Master Portrait Prompting with These Techniques - Midjourney Tutorial
Вставка
- Опубліковано 14 чер 2024
- Help keep the Channel Sponsor-free ➡️ ftp.live/patreon
In this video, I'll take a deep dive into the art of portrait prompting and show you all the techniques you need to know to master the generation of stunning AI portraits in Midjourney. From poses to color tones, fashion, and aspect ratios, I'll demonstrate the formula and everything you need to know to get the most out of Midjourney V5
FREE PDFs
cameralens.futuretechpilot.com/
200styles.futuretechpilot.com/
cheatsheet.futuretechpilot.com/
Other Links:
Merch - FutureTechPilot.threadless.com
Discord - / discord
Creative Fabrica - www.creativefabrica.com/spark...
My MJ Gallery - www.midjourney.com/app/users/...
Wow, what an amazing and value-packed tutorial! I think you're the best teacher on UA-cam for Midjourney because you go so deep but keep it simple enough for us to understand with all your dozens of screenshot examples. I copy and paste the screenshots underneath the sections so I don't have to rewatch the videos when I forget what settings to use. Thank you and great work!
This was excellent. It's absolutely perfect in the explanation, recommendations, and level of detail. Your free PDFs are also extremely helpful.
Well thanks a lot for the feedback! I'm glad my style works for you
Awesome! Thank you so much for sharing that info
I'm glad I could help! Thanks for the comment
Thanks for explaining the double AR and V tags, I was never sure how that worked.
haha no problem! Glad I could clear that up
Great tips! Most my images I like 16:9 and have been manually typing sooo many so this prefer tip a new fav
Workflow game-changer!
Paste the prompts, is the most important for users..
Thanks!
the prefer usage is brilliant, need to use this in my own prompts.
Which part?
the ability to auto add the suffix ar values in my prompts.
@@rongc4105 Yes! Super useful. Enjoy!
Great video, thanks a lot
Thanks for the comment, glad I could help!
great vids, alot of small stuff I didnt know
Awesome, happy to help!
What a great tutorial on techniques. Sub gained... 👍👍
I appreciate it!!
thanks!
Cheers!
Your dictionary is fantastic! I was just wondering if there was a directory of styles somewhere that I can get. Then Wow I found your UA-cam video. Thank you.
hahah funny how that works out... I'm glad I could help!
I don’t know you, but I really love you. Thanks so much for the help.
😂
Using a image posing is great too
Yeah for sure!
Great video! Lots of things to try. One thing that was missing a bit was: how to pose the model towards the camera and how close-up or full body you want it. Like side-view, (en profile?) front facing or how to specify a portrait from the waist up etc. I know there are some professional terms that may help with that... from movie making etc. (I can't remember them all and don't know how good MJ responds to them.)
It seems like there are still some things to figure out with regards to posing the model for different camera angles and shots
It's a good question! I'll definitely look more into it, I'll need to experiment more.
Actually, here is something that might help. Describe your prompt and then include '--s 0' at the end
@@FutureTechPilot Thanks, I'll give that a try!
Hii what does - s 0 mean :)
for me the iw attribute has been one of the most powerful ways to place characters effectively
Yeah for sure!
one correction to something you said about the "style" parameter:
a lower "style" value actually gives you more photorealistic and "dry" pictures. the higher the "style" value, the more stylized and artistic the image becomes.
It's not more realistic, it has to do with your prompt. From the user guide: "Low stylization values produce images that closely match the prompt but are less artistic. High stylization values create images that are very artistic but less connected to the prompt."
Great video, lots of great tips, love the ART-AESTHETICS pdf. Do you have one for posing subject?
No I don't, but thank you for the idea! I wrote it down and I hope I can work on it sometime soon
@@FutureTechPilot Thanks, Love your channel
Wondering if you’ve checked out the newly added feature /describe? You enter /describe and add an image to /describe and Midjourney generated 4 prompts. Check it out, interesting stuff. Thanks for this video!
