Unlock the Power of Midjourney v5: A Prompting Guide!
Vložit
- čas přidán 30. 06. 2024
- Midjourney version 5 is out for alpha testing (March 15, 2023) and I've got good news and bad news. There's some great news if you're a creator who likes realism. You're going to be squealing like a happy pig! But there's some bad news if you like MJ's default style.
The style is gone - or maybe just hidden away in a cupboard while MJ team plays with a shiny new AI camera.
Let's take a look at how v5 is different from previous versions and how you might need to change your prompts to get the best results.
**Share your favorite things in v5 and some of the parts you don't like comments below.**
Subscribe for more tips like these bit.ly/MtP-Subscribe
If this video was helpful, remember to like the video and subscribe to the channel. And a Thanks would be much appreciated!
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
VIDEO TIMELINE
0:00 Start
0:29 How to turn on v5
1:00 Default v5 style: Realism
1:24 Amazing hands!
2:49 Details in v5
3:55 How to prompt style in v5
5:31 Prompting tips for v5
6:53 Banned words in v5
7:39 Clean up prompts for v5
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
GET MORE!
►MIDJOURNEY
www.midjourney.com
►20 Art Terms you should be using in Midjourney
• 20 Art Terms you Shoul...
►Secrets to Creating Stunning AI Images: Expert Prompts
• Secrets to Creating St...
►UPCOMING VIDEO "More Magic Midjourney Prompts: 20 MORE Art Styles"
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
MY FAVORITE TEXT-TO-IMAGE GENERATORS
►MIDJOURNEY
www.midjourney.com
►Leonardo AI
leonardo.ai
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
GETTING STARTED VIDEOS
►►VIDEO: Starting Your Midjourney
• Starting Your Midjourney
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
FOLLOW
►Follow on Facebook / witsendphoto
►Follow on Instagram / jennatwitsend
►Follow on Flickr www.flickr.com/photos/jae_at_...
►Follow on Midjourney (subscribers) www.midjourney.com/app/users/...
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
MAKING THE VIDEO
Images by Jenn Mishra via Midjourney
Music by Jenn Mishra via GarageBand
Video created with Canva, RunwayML & Audacity
►CANVA
www.canva.com
►AUDACITY
www.audacityteam.org
►RUNWAY
runwayml.com
►GARAGEBAND
www.apple.com/mac/garageband/
STUDIO EQUIPMENT
►MacBook Pro amzn.to/3KR5awo
►Audio Technica 2020 usb microphone amzn.to/3Zhy1hV
►Stedman PS101 Microphone Pop Filter amzn.to/41BhE10
►BenQ SW271C 27 Inch 4K Photo and Video Editing Computer Monitor
amzn.to/3Ys6Z6g
►Sony A7RV amzn.to/3y8Z175
►Sony FE PZ 16-35mm F4 G amzn.to/3SLmncu
►Neewer Ring Light amzn.to/3mjAxW9
Some of these links are affiliate links. I earn a commission when you purchase products. - Jak na to + styl
Thanks again Jenn, really insightful and informative video with great examples. And, regardless of v4 or v5, there are still moments when I can't believe that this is actually happening, that AIs have gotten so good. It's unbelievable, I feel like I don't really know what's possible or impossible anymore.
I'm off to watch your Adobe FireFly video! 🤓
😊Thanks! I love being creative with MJ. I don't have access yet to Firefly. But Adobe's announcement looks amazing!
Thanks a lot! Many more useful tips than your competitors. The tip for not having to type stuf like "insane detail" etc. saves me a lot of time Thumbs up!
Glad it was helpful! It's hard to keep up with the changes.
Always love your video's, Jenn! Great tips and to the point. Very helpful, keep 'em coming!
Thank you! Will do!
You've done such a great breakdown and comparison between version 4 and 5! I resonated a lot with the "party trick" idea you mentioned, v4 was more geared towards adding amazing elements to make the image pop. With version 5, we need to basically be more literal with what we want and while it does make it harder for us non-artists, it also gives us the freedom to be more specific with what we want. One thing that did help me though, is using the --stylize 300-1000 to the end of v5 prompts. I got much more detail and flair, it was very impressive! Thanks again
Thanks! V5 just needs a few style prompts and it makes a huge difference. But, yes - that's hard for those of us who haven't gone to art school 😀 I've been playing with the parameters, too.
