@@fatjay9402 Ahhh, thanks so much! Yeah, I'm really trying not to fill these videos up with a lot of fluff. The thing I really hate is when people make MJ videos and don't cut out the generation time. Just EDIT IT!!!
I watched Tom Cruise's acceptance speech for the PGA awards today and was awed by it. He spoke about being 18 years old and beginning a lifelong quest into learning as many elements of cinematography from as many teachers as he could find, and humbly thanking each one of them for what they shared with him along the way. I'm in my mid 70's now and am finally retiring from one career and trying to become savvy about this ai world of word generated images, and trying to find a path to get into it more effectively. I've been searching for videos like this one that can give me a more complete picture of how to get as much controllable imagery output from Midjourney as is reasonable. My goal is to illustrate my fantasy writings that have sat hidden in my closet for waaay too long. Thanks so much for your video. I'm look forward to putting it to work to the best of my ability.
Thank you so much, Wayne! That’s really fantastic, and honestly your comment means a lot to me! The channel started to take focus toward Creative Ai tools late last year. I kinda saw it coming just before ChatGPT, and I knew this was going to be a game changer for independent artists. Midjourney, and many of the other things I’m going to be covering, truly allow us to finally realize many of our dream projects. Of course, the thing that I always like to stress is that they assist us in creation. It always comes down to the human directing the machine. I like to see it as Assisted Intelligence, as opposed to Artificial. Can’t wait to read your Fantasy works as well! Growing up I was a big fan of everyone from Tolkien to Terry Brooks. I’ll also admit to being a total nerd in high school and reading the Forgotten Realms D&D novels…there were actually a couple of good ones in there!
I have been an indie filmmaker and 3D animator for over 2 decades and while I like to think I have as good a handle on the A.I. wave as anyone, having used A.I. in production for several years BEFORE it became the current iteration, YOU Sir have defragged and decoded many things that are vital and yet rarely if ever addressed by anyone. (Even creative A.I. is too fluid for ANYONE to ever truly be able to call themselves a master or power user of the tech IMO)The depth of field usage alone can make ALL the difference and since the Genie isnt going back in the bottle, the difference between 'likes n' shares' and being able to secure a niche that generates a revenue, will be those that go beyond the cutting and pasting prompts. OUTSTANDING presentations with regard to some best practices and/or new techniques I have seen so far on your channel, easy to comprehend and apply. Glad I found your channel and I am now a subscriber.
Super useful channel with an excellent signal to noise ratio, well done. Here's the basic prompt formula from your video: /imagine (cinematic still), (film by), (scene/subject/action, set), (link), (shot), (dash dash) And that list of composition call outs you mentioned: Wide-angle Ultra wide angle Satellite Eye Level Low angle Full Body Close up Over the shoulder Two shot Based on the date of this video, I'm assuming this is MJ V4 vs. V5. It might be a useful exercise (well, for me) if you go through this video with the release of V5 highlighting what you do/don't need. Thanks again for the videos and for speeding up my learning by months!
Excellent to hear, Rick! And yeah-- This video was done at the tail end of v4. I'll update it at some point with a refresh...haha, probably just as v6 comes out! Ugh!
@@SCWhiteJazz No Dumb Questions here! So, the commas: yes. That lets midjourney know to separate ideas. Although you'll want to use it as follows: "A city sidewalk, raining, a man with a yellow umbrella" That lets MJ know these are separate elements in your image. The Parentheses you can totally ignore-- that was just for illustrative purposes in the video!
Holy shyte. I'm new to all this A.I.stuff from a CNN story. Is this shit really for real? And this can be used to make fully blocked out scenes? Not just stills?
Omg Thank you so much for making a style guide. Just as I was thinking “boy it sure would be handy to have a guide to directors styles” you mentioned it. Subscribed! Thanks again
Oh yeah-- I know. I gotta get to that at some point. Around that same time, the channel took off and I got super busy. It is on the docket, I'm just not sure when I'll get to it.
12 minutes and you hit this tute out of the park. Brevity, no long intros, great huge fonts with prompts, and background info without "being up yer own a** TH-cam style". I took notes and want to see what the rest of your channel is about. Checking out that Carpenter short as well. Thanks!
According to chatgpt, these are the corresponding lens lengths to the shots discussed. - Wide angle shot: Typically uses a lens with a focal length of 24mm or wider, which creates a sense of spaciousness and can make the subject appear further away. - Ultra wide angle shot: Uses a lens with a focal length of 14mm or wider, which exaggerates depth and creates a dramatic, sweeping effect. - Satellite view: This type of shot is usually achieved with a drone, and the lens length will depend on the altitude and the desired field of view. However, a popular option is to use a 16mm lens to capture a wide-angle view with a sense of scale. - Eye level shot: Generally uses a lens with a focal length of around 50mm, which most closely matches the perspective of the human eye. - Low angle shot: Often uses a wide-angle lens with a focal length of around 24mm to create a sense of height and make the subject appear larger than life. - Full body shot: Typically uses a lens with a focal length of around 50-70mm to capture the subject in full without distorting their proportions. - Close up: This shot is achieved with a lens with a focal length of around 85-135mm, which can create a sense of intimacy and highlight details on the subject's face or body. - Over the shoulder: Typically uses a lens with a focal length of around 50-70mm to match the perspective of the human eye and create a natural, balanced image.
@@TheoreticallyMedia Brazilian filmmaker here - Can't wait for the prompt guide, if it is as well presented as this video I would be happy to pay a pizza for that ;)
@@TheoreticallyMedia quick question if I put these prompts in it definitely creates the image the way youve described but they are rarely photorealistic - still look like game characters - any prompt suggestions to make them feel real real
Great prompt guide well done. You mentioned on March 3 in your description: MIDJOURNEY CINEMATIC STYLE GUIDE: (Sorry, link coming soon!) Its almost 6 weeks now, would still love to see it.
Thanks for this i shall try them as i have had no luck getting a wide DOF. I was designing a room with a sofa in the middle but wanted the background shap and in focus but struggled to get it. MJ seems to like to use the subject as the focal point which is fair but at the expense of everything behind it. I would have though f/22 would have fixed that but no. Thanks for the great video !!
You might want to try going with something a little more on the “conversational” side. Something like -no background blur. Or “Sharp Focus” They’re constantly playing with the language model, so a lot of times the results can vary.
Great videos, just stumbled onto your channel. I am an older dude (over 50) who has just started with midjourney and your explanations are really good. Still trying to understand the styles and such in prompts but I am furiously scribbling down points from your videos.