I'm playing with /describe right now! Hopefully I can have a video about it tomorrow
I get a lot from your videos. I’ve created over 40,000+ images in Midjourney and since I experiment a lot, many of them are “mistakes” to put it politely. I’ve gotten a lot of solid, valid tips from you that have allowed me to get some really awesome stuff. If I could erase I’d pare my archive down to manageable levels, but thanks to some of your amazing tips and lots of work I’ve been able to generate some beautiful stuff.
Plz make a video of doing dual portrait sir,it will be very useful, thanks in advance 🙏
What exactly do you mean?
Great stuff. Thank you! Haven't tried the '-0.75' weightage before. 👍
I've seen the 'Instagram model' phrase in a lot of places and always wondered how that's different from 'beautiful woman' or 'handsome man' etc.
Also, another thing to mention which I know affects things - if you add an adjective like 'beautiful' before green eyes, you'll notice it tends to focus on the details in the eyes and the face, in general, more than the rest of the portrait. I added 'intricately detailed' before 'hazel eyes' and at --s 700, I actually saw arteries in the eyes! 🤯
v5 has really improved with displaying emotions more accurately. And teeth - teeth are human now! Oh, and ethnicities are far more achievable too.
Thanks for sharing your experience and insights on the topic, it's interesting to see how small adjustments in the text input can affect the output of the AI-generated image, and it's amazing to see the progress in the accuracy of emotions, teeth, and ethnicities in the latest version of the tool. Keep experimenting!
Woah, cool tip about describing the eyes!
And as far as instagram model goes ... I just try to rationalize how the dataset was probably tagged. Like, lots of things are beautiful but not everything is an instagram model ... do you know what I'm trying to say?
So even though beautiful and handsome would work, instagram model brings a specific look with it and I think that helps with composition + details in aesthetics (but maybe I'm wrong and the words are interchangeable lol)
Excellent explanation on prompting. I just subscribed because you are so clear and thorough.
Been using Midjourney only a month and love it for thumbnails to show my models at my photoshoots. Im a professional portrait figurative painter that was a 3d animator so finding this easy to create what I see in my head but whacky or repetitive.
I’m already using a lot of the words and prompts because of the photography and film background with the painting background. I wish people like me would not be afraid to use this to inspire new ideas to them paint will create the photo shoot. I mean we are creating it without promise we see it, right ?
My question to you is organizing. I have different texts set up for different projects. I’m working on is there a way to save my props instead of a notepad and then copy and pasting them into the imagine prompt? I also would like to have a ““ folder For different settings and art styles that I can color. What I use category or text?
I'm glad my style works well for you, and I completely agree about how more artists should be inspired by this technology rather than dismissive of it!
As far as organizing goes? I don't have a good solution for you. I personally use a note taking app called Obsidian, and it works great for me but I think you could get the same results with something like Google Docs or Notion.
If you happen to find a better way to organize your styles, please let me know!
Hello so far this is the first video very detailed i found. I am going to pause the video and follow step by step to see the outcome. But my question is do we need to subscribe to midjourney to use the version 5? Or the discord free can work?
You need a Midjourney Subscription to generate images. If you have a subscription, you can use any of the algorithms like Version 5
Very useful video! How to make MJ's portrait image similar to the portrait image in the original link?
I'm not sure I understand your question. Can you explain what you mean?
Awesome video chief! please, whats a good aspect ratio for instagram?
I think --ar 1:1 (the default square ratio) is probably the perfect one for instagram pictures? but 2:3 might work better for vertical reels or whatever
Hey brother. About the DSLR prompt issue, maybe adding a "in the style of DSLR photography" or "in the style of DSLR camera", could work around? What do you guys think?
Anyways, I will stick to Nikon cause I´m a nikon photographer. But, using 85mm makes a diference, IMHO. Also, 85's are kinda the best bokeh and angle lens... BUT - if you think that faces are getting toooo "squared", use 70mm instead.
IT WORKS! I just tested it! But it sucks, somehow. It looks an ordinary photo, lacking style. I used the following prompt:
She looks awesome, in the style of dslr photography, contemplating pose, ::in the style of moody chiaroscuro lighting:: 2 --ar 2:3
I appreciate the tip!! Writing it down now, thanks
@@FutureTechPilot Im the one that Thank you. I got many tips Just in this video. Anyways I was surprised that It Works - or not. Cause midjourney can Just ignore the "in the DSLR style". How to know? Maybe whith the "shorten" command? To see If is relevant?
Thanks for the great chart and this video. I can hardly ever get full body to give me a full body. I try describing shoes and sometimes that works. Other times it gives me a head sticking out of a massive shoe.😂
hahah Midjourney is full of surprises eh! Let me know if you find any good words for particular poses.
❤❤❤
🤝
Really helpful, i try get the most realistic images for my animal blog.
It's quite hard, to get these artists styles out of it.
Someone got any Tips on how to get realistic Animal Photos specific?
Try the word 'unsplash'
vertical portion is good in various cases, you all use for what purpose?
I try to think of how I want the subject to be framed. If someone is standing up, I would use vertical. If I want them sitting down, I think I would use landscape
Hey Nolan, I have a challenge … I cannot get a cyclops image… as in good old fashioned Greek mythology, so far, no matter what I prompt I get two eyes! Have not tried multi prompt yet… any tips/insight? Keep up the great work
Thanks for the prompt problem! Sounds cool. The first thing I'll say is that it could just be a dataset issue. Like, if for some reason, Midjourney didn't include a cyclops in its training, then it's not going to know what you're asking for. However, that does seem like a strange thing to be excluded ... so I would try image prompting a picture of a cyclops along with your prompt. (I don't think that will 100% work but it's the first thing I would try)
And as far as multi-prompting goes ... Off the top of my head, I'm not sure what words I would use in a multi-prompt or negative-prompt. Like, you could mention one eye but I don't see Midjourney doing anything other than removing one eye, rather than making it one eye in the middle like a cyclops.
Let me know if you crack that problem!
ancient greek cyclops with one giant eye --ar 4:5 --seed 1777668814 --v 5 ... check out #4
I'm not sure what the lense mm does. Like a portrait taken with a 10 mm lense would look vastly different than the results, it would distort the face showing a huge nose for instance, more like a fish eye.
Yeah I don't think the lens mm matters right now
Hey, I gave the '::' a shot with a negative number but it didn't quite work out. The darn bot gave me an error message saying the attribute weight can't be negative. I'm stumped, but I'll figure it out eventually!
You have to factor in all the other weights. It has to end up having a positive value. So if you only put a negative weight on one term, you have to add ::1 to something else to have a positive .5... Does that make sense?
If you put a negative two for instance you would have to add a positive two at least to make the composition zero.
Strange! I wonder why ... let me know if you figure out what went wrong
@@michaelbrown6398 perfectly, but from the video it seems to me that there is not a positive quantity but only a ::
@@FutureTechPilot Michael is right. MJ asked me for a positive amount to be added
I am writing the prompt with the --ar16:9 ... It says invalid parameters. -- test is not compatible with --version 5. What is the wrong part of my prompt
There needs to be a space between 'ar' and the ratio --> --ar 16:9
@@FutureTechPilot thanks I had to upgrade to midjourney basic plan before it works. But I generated some good portrait but when I set the quality to 4k it start generating cartoon skin for the model. Do you know how to fix that?
@@floridehounkpatin2110 You can't set the quality to 4k. If you put 4k in your prompt, it will generate something like a 3d graphic render. That's probably why it looked cartoony
@@FutureTechPilot aah ok I. Understand now. So I maybe after generating I need to go to Photoshop to get high quality images
@@floridehounkpatin2110 exactly!
9:16 ?
hmm?
👋
🤝
FYI- the term ‘Instagram model’ is not accepted. It creates a ‘terms of service’ error.
no way. is that cause of the new moderator? What happens if you appeal it?
I deeply apologize Mr. FTP. It's Cee -- Pee --Ah
😂 I had never heard anyone say it until the other day. Oops!