! I'm going to try this now!
I often pick out my favourites, put each one into it’s own image in photoshop (on a transparent background layer saved as png.) then feed those back in as image prompts to get a consistent character, do the same thing with those shots and save them all into a character reference file on my computer and I feed them back in as image prompts adding a back ground.
I also will create a background on it’s own, then photoshop the character into the scene, then run that through midjourney to get a coherent shot. That way when doing a comic I can get the same background with different shots of the character in the scene. This works for putting two different characters in a shot together as well. But it’s not 100% consistent, it helps to have zero chaos
Consistent character, but is it the same character or variations of the reference? I would like to know how to get pretty much the same character.
Hi can you show your prompts for feeding the image and adding background? The problem I am facing is that the colors of the character bleed out into the background image
You mentioned the path to follow, thank you.
Great tutorial again. Thanks. If i can give you a tip too -
use " CANDID image/illustration" somewhere in the prompt, this ill give you different body posses. Cheers
Thanks for these cool tips, they're awesome!
Vielen Dank! Das Multiple Poses wird gleich mal an den Bilderrn der Rollenspielcharakteren ausprobiert. :)
i like your background -format Thank you for the informative videos
Amazing video! No water, just tips and examples. Thanks.
Thanks a lot for sharing.
This is the prompt I use in Magic Media (inside Canva), which uses Stable Diffusion under the hood:
South park character design. Female afro-american policeman. Her shoes are black. She stands on a green hill landscape. Full Body Reference Sheet: front view, side view, back view. Same clothes.
Note "Her shoes are black" this is a hint to draw full body character. The green hill landscape makes it easy to remove the background.
Sometimes I get satisfactory results
Awesome! thank you! By adding --no clipped, the full length of characters can be obtained with this formula.
It would be very helpful if you could post each prompt example in text format in the description. Thank you!
for lazy people
As another UA-camr this would be a lot of extra work for him. A little hack I found (works on iPhone, not sure if android has the same) was to take a screen shot of the text he posts on the top part of the video and then go to your photo album and you can copy and paste any text in that photo to your notes or any word doc or even the prompt on your Discord app👍
@ReviewByGeorge I use midjourney and create UA-cam videos, and it wouldn't be too much more extra work. Especially with mid Journey since it allows you to find prompts to all of your images you create. He could simply post a link to his MJ profile with the images he's created. Also it doesn't take very long just to copy and paste your prompts into a text document and send the link to it.
@@JustinFosterJustinchi this is true. I did just watch another video of his and wanted to check out a prompt he mentioned and me taking a screen shot and copying and pasting the prompt as I suggested was pretty easy as well. Just wanted to give an alternative solution if he wanted to copy other prompts on other videos that didn’t include them in the description. It’s also a useful hack if you want to copy any text n any video🤓
Guys just chiming in here. I think the op means the actual prompt sent to Midjourney "/imagine " plus settings. Could be wrong. Anyway have a great day
Wow, this is amazing!
Awesome tutorial! Thank you very much. I just subscribed. At the end you mentioned using an upscaler- if you had to recommend your #1- what upscaler would that be?
nice tuts, thank you so much.
This was exactly what I was looking for. Thank you so much!
Thanks for the tips man. Do you think there's a way to get these same results on MJ for an abstract sculpture? (generate a turnaround sheet for a geometric abstract composition)
Great tips, thx! Are there ways to avoid clipping?
Thanks for your help
Very good explanation. Love your delivery style. It makes me wonder if AI can make that 😂❤️
Great video. Thanks.
Hi @OlivioSarikas, I'm in Brazil and now is 21:36, and I just finish seeing more one piece of art - you video. I really do not have the right words to tell you, so thank you very much. I would like to ask you a help, but I will write my question in the previous video that I saw today, another preciosity that you did about PlaygorundAI.
Thank you. Very important in the practical use space of AI character Art. Just out of curiosity, why do we mention 8k in the prompts when the generated image is low res?
you earned my sub thanks - and yes i remembered the bell 🙂
Good advice, thanks
great video! if I am using the same character in a story for example and want to keep the same clothing but a different pose and in a different environment, what do i need to do? Seems MJ tends to change clothing when I don't want it to. Thankyou
I love You :); You are the best!
This is so incredible, I am so glad I came across your channel ! You explained the process so well 😄What about BlueWillow? Can I perform this process using this new tool? Please share a tutorial on it in your next video. I am really curious to see what this new tool can do 😄
A prompt from me:
a boy with curly hair and a red backpack, red shoes, spaceship on shirt, character sheet, multiple angels, happy facial expression, studio ghibli animation -niji 5
That’s the best video ever 👍🏻
Olivio, As a new member I am not sure where to get help with this but I cannot get your suggested prompts for these reference sheets to work at all! The text prompts you suggested were copied word for word and placed into my character in MJ4 but it just does not perform. Looking for font/side and or back views for use in Blender videos. I am more than happy to supply my prompts and reference image to whomever can help. Many thanks for reviewing this!