Awesome to have you on board! I think next week I'm going to circle back for more of a "starters" series-- I think that's been something that the channel has been lacking. Hopefully I can fill any gaps for you!
@@TheoreticallyMedia a mate that would be awesome, your videos are really well structured so I feel I will get a lot from a starter series. I really bumble through the prompts so will thrive on a starter series :)
That style guide is still on the “to do” list. Sorry. The channel got really busy and I have had a free minute to get back to it. At some point I’ll release it, I promise! Every week, I say: this is the week! And then something crazy happens!! Ugh.
Thank you! I've got an update on it coming soon. I mentioned the Prompt Style guide in the video, and I haven't finished that yet-- but once I do, I have another full video on this!
Thanks so much for watching, Mathieu! I think at the time of this video, Midjourney was limited to which aspect ratios you were allowed tto use, and 2:1 was as close as I could get to that "Cinescope" look-- ALTHOUGH, with v5 we now have any aspect ratio we want! Which is INSANE! I cover it in this video: th-cam.com/video/wqC9---Xmw8/w-d-xo.html where I play around with 2.35:1, plus do a bunch of other wacky ones, like 10:4, or 13:2-- just plugging in numbers to see what will happen. It's actually remarkable at how MJ will manage to maintain a good composition, no matter what aspect ratio you throw at it! Thanks so much for the comment!!
That’s awesome! A little while back, I was using MJ to do faux documentaries about films that never existed. The possibilities are really endless. And v5 looks crazy good!!
That's awesome to hear! I'm doing a revised version soon, since this was actually done before v5 was released, and I'll make sure to add a lens section into it!
This video is excellent. A camera and lighting tutorial would be great. How to call lenses as well as specify lighting setups. All the best. Great channel!
Thanks so much! Yeah, I need to cover that in the (ever growing and still not done) Style Guide that I promised in this video. It WILL happen at some point!
So it is: dash dash space AR space number(colon)number No () but just wanted to indicate there is no space in the number to the colon. Let me know if you keep running into issues!
@@TheoreticallyMedia thank you! I figured it out, I was putting a comma before the dash dash. After I removed that it started working just fine. Thank you for your help! I will be watching your content for sure!
I still come back to this months later! Don’t suppose you’ve had a chance to make a start on that guide? Or even have rough notes/other resources to help with choosing cameras, different types of shots, shot length etc?
So, it's kind of there in the background as a project, but at this point I'm thinking about waiting until MJ v6 drops, since that should be within the next few weeks (to about a month)-- It just makes sense to have a guide with the newest version, y'know? This one, oddly enough, dates back to v4, if you can believe it!
@@TheoreticallyMedia yeah totally makes sense. In the meantime if you know of any resources for camera angles/lighting/cameras to use when experimenting with prompts, please let me know! Thanks
That's excellent to hear! You might be interested in checking out the video I did here on Midjourney to AI filmmaking with Gen2: th-cam.com/video/PlZkWcnnRt4/w-d-xo.html We're ALMOST there!
Not...yet. There are some tricks, but nothing that actually gets a real consistent character across multiple panels/pages ect. MJ says they're working on it-- so hopefully we'll see that feature in v6!
This is pure awesomeness! So interesting how a simple prompt with theright information can definitely pull something awesome out of AI. Thanks for sharing.
Thank you! Couldn't agree more. I also love how changing just one word can result in a drastically different image. It's so fun to quickly iterate on ideas!
I am confused by "action" and "set". For set, your example was "holding an umbrella", wouldn't that be an action? Are they interchangeable or would it be, scene "waiting for a bus", set "city street after it rained" ?
So, I probably need to revise this-- and on the plus side, since v5 came out, I actually have a reason to! The original idea is Scene/Subject/Action/Set were all sort of one block. But, I don't think I was clear enough in the video, they are sub-sections within that block. For the revision, I'm going to structure it more like this: (Film by) (Character/Action) (Scene/Location) (Link) (Shot) (Dash Dash). A prompt like that might read more like: "Cinematic Still, Film by Wes Anderson, A man in a yellow raincoat, holding an umbrella, at a bus stop, city street, morning, cloudy skies, wide angle" (Ha, I just ran that prompt and it came out kinda cool!)
excellent! Thank you! There's a bit of an updated version here, at least on prompting, if you'd like to check it out: th-cam.com/video/j-GZvXgfgaY/w-d-xo.html
@@TheoreticallyMedia hahah no pressure to respond to them! When I find a channel that I enjoy I like to make sure I comment to boost the algorithm, I figure it is the least I can do.
I really love you channel and your MJ videos!! Thank for putting out such great content for the AI community :) when are you releasing your cinematic style guide for MJ?
Thank you so much for your presentation. It's the most informative midjourney training I've come across that's relevant to film making. Can you ask Midjourney to create a 3D video clip of a scene or does midjouney only create still frames?
Thank you! Thus far, we haven't had full 3d. It's mostly more in what would be considered 2.5D animation. I did a video detailing some of those techniques here: th-cam.com/video/Hzg06qJQkpw/w-d-xo.html
Hey so don't sound rude but where is the style guide you mentioned it would be liked bellow but I can't find it 😅 not sure if it's been removed maybe I've been experimenting alot with film stills in v5 without really mentioning directors much and got amazing results I'd love to experiment with some of these idea. 😊
Ha! No rudeness taken! It's my fault more than anything-- I'm still working on it. Per usual, I bit off more than I could chew-- and once I started on the project, I just kept adding more and more stuff to it. I am thinking about scaling it back and releasing a "simple" version soon...Sorry, just trying to find the balance of something that is useful to you, but also isn't going to take me a year to finish!
@@TheoreticallyMedia that's understandable can't complain about how long it takes when your spending so much effort on making sure it's comprehensive 😃
Contradictory prompts can provide unique, and sometimes, excellent results. I've combined 2 types of film, or multiple camera types, for a unique effect. For example, in doing still shots, combining a Kodak Porta high ISO film with Polaroid film gives me an image with realistic skin tone and features. I've done the same with using 2 camera types in the same prompt. It gives different results than the individual cameras used on their own.
Thanks for great video! I tried the Cinematic still prompt but kept ending up with images of the director behind the camera rather than "a leading actress". Tried out different things like instead of just typing "cinematic still, James Cameron, a leading actress..." I tried with "cinematic still, a James Cameron movie, a leading actress..." but still ended up with James behind the camera, or Tarantino or whoever. The only one I got working was Wes Anderson. Any ideas?
