Huge Midjourney Update: Consistent Characters, Step-by-Step Tutorial
Vložit
- čas přidán 28. 06. 2024
- In this video, I will provide a step by step tutorial on how to use Midjourney's new character reference and weight parameters, --cref and --cw, respectively.
0:00 - INTRO
0:15 - Choosing a Base Image
0:29 - Setting a Custom Option for Character Reference
1:30 - Example Images
2:16 - Introducing Character Weight
3:19 - Multiple Image References
3:58 - Creating a Character Sheet
4:30 - Image Style Reference
5:40 - Limitations
6:09 - Wrap-up
finally a perfect video, thank you so much
You're welcome! Glad you liked it.
Fantastic video, def going to practice with
Thank you.
Man the amount of information in this video is mindblowing! This has 100x'd my MJ skills
Thank you.
Love your videos so much! Can you make a tutorial video on FlexClip’s AI tools? Really looking forward to that!
Thank you. I'll look into it.
Love the information given! what would be the MJ tips for beginner like me? I've been using Stylar for a while and its easy to use.
If you are just getting started with Midjourney, I suggest reading the official documentation on MJ's website, watching tutorials on youtube, and just experimenting with different prompts and settings.
Thanks for this great video! Will support your channel and watch you grow. Do you know if Midjourney is good with image to avatar and then making consistent characters out of the avatar. Would like to try and make a pixar avatar short story of myself in different backgrounds to create one story. What would you recommend?
I'm glad you enjoyed the video. I think Midjourney is great for what you want to do, but you will have to use another tool to animate your avatar. I have a video on how to create talking avatars using the artflow ai tool. One of the images I used in that video, the African American woman, was created in Midjourney. You mentioned you wanted an avatar of yourself. Remember, Midjourney cref feature works best with images generated with Midjourney, so you probably shouldn't use a photo of yourself.
Edit: I misread your post, I thought you meant a talking avatar. Sorry. Nevertheless, I think Midjourney would be great for what you want to do.
Great video, but for some reason, this method or any other method describeb in other videos work, can you are there any settings that need to be set, before attempting --cref and --cw? Thanks for the great video!
It's hard to say why you are not getting the results you want without more information. I didn't set any special setttings beforehand. Are you using style raw? Some tricks I discovered after I posted this video: 1. If you are getting inconsistent resuls, try including the original prompt along with the cref parameter - remove anything you would like to change, like clothing description 2. Use the same parameters across generations, for example if you set the style parameter to 100 in your reference image, use this same value for the other generations. 3. Maybe try using style raw so Midjourney adheres more closely to the prompts. I hope this helps.
Beautiful and, above all, VERY USEFUL video tutorial... Spectacular, THANK YOU SO MUCH.
One question, though.
To access Midjourney, do you have to use the discord? Or, what is the right option?
Kindly, can you tell me what it is or what the links are?
Thank you infinitely in advance... You are great
Thank you for the kind words. Midjouney can be used on their website. I think using the website is only available to users who have generated at least 1000 images, but I'm not sure if that is still the case. The website is www.midjourney.com, and you do need a paid subscription.
I keep getting "custom option removed," and I THINK I'm following the tutorial. Will keep working on it.
Make sure you are entering something in the value text box before pressing enter.
Yeah this is not working for me. Unfortunately. Each time the character is different. I’m not a novice user but thinking it’s the level of detail in my character (specific color of glasses, etc.)
Midjourney states that precise details likely wouldn't be copied. I usually get good results as far as the face and overall outfit go, but sometimes I have to reroll the prompt many times before I get a good result, and usually there are some minor differences. Have you tried using Midjourney's vary region tool?
Hi ! Do you know if there's a way to add 2 characters ?
Thank you for this video!
I haven't tested multiple consistent characters using the cref parameter, but I might do so in the near future.
could you please guide me how to change face impression & turn the face right to left in Consistent Characters ?
Try phrases like "sad expression", "happy expression", "facing left", "head turned right", etc.