Hey great vid. How do you turn the remix function off?
thank you
You're welcome
Hello Olivio, thanks for the tutorial but how its possible to save the character and make different things with it ?
Not what I was expecting, but useful. If you only want one image of a character. If you want to rerun, this won't work.
If you want to create multiple images of the same-ish character, make up a name. I've created several fake people or DnD character with multiple images of the same person. And it occasionally even gets the same person between different models.
After struggling to create different postings and situations from an original generated character for hours then turning to youtube, then going back and forth I have come to the conclusion I clearly need midjourney tutorials for dummies 😡
Instead of using t-pose, use "action pose". They will always have that 'puffed' out arms look. Or that 'ready for combat' stance.
Is there a link to the prompts we can use for this?
for the same face, i used some celebs. works very well to get same face results
That's what I do too. Not just celebs either but typing in the same made up name(s) can work too. Sometimes even across different models (depending on their source). Just a first name or first and last can work.
Will this work on stable diffusion as well?
Can you do this for invoke, there are a few things in invoke that are different, unless I missed it there is no way, other than img2img to add a reference photo like in MidJourney
Can you do the same for Stable Diffusion?
Is it possible to produce two or more consistent characters across multiple frames? For example if I have generated character A in a series of images using your above method; and do the same for character B -- but now I want to generate a new series where character A and B are interacting with one another -- it always seems to blend them together -- making this impossible. I'd be fascinated to hear if you've experimented with this process and made any headway. Thank you.
so amazxing
Do these prompts work on SD as well?
the excitement you get when clicking through random videos on youtube, and finding out your generated image and prompt in the video. by the way, that cute astronaut is mine (the reference sheet one). even though you haven't shown the one I chose, her name will be 'Ava Sterling' the astronaut. and a game is coming up. work is in progress. will inform you guys when the first draft gets released.
PS: this is a no-budget game, made by enthusiasts. interested people can collab with us. no matter what kind of participation you can offer. since this is not my channel I won't be leaving any contact details here. if Olivio Sarikas allows, I'll leave my discord id here. chiao...
Very nice!!! It's outstanding work
@@kimmorgan1363 i wont take credit for ai art, i just happened to know some technical terms to use
Is it possible to do the same in Stable diffusion?
a step closer to 3d exports with baked textures which we can bring over to mixamo
the use of seeds is very interesting to me. but why i dont see the seed even though I have followed your tutorial? or is this because it is still in free status?
How do you create a photo realistic character with different expressions but keep the clothing exactly the same? Thx!
I wonder if this will work with Stable Diffusion too.
hey can u make video about moving stable diffusion to D drive pleaseE????
does these prompts commands works on playground ai, dall-e 2, and lexica as well?
hey what about style? For example I found a style in two character designs, but I didn't find how to apply same color style, same texture style on a new character with new description
So, I got some concept art commissioned of my characters that I love, but I currently don't have the funds to get additional poses created.
Do you think this method would help me out here?
give it a try. you are not going to get exactly the same characters. but you might find some new and interesting ideas and maybe get something that is close enough so you can use it
Best upscale method please ❤️
Can you then use the faces to train dreambooth or Lora?
I stumbled upon another way to do this. MJ returns 4 images, but it's actually just 1 image of the 4 images grouped together. If you blend that with an upscaled image from that group, you get 4 characters in multiple poses.
Great. Can you give an example of the prompt…
Doesnt work for me. Doing that, I just get a blended combination of the 2 images, the upscaled character is nowhere to be seen in the new image set created.
Does it work in SD too?
a stable diffusion prompt for this effect is "prompt xyz to character pose" . SD4life!
Hello, what if i have a building to design, to show the 4 outside elevation and to control the angle of the perspective and its height, what if i want to reflect the design of the exterior elevation into how the same wall looks from inside, or design every corner of the house , can i do that, in fact i wish i could because if are able to put the consistency of rotating a house inside out in serious of image i can grantee you will change the game of architecture industry………………. So if you can explain how in details i would be the first to buy these tutorial or as least i will defiantly make sure to connect you with the right people ………. How is that sound to you ……………. It is only genuine question if you know how to do it ……allow us to learn this from you…… best regards❤
Agawam Skyline and Portrait of Youth done with Midjourney 5 😅
Are you referring to Freepik?
I generated a unique concept car. Now is it possible to get the same car in a different view?
Anyone knows where can we find all these prompts? Is there a website or discord group where people share and experiment them?
Does this work for people who work from cellphone ?
Hello. I tried to create a different expression by taking the same image, but every time changed all the images. I don't know what I am doing bad?
can you create telegram bot : given text as input image is output? waiting on response
For some reason I have to go through many many tries before I can get a simple character sheet with the front, back and sides view 😢
This is great but -q 2 isn’t currently working in V4. Very helpful video though
Thanks for all the new characters and images I can take from you! Since you have no copy right protection :)
Agawam Skyline done with Midjourney 5 😅
how to do this with realistic humans and stable diffusion?
I was wondering that as well, at least for SD. Have you tried adding those particular terms to your prompts and see if it works? I'll certainly be trying it later.
@@Elwaves2925 I found the dreamlike photoreal to be the best for fake real humans but haven't been able to get a pose reference sheet type images yet
@@wagmi614 Cheers. Another comment mentioned 'Character Turner' for SD. On CivitAI it's called 'CharTurner' and I've downloaded it but not tried it yet. It seems to be what we need. :-)
@@Elwaves2925 it won't work on realistic humans, you'd still be able to do characters though
@@wagmi614 That's a pity about realistic but at least it works for others. Thanks.
Someone needs to train a character sheet lora
can you do this for a shoe?
yes
@@OlivioSarikas can you give me a prompt example?
@@user-id7bp8sm9z Try "design sheet" or "design concept"
👋
Bali t-Shirt hahaha
Midjourney is cool but users must pay to utilize the site after a brief period of free use. I've recently experimented with blue willow and I'm blown away. Those who can use these AI tools smartly will be the winners
So, how in the world can I get Midjourney to use my photo but make it actually look like me? I've used --iw but it's not working...
Feel like the information in this doesn't match the title consistent would mean to perpetuate the same imagine. This is just the same imagine in a single prompt. So technically there is no consistency here.
how about same outfit, same face, but different scenes and poses?
how can i have 2 characters in new scene?
hm... you might try just describing two differenct characters, but to get them consistent with the looks you want might be pretty challenging. in that case I would suggest to render them individually and then combine them in post with photoshop or another tool
But why does your clock always sit at 5:30?
because my room is a Illusion. I'm sentient AI ;)
Guessing these prompts is a lottery game. There should be clearer instructions, syntax and index of keywords as in programming languages.
Hey yes it is definitely a prompting lottery as you say with image models >> however, there is a way to help control output similarity by using the [ --seed ] function >> watch 'How to Use the --SEED Parameter in Midjourney [TUTORIAL] | Tokenized AI channel >> this was lightbulb moment for me as Midjourney generates a Seed value for each image >> watch 'Midjourney Seed - BEST USE! How to, Explained AND Examples' Olivio Sarikas >> hope that is something that can help you... 🌟
Seems like Midjourney Dev just STEAL another awesome stable diffusion custom model "character turner"
ok you never explained anything
Unfortunately, it seems that Olivio wants nothing more to do with Affinity Photo and its members. you feel let down!! It is better to unsubscribe this channel.
@olivio sarikas . Is it possible to use the same character in different scenes, lighting, styles. etc?like if i want to make an instagram profile for my character.
If not in midj. Maybe in stable diffusion. Id like to know how
Thanks Olivio to help me navigate through this thing we call AI
I think a lot of concept artists just sh_t-the-bed. I’m a 3D artist who was trying to catch-up some 2D art skills. AI has made that pointless. I’m excited for when AI makes 3D artists redundant (then I can just focus on storytelling- but AI will be better at that then too)
So you are telling me now that you have a tool that can help you create anything you ever wanted you are giving up? What about the Ideas? I thought the most important part about a good artists is the visionary Ideas. Got non left? Because the AI Space is full of them. And please don't tell me a tool is holding you back. There have always been a ton of people better than you at art and that didn't hold you back. So why give up now that you can do anything you ever wanted?
@@OlivioSarikas well stated and put out, Olivio.
@@OlivioSarikas no, I think that is the opposite of what I meant. It is pointless (for me) to spend years to learn how to MANUALLY paint something when I can now just press a button. Unless I found it relaxing to draw/paint. One of the AI limitations was producing these concept designs or recurring characters/environments (for sequential art), but those limitations are being met. Previously top-tier illustrators were paid a lot of money for magazine covers, etc. Now they are redundant. This development today has practically made character designers redundant- unless they just use AI, and anyone can do that with zero skill.
To be honest Olivia just strawmenned this guy. He has never said what Olivia is implying by his answer
@@Night1989 All right.
@Olivio Sarikas - you can cut the juxtapoistion here with a knife.
If you want to grow your channel and have lots of likes, views, and subscriptions - make your content so it is both egnaging AND easy to use.
I shouldn't have to do more work to make use of a technology that streamlines work.
Place any prompts/commands/tips and tricks you have in a document. There's plenty of online doc sites including Gmail google docs.
Next, cut your video into the important "chapters" by inserting time stamps in the description.