I think MJ might be futzing with the language model, you aren't the first person I've heard having this issue. One thing that seems to work is by naming your Leading Actress. I think when you leave her unnamed, MJ just sees James Cameron and says: Ok, I know that dude! This one worked for me, give it a shot and see how it goes: cinematic still, SciFi film by James Cameron, Scarlett Johansson as a Space Marine, holding a Flamethrower, on a Spaceship, Horror Movie Lighting, Wide Angle --ar 2:1 Note that I did add in "Sci-Fi Film" to James Cameron.
Soon? Sorry-- got caught up with a bunch of other videos. I do need to circle back to this. I'm working on it here and there, but the problem is always time. I'm actually going to do some work on it today and then do a refresh video on the whole shebang very soon!
Hey, I loved the documentarian style of the images for your John Carpenter video. How did you make those pictures so grainy and gritty looking? Was that done in midjourney, or did you add grain and unsharpness in post?
Oh, thank you so much! Yeah-- there was probably a lot of futzing in Premiere when I was making the video. I think I was putting a bunch of light leaks and film grain adjustment layers on most of the images, just to get them to pop a little more.
@@TheoreticallyMedia Thanks! Yeah I thought so, midjourney doesn't really seem to do things like noise and film emulation on a really detailed level yet. I'm waiting for the first really great color grading, texture & film emulation AI to come out. With the detailed midjourney images I have not really seen anything yet that could really fool me that it was real (especially with people in them), but as soon as you artistically destroy the image like you did, stuff can already look very real. I personally am much more interested in that than all the weird stuff
Hi! Great channel & tanks for the great videos. Great tips on formulating prompts. Can I ask a question? Is there an AI that does the reverse? I mean an AI that creates a prompt or a description from an image?
Hi! Thanks for the comment! There actually IS a tool for that, but I'd say don't expect a one-to-one just right now. The tool is free and you can use it here: huggingface.co/spaces/fffiloni/CLIP-Interrogator-2 Now, that said-- It really just captures an overall vibe in the text prompt, so you probably aren't going to be getting poses, or an exact clone of the source image. But, I think it's a really great way of getting in the ballpark of what you're trying to create, and from there you can hone in. Let me know how it works for you!
Love this channel, but what if we don’t want a cinematic look but want to mimic stock photo uploaded by amateur photographers to sites like Unsplash. In other words the images is not to polished and therefore looks like someone took a pictures on their DLSR or iPhone.
It’s funny, I’ve been hearing that more and more. I’ve been thinking about it too. I’ll let you know if I find a solution. That’s one thing I’d love to see, something where it doesn’t look perfectly composed.
Haha! Soon? It turned into a bit of a grind. What started off as a simple spreadsheet has gotten a little out of hand. Per usual: I bit off more than I should have! Still working on it though!
Really enjoyed this training. I like how you present your information and I have seen a lot of benefit in my rendering. II did try the "film by Quinten Tarantino" suggestion, however all I was to get was renders of Margot Robbie's bare feet smashed up against a car windshield. =)
Great! I did some Kurosawa Star Wars around a month ago.Turned out some really great stuff, I didn't do anything too different to the things you have set out and it was nice to see your examples.Placing the year to 1963 turned them black and white and gave them a very old Japanese movie feel. Are you on twitter?
Oh, that's awesome! Kurosawa and Star Wars, like Peanut Butter and Sashimi! Wait...no, that sounds gross! Yeah, the year modifier is something I'm going to get into in a follow up video, mostly talking about blending styles, which I think is something Midjourney excels at-- more so than any other Text to Image generator. I'll actually be on Twitter in a few days...I was on from fairly early in, but left when things started to get...ugly? That said, with the channel, I'd like to be back on with a fresh account, and since it'll be more focused on the channel, I think it'll be easier to keep the toxic rambling at bay! Drop me your handle and you'll be my first follow!
Thank you very much! Great professional content, great attitude = loyal subscriber. Would love to see a video about the interaction of MJ with photographic terms! Keep up the great work!
Thanks so much!! So, I think of "Link" as additional atmosphere details for "Scene"-- like, "Scene" might be a Beach, but "Link" will have Sunny Day, or Moonlit Night. I probably should revise the prompt structure to something more like: (Film by) (Character/Action) (Scene/Location) (Link) (Shot) (Dash Dash). I am going to circle back fairly soon (once the stupid style guide is done), and I'll add that in! Thanks so much for the question!!
Thanks so much, Dwayne! I do need to do that at some point, but I'd like to approach it more from more of a tutorial perspective-- less of a "formula" (which, ironically I use in the title of this video) and more as a perspective. I have been putting up PDFs of later videos and offering them for free/Donation-- and I'm very happy to say that the donation response has been super generous. Can't thank those people enough!
Happy to help! But do you have the correct time stamp? At 9:09 I am apparently reading a prompt and the text appears on screen. Happy to clarify anything you need though!
Oh man, still working on it. Sorry-- It turned into a bit of a monster (my fault). I wanted to kick it up a notch, and it ended up being a bit more than I could chew. I'll get it there soon though!
Hey Tim, and all. I was not sure where to post this so thought I would do so on this video since Tim is talking about Cinematic prompts here. And this is a real business application where I am building a business around AI Character-driven book trailers. Here is one I did - "The High Priestess". I used Midjourney to generate the images, then morphed the faces to get consistent characters with Patreon, then used Photoshop for final clean up (eye and hair color for the most part), and then pulled into TensorPix to upres some of the final stills. For the second to last shot I used Runway to generate the video (from a Midjourney generated still). All then edited in Adobe Premier Pro. Enjoy: th-cam.com/video/UMLpGROtrxQ/w-d-xo.html
@@ryanhowell4492 To be honest, we all do! This is such new technology that we're all learning on the fly! New stuff is popping up every day, leading to new techniques and approaches. And we get to be a part of it all! It's a pretty exciting time!
Perhaps try "firing at a lovecraftian monster" rather "shooting at a lovecraftian monster". I think Midjourney will associate the word shooting with a camera especially if you've specified a camera or lens type.
Haha, it so true! And watching Ashoka so far, it’s clear that they’re leaning into the Kurosawa aspects in that show. I know a lot of people are complaining the show is “boring” and I’m like: no, dude- it’s a 60s Samurai movie! I mean…in space. With space wizards. But still!!
