Personalized Models in Midjourney!

[@poppingjaz]

This is great. You can also combined SRefs now similar to blend. For example [subject] sref1 sref2 sref3 etc. There's no limit to what you can come up with.

[@taoprompts]

Yeah, I saw the update. There's so many possible visual effects now.

[@jamilgotcher365]

Which image generator do you recommend where you can use your own photography of people and generate a new background or new outfit? I'd like to keep the face of the photo of the person with very little change. I've already used Photoshop's Gen Fill but the results are hit and miss.

[@taoprompts]

The best results will be using stable diffusion + inpainting. That will require some extra setup on your end. You could also try using the insight face swapper with the Midjourney in discord.

[@baikastudio]

Amazing explanation as always. Thank you!

[@taoprompts]

Thanks, I really liked the way the style code turned out.

[@masterzed1]

Thank you very much for your videos, they really have a lot of content and are of great value. i cant believe this is free. TYTM PD:I am using stable diffusion but it works the same way in most cases

[@taoprompts]

Stable diffusion is great, it's got so many features & extensions that can be used.

[@chariots8x230]

I wish it was possible to create multiple personalization styles. I can’t see myself sticking to a single visual style.

[@taoprompts]

You can always try other people's styles, this feature is still very new so I'm sure they will add more to it.

[@chariots8x230]

@@taoprompts That’s an option, although I’d prefer to create my own.

[@chariots8x230]

@@taoprompts Besides, some people use Midjourney for client work, in addition to their own personal projects. So, their Midjourney account style will be inconsistent with their preferred style, due to the influence of the irrelevant styles they’re using for their client work.

[@nolow_life]

​@@chariots8x230 The personalization is for rated jobs though. If you have one account that you use for personal and work, I don't see why the rated tasks would reflect your work 'tastes' .. it would just be your personal preferences. 🤷‍♂️

[@thewebstylist]

Awesome video and walk thru as always brah! I just rated a few hundred from my iPhone but it doesn’t have the text showing the amount rated on mobile unfortunately

[@taoprompts]

Thanks man. Yeah, the website was a bit buggy when I was trying but it seems to work fine as long as you rate enough images.

[@susanne.mcmillan]

Did that and now am getting more artsy/painterly images which is what I am after most of the time

[@taoprompts]

That's great! This feature actually works really well if you are careful about ranking the images, most people are able to produce a personal style they like

[@sakalpaul]

Thanks for your video. love it.

[@taoprompts]

Thanks man, appreciate the support 👍

[@ActiveandMarketing]

How do you get to the personal code of others?

[@taoprompts]

I think you can only get it if they share. Or maybe on the Midjourney explore page you can see what prompts other people are using.

[@VulgarTruth]

guys.. I always have problem to generate a shot of full body of a person to take the whole frame.. lets say a man in a suit brown leather shoes ... what prompts should i use.. i need them for my animationss . cheers

[@taoprompts]

Include details about both the head and also the feet in the prompt

[@VulgarTruth]

@@taoprompts thanks

[@YouUsAndAI]

Can this be used in Alpha? I've been using it mostly.

[@taoprompts]

Yeah, you just need to add --p in the prompt and it will work

[@YouUsAndAI]

@@taoprompts Thank you!

[@LASartor]

Where is our code for personalization? And I’ve recently seen some folks who have multiple personalization codes…how do they get those? EDIT, what I think I’m seeing is adding other people’s codes in addition to their codes? Is that possible?

[@taoprompts]

You can use other people's personalization codes. You can also have multiple codes yourself. Here is the documentation for it: docs.midjourney.com/docs/personalization

[@DavidOliverbuck]

Thanks, Tao, my bro.

[@taoprompts]

I got you man 🙏

[@midjourneyaiarts]

MJ releasing model personalization. Get at least 200 image ratings and then add --p after your prompt. Personalization completely changes the way our system interprets your prompts and MJ think you'll really like the new direction it takes. Have fun! 😂❤

[@Steve.Jobless]

It's basically like style alignment in ComfyUI and SD WebUI. 👍

[@taoprompts]

I haven't tried that, but it's good to know Midjourney is trying to integrate more features from other models.

[@YouUsAndAI]

That's soooo cool!!

[@wonkygustav4457]

The idea is good but I’d like to be able to reset my rank preferences as I didn’t really take it that seriously before just wanted to free hour lol

[@taoprompts]

If you keep ranking them, eventually it should it correct itself to your ideal preferences.

[@wonkygustav4457]

@@taoprompts that’s a lot of ranking though. But I appreciate your point

[@LASartor]

This is what I was doing, not for the free hours, but which I like better and there was little consistency in my choices, or so it seems.

[@1trxnt]

Can you do this on the website?

[@taoprompts]

yeah just add "--p" to the end of your prompts.

[@nhtna4706]

Does it mean the image u generated is not accessible to others?

[@taoprompts]

You can share your oersonalized style code with other people.

[@nhtna4706]

@@taoprompts didn’t get u, u mean it’s publicly visible to others when u use the command of p?

[@taoprompts]

@@nhtna4706 You can take your personalized style code I show at 1:05 of the video for other people to use if you want.

[@darkregentt]

@@nhtna4706 I think right now its not, I dont see any images with personalisation on Explore page. But I think its visible if somebody open your profile on the website

[@Ramfire]

Whoa, anyone ever tell you you sound and kind of look like Jason from The Good Place?

[@taoprompts]

Yeah I got some other comments on that lol

[@kekw21234]

I don't see the point of this, the tasks you are rating are random, so it's not really catered to you but to that source image instead.

[@taoprompts]

it supposed to pick up on small preferences you have based on the images you pick. It's worth trying out, you might get some really cool results