great job ! there are tons of resources available on prompts but urs stands out for berivity and no nonsense approach , keep up the good work we need people like you
Ah, thank you so much!! To be honest, that’s something I sort of position the channel as: I don’t want to spend 30 minutes waffling on about prompts! Much better to show the underlying logic and get you all out and creating as quickly as possible!
Fantastic to hear! The kind of unofficial sequel to this video is the one on "dash dash" commands-- you might find that helpful as well: ua-cam.com/video/56TnVebX6HI/v-deo.html And as always, if you run into any problems, please reach out! Always happy to help!
Based on the transcript you provided, here are some key pieces of advice and concepts related to prompting for image generation: Brevity: Keep prompts relatively short due to token limitations (around 77 tokens). Medium: Experiment with different mediums such as photo, painting, TV show era, comic book illustration, etc. Style and Composition: Explore different styles like Pixar, Tim Burton, etc., and consider how composition affects the overall look. Camera Angles and Shots: Use specific camera angles and shots to direct the image generation process. Scene: Describe the scene with details like actions, props, and location to influence the generated image. Atmospheric Effects: Specify atmospheric effects like lighting, fog, weather, time of day, and seasons to set the mood. Chaos Command: Utilize the chaos command to introduce variability and stimulate imagination. Experimentation: Encouragement to experiment and have fun with different prompts and styles. Some key words and concepts involved include: prompt framework, medium, style, composition, camera angles, shots, scene description, atmospheric effects, chaos command, experimentation.
It would be exceptionally helpful if you showed the prompts you were using on the screen in text versus just voice. Other than that, excellent video. Thank you.
Thanks very much! I thought about that actually- the prompts are all in the PDF, I kind of wanted to keep the image example textless. Although, my dumb mug shows up from time to time!
I really enjoy seeing these template things. I independently came up with the following for art prompts: [art style][artist][medium] of [subject][descriptor 1,2,3][technical parameters] where descriptors are things which add information such as what he gave, but some other examples would be time of day, sky information (e.g. Golden hour, interesting clouds), weather, texture/pattern, emotion, season, color palette, blur, lighting, etc. For photos it’s very similar, but I keep the descriptors to just two, and add [camera info] as a section, which includes as required the camera, film, and lens, but you could add more optional camera info to this as desired.
Love that! And it really goes back to what I always say: there is no right or wrong way to prompt. The whole “prompt guru” thing that was popular awhile back has thankfully phased out, since I think most people see the snake oil. It really all come down to concepts and tokens. The vocabulary to express what you want to see, and to some degree, the order in which you present it to the bot. That’s the fun and art of it!
@@TheoreticallyMedia MJ is very quirky thing, so sometimes the weirdest things give great images. Your point is right on, that understanding how it works is what matters.
Fantastic to hear! Please do keep experimenting though! I always urge people to play around with this format and make it your own! Consider it a foundation to build your own castle!
As a rank beginner learning about prompting is the big hurdle. This video gave me ideas of how it is actually done. Can't wait to see more of your videos. Thank you for your time and effort.
I've been loving using the Bing prompt "A 1950's painted science fiction book cover." It's fun using different decades, and "science fiction" might put in neat things like planets and robots and such.
I've been using many of these AI portals for some time now, and believe the real key is a creative imagination. The more you can dream up, wildly dream away, the "better" your images will be. Really enjoyed this video. Your clarity is better than I could have explained it, and there are details here and on your other videos I was not aware of, or clear on.
happy to have you aboard! Hoping to have some new Midjourney videos soon! In the meantime, I've been covering a lot of AI Video and Music as well, if you're interested! Once MJ drops V6, there is going to be a LOT of /imagine!
This video came out right when I needed it. I’ve been struggling with my midjourney prompts. It seems like most of the times it does the opposite of what I asked for lol I’ll definitely be experimenting with this structure. Thank you for sharing your knowledge sir🙏
excellent to hear! If you have any questions, or would like any clarifications etc-- please swing by the Discord! Or, hey: You can just pop up images you like in the AI Art Showcase!! discord.gg/6PMENW3k
Another tip and demo-filled video with just enough side remarks to make it entertaining and not annoying. I am so very appreciative of your guides and will add them to my notes. I even dig your font choices! Glad I subbed. Keep rockin'.
Oh man, thank you so much! Haha, it’s a real balance with the side comments. One day I’ll release all the outtake jokes that don’t make the cut as an epic blooper reel. It’ll be even more baffling, since no one will know the context of any of the jokes! But for real, thank you very much! And I’m really glad the guides are helpful, I find them weirdly fun to make.
@@giri.goyo_yt Thank you so much! And yeah, Gumroad seems really solid thus far. I know I'm not using 70% of what it offers-- I need to dig into all that at some point. Let me know when you're getting set up and maybe we can chat up some ideas about it!
For the dash dash video, one thing I like to do is reroll blends stepping up the chaos every time (15, 25, 35, 50, 75 etc) it's great because you get more variety than you get from just rerolling without adding prompt direction
Hey Tim im so interested in getting a mid journey plan, and I was looking around for someone great here on UA-cam to explain for me. And I can say you seam really cool smart and interesting. Thanks!
Merci beaucoup! J’espère que cette traduction est exacte, mais j’apprécie vraiment le commentaire, et le fait que cette vidéo ait fait le tour du monde! En espérant visiter bientôt, et manger tout votre pain!
(Medium) (Style/Composition) (Scene) (Modulate) (Dash, Dash) (Cinematic still) (Style by the cohen brothers, long shot) (business man walking down a busy city street ) (Green and yellow color pallet) (--ar 16:9)
excellent! You were exactly who I was hoping would see this! Please feel free to poke around the channel, there's a lot of other MJ tutorials on here as well!
Tim, thank you for the video and the summarizing PDF. You know how much I appreciate the intelligence and professionalism you bring to your videos. On that note, it wouldn't be a bad idea for you to cover the /describe command of Midjourney. I've been using it to give a defined style to all the footage on my UA-cam channel, and it seems to be working well for me
“Sir, should we activate the Voltron protocol?” “I don’t know. He’s not really knocking down buildings or anything. I think he’s actually kinda depressed…”
There were just _way_ too many priceless lines in this one! Loved it. Thank you for the video. 👍 Might I strongly suggest people be very careful with the chaos value if they're using fast hours! _Yeah, the K. Effect was definitely working when I saw Hitchcock and then the nice young lady on the grass._
Ah, true, I should have mentioned to use chaos with Relax. I'll pop that into the followup on dash dash! And the Hitchcock thing is such a funny example-- particularly considering he's the one that wrote it! He used that very set of images explaining editing in a CBC news program! Hitchcock had a pretty great sense of humor.
@@TheoreticallyMedia I didn’t have the courage to mention the prompt I wrote that produced a woman with 3 arms and legs. Even “free” (Leonardo AI) costs tokens. Thanks.
7:42 how do I ensure that the results that I get include the character with their back to the camera. I find myself putting several prompts in to say character should have their back to the camera, but I still get faces.
one trick there-- and this is kind of funny, considering in the early MJ days, we were always trying to do the opposite (since characters always stood with their backs to camera and we were trying to get them to face camera!)-- But, let's try something like this (using our pal, the Man in the Blue Business Suit: /imagine Rear view, from behind, a man in a blue business suit. The suit has a vent in the back that is slightly open, suggesting movement. The back of the suit jacket is smooth and well-fitted, with a noticeable sharp crease down the sleeves. His posture is upright. The back of his neck is visible. By reenforcing "back" and giving details about the back of the suit, we're basically telling MJ, "give me a back shot"-- It might not be the most exciting shot in the world, but it'll work! Color as needed!
@@TheoreticallyMediaThanks so much. I currently use it to create flat vector illustrations for minimal wall art. I want a prospective buyer to “see themselves” so I don’t usually want a face or too many details in the final result. I will try to use some of these principles. Thank you so much for responding. Much appreciated.
Hi. An excellent video. Very informative. I am making audio songs with just one cover picture. What I want to do is find an image generator software that would show this female dancing for between approx the length of my songs rating from 20 seconds to 4 minutes who might mean looping. Any advice would be very welcome. Ray
Grab the Free PDF Guide Here: theoreticallymedia.gumroad.com/l/Imagine If you like the work I'm doing on the channel, please consider joining the Patreon! patreon.com/TheoreticallyMedia
Hi Tim, thank you for all the videos. They help me a lot! Would be possible make a video talking about the control of the image? For example: if I try generate "nina simone wearing a floral dress in front of a striped wallpaper background" (or something like that), MJ just mess things up. Like striped dress and floral background. I tried photobashing, ::, --iw, reference link, and all the parameters. What I mean is that I feel theres no way to reeeeeally get the control of it. If you ask for a specific color or pattern, etc MJ just tend to ignore others words of the prompt. Thank you!
Hi Tim: I very much enjoy your videos and I know that your prompt tips will go a long way to getting me off to a good start. After having studied so many videos on UA-cam and Vimeo I am fascinated by those that I see by creators like "Friendly Advice", "The Panavision 70", and others of course. Not necessarily the 'big buxom' aspect of these videos but the whole look, the slow motion movements, the highly stylized futuristic themes. It is simply amazing but I do not know where to start. I can find no guide on what the general workflow would be to go from creating the image to to then having a stunning video clip. I am not even sure if they are using Midjourney and if so in what capacity as it relates to other AI tools in the chain. Any push in the right direction would be greatly appreciated. Thank for the work you do!
In midjourney i have used "facing camera" and "back to camera" successfully. Also do you know how to do get max Depth of Field? I designed a room in MJ but i have tried everything i can think of without success. No blur, no bokeh, f/22, everything in sharp focus etc etc. Any ideas please?
you can try going simple, like "blurred background" or I have used f2.8 in the past and it has worked. Hop over to the Discord and we'll see what we can do!
im confused on how to add sections? obviously when you press enter it sends the prompt. looks like you've arranged the prompt using enter, so how do we do so please?
there are no rules with prompting but 'the Parameters are always added to the end of a prompt' --Max Aspect Ratio,--Chaos,--Image Weight,--No,--Quality,--Repeat,--Seed,--Stop,--Style,--Stylize,--Tile
Oh yeah, that is for sure a rule! I mean, I wonder if anyone has tried putting the parameters at different places in the prompt? Would it work? Only one way to find out!
Hi there - great video - I went to your site to buy the pdf - it won't take my card, I am guessing that is because I am using a visa gift card I have - is there another way to get the pdf? Thanks!
Oh totally, just put 0 in for the amount and you should be able to download it. If not, drop me a line at theoreticallymedia@gmail.com and I’ll email it over to you!
... well, actually there's a snow scene in the Blade Runner 2049 movie ;) Thank you very much for all these awesome tips!! I'm very excited to start experimenting with AI.
I'm still kicking myself that I forgot about the Blade Runner scene! Welp, time to go watch Blade Runner again!! And thank you! There's a lot of resources here on the channel, so please do poke around! Ley me know if you have any questions!
User we dont have a 'chaos' command with Dall-E but if i want randomness i'll tell you. So we could start with me giving you carte blanche in choosing a 'Hero' for our scene, the location , the atmospherics etc. Your constraints are 1- follow the recommendations from the tutorial and 2- the prompt produced must be maximum 480 characters. ChatGPT Understood. Here's a prompt following the recommendations from the tutorial with a maximum of 480 characters: "Photograph medium shot of a courageous astronaut exploring a mysterious alien forest under the eerie glow of bioluminescent flora. Blue color palette enhances the otherworldly atmosphere. Camera angle: Close-up. Atmospheric effect: Misty fog surrounds the astronaut, adding to the sense of wonder and discovery."
This channel has become my top preference for MJ. I have a question regarding the creation of a detailed scene or photo featuring multiple characters. For example, I want to create an image with 2 couples. Specifically, I would like to know how to accurately depict the height, physique, hair color, age, and positioning of each person in the image. Lately, despite the last two updates, I have been experiencing issues where the generated images consistently depict anything but 2 couples I adked for. Instead, I get, three men or two women, and one man, rather than the 2 couples I inputted. Any tips on how resolve?
Thanks so much! I’ll say, positioning/posing can get a little tough in MJ. Stop by the Discord (I think I linked it here) and hop in #Projecthelp and let’s see what we can do.
Greetings. My question is, If I use a picture of someone's face, how can I force midjourney to stay loyal to the features of the person? Ive used real pictures of people but the result changes the person to the point it's just someone else, sometimes with too much filter. For example Ive used the picture of a man with a beard and the beard looks almost like makeup or airbrush but not real facial hair.
Yeah. That’s a “feature not a bug”’of MJ. Mostly to avoid abuse with face swapping and creating harmful images. So, I kinda get it. You can use a tool like InsightFace to accomplish it though, you can check out a video I did on it here: ua-cam.com/video/a3MOPd6pEOg/v-deo.htmlsi=wDbTyKbXN9-WuiUn
I am constantly getting an Invalid parameter Unrecognized parameter(s): AR /imagine Photograph, business man walking down a busy street, Blue color palette --AR 16:9 --s 250 the --s250 was added by mJ I am using model 5.2 / stylize high / public mode / remix mode / low variation mode / relax mode a little help thank you
I tried everything. What did it and made it work was a change in my settings from model 5.2 / stylize high / public mode / remix mode / low variation mode / relax mode - to - model 5.2 / {stylize med} / public mode / remix mode / low variation mode / relax mode. What or why I have no idea. Thanks man just getting started on this journey. I want to create scenes to use as a story board then set up and photograph. I was a photographer, then went to the dark side as a retoucher and now as AI and photoshop have become so user friendly. I think a Photographer again , but not assignment work. I think it will have to be film. Wish I had not liquidated my med format equipt. Thank you for your info @@TheoreticallyMedia
Thank you! Yeah, I try to avoid the 30 minute waffling, or just chatting with no editing. That “podcast” type video style works for some, but it generally drives me insane.
To be honest, I keep these pretty simple, it’s usually just “Stunning” followed by a word that is somewhat related to the topic. But yeah, Stunning seem to trigger this as an MJ house style.
I did the prompt on that one-- Usually my format for these types of thumbs are "Beautiful + whatever the topic of the video is" It'll often taken a number of rolls and I need to play around with variations on the how to word the topic, but eventually I find something I like.
Is there any way to preserve a model and make variations with the same face? Apart from the swapid engine, is there a prompt or another engine like swapid that allows me to always generate the same model for everything or? Is another external program required?
Not...yet. There are some tricks to get a similar output, but MJ has been weirdly steadfast on not being able to repeat the same character. They've been talking about "Storytelling" tools in development, which I think solve the problem. My guess is that it will be a suite of features that will consist of Character, Location, and Emotion...but we'll see. In the meantime, I generally will use a combo of --seed and prompts that direct a very simple character: "Woman, mid 20s, blonde hair, black dress" -- That simple "Blonde Hair/Black Dress" will usually get you in the ballpark with a few re-rolls, and then you can face swap from there.
The human's eyes look dead, or like the human is in a daze. The dead eyes makes the whole image look fake. When I have tried to give Midjourney directions on how to imagine the eyes, the whole composition changes to an extreme closeup of the eyes.
I tried these prompts today, and I do notice the difference. I am able to get a lot closer to what I'm looking for in an image. You mentioned something about the order of things, so I rotated the order to get a more historical feel for the shots. I guess I'm wondering where in the prompt the command starts to trail off into something I didn't ask for? Do you have suggestions that would help keep images more attuned to their time in history?
So glad that is helping!! So, to answer your question, check out this video on 5.2 and take a look at the Shorten command. There's some clues in there w/ how MJ "sees" your prompt: ua-cam.com/video/1tIXOwNQuT8/v-deo.html
thank you for this video. I subscribed to your channel. I've used midjourney for coloring pages but not getting the results I want. Can you offer any suggestions? I am also learning photoshop so I can convert if needed.
I watched the above video of yours and find some of it confusing. You list your prompt framework like this: (medium)(style/composition)(scene)(modulate)(dash-dash). When you talk about styles in the video, you basically replaced the Medium (photograph) with the Style by Pixar and did the same thing with By Tim Burton and stopped there. If that is the case, why doesn't your prompt framework look like this: (medium/style)(composition)(scene)(modulate)(dash-dash). I'm really puzzled by your prompt framework based on how you explained it in the video, which left me wondering what other Mediums are there besides the ones you talked about in the video. I wish there was a list of Medium types, like your list of Long Shot types. Actually, from all the prompts I studied I think you are only one using the word Medium while other frameworks seem to use Subject. Also, the word photograph seems more like a style than Medium. I tried to google medium types and it didn't help. But if I googled for styles I got this: isometric anime analytic drawing infographic drawing coloring book diagrammatic drawing diagrammatic portrait double exposure 2D illustration isometric illustration pixel art futuristic style ornamental watercolour dark fantasy paper cut craft paper quilling patchwork collage iridescent ukiyo-e art watercolour landscape op art Japanese ink pastel drawing dripping art stained glass portrait graffiti portrait winter oil painting anime portrait cinematographic style typography art one-line drawing polaroid photo tattoo art If I understand your video you would call the above Medium while others seem to call them styles. Any thoughts?
Well, I'll say that sometimes I get fuzzy with my prompting. I like to think of the framework as a basic guideline, almost like seasonings in a recipe. I think in this video I mention there is no right or wrong way to prompt, just an organization that I think helps Midjourney understand what to prioritize in your prompt. That list is pretty great! A fun thing to do is mix them together as well, say "Japanese Ink and Graffiti Portrait" hmmm, might have to go try that!
@@TheoreticallyMedia Here is my attempt to come up with a prompt guide cheat sheet and it's components. drive.google.com/file/d/1vl2u21FN8ZHdNcNSPXPP6w65Ih0RRjmV/view?usp=sharing
From my experiments, it is. Although things are changing, this still works. The language model is smart enough to know what you’re looking for via this prompt format. But, on the plus side: I think it’s capable of understanding even more detail now- so you could add some extra adjectives into your various sections for ever more control.
Definitely going to try this prompt structure out with some of my custom words. Thanks for the breakdown and the resources. Incidentally, David Fincher coined the term "Dark Clarity" for his visual style and Midjourney knows what it means. Also, I'm totally considering the coverage of chaos as a shout out to my Twitter profile. 🤣
Your video is great - and I paid $10 for your PDF - and expected much more - instead it was only 12 pages - and were the pages you used in this video. I hope you can send me some more PDFs wit additional information.
Hey! Really glad you found the video helpful! Sorry that the PDF didn't have more info than what I presented, and I very much thank you for the donation. Drop me a line at theoreticallymedia@gmail.com and I can send you the PDFs to my other videos-- at the very least, I'd like to load you up with as much info as I can!
THIS is a fantastic video tut. Thank you for creating. Instant sub and like and follow blabbity blah from me. I"ve only just started with Midjourney and this takes it to another level. Thank you.
excellent to hear! There's a LOT of Midjourney content on the channel-- various levels, but I think you should be able to pick up on everything as you go along-- and if not, please don't hesitate to reach out! (there is the Discord link as well!)
Great video as always, man! Learned some good tidbids. One point I have with using the chaos parameter, I did a bunch of tests with and without it cranked to 0, 50 and then 100 and every time without fail I got significantly better output the lower the chaos was set. I don't use it at all anymore so it was interesting to hear that you use it constantly. Maybe I'll give it some more testing.
interesting! I'll have to experiment with some lower values! And yeah, I wouldn't say I always use Chaos, but I do find it helpful when you're searching for something you can't quite put your finger in. I almost treat it like free association-- And I'll say, it often leads me to some unexpected ideas, or inspires me to take a new approach to an idea that I've been working on. It can be a dice roll though! And just like real life: The dice aren't always in your favor!
Thanks, Tim. Yours are some of the most lucid tutorials I have come across. Please keep doing what you are, it's very insightful and helpful.
Thank you so much!! Super awesome of you! And I shall do! And believe me, comments like your keep me motivated when schedules get hectic!!
Thanks! Great info and included content!
great job ! there are tons of resources available on prompts but urs stands out for berivity and no nonsense approach , keep up the good work we need people like you
Ah, thank you so much!! To be honest, that’s something I sort of position the channel as: I don’t want to spend 30 minutes waffling on about prompts! Much better to show the underlying logic and get you all out and creating as quickly as possible!
You are doing such a great job Tim. You are becoming my favourite Midjourney educator. I really appreciate your efforts!
Oh wow, thank you so much! That REALLY means a lot!
I never comment on videos. But this is so so so exciting! I’m inspired to get started on midjourney myself
Fantastic to hear! The kind of unofficial sequel to this video is the one on "dash dash" commands-- you might find that helpful as well: ua-cam.com/video/56TnVebX6HI/v-deo.html
And as always, if you run into any problems, please reach out! Always happy to help!
Based on the transcript you provided, here are some key pieces of advice and concepts related to prompting for image generation:
Brevity: Keep prompts relatively short due to token limitations (around 77 tokens).
Medium: Experiment with different mediums such as photo, painting, TV show era, comic book illustration, etc.
Style and Composition: Explore different styles like Pixar, Tim Burton, etc., and consider how composition affects the overall look.
Camera Angles and Shots: Use specific camera angles and shots to direct the image generation process.
Scene: Describe the scene with details like actions, props, and location to influence the generated image.
Atmospheric Effects: Specify atmospheric effects like lighting, fog, weather, time of day, and seasons to set the mood.
Chaos Command: Utilize the chaos command to introduce variability and stimulate imagination.
Experimentation: Encouragement to experiment and have fun with different prompts and styles.
Some key words and concepts involved include: prompt framework, medium, style, composition, camera angles, shots, scene description, atmospheric effects, chaos command, experimentation.
It would be exceptionally helpful if you showed the prompts you were using on the screen in text versus just voice. Other than that, excellent video. Thank you.
Thanks very much! I thought about that actually- the prompts are all in the PDF, I kind of wanted to keep the image example textless. Although, my dumb mug shows up from time to time!
The PDF is excellent, thanks!@@TheoreticallyMedia
Thanks a lot man. This is an incredible start.
Thanks!
Oh goodness! Thank you so much!!
I really enjoy seeing these template things. I independently came up with the following for art prompts: [art style][artist][medium] of [subject][descriptor 1,2,3][technical parameters] where descriptors are things which add information such as what he gave, but some other examples would be time of day, sky information (e.g. Golden hour, interesting clouds), weather, texture/pattern, emotion, season, color palette, blur, lighting, etc. For photos it’s very similar, but I keep the descriptors to just two, and add [camera info] as a section, which includes as required the camera, film, and lens, but you could add more optional camera info to this as desired.
Love that! And it really goes back to what I always say: there is no right or wrong way to prompt.
The whole “prompt guru” thing that was popular awhile back has thankfully phased out, since I think most people see the snake oil.
It really all come down to concepts and tokens. The vocabulary to express what you want to see, and to some degree, the order in which you present it to the bot. That’s the fun and art of it!
@@TheoreticallyMedia MJ is very quirky thing, so sometimes the weirdest things give great images. Your point is right on, that understanding how it works is what matters.
THANK YOU!!! THANK YOU!!! I have been using MidJourney successfully for months in a different format, but yours is far better. Thank you again 👍
Fantastic to hear! Please do keep experimenting though! I always urge people to play around with this format and make it your own! Consider it a foundation to build your own castle!
As a rank beginner learning about prompting is the big hurdle. This video gave me ideas of how it is actually done. Can't wait to see more of your videos. Thank you for your time and effort.
I follow your page regularly. Thus is your best one so far. Looking forward to the the dash jawn.
Oh, dude-- that means a lot! Thank you so much! Haha, it's all downhill from here!!
Thanks!
Thank you so very much!!!
Excellent approach for producing prompts that work! It is simple to understand and effective for both beginners and experts.
Excellent video, the best I've seen on how to play with the styles, great job
Thank you so much!
I've been loving using the Bing prompt "A 1950's painted science fiction book cover." It's fun using different decades, and "science fiction" might put in neat things like planets and robots and such.
MOST EXCELLENT, great overview Tim.
Thank you soooo much!
I've been using many of these AI portals for some time now, and believe the real key is a creative imagination. The more you can dream up, wildly dream away, the "better" your images will be.
Really enjoyed this video. Your clarity is better than I could have explained it, and there are details here and on your other videos I was not aware of, or clear on.
תודה!
Thank you soooo much! Super appreciated!!
Awesome tutorial! Your framework for MidJourney prompts is a lifesaver for both beginners and advanced users. Thanks for the tips! 👌
That’s awesome to hear! Thank you so much!!
I just discovered your channel through this video and wow, you rocked. I subscribed.
happy to have you aboard! Hoping to have some new Midjourney videos soon! In the meantime, I've been covering a lot of AI Video and Music as well, if you're interested!
Once MJ drops V6, there is going to be a LOT of /imagine!
This video came out right when I needed it. I’ve been struggling with my midjourney prompts. It seems like most of the times it does the opposite of what I asked for lol I’ll definitely be experimenting with this structure. Thank you for sharing your knowledge sir🙏
excellent to hear! If you have any questions, or would like any clarifications etc-- please swing by the Discord! Or, hey: You can just pop up images you like in the AI Art Showcase!! discord.gg/6PMENW3k
Another tip and demo-filled video with just enough side remarks to make it entertaining and not annoying. I am so very appreciative of your guides and will add them to my notes. I even dig your font choices! Glad I subbed. Keep rockin'.
Oh man, thank you so much! Haha, it’s a real balance with the side comments. One day I’ll release all the outtake jokes that don’t make the cut as an epic blooper reel. It’ll be even more baffling, since no one will know the context of any of the jokes!
But for real, thank you very much! And I’m really glad the guides are helpful, I find them weirdly fun to make.
@@TheoreticallyMedia Ha! Excellent! Can't wait!
Just purchased the pdf and tipped you for some coffee. I am going to do a Gumroad soon and want to support others. Killer pdf, man.
@@giri.goyo_yt Thank you so much! And yeah, Gumroad seems really solid thus far. I know I'm not using 70% of what it offers-- I need to dig into all that at some point. Let me know when you're getting set up and maybe we can chat up some ideas about it!
For the dash dash video, one thing I like to do is reroll blends stepping up the chaos every time (15, 25, 35, 50, 75 etc) it's great because you get more variety than you get from just rerolling without adding prompt direction
Tim, thank you so much for the best video on youtube 🥰
Way late on this reply, but THANK YOU!
Wonderful source. Can't wait to dive in. Cheers.
Great video! The business meeting line was hilarious 😂
He is sick of the weekly 10am cadence meetings! This could have been an email!!!
really great recap and pointers again Tim! Looking forward to the -- part :)
Thanks so much!! Hopefully shooting it tomorrow morning!
Excellent! The PDF is fantastic. Your stuff is always some of the best around.
Oh thank you so much! and really happy the PDF was helpful!
Thanks for the very educational and well produced video Tim!😊
Thank you very much! Really appreciate that!
Hey Tim im so interested in getting a mid journey plan, and I was looking around for someone great here on UA-cam to explain for me. And I can say you seam really cool smart and interesting. Thanks!
Bien que francophone , j'ai aprécié la clarté de votre exposé sur la construction d'un Prompt
Bravo et many thanks.
Merci beaucoup! J’espère que cette traduction est exacte, mais j’apprécie vraiment le commentaire, et le fait que cette vidéo ait fait le tour du monde! En espérant visiter bientôt, et manger tout votre pain!
@@TheoreticallyMedia et nos croissants !
(Medium) (Style/Composition) (Scene) (Modulate) (Dash, Dash)
(Cinematic still) (Style by the cohen brothers, long shot) (business man walking down a busy city street ) (Green and yellow color pallet) (--ar 16:9)
Do you actually use the parentheses?
WOW! These are amazing prompts. So helpful for someone starting on with MJ. Thank you for sharing such great content.
excellent! You were exactly who I was hoping would see this! Please feel free to poke around the channel, there's a lot of other MJ tutorials on here as well!
Excellent learning video sir
I am truly grateful for it ☺️🙏🏻
That cyberpunk snow shot was awesome!!!
Tim, thank you for the video and the summarizing PDF. You know how much I appreciate the intelligence and professionalism you bring to your videos. On that note, it wouldn't be a bad idea for you to cover the /describe command of Midjourney. I've been using it to give a defined style to all the footage on my UA-cam channel, and it seems to be working well for me
covered it!
ua-cam.com/video/cH8UdeaYQls/v-deo.html (plus some stuff on Tokens in this one!
Hi Tim, your tutorials are incredibly helpful and your generated images just breathtaking! You really nailed it. 1.000 thanks for your generosity!
Oh, thank you soooo much!!!
Thank you! Just discovered you and subscribed. The simplification of a prompt formula is exactly what I needed!
"… The most boring kaiju movie of all time" - You just made my day.
“Sir, should we activate the Voltron protocol?”
“I don’t know. He’s not really knocking down buildings or anything. I think he’s actually kinda depressed…”
Such a great explanation of higher level prompting. Thanks.❤
Amazing video. Thank you, I have just started out with Midjourney
thanks. i am your big fan. i have learn lot from you. and write amazing hundreds of prompts.
Wow! Tim! thank you for helping me to start out on my journey.
So glad to hear! Lots of resources on the channel, but feel free to reach out if you have any questions!
There were just _way_ too many priceless lines in this one! Loved it. Thank you for the video. 👍
Might I strongly suggest people be very careful with the chaos value if they're using fast hours!
_Yeah, the K. Effect was definitely working when I saw Hitchcock and then the nice young lady on the grass._
Ah, true, I should have mentioned to use chaos with Relax. I'll pop that into the followup on dash dash! And the Hitchcock thing is such a funny example-- particularly considering he's the one that wrote it! He used that very set of images explaining editing in a CBC news program! Hitchcock had a pretty great sense of humor.
Excellent video, thank you - subbed 👍
Fantastic to hear! Thank you so much!
thanks from France . have a nice day :)
merci! je veux visiter et manger tout ton pain!
Thank you for “Prompt 101.” Speaking as a person who has flailed with prompts.
excellent!! You are EXACTLY who I was hoping was going to find this video!
@@TheoreticallyMedia I didn’t have the courage to mention the prompt I wrote that produced a woman with 3 arms and legs. Even “free” (Leonardo AI) costs tokens. Thanks.
7:42 how do I ensure that the results that I get include the character with their back to the camera. I find myself putting several prompts in to say character should have their back to the camera, but I still get faces.
one trick there-- and this is kind of funny, considering in the early MJ days, we were always trying to do the opposite (since characters always stood with their backs to camera and we were trying to get them to face camera!)-- But, let's try something like this (using our pal, the Man in the Blue Business Suit:
/imagine Rear view, from behind, a man in a blue business suit. The suit has a vent in the back that is slightly open, suggesting movement. The back of the suit jacket is smooth and well-fitted, with a noticeable sharp crease down the sleeves. His posture is upright. The back of his neck is visible.
By reenforcing "back" and giving details about the back of the suit, we're basically telling MJ, "give me a back shot"-- It might not be the most exciting shot in the world, but it'll work! Color as needed!
@@TheoreticallyMediaThanks so much. I currently use it to create flat vector illustrations for minimal wall art. I want a prospective buyer to “see themselves” so I don’t usually want a face or too many details in the final result. I will try to use some of these principles. Thank you so much for responding. Much appreciated.
@@MsAnderson- "facing away"
Hi. An excellent video. Very informative. I am making audio songs with just one cover picture. What I want to do is find an image generator software that would show this female dancing for between approx the length of my songs rating from 20 seconds to 4 minutes who might mean looping. Any advice would be very welcome. Ray
Grab the Free PDF Guide Here: theoreticallymedia.gumroad.com/l/Imagine
If you like the work I'm doing on the channel, please consider joining the Patreon! patreon.com/TheoreticallyMedia
👋
Thank you!
@@JanneWolterbeek 100%!!
Thanks for the great tutorial. How can you get a consistent character if you want to do some type of series?
Thank you very much for this great explanation
Hi Tim, thank you for all the videos. They help me a lot! Would be possible make a video talking about the control of the image? For example: if I try generate "nina simone wearing a floral dress in front of a striped wallpaper background" (or something like that), MJ just mess things up. Like striped dress and floral background. I tried photobashing, ::, --iw, reference link, and all the parameters. What I mean is that I feel theres no way to reeeeeally get the control of it. If you ask for a specific color or pattern, etc MJ just tend to ignore others words of the prompt. Thank you!
Maybe stop by the Discord and I'll see what I can do in #project help!
hi, how would I get it to create a 5:7 ratio? Is it possible for it not to do landscape? thanks
Thanks very much, also for the PDF. As a beginner trying to get my head around Midjourney this helps, a lot. Keep up the good work 👏
Great work!!
Thank you so much!!
Just got it for $1 awesome work.
oh man, thank you very much!!
The PDF file is not available to download for free. Please provide us with the access. Thank you so much!
Sick! Thankyou brother!
Hi Tim: I very much enjoy your videos and I know that your prompt tips will go a long way to getting me off to a good start. After having studied so many videos on UA-cam and Vimeo I am fascinated by those that I see by creators like "Friendly Advice", "The Panavision 70", and others of course. Not necessarily the 'big buxom' aspect of these videos but the whole look, the slow motion movements, the highly stylized futuristic themes. It is simply amazing but I do not know where to start. I can find no guide on what the general workflow would be to go from creating the image to to then having a stunning video clip. I am not even sure if they are using Midjourney and if so in what capacity as it relates to other AI tools in the chain. Any push in the right direction would be greatly appreciated. Thank for the work you do!
In midjourney i have used "facing camera" and "back to camera" successfully.
Also do you know how to do get max Depth of Field?
I designed a room in MJ but i have tried everything i can think of without success.
No blur, no bokeh, f/22, everything in sharp focus etc etc. Any ideas please?
you can try going simple, like "blurred background" or I have used f2.8 in the past and it has worked. Hop over to the Discord and we'll see what we can do!
im confused on how to add sections? obviously when you press enter it sends the prompt. looks like you've arranged the prompt using enter, so how do we do so please?
So, I actually don’t normally do that, but you can (I believe) by hitting shift-enter.
I learn a lot 💪🏽
Thank you so much!!
there are no rules with prompting but 'the Parameters are always added to the end of a prompt' --Max Aspect Ratio,--Chaos,--Image Weight,--No,--Quality,--Repeat,--Seed,--Stop,--Style,--Stylize,--Tile
Oh yeah, that is for sure a rule! I mean, I wonder if anyone has tried putting the parameters at different places in the prompt? Would it work? Only one way to find out!
Hi there - great video - I went to your site to buy the pdf - it won't take my card, I am guessing that is because I am using a visa gift card I have - is there another way to get the pdf? Thanks!
Oh totally, just put 0 in for the amount and you should be able to download it. If not, drop me a line at theoreticallymedia@gmail.com and I’ll email it over to you!
Push 1 in the background?
... well, actually there's a snow scene in the Blade Runner 2049 movie ;) Thank you very much for all these awesome tips!! I'm very excited to start experimenting with AI.
I'm still kicking myself that I forgot about the Blade Runner scene! Welp, time to go watch Blade Runner again!!
And thank you! There's a lot of resources here on the channel, so please do poke around! Ley me know if you have any questions!
thank you so much!
1000% welcome!!
User
we dont have a 'chaos' command with Dall-E but if i want randomness i'll tell you. So we could start with me giving you carte blanche in choosing a 'Hero' for our scene, the location , the atmospherics etc. Your constraints are 1- follow the recommendations from the tutorial and 2- the prompt produced must be maximum 480 characters.
ChatGPT
Understood. Here's a prompt following the recommendations from the tutorial with a maximum of 480 characters:
"Photograph medium shot of a courageous astronaut exploring a mysterious alien forest under the eerie glow of bioluminescent flora. Blue color palette enhances the otherworldly atmosphere. Camera angle: Close-up. Atmospheric effect: Misty fog surrounds the astronaut, adding to the sense of wonder and discovery."
thank you for sharing
100% and thank you for the comment!
This channel has become my top preference for MJ. I have a question regarding the creation of a detailed scene or photo featuring multiple characters. For example, I want to create an image with 2 couples. Specifically, I would like to know how to accurately depict the height, physique, hair color, age, and positioning of each person in the image. Lately, despite the last two updates, I have been experiencing issues where the generated images consistently depict anything but 2 couples I adked for. Instead, I get, three men or two women, and one man, rather than the 2 couples I inputted.
Any tips on how resolve?
Thanks so much! I’ll say, positioning/posing can get a little tough in MJ. Stop by the Discord (I think I linked it here) and hop in #Projecthelp and let’s see what we can do.
Greetings. My question is, If I use a picture of someone's face, how can I force midjourney to stay loyal to the features of the person? Ive used real pictures of people but the result changes the person to the point it's just someone else, sometimes with too much filter. For example Ive used the picture of a man with a beard and the beard looks almost like makeup or airbrush but not real facial hair.
Yeah. That’s a “feature not a bug”’of MJ. Mostly to avoid abuse with face swapping and creating harmful images. So, I kinda get it. You can use a tool like InsightFace to accomplish it though, you can check out a video I did on it here:
ua-cam.com/video/a3MOPd6pEOg/v-deo.htmlsi=wDbTyKbXN9-WuiUn
Question … what camera are you using for this video?
Canon R6. The old one, not the mk2
can you show me how to make oracle cards in this platform if that's possible thanks
I am constantly getting an Invalid parameter
Unrecognized parameter(s): AR
/imagine Photograph, business man walking down a busy street, Blue color palette --AR 16:9 --s 250
the --s250 was added by mJ
I am using model 5.2 / stylize high / public mode / remix mode / low variation mode / relax mode
a little help thank you
try lower case? It shouldn't matter, but give that a shot. Spelled out: dash dash (space) ar (space) number colon number
I tried everything. What did it and made it work was a change in my settings from model 5.2 / stylize high / public mode / remix mode / low variation mode / relax mode - to - model 5.2 / {stylize med} / public mode / remix mode / low variation mode / relax mode. What or why I have no idea. Thanks man just getting started on this journey. I want to create scenes to use as a story board then set up and photograph. I was a photographer, then went to the dark side as a retoucher and now as AI and photoshop have become so user friendly. I think a Photographer again , but not assignment work. I think it will have to be film. Wish I had not liquidated my med format equipt. Thank you for your info @@TheoreticallyMedia
Could the tim burton prompt not work due to copyroghts?
Your video was amazing! How about Stylar AI? A web-based tool for logo, 3D, and interior design, still in early stages but seems intriguing.
what does the --chaos prompt do?
Great content
Thank you so much!!
Thanks for the video. Looks like the link to the PDF isn't working.
Thank you! I was futzing around with some stuff on Gumroad today-- I think I broke it! Should be fixed now!
Worked. Thanks again@@TheoreticallyMedia
Great content. Most importantly no BS.
Thank you! Yeah, I try to avoid the 30 minute waffling, or just chatting with no editing. That “podcast” type video style works for some, but it generally drives me insane.
What is the name of the artist or the type of prompt to create the style of the video cover?
To be honest, I keep these pretty simple, it’s usually just “Stunning” followed by a word that is somewhat related to the topic.
But yeah, Stunning seem to trigger this as an MJ house style.
Does --chaos in v5.2 replace --creative in 5.1? I miss that....loved the video, new to midjourney and you're videos are a big help. Thanks
The discord invite is no longer valid. Do you have a new one?
Can you use your own picture for an image?
You can, but Midjourney will change it around. It’ll basically come out as a new image “inspired” by your source picture.
Does anyone know who the artist of this video’s thumbnail is?
I did the prompt on that one-- Usually my format for these types of thumbs are
"Beautiful + whatever the topic of the video is" It'll often taken a number of rolls and I need to play around with variations on the how to word the topic, but eventually I find something I like.
Is there any way to preserve a model and make variations with the same face? Apart from the swapid engine, is there a prompt or another engine like swapid that allows me to always generate the same model for everything or? Is another external program required?
Not...yet. There are some tricks to get a similar output, but MJ has been weirdly steadfast on not being able to repeat the same character. They've been talking about "Storytelling" tools in development, which I think solve the problem. My guess is that it will be a suite of features that will consist of Character, Location, and Emotion...but we'll see.
In the meantime, I generally will use a combo of --seed and prompts that direct a very simple character: "Woman, mid 20s, blonde hair, black dress" -- That simple "Blonde Hair/Black Dress" will usually get you in the ballpark with a few re-rolls, and then you can face swap from there.
@@TheoreticallyMedia Thnk
The human's eyes look dead, or like the human is in a daze. The dead eyes makes the whole image look fake. When I have tried to give Midjourney directions on how to imagine the eyes, the whole composition changes to an extreme closeup of the eyes.
I tried these prompts today, and I do notice the difference. I am able to get a lot closer to what I'm looking for in an image. You mentioned something about the order of things, so I rotated the order to get a more historical feel for the shots.
I guess I'm wondering where in the prompt the command starts to trail off into something I didn't ask for? Do you have suggestions that would help keep images more attuned to their time in history?
So glad that is helping!! So, to answer your question, check out this video on 5.2 and take a look at the Shorten command. There's some clues in there w/ how MJ "sees" your prompt:
ua-cam.com/video/1tIXOwNQuT8/v-deo.html
@@TheoreticallyMedia Good point. I watched that piece of your video, and I'll try some longer prompts to see what MJ "sees".
thank you for this video. I subscribed to your channel. I've used midjourney for coloring pages but not getting the results I want. Can you offer any suggestions? I am also learning photoshop so I can convert if needed.
Pop over to the discord and we’ll see what we can do!
@@TheoreticallyMedia I just got your reply. How can I make a better prompt?
I watched the above video of yours and find some of it confusing. You list your prompt framework like this: (medium)(style/composition)(scene)(modulate)(dash-dash). When you talk about styles in the video, you basically replaced the Medium (photograph) with the Style by Pixar and did the same thing with By Tim Burton and stopped there. If that is the case, why doesn't your prompt framework look like this: (medium/style)(composition)(scene)(modulate)(dash-dash).
I'm really puzzled by your prompt framework based on how you explained it in the video, which left me wondering what other Mediums are there besides the ones you talked about in the video. I wish there was a list of Medium types, like your list of Long Shot types. Actually, from all the prompts I studied I think you are only one using the word Medium while other frameworks seem to use Subject. Also, the word photograph seems more like a style than Medium. I tried to google medium types and it didn't help. But if I googled for styles I got this:
isometric anime
analytic drawing
infographic drawing
coloring book
diagrammatic drawing
diagrammatic portrait
double exposure
2D illustration
isometric illustration
pixel art
futuristic style
ornamental watercolour
dark fantasy
paper cut craft
paper quilling
patchwork collage
iridescent
ukiyo-e art
watercolour landscape
op art
Japanese ink
pastel drawing
dripping art
stained glass portrait
graffiti portrait
winter oil painting
anime portrait
cinematographic style
typography art
one-line drawing
polaroid photo
tattoo art
If I understand your video you would call the above Medium while others seem to call them styles. Any thoughts?
Well, I'll say that sometimes I get fuzzy with my prompting. I like to think of the framework as a basic guideline, almost like seasonings in a recipe. I think in this video I mention there is no right or wrong way to prompt, just an organization that I think helps Midjourney understand what to prioritize in your prompt.
That list is pretty great! A fun thing to do is mix them together as well, say "Japanese Ink and Graffiti Portrait" hmmm, might have to go try that!
@@TheoreticallyMedia Here is my attempt to come up with a prompt guide cheat sheet and it's components. drive.google.com/file/d/1vl2u21FN8ZHdNcNSPXPP6w65Ih0RRjmV/view?usp=sharing
Would be really helpful to see the actual prompt you used for each picture
So, if you download that PDF, all the prompt are there!
Is this still relevant for v6? Thanks
From my experiments, it is. Although things are changing, this still works. The language model is smart enough to know what you’re looking for via this prompt format. But, on the plus side: I think it’s capable of understanding even more detail now- so you could add some extra adjectives into your various sections for ever more control.
Definitely going to try this prompt structure out with some of my custom words. Thanks for the breakdown and the resources.
Incidentally, David Fincher coined the term "Dark Clarity" for his visual style and Midjourney knows what it means.
Also, I'm totally considering the coverage of chaos as a shout out to my Twitter profile. 🤣
I had not heard that Dark Clarity thing, that’s a perfect descriptor for his films!
And haha, yeah twitter loves chaos! A perfect match!
Your video is great - and I paid $10 for your PDF - and expected much more - instead it was only 12 pages - and were the pages you used in this video. I hope you can send me some more PDFs wit additional information.
Hey! Really glad you found the video helpful! Sorry that the PDF didn't have more info than what I presented, and I very much thank you for the donation. Drop me a line at theoreticallymedia@gmail.com and I can send you the PDFs to my other videos-- at the very least, I'd like to load you up with as much info as I can!
THIS is a fantastic video tut. Thank you for creating. Instant sub and like and follow blabbity blah from me. I"ve only just started with Midjourney and this takes it to another level. Thank you.
excellent to hear! There's a LOT of Midjourney content on the channel-- various levels, but I think you should be able to pick up on everything as you go along-- and if not, please don't hesitate to reach out! (there is the Discord link as well!)
your discord link is broken :(
Great video as always, man! Learned some good tidbids.
One point I have with using the chaos parameter, I did a bunch of tests with and without it cranked to 0, 50 and then 100 and every time without fail I got significantly better output the lower the chaos was set. I don't use it at all anymore so it was interesting to hear that you use it constantly. Maybe I'll give it some more testing.
interesting! I'll have to experiment with some lower values! And yeah, I wouldn't say I always use Chaos, but I do find it helpful when you're searching for something you can't quite put your finger in. I almost treat it like free association-- And I'll say, it often leads me to some unexpected ideas, or inspires me to take a new approach to an idea that I've been working on. It can be a dice roll though! And just like real life: The dice aren't always in your favor!
I cannot get consistent wide shots no matter what I do, its maddening, Ive tried countless prompts and its still waist-up portraits
Wide angle/full body? That one can be tricky. Stop by the Discord (link in the description) and we can work on it!