I 100% agree, I like to use clear and concise direction, you might get something you like from spamming the prompts, but you're just playing the odds at that point hoping to get something cool. For more control, less is more. Also, wanna let you know that I'm going to try out your "non-Star Wars Star Wars prompt" in GEN 2 lol I might change it up as GEN 2 isn't midjourney level, but I gotta say I've been getting pretty decent stuff. Mostly through proper prompts and getting the settings just right. Let's see how I do, I'll post it for you in this comment. Thank for another great video, you videos have been a great help, so much appreciated!
as usual, I see folks uing MJ and getting cool results. But I want to see someone getting results that were intended or changing the prompt to get the right results
@@TheoreticallyMedia I often see huge and probably redundant prompts people make and MJ will ignore basic requests like giving closeups instead of full body. Trying to get things closer or further away in frame. Getting MJ to put a background that YOU request. Not putting a tail on something when to ask for no tail
@@DKGCustom Oh, I get ya. It's funny, I just made a video about that after someone commented something very similar: th-cam.com/video/j3LiGOBglCs/w-d-xo.html Long story short, there are some tricks to get what you're looking for-- but there are other times that, yeah-- MJ is just gonna ignore you. Or give you some really terrible results!
I ask this in the least condescending way I can, but what about this process do people like? Why do people like assembling pixels this way, why are these even being made? I can't really warp my head around what's enjoyable about making and viewing these, for a lack of better terms, "images". Looking for a real conversation
Thank you for your question! It's great to see that you're interested in understanding the appeal behind AI-generated images, and I'd be happy to provide some insight. Here's a few reasons, I'm sure other people can add to this list. Creative exploration: AI-generated art offers a unique opportunity to explore new and unexpected visual realms. By training AI on a diverse range of styles and visual elements, it can generate images that blend these influences in novel ways, resulting in truly unique pieces. This can be a source of inspiration for artists and art enthusiasts alike. Collaboration between humans and machines: Many artists have embraced AI-generated images as a way to augment their own creative process. By using AI as a collaborator, artists can push the boundaries of their work, blending human intuition with machine-generated suggestions to create innovative pieces that might not have been conceived otherwise. Entertainment and fun: Some people find enjoyment in the surprise and novelty of AI-generated images. As the AI learns from its training data, it can produce images that are unexpected, strange, or even humorous. This can be entertaining for users who appreciate the unpredictability of the results. Customization and personalization: AI-generated images can be tailored to individual tastes and preferences, making them an appealing option for personalized art and design. For example, users can input their favorite colors, styles, or themes, and the AI will generate a piece that incorporates those elements, creating a one-of-a-kind work of art. Practical applications: Beyond entertainment and artistic expression, AI-generated images have numerous practical applications as well. They can be used for tasks such as generating textures for video games and movies, creating advertising materials, or producing visually engaging content for social media platforms. Accessibility: AI-generated art democratizes the art-making process, allowing people with little or no artistic skill to create visually appealing images. This can be empowering for those who might have previously felt excluded from the world of art due to their perceived lack of talent or expertise. Research and innovation: AI-generated images are not only a form of artistic expression but also an area of active research in the field of artificial intelligence. As researchers continue to develop and refine these technologies, we can expect to see new and innovative applications emerge, pushing the boundaries of what's possible in the realm of AI-generated art. I hope this helps you understand the various reasons people enjoy creating and viewing AI-generated images. It's a fascinating intersection of art, technology, and human creativity that has captured the imaginations of many. If you have any further questions or would like to discuss this topic more, please feel free to ask!
Hey man, thanks for the thoughtful question-- and thank you for your opening, I think that sometimes the tone of questions come off wrong in a text/comment format, so I greatly appreciate your clarification at the top! Anyhow, I do hope you get a few different answers, because honestly, it is different for everyone. Since starting the channel, I've heard all kinds of reasons-- one person that always wanted to visualize a dream they had as a child. Many like creating these cinematic images for mood-boards, or creating images from a film they wished had been made. There are a few comic book authors (although, I still personally feel that Midjourney is the wrong tool on that front), folks that want to make books for their kids...The list goes on. I got really into it via a fake-documentary I made using Midjourney, detailing a 1977 John Carpenter film that was Alien, before Alien. If you're interested you can check it out here: th-cam.com/video/BbeQDnBSjD8/w-d-xo.html Ultimately, I think the real joy of it all is thinking "I wonder what X would look like?" popping in a prompt and seeing it. Once you go down the rabbit hole, you'll see that first image, and a bunch of other ideas will come to you. Before long, you're coming up with specific ideas, and that's where the process factor kicks in. Thanks again for your comment! If you have any questions, please don't hesitate to reply!
It’s a fair question and I can answer it from a practical point of view as someone who uses Midjourney for work purposes. I make thumbnails for TH-cam and the amount of time it saves is phenomenal. A client could give me a really obscure piece of work which could lead to me spending hours and hours looking for the right assets to make the thumbnail. Looking at different assets with different backgrounds, lighting, composition, etc can be so draining if you can’t get it right. With AI tools I can prompt exactly what I need and make minor adjustments instead of spending hours just searching for assets. This allows me to spend more time perfecting the lighting, blending, etc instead of that time going on searching for assets. That’s my take on it anyway.
@@James777x Thanks for the deep and thoughtful response! Entertainment and fun, I get that. I'm sure it's fun to see all these things come to life almost instantaneously. I guess where I get tripped up is the whole 'art' thing. You say "it" a lot and not "me" or "you". The idea of a person taking credit for this as art based on typing a few words into, essentially an intake form, is a little bit unsettling to me. While I'm not Wes Anderson, although I hope to be at that level some day, I don't think I'd appreciate someone saying "make me a Wed Anderson style photo" having it generate instantly then going to take credit for it as art. I would argue that had nothing do to with the "artist" and 100% to do with the machine biting his style. I think creative exploration is more about thinking and playing and working it out, which in turn will inevitably force the artist to put their own spin on it, not just waiting of AI to do all the work for you. I have hope that society will reject it as "art", but that really isn't up to me and we'll just have to wait and see I guess!
Oh, it’s crossed my mind more than once. The really weird part is when you run across an AI character that looks like someone you know in real life. Like when there is a background character and you’re like “geez, that looks like Jeff in accounting!” That’s weird. It also might be Jeff’s fault that he’s so bland looking. For god sakes, Jeff, get a personality!!
Very helpful video thanks. I've been trying this whole AI thing over the last few days (video below). But irs really the prompts that get me confused. It seems the order is very important. The first item mentioned is a big deal apparently. I was trying to recreate a Moses character and wanted a wide angle shot but with a bad order Moses appeared with a DSLR to his eye! 😂 th-cam.com/video/YEYkcwJrNrA/w-d-xo.html
Ha! Yeah, the camera thing happens pretty often. You can try dropping specifics on the camera model if it keeps happening, or try something like “style by DSLR” and see if that helps! Oh, and order is 100% clutch!