Yeah, version 5 is going to take some relearning how to prompt for it. I found myself going back to version 4 largely because I prefer the more artistic style that midjourney is known for.
I like v4, but as I get used to v5 prompting, I'm seeing the amazing image emerge. I'll keep learning what makes v5 tick.
Agree there, I have been going back to this too
Me too! Can't stand MJ5 and now a person can't type "Chest"? Are they insane?! Why not just say "We won't render the human body accurately" Chest could me wooden chest or Breast could mean "Breast Plate" - You can not create an accurate human without knowing the medical terms of each part of the body. The renderings don't create anything "artististic with alternate versions" so the artist can get new ideas. Like one critic put it, up until now, MJ was a great source of inspiration and as she put it, she would run MJ5 and within minutes, get bored of it and say "big deal" an not respond. Not all of us want photorealism. In fact I am sick of photographic quality or photographs for everything.
Thanks so much, This video was quite helpful. I especially appreciated seeing MJ going wrong so many times, which is my usual experience. I recently used Blue Willow that's completely free and super cool
You're very welcome! I've been exploring BlueWillow a lot since their v2 update.
Very informative. Thank you.
No problem!
Awesome, great summary! This stuff is way too fun.
Glad you enjoyed it! I'm having a good time!
Fantastic video and tips!
Thanks so much!
Never used v4, just started using Midjourney 2 days ago. Thank you for making this video, I’ve noticed the option to use previous versions in settings…you’ve inspired me to try out 4🙂
Welcome! If you're not adding --v 5 to the end of your prompt, then you're using version 4 by default.
Another fantastically helpful and perfectly explained description of the differences. You are not a Goddess in the mind blowing world of AI art and evolution, you areThe Goddess. Hope you get 100k subs soon.👏🌟👍
Wow, thank you! Tell 99,999 of your friends! 😜
I like midjourney_v5 I have more control over the final result. Thank you for the video.
Thanks for the feedback! V 5 does respond to really well to the right prompts.
I like the idea of a realistic image as the starting point. For me, it makes things more predictable. Then I can modify with additional prompts/commands as needed.
Good point. The image is the starting point. Personally, I can make the realistic images. I have a soft spot for the digital art look in AI.
Somehow I have SO many more difficulties getting v4 to do what I want than v5. And recently, i discovered v3, and somehow it's the best thing that ever happened to me. My entire life is upside down
😀It sounds like you're having fun exploring all the different looks in Midjourney! Hey, roll with your vision! I like v3 style a lot. If you want one more thing to explore, try making something in v3. Then remixing into v4 or v5. Let me know how it works!
I love it when a new ai model releases and it makes the other one look like a toy.
Just when I think it can't get any better.... WOW!
Thanks so much for making this video, Jenn. I knew there was something I felt uncomfortable about with V5, and you've hit the nail on the head - it's the lack of playful imagination, and the higher barriers that you have to navigate to get "art" rather than uncanny photos. It's still great, it's just lost it's mysticism, sadly. I don't mind working harder, it's just sad to feel like I've lost my creative partner ("Midj") to a more serious "realist". Lol. I wonder if your previous ChatGPT scripts will be more helpful at creating more specific prompts that help us tease out Midj's now more "hidden" imagination?
Thanks for the feedback! I'm sure the MJ style won't stay this way - and I was happier when I found that a few style prompts made v5 images sing! But it's fun to put in some random nonsense or an idea like "the apocalypse" and have MJ come up with something amazing!
The decision to make V5 less opinionated is right on point. It will encourage people to think more artistically and expand their knowledge of styles, composition etc. The spoon fed option is always there for use in V4 for beginners who aren’t necessarily interested in developing their own artistic integrity and just want a quick computer iterated pic so they can share on Insta and claim as “all their own work.”
Yes! Being able to talk about art and visuals is key to AI. That's why I do my series on prompting art language. Another will come out soon! AI is making us all a bit more thoughtful about what we see.
I agree. I've only just started with MidJourney a few weeks back, so was "brought up" on v4. I've now played around with V5 and the details are incredible. I've also built a "prompt generator guide", nothing amazing, it's an excel spreadsheet afterall, but I've added various prompts in a dropdown that allows me to access things like lighting, artist, Style etc. and it's actually made a big difference to the outcome of the images. Quick output of an image in the style of Jackson Pollock, or H.R. Giger, and you can see the results. It's also forced me to go read up on my favourite artists, getting to know them a bit better, so there's some art history classes included if you MJ growth curve.
Ultimately, you need to get to know the system and fine-tune your style - it also helps to document what works, and what doesn't, to avoid future failures.
I started on v3, so this is my second big change in prompting! I'm learning a LOT of art history!
Great job! :) It's funny how learning how to talk with AI models also increase your vocabulary. Even if creating the prompt is a silly task you still have to put some thoughts on it
I get your point but I don't know why your comment became bashing. There is a market for making quick images when the art is not as important as the story that's being told with it. Open your mind my friend.
The stylize and chaos parameters also introduce more artistry. And the current version seems to be a mode you can set to be the default or not. So it will all be different again on a short notice at any moment in the coming weeks, probably.
Thanks! This video got long so I'm covering the parameters in a separate video. I'm not noticing any difference in the --S and --C commands. But maybe they just aren't turned on right now.
We searched for more info on MidJourney V5 and discovered its currently running in Pro mode that will be an option after V5 non pro mode becomes the default. Yes the Pro mode dose require more descript prompts with more control over your images. V5 Pro is the future of AI in our world. Keep Creating your changing the world at the speed of light.
Thanks! Yes. There is talk of a "friendly" v5 that brings back more of that smooth MJ style!
Artists prefer this way... Now it's up to you to craft the perfect prompt. I love it personally.
Great! Yes, v5 needs some prompt craft. There's talk of a friendly MJ model. That would be great for the beginners. Then we can move to the pro mode for more advanced prompting.
@@MakingPhoto beautiful compromise!
Thanks!
I thought v5 was terrible for print on demand at first but now the stickers and tshirt designs end up coming up way better in most cases, in some i'll use v4 instead but in most if you craft the prompt right they come out way cleaner and closer to what you're looking for.
I agree! The trick is in the prompt. You could also try making images in v4 and remixing them into v5. See if you can take advantage of both.
So cool...very nice video....
Thank you 🤗
It might not be for others but I prefer it works like this, where you can to say the style you want. I'm not finding I have to add too complex prompts to get back to the digital illustration style of v4. Don't forget we can use the --s and --c parameters. I prefer having more control. If you prefer v4 then v4 won't be going anywhere so just put --v4. It's good to hear a "friendly" v5 version is coming I didn't hear about that so that will help all sides happy. :-)
The banned words is going up but the mods said they really want to cut it back so watch this space I guess.
Thanks, Kol! I'm jumping back and forth between v4 & v5 right now. I think a few style words work wonders in v5. I just finished a video taking a deep dive into the v5 parameters. But I'm not seeing a difference with the --S and --C commands. Maybe they just aren't turned on yet.
@@MakingPhoto I tend to forget about them and just use words so that could be the case.
it is very different indeed, but as a photographer i can't complain, thanks for the video it was fun watching 🤓
Glad you enjoyed it! I'm a photographer, too! I've been trying to prompt photos since v3! 📸
@@MakingPhoto started again 3 days ago with v5, the eyes and hands are still a problem sometimes but other than dat it's amazing 🤠
I still get a few off hands and faces. But v5 is so much better! I was just playing with Adobe's Firefly. The faces can be a problem. But when I ask for hands - I actually get pretty good hands!
Nice video...again!
It's definitely a little struggle to get comparable results witch v5, because you (again) need to "re-learn" prompting... Still a thing that is probably a no-go for professional use. at that point...hey professionals...have a look at Adobe's new ai image generation tools...that is probably what you have been waiting for ;-) but when you get your v5 prompt right, it's mostly awesome 🎉
Thanks! I think we'll catch on to v5. We just need to remember to add a few style prompts. That seems to be the key to opening the v5 power!
I prefer this, it is a lot more like SD, you have to KNOW what you want and be more specific--yes that is more challenging for beginners, but it also takes off the midjourney training wheels so to speak.
You're so right, Joshua!
I used the word chest yesterday and I had no problem, today I see your video and it's about march actualization, maybe it was unbanned?
Yes - since this video, Midjourney has completely changed the way they treat banned words. They've gone to an AI version (not a list). This has helped a lot. I occasionally still trip the censor, but I can easily appeal the decision and my prompts always go through.
Hi, a newbie here thankyou for all your insights. I found 'no' in certain area refuses the prompt . A recalibration of wording is necessary in each are. Using V 5 and the added indigineous dot art is crazy with animals as hand feet are all over the place.Omg as I was just getting used to version 4 and niji for illustrations. Faces with indigeneous V5 does not give me the variation of faces in Australian Aboriginal only few and looks like american actors black, lol. Also worst is the Copyright Criss Cross Stock Photo is all my aboriginal animals style art with signage and words . It is like all the signaures came together...
As an artist I feel merging a photo will most likely work better which I will try later. Thank you for reading.
Thanks for watching! V 5 is sometimes TOO real. You can prompt in v4 and then try remixing into v5 to see
What that does. It might be the best of both worlds.
V5 seems slower because it's upscaling all images in the background before they show the thumbnails. AI technologies like Blue Willow will become more and more popular over time. Blue Willow recently launched V2 with a brand new model upgrade. Now I got more better-quality images with more aesthetic outputs
Just today the MJ team announced some changes that should make v5 faster. I'm almost out of fast hours for the month! Will have to do videos on other AI!
Time traveling penguins.. 🤔... that explains a lot of this world. Lol
😜 There should be a movie!
We thought that one think that can never be replicated is the human touch. Turns out it was one of the easy ones.
It's still hard on other AI image generators. Midjourney just has some secret sauce 😀
Hey hey, im new to this, I hope someone can help me. You all just use less words and get stunning results. When I do it, it does not work well. I used the same prompts as a friend, but my picture looked like drawn by a child, also the perspectives were weird?
The prompts were: unique flower field, colorful, sun.
Question: Is it nececcary to use all the codes like '-- quality' or whatever, or are they already 'right' in the default mode?
Hi! Welcome to AI image generation! It's a lot of fun, but there are some tricks to getting good images (that's what we do here!) Did you use the same AI AI image generator as your friend? That makes a difference. Midjourney AI and Leonardo AI have their own styles that they add to your prompt. So, you can use simpler prompts and get a good result. Some other AI image generators are based on Stable Diffusion and don't have a base style.
Booty cache. 🤣I'm going to have to start using that.
😜 Let me know how it turns out!
Surprisingly both 4&5 still can't create natural looking red eyes. I had fun inputting names and changing versions as well as detail setting to see what would come up.
Whoa! Show me the red eyes!
Its easy .. you want realistic photos = v5
Imagination photos =v4 ..
And i think when u use the parameters they talked about will change the image style .. and its beta version so its not the final release.. thank you
Yes! In office hours today, David was talking about the style. The initial ratings for v5 favored realism. But creators are making the style more artistic. There will be some changes over the next few weeks.
Is there a collection of the banned words somewhere?
No. The Midjourney team has been asked about this a lot, but they don't want to release the list - it may motivate people to come up with work-arounds. I don't often run into banned words, but it depends on what type of images you're trying for.
BTW, did to see Adobe Firefly announcement yesterday? Wow!
I did! What do you think? I put an announcement out but I'm still on the wait list. I get all sorts of communications from ADobe and only saw the announcement yesterday before the launch!
@@MakingPhoto It looks stunning. Does a lot of what Stable Diffusion does, including some of the extensions. I believe the beta won't let us use our own images so presume it will only be images created within Firefly. So that's a shame.
@@KolTregaskes thanks! I waited until the launch yesterday to get on the waitlist -- grr now I wait!
Nice video
Thanks!
I think v5 seems slower because it's upscaling all images in the background before they show the thumbnails. AI technologies like Blue Willow will become more and more popular over time. Blue Willow recently launched V2 with a brand new model upgrade. Now I got more better-quality images with more aesthetic outputs
Thanks! I play on BlueWillow a lot and with it's new version, It's probably time to take a deep dive. I'm still seeing some typical SD things with BW. I haven't noticed a slow down in MJ. V5 is not actually upscaling - but it is making larger images to start with.
I do very much prefer Midjourney v5 default photo settings, over the digital art style of v4. 😊
I'm torn. I like the v4 style. But v5 is so much more realistic!
@@MakingPhotoThanks for your reply Jenn, it really is appreciated.
Personally I do like the hyper-realism of Midjourney v5, as I am trying to get my AI painting, and also hand-drawn painting as photo- or hyperrealistic as possible.
I am very much also influenced by the artwork of an Italian hyper-realistic painter called Marcello Barenghi, who does extremely accurate hand-drawn paintings that actually look like photographs.
I am currently trying to do some photorealistic “Dungeons & Dragons” fantasy art, with the help of mixing ChatGPT and Midjourney v5. However, due to it being fantasy and set in ancient times, rather than modern, the AI always seems to produce “painterly” (similar to a “Dungeons & Dragons” manual), rather than photo- or hyper-realism artwork (like a “National Geographic” photo), as it seems to do with 1950s “James Dean” post-camera, modern or futuristic subject matter.
Hopefully you could do a Midjourney tutorial on photo- and hyper realistic “Dungeons & Dragons” fantasy artwork, with the right prompts to use for photo-realism, to break out of the “painterly” mode the AI seems to have for this subject matter.
All the best for the future...😊
Can I continue to use V4 in the future?
You should be able to. We can still use V3 (and I think v1 & 2)
A new decision has been issued now
midjourney error:
You are not allowed to use the relaxed flag 👎
Hmmm. They were having problems with GPUs. The v5 images are so much larger than v4. Give them a little time. This might be temporary.
@@MakingPhoto Yes it's back to working normally now.
Can't we simply keep using v4 by adding the prompt --v 4?
Yes. Right now, v4 is still the default model so you don't need to add anything. But soon v5 will become the default. Then, you can add --v 4 to keep using the earlier version. And --style 4a or 4b if you want to use the even earlier version.
She said Midjourney V5 is more detailed and less realistic...I thought it was more realistic because everything looks photo-realistic now with this latest version. 🤔🙃🤷🏽♀
I don't think I said it was less realistic. It's very photo-realistic as a base model.
So you can't just pick v4 from the settings menu and go on as you were?
You can! You can make v4 your default
Right now, v4 is the default. If you don't do anything, you'll get a v4 image. If you want to use v5, type --v 5 after your prompt. And as @nahiddotai says, you can make either your default by typing /settings and choosing the one you want.
@@MakingPhoto
If you type /settings and hit enter, V5 will be there, just click to highlight it and it's set by default.
@@E_Stew Yes! Great tip!
Can we still use version 4?
You bet! Prompt --v 4
can v5 replace prof photographer?
Well, I'm a professional photographer and I've been using AI for years. We probably don't need stock photographers any more - but they've not been earning very well the last few years anyway.
How to stop a previous prompt from appearing at the end of a new prompt? Help, please!
Hmmm. The only thing I can think is that you have something in your /prefer settings. Type "/prefer suffix" then hit enter. If this doesn't work, can you send me the prompt. Maybe I can figure it out.
@@MakingPhoto Will try...
I tried your suggestion and am happy to admit it corrected the issue. I will subscribe to your channel. 😎
Great to hear!
wait chest wasn't already banned?!
I think it's new. I hadn't actually tried prompting for it until I heard the buzz. The MJ team is pretty open about the longer banned word list.
happy pig
😀😀
As someone who doesn't use Midjourney, but is into it and uses other AIs, all I can say is that most "artists" or people who use Midjourney don't even want to make the slightest effort to generate proper prompts. On the one hand very amusing to observe how exactly these people are now upset about having to get creative, on the other hand it's quite sad.
In art, you have to accept people where they are. Some people can write long prompts and talk about art. Others didn't have the education. I love a tool that meets people where they are.
@@MakingPhoto YOu might be right. But there are way too many people taking advantage of a tool to call theirself "artists". You can see it daily on Instagram. And I'm not gonna lie when i say, that i am happy to see those people struggle.
I'm a professional photographer. Everyone takes photos with their mobile phones. So, anyone can say they are a photographer. I don't really care. If they feel like a photographer, then they are a photographer. You're really talking about the quality of the final work. There are photographers way better than me. And I'm way better than a lot of professional photographers. There are photographs that are right for THIS situation. But wrong for THAT situation. This is always the way in art. It's subjective and variable.
@@MakingPhoto The difference here, however, is that you can always tell a good photographer from a not-so-good one. It is also very difficult to recreate or even steal a good photo and present it for your own work. AI Art tools like Midjourney require little to no knowledge to create a really good subject or image. It's different with Stable Diffusion, but even there it's relatively easy. I respect people who dive into a subject and get rewarded for their work and dedication. And then there are always the people who steal prompts or just take advantage of a tool and pass it off as their "work". But I understand your reasoning and by now I don't get so upset about it either. Everyone as they wish
I don't even know what a prompt is?
Ah - you need to start here czcams.com/video/zPcnHFYxXxk/video.html
AI camera
Yeah - sort of. :)
resident evil 4 remake walkthrough
Not being able to use words to correctly identify medical terms of the human body like breast I might be able to overlook, but "Chest" is taking it too way far as what are we now a "Puritanical Church with Beavis and Butthead" leading the choir!? "She said "Chest"! Heh, heh....." how sophomoric. Fine, I'll use words like upper Pectoralis Major and Minor Muscles just above the sternum and breast bone - oops (can't say "breast" bone). Midjourney needs to decide to either be an artistic masterpiece tool, or just another plugin for Photoshop Photos in LightRoom. Blue Willow and even Adobe are coming up fast and Blue Willow is 100% free where I'm paying $60 a month for the use of MJ. I hope they come off their high horse about words as there are millions of sites to see those other types of images.
Good point! This isn't limited to Midjourney. Most of the AI have some sort of blocks unless you're making models in SD. MJ is mostly responding to some bad actors.
@@MakingPhoto Exactly! A few ruin it for millions.
It needs to take into consideration the context in which these banned words written in. It's extremely annoying!!
I think that's the goal, Benji. But the team is struggling to keep the content PG. We just need to find some work-arounds right now.
idk what i want
Keep creating until you do!
@@MakingPhoto thanks for the reply! wish i would have asked a proper question
You think prompting creativly is hard ? try make the are itself yourself ...
I do. Look at my bio.
Funny to hear all the people who suddenly found a "digital artist" in themself whine how troublesome and boring it is to actually spend TWO MINUTES to word your prompt properly, instead of just tossing around sounds and buzzwords, generating clichés in teal and orange on a never-ending dopamine ride. How disappointed they are that "art" isnt about just pressing the button and owning the result anymore.
I'm a classical musician and professional photographer - I never hear artists whine about anything they have to do.😜 I have 2 hours of practicing to do today and a 3 hour rehearsal. What are you doing today?
@@MakingPhotoYou just did - what you call "creativity" is a meatgrinding generator photobashing somebody elses work for you.
@@f4ust85 You shouldn't use words if you don't know the meaning. I've never used any AI art program but I've spent 10 minutes to read how it works. This is not photobashing. Unless you don't know the meaning of 'photobashing'.
@@Matt-bp5vy Thanks captain Oobvious. I know very well how it works. Its not really "AI" either. Call it what you wish, thats your copium. Its still cooking a smelly goulash from the dataset of Pinterest and Artstation.
Sadly Not really helpful . I thought you will explain POWERFUL prompts :(
What do you mean "powerful"? Try my video with lighting and sensory language tips. Secrets to Creating Stunning AI Images: Expert Prompts czcams.com/video/Od3FRMLqwFk/video.html
All this prompt re-learning costs time and of course money.. do they do it on purpose ? why the fuzz >?..anyway thanks for your informative video.
You're welcome. The images are improving, but we have to learn with the AI.
yes because midourney was using working of other millions of artists out there.
OK - so how do you think artists learn to paint?