Hi ! I followed all steps but... no results, they say ---cref is not compatible with V 5.0😔
It's only compatible with version 6 and Niji 6.
It becomes expensive to generate one that we want. So many trials and errors to get it right.
I hear you. If you have the $24/month plan you can generate unlimited images in relax mode. Also, have you tried Leonardo AI's character reference tool? I posted a video about it, and the results were pretty good.
@@aiconcoction I haven't. I am looking for something to generate consistent characters and stuff around as well. For story telling, I need multiple characters in the same image. Also the house, the furniture, the horses, the landscape has to remain consistent. Nothing close yet.
Yeah, I haven't encountered an AI tool that can do that, especially with multiple consistent characters. You might have to create the characters and backgrounds separately, then use an image editing software to put it all together. Or create consistent images of one character in different scenes, then add other characters using tools like inpainting. If you want a consistent style across images, use the style reference parameter in Midjourney or Leonardo AI.
could you please tell me how to remove value/ tom1 ? i'm adding another character when i enter after option i'll direct to tom1 how to delete it thanks
Just leave the value empty and the option will be removed.
im trying to right click but Im not getting the same options, is there another way
Are you trying to copy the link or download the image? Try clicking on the image. Then right click on the popup.
how do you get --tom1 yours showed up in a dropdown box for you to select mine showed nothing
If nothing is showing, you probably didn't set any custom options. When using the prefer option set command, make sure to enter the name of the option, click +1 more, select the value option from the dropdown, type a value in the text box, press enter. If everything goes well, you'll receive a message that the custom option was set to the value you entered. Make sure to click +1 more then value, otherwise if you press enter, the custom option will be deleted. If you've done all of this and you still don't see your option in a dropdown box, I don't know what's the problem. I hope this helps.
I need help with the options part. I'm brand spanking new so I didn't understand the tutorial. Is midjourney for people who can code? This seems foreign to me
You don't have to know how to code to use Midjourney, just prompting. The prefer option set is just a command that allows users to create custom options that can be used as placeholders, so you don't have to retype parts of your prompt that you use many times. There are many Midjourney tutorials on the web. I would start with Midjourney's documentation on their official website. Just play around with prompting and watch as many tutorials as you can, you'll eventually get the hang of it. Good luck.
so which one do you prefer and think is more consistent the midjouney or the chat gptt?????
I think it depends on what you want. They both are generally good at generating consistent characters, and both are not perfect. Midjourney is better in generating photo realistic images, while Dalle-3 is better with illustrations. For instance, if you want to create a consistent Instagram influencer, go with Midjourney, but if you want more child-like consistent characters, Dalle-3 may be better.
@@aiconcoction much appreciated 👍 that helps a lot!
You're welcome.
I get a message "Custom option Nina removed!" When I set the option. Why is it?
You have to enter a value in the value field, otherwise the option will be deleted if it exists.
How did you get - copy link - as an option?
If you are using Midjourney in Discord, right click on the image then click copy link. If you don't see that option, try left-clicking on the image, then right click on the popup. If you are using it on Midjourney's website, right click on the image then select copy image address.
Is it possible to hire you for certain images?
Unfortunately, I'm not doing any freelance work right now.
Why would you pay him? Just prompt your own picture, after all that's the entire point right?
contact me
It does not work. No character consistency.
I'm sorry it doesn't work for you, could you give more information about your prompt and settings?
Is it free?? Making or any limitations
Midjourney is not free. You can find the subscription plans here: docs.midjourney.com/docs/plans. There are free alternatives available.
Your video is great. I'd like to collaborate with you. How can I contact you?
Thank you. Unfortunately, I'm not doing collaborations at this time.
HI Im new to these. What is the link to the Midjourny tool, I was directed to a web tool called OpeArt. is that the same?
You will need Discord to use Midjourney. Follow the instructions here: docs.midjourney.com/docs/add-the-bot. If you have created over 1,000 images with Midjourney, you can use it on their website: www.midjourney.com/explore. I'm not familiar with Opeart.