Oh, I can’t say I disagree with you. There were obviously women in the example images, but they were mostly Caucasian. But I think that has more to do with the directors chosen in the image prompts. There aren’t many people of color in Tim Burton or David Fincher films, right? And Midjourney is basically taking the median of character that appear in those films and creating a person who might be cast. I will say that the entire back half of the video contains examples from Akira Kurosawa films, so there is a pretty good amount of Asian representation in this tutorial. But, you bring up a point that I’ve been mulling over for the last week or so, and I think your comment finally pushed me over the edge to make a video about it. There’s a problem with both body type and representation in AI imagery, and we should start having a meaningful discussion about it. Thanks very much for your comment.
This is absolutely inherent in midjourney, it’s amazing. You HAVE to specify a specific race to get anything but white people. Systemic biases illustrated for all to see! 💔
What about doing your own Midjourney stuff and keep your vertue signaling and "anti"racist rants for yourself ? He could have shown only whites or blacks or whatever, what's the deal ? You get a free video with lots of cool tips, that doesn't talk a single second about race, and this is the only comment you could think of ? What about thanks for the content ?
Yeah, I'm not wading too deeply into those waters. I'll say this though: David Bowie had a quote that I always liked (I'm paraphrasing it here): "There's no such thing as an original artist, just very tasty thieves" AI is not very tasty. It takes human input to tell it what to create. AI is very good at taking two or more styles and blending them together to create a new look, but it still takes a person to decide what those things are. The thing that I'm always pushing is: We have these tools now that will allow you to create anything you want, go MAKE something with them. Make a short film, make a music video, make a comic book, make a video game. To me, that's where a creative person's style will evolve from all this.
How do you figure? If someone is new to creating images, are they expected to step in and create the most masterful prompt on their first go around? You might be thinking of babies. Babies have zero skill and when they try to do something it is hilarious.
Do you have any good Midjourney Prompt Tips? Please let us know by leaving a comment! Or...y'know, just say Hi. It's always nice when people say hi!
Your videos a great! good info .. no the normal AI bullshit most people put in! Thx
@@fatjay9402 Ahhh, thanks so much! Yeah, I'm really trying not to fill these videos up with a lot of fluff. The thing I really hate is when people make MJ videos and don't cut out the generation time. Just EDIT IT!!!
Great video! Were you able to every create and post the cinematic style guide? Thanks!
Where is the prompt?!kindly provide it ❤❤
I watched Tom Cruise's acceptance speech for the PGA awards today and was awed by it. He spoke about being 18 years old and beginning a lifelong quest into learning as many elements of cinematography from as many teachers as he could find, and humbly thanking each one of them for what they shared with him along the way. I'm in my mid 70's now and am finally retiring from one career and trying to become savvy about this ai world of word generated images, and trying to find a path to get into it more effectively. I've been searching for videos like this one that can give me a more complete picture of how to get as much controllable imagery output from Midjourney as is reasonable. My goal is to illustrate my fantasy writings that have sat hidden in my closet for waaay too long. Thanks so much for your video. I'm look forward to putting it to work to the best of my ability.
Thank you so much, Wayne! That’s really fantastic, and honestly your comment means a lot to me!
The channel started to take focus toward Creative Ai tools late last year. I kinda saw it coming just before ChatGPT, and I knew this was going to be a game changer for independent artists. Midjourney, and many of the other things I’m going to be covering, truly allow us to finally realize many of our dream projects.
Of course, the thing that I always like to stress is that they assist us in creation. It always comes down to the human directing the machine. I like to see it as Assisted Intelligence, as opposed to Artificial.
Can’t wait to read your Fantasy works as well! Growing up I was a big fan of everyone from Tolkien to Terry Brooks. I’ll also admit to being a total nerd in high school and reading the Forgotten Realms D&D novels…there were actually a couple of good ones in there!
I have been an indie filmmaker and 3D animator for over 2 decades and while I like to think I have as good a handle on the A.I. wave as anyone, having used A.I. in production for several years BEFORE it became the current iteration, YOU Sir have defragged and decoded many things that are vital and yet rarely if ever addressed by anyone. (Even creative A.I. is too fluid for ANYONE to ever truly be able to call themselves a master or power user of the tech IMO)The depth of field usage alone can make ALL the difference and since the Genie isnt going back in the bottle, the difference between 'likes n' shares' and being able to secure a niche that generates a revenue, will be those that go beyond the cutting and pasting prompts. OUTSTANDING presentations with regard to some best practices and/or new techniques I have seen so far on your channel, easy to comprehend and apply. Glad I found your channel and I am now a subscriber.
Super useful channel with an excellent signal to noise ratio, well done.
Here's the basic prompt formula from your video:
/imagine (cinematic still), (film by), (scene/subject/action, set), (link), (shot), (dash dash)
And that list of composition call outs you mentioned:
Wide-angle
Ultra wide angle
Satellite
Eye Level
Low angle
Full Body
Close up
Over the shoulder
Two shot
Based on the date of this video, I'm assuming this is MJ V4 vs. V5. It might be a useful exercise (well, for me) if you go through this video with the release of V5 highlighting what you do/don't need. Thanks again for the videos and for speeding up my learning by months!
Excellent to hear, Rick! And yeah-- This video was done at the tail end of v4. I'll update it at some point with a refresh...haha, probably just as v6 comes out! Ugh!
ok, just too curious to not ask this probably dumb question: should I always include the "()" and the "," in midjourney?
@@SCWhiteJazz No Dumb Questions here! So, the commas: yes. That lets midjourney know to separate ideas. Although you'll want to use it as follows:
"A city sidewalk, raining, a man with a yellow umbrella"
That lets MJ know these are separate elements in your image.
The Parentheses you can totally ignore-- that was just for illustrative purposes in the video!
Holy shyte. I'm new to all this A.I.stuff from a CNN story.
Is this shit really for real? And this can be used to make fully blocked out scenes? Not just stills?
Omg Thank you so much for making a style guide. Just as I was thinking “boy it sure would be handy to have a guide to directors styles” you mentioned it. Subscribed! Thanks again
Oh… and then I see you didn’t actually post the link…. Mmmmmmm
Oh yeah-- I know. I gotta get to that at some point. Around that same time, the channel took off and I got super busy. It is on the docket, I'm just not sure when I'll get to it.
12 minutes and you hit this tute out of the park. Brevity, no long intros, great huge fonts with prompts, and background info without "being up yer own a** TH-cam style". I took notes and want to see what the rest of your channel is about. Checking out that Carpenter short as well. Thanks!
This is really beneficial! With Midjourney, I'm excited to explore with these cinematic trigger formulas to produce amazing outcomes.
According to chatgpt, these are the corresponding lens lengths to the shots discussed.
- Wide angle shot: Typically uses a lens with a focal length of 24mm or wider, which creates a sense of spaciousness and can make the subject appear further away.
- Ultra wide angle shot: Uses a lens with a focal length of 14mm or wider, which exaggerates depth and creates a dramatic, sweeping effect.
- Satellite view: This type of shot is usually achieved with a drone, and the lens length will depend on the altitude and the desired field of view. However, a popular option is to use a 16mm lens to capture a wide-angle view with a sense of scale.
- Eye level shot: Generally uses a lens with a focal length of around 50mm, which most closely matches the perspective of the human eye.
- Low angle shot: Often uses a wide-angle lens with a focal length of around 24mm to create a sense of height and make the subject appear larger than life.
- Full body shot: Typically uses a lens with a focal length of around 50-70mm to capture the subject in full without distorting their proportions.
- Close up: This shot is achieved with a lens with a focal length of around 85-135mm, which can create a sense of intimacy and highlight details on the subject's face or body.
- Over the shoulder: Typically uses a lens with a focal length of around 50-70mm to match the perspective of the human eye and create a natural, balanced image.
Thanks for that! Still working on the document for the Cinematic Prompt Guide-- I'll for sure pop that in!
@@TheoreticallyMedia Brazilian filmmaker here - Can't wait for the prompt guide, if it is as well presented as this video I would be happy to pay a pizza for that ;)
@@TheoreticallyMedia quick question if I put these prompts in it definitely creates the image the way youve described but they are rarely photorealistic - still look like game characters - any prompt suggestions to make them feel real real
Great prompt guide well done. You mentioned on March 3 in your description: MIDJOURNEY CINEMATIC STYLE GUIDE: (Sorry, link coming soon!) Its almost 6 weeks now, would still love to see it.
Wow! This was so much great information! Been on the fence about trying Midjourney due to being intimidated by it. New sub here!
excellent to hear! There is an updated (sorta) version of this on Prompting. Might be helpful: th-cam.com/video/j-GZvXgfgaY/w-d-xo.html
we need that link to the style guide please :) love the content ive seen so far keep up the good work
Thanks for this i shall try them as i have had no luck getting a wide DOF. I was designing a room with a sofa in the middle but wanted the background shap and in focus but struggled to get it. MJ seems to like to use the subject as the focal point which is fair but at the expense of everything behind it. I would have though f/22 would have fixed that but no. Thanks for the great video !!
You might want to try going with something a little more on the “conversational” side. Something like -no background blur. Or “Sharp Focus”
They’re constantly playing with the language model, so a lot of times the results can vary.
Great videos, just stumbled onto your channel. I am an older dude (over 50) who has just started with midjourney and your explanations are really good. Still trying to understand the styles and such in prompts but I am furiously scribbling down points from your videos.
Awesome to have you on board! I think next week I'm going to circle back for more of a "starters" series-- I think that's been something that the channel has been lacking. Hopefully I can fill any gaps for you!
@@TheoreticallyMedia a mate that would be awesome, your videos are really well structured so I feel I will get a lot from a starter series. I really bumble through the prompts so will thrive on a starter series :)
Great Video!! Thx for sharing. Also, did you ever release the "Style Guide" ? Thx in advance.
That style guide is still on the “to do” list. Sorry. The channel got really busy and I have had a free minute to get back to it. At some point I’ll release it, I promise! Every week, I say: this is the week! And then something crazy happens!! Ugh.
WOW, what a great tutorial!! Many thanks for sharing your fantastic formula!
Thank you! I've got an update on it coming soon. I mentioned the Prompt Style guide in the video, and I haven't finished that yet-- but once I do, I have another full video on this!
@@TheoreticallyMedia Thanks, looking forward to your next video.
Great video and explaing the details, appreciate your style! any chance of getting the link for the guide? Awesome work and pace!Thank you!
Do you still have a link to the directors that MJ seems to recognize? I don't see it in the video info.
I’m working on a full revamp of this video, and this time I’ll have the accompanying document ready to go! Ha, sorry- just a little longer!
@@TheoreticallyMedia All good, thank you for everything you do, I've found it really helpful!
Thank you so much for this! Looking forward to using it with my film students. Is the style guide out yet?
It is not…ha. Sorry. I do plan on revising this video soon, so I’ll make sure to include it when I do!
Thanks for this video ! To get a ratio closer to a real cinematographic format, it is possible to multiply it by 10 or 100, example: --ar 235:100
Thanks so much for watching, Mathieu! I think at the time of this video, Midjourney was limited to which aspect ratios you were allowed tto use, and 2:1 was as close as I could get to that "Cinescope" look-- ALTHOUGH, with v5 we now have any aspect ratio we want! Which is INSANE! I cover it in this video:
th-cam.com/video/wqC9---Xmw8/w-d-xo.html
where I play around with 2.35:1, plus do a bunch of other wacky ones, like 10:4, or 13:2-- just plugging in numbers to see what will happen. It's actually remarkable at how MJ will manage to maintain a good composition, no matter what aspect ratio you throw at it!
Thanks so much for the comment!!
Loved the Video! I visualized a complete series bible with midjourney these days, it's just amazing.
That’s awesome! A little while back, I was using MJ to do faux documentaries about films that never existed. The possibilities are really endless. And v5 looks crazy good!!
This is such a fantastic guide, thanks so much for your hard work on this! Also, I'd LOVE to see a lens test...
That's awesome to hear! I'm doing a revised version soon, since this was actually done before v5 was released, and I'll make sure to add a lens section into it!
I've been trying to get consistent oil painting portraits in the style of Diego Velazquez, Chiaroscuro, Caravaggio. Any tips?
Pop over to the discord and drop a line in ProjectHelp and let’s see what we can do:
discord.gg/bZJmJK9y
This video is excellent. A camera and lighting tutorial would be great. How to call lenses as well as specify lighting setups. All the best. Great channel!
Thanks so much! Yeah, I need to cover that in the (ever growing and still not done) Style Guide that I promised in this video. It WILL happen at some point!
Hi, this was very hlepfull, but I can't get the AR prompt to work? Any help is appreciated!
So it is: dash dash space AR space number(colon)number
No () but just wanted to indicate there is no space in the number to the colon.
Let me know if you keep running into issues!
@@TheoreticallyMedia thank you! I figured it out, I was putting a comma before the dash dash. After I removed that it started working just fine. Thank you for your help! I will be watching your content for sure!
I still come back to this months later! Don’t suppose you’ve had a chance to make a start on that guide? Or even have rough notes/other resources to help with choosing cameras, different types of shots, shot length etc?
So, it's kind of there in the background as a project, but at this point I'm thinking about waiting until MJ v6 drops, since that should be within the next few weeks (to about a month)-- It just makes sense to have a guide with the newest version, y'know?
This one, oddly enough, dates back to v4, if you can believe it!
@@TheoreticallyMedia yeah totally makes sense. In the meantime if you know of any resources for camera angles/lighting/cameras to use when experimenting with prompts, please let me know! Thanks
What if I want my subject to fit further into the scene where it's more full body, than portrait? I don't even prompt portrait either.
Thanks for the great info! I'm just learning all these cool new things with A.I. to use with my own screenplays!
That's excellent to hear! You might be interested in checking out the video I did here on Midjourney to AI filmmaking with Gen2: th-cam.com/video/PlZkWcnnRt4/w-d-xo.html
We're ALMOST there!
Is it possible to create sequential images in midjourney in such a way that you can create a story that maintains characters and themes
Not...yet. There are some tricks, but nothing that actually gets a real consistent character across multiple panels/pages ect. MJ says they're working on it-- so hopefully we'll see that feature in v6!
This is pure awesomeness! So interesting how a simple prompt with theright information can definitely pull something awesome out of AI. Thanks for sharing.
Thank you! Couldn't agree more. I also love how changing just one word can result in a drastically different image. It's so fun to quickly iterate on ideas!
How do you create an image greater than 1024x1024 72 DPI in MidJourney?
Did I respond to you on another video comment?
I am confused by "action" and "set". For set, your example was "holding an umbrella", wouldn't that be an action? Are they interchangeable or would it be, scene "waiting for a bus", set "city street after it rained" ?
So, I probably need to revise this-- and on the plus side, since v5 came out, I actually have a reason to! The original idea is Scene/Subject/Action/Set were all sort of one block. But, I don't think I was clear enough in the video, they are sub-sections within that block.
For the revision, I'm going to structure it more like this: (Film by) (Character/Action) (Scene/Location) (Link) (Shot) (Dash Dash). A prompt like that might read more like: "Cinematic Still, Film by Wes Anderson, A man in a yellow raincoat, holding an umbrella, at a bus stop, city street, morning, cloudy skies, wide angle"
(Ha, I just ran that prompt and it came out kinda cool!)
@@TheoreticallyMedia I'll give it a try!
Amazing video- could you please link the style guide as google docs even if a work in progress, thats. The images are amazing
Thank you! I’ll get to it soon, promise! Maybe I’ll start working on it in the Discord. Might keep me more on my toes!
Finally someone who just said it free just go grab it. Kudo!!!
Kudos, indeed!
On the film by part, on the prompt do you put in "film by Quentin Tarantino" or just straight up "Quentin Tarantino"?
I’ve found that sometimes if you don’t add that “film by” you end up with QT in the shot. Hah.,granted Quentin DOES love to cast himself!
@@TheoreticallyMedia thank you 😁👍
Great video! Really well explained!
excellent! Thank you! There's a bit of an updated version here, at least on prompting, if you'd like to check it out: th-cam.com/video/j-GZvXgfgaY/w-d-xo.html
Currently on my 10th or so video of yours for the day
Haha! I’ll get caught up on your comments a little later! Welcome to the party! Great to have you here!!
@@TheoreticallyMedia hahah no pressure to respond to them! When I find a channel that I enjoy I like to make sure I comment to boost the algorithm, I figure it is the least I can do.
I really love you channel and your MJ videos!! Thank for putting out such great content for the AI community :) when are you releasing your cinematic style guide for MJ?
Hopefully next week? I keep saying that, but it IS on the docket/plan for next week!
@@TheoreticallyMedia 😀😌🙏🏻🙌
@@TheoreticallyMedia hey Tim, I hope I don't bug you, but will you still release a pdf on this matter? :) would love to see it
Thank you so much for your presentation. It's the most informative midjourney training I've come across that's relevant to film making. Can you ask Midjourney to create a 3D video clip of a scene or does midjouney only create still frames?
Thank you! Thus far, we haven't had full 3d. It's mostly more in what would be considered 2.5D animation. I did a video detailing some of those techniques here: th-cam.com/video/Hzg06qJQkpw/w-d-xo.html
Hey so don't sound rude but where is the style guide you mentioned it would be liked bellow but I can't find it 😅 not sure if it's been removed maybe I've been experimenting alot with film stills in v5 without really mentioning directors much and got amazing results I'd love to experiment with some of these idea. 😊
Ha! No rudeness taken! It's my fault more than anything-- I'm still working on it. Per usual, I bit off more than I could chew-- and once I started on the project, I just kept adding more and more stuff to it. I am thinking about scaling it back and releasing a "simple" version soon...Sorry, just trying to find the balance of something that is useful to you, but also isn't going to take me a year to finish!
@@TheoreticallyMedia that's understandable can't complain about how long it takes when your spending so much effort on making sure it's comprehensive 😃
Contradictory prompts can provide unique, and sometimes, excellent results. I've combined 2 types of film, or multiple camera types, for a unique effect. For example, in doing still shots, combining a Kodak Porta high ISO film with Polaroid film gives me an image with realistic skin tone and features. I've done the same with using 2 camera types in the same prompt. It gives different results than the individual cameras used on their own.
Is the cinematic style guide link available yet please?
Ugh…soon? Sorry, it got a bit out of hand with juggling too many things. It is still on the docket to get done though!
Awesome video. Thanks a lot.
Thanks for great video! I tried the Cinematic still prompt but kept ending up with images of the director behind the camera rather than "a leading actress". Tried out different things like instead of just typing "cinematic still, James Cameron, a leading actress..." I tried with "cinematic still, a James Cameron movie, a leading actress..." but still ended up with James behind the camera, or Tarantino or whoever. The only one I got working was Wes Anderson. Any ideas?
I think MJ might be futzing with the language model, you aren't the first person I've heard having this issue. One thing that seems to work is by naming your Leading Actress. I think when you leave her unnamed, MJ just sees James Cameron and says: Ok, I know that dude!
This one worked for me, give it a shot and see how it goes: cinematic still, SciFi film by James Cameron, Scarlett Johansson as a Space Marine, holding a Flamethrower, on a Spaceship, Horror Movie Lighting, Wide Angle --ar 2:1
Note that I did add in "Sci-Fi Film" to James Cameron.
@@TheoreticallyMedia Ok cool, I’ll try that! I thought that by mentioning an actual actress or actor by name would add more confusion.
Great video but I am struggling with the first one. Cinematic still. Is this literally all we are typing before going to film by?? I’m so confused
Yes "cinematic still" then "film by"
Thank you, François!
Hey any news on the guide link :)?
Bump
This video was uploaded a month ago and there's still no style guide attached to it?
Soon? Sorry-- got caught up with a bunch of other videos. I do need to circle back to this. I'm working on it here and there, but the problem is always time. I'm actually going to do some work on it today and then do a refresh video on the whole shebang very soon!
Hey, I loved the documentarian style of the images for your John Carpenter video. How did you make those pictures so grainy and gritty looking? Was that done in midjourney, or did you add grain and unsharpness in post?
Oh, thank you so much! Yeah-- there was probably a lot of futzing in Premiere when I was making the video. I think I was putting a bunch of light leaks and film grain adjustment layers on most of the images, just to get them to pop a little more.
@@TheoreticallyMedia Thanks! Yeah I thought so, midjourney doesn't really seem to do things like noise and film emulation on a really detailed level yet. I'm waiting for the first really great color grading, texture & film emulation AI to come out.
With the detailed midjourney images I have not really seen anything yet that could really fool me that it was real (especially with people in them), but as soon as you artistically destroy the image like you did, stuff can already look very real. I personally am much more interested in that than all the weird stuff
I appreciate your videos so much. Thanks
Thank you! Really appreciate your comment!
Thank you so much for the comment, means a lot!
Hi! Great channel & tanks for the great videos. Great tips on formulating prompts. Can I ask a question? Is there an AI that does the reverse? I mean an AI that creates a prompt or a description from an image?
Hi! Thanks for the comment! There actually IS a tool for that, but I'd say don't expect a one-to-one just right now. The tool is free and you can use it here:
huggingface.co/spaces/fffiloni/CLIP-Interrogator-2
Now, that said-- It really just captures an overall vibe in the text prompt, so you probably aren't going to be getting poses, or an exact clone of the source image. But, I think it's a really great way of getting in the ballpark of what you're trying to create, and from there you can hone in. Let me know how it works for you!
Just wait for the release of chatgpt 4, it can describe image in details
@@Emsyaz Yes, I just tried it! Works quite well. Makes some obvious mistakes but I guess it will improve over time.
Wow, great video- thank you.
Thank you for the watch and the comment! Means a lot!
Love this channel, but what if we don’t want a cinematic look but want to mimic stock photo uploaded by amateur photographers to sites like Unsplash. In other words the images is not to polished and therefore looks like someone took a pictures on their DLSR or iPhone.
It’s funny, I’ve been hearing that more and more. I’ve been thinking about it too. I’ll let you know if I find a solution. That’s one thing I’d love to see, something where it doesn’t look perfectly composed.
drop that cinematic styleguide!!!
Haha! Soon? It turned into a bit of a grind. What started off as a simple spreadsheet has gotten a little out of hand.
Per usual: I bit off more than I should have! Still working on it though!
Love this.Thank you
Thank you so much! And thank you for the comment as well!
🤩 Well done m8. Thank you!!
Thank you so much for watching, and the comment! Really makes my day!
Really enjoyed this training. I like how you present your information and I have seen a lot of benefit in my rendering. II did try the "film by Quinten Tarantino" suggestion, however all I was to get was renders of Margot Robbie's bare feet smashed up against a car windshield. =)
Haha! Then it is working EXACTLY as QT would have wanted!
Thank you!
100%! There's a sorta update on the channel now as well: th-cam.com/video/j-GZvXgfgaY/w-d-xo.html you might be interested in it!
Great! I did some Kurosawa Star Wars around a month ago.Turned out some really great stuff, I didn't do anything too different to the things you have set out and it was nice to see your examples.Placing the year to 1963 turned them black and white and gave them a very old Japanese movie feel. Are you on twitter?
Oh, that's awesome! Kurosawa and Star Wars, like Peanut Butter and Sashimi! Wait...no, that sounds gross!
Yeah, the year modifier is something I'm going to get into in a follow up video, mostly talking about blending styles, which I think is something Midjourney excels at-- more so than any other Text to Image generator.
I'll actually be on Twitter in a few days...I was on from fairly early in, but left when things started to get...ugly? That said, with the channel, I'd like to be back on with a fresh account, and since it'll be more focused on the channel, I think it'll be easier to keep the toxic rambling at bay!
Drop me your handle and you'll be my first follow!
good stuff. thanks for the images.
Thanks for the watch and comment!!
Great video, Homeboy! ^_^ X
Thank you very much! Great professional content, great attitude = loyal subscriber. Would love to see a video about the interaction of MJ with photographic terms! Keep up the great work!
Thank you so much!! And that's a great suggestion! I'll jot it down on my list of upcoming topics! And really, thank you! Your comment means a lot!!
Thanks 🙏
your welcome!!
thanks, this is so good.
Thanks so much for watching!
No Style Guide yet?
Working on it!!
I use 2.4:1 all the time. Just express it as 24:10 as MJ only understand integers not decimals
Oh, that's awesome-- As a note, this was pre v.5. Since we now have ALL the aspect ratios, that ending part is now pretty much outdated.
@@TheoreticallyMedia Unfortunately it cant do 2.35:1 though, as 235:100 is TOO BIG an integer for MJ! 😁For now....
If I REALLY want 2.35:1 I can always crop a tiny bit off the sides
I just started my MJ journey. Your videos of great help. Honest and simple. Thanks for helping hobbits.
Happy to help! By the way, are you going to eat that? I missed Second Breakfast!
@@TheoreticallyMedia 😂😂😂 good one with my misspelling
How is "Link" different from "Scene" then?
Also thanks for these videos, they are super helpful!
Thanks so much!! So, I think of "Link" as additional atmosphere details for "Scene"-- like, "Scene" might be a Beach, but "Link" will have Sunny Day, or Moonlit Night.
I probably should revise the prompt structure to something more like: (Film by) (Character/Action) (Scene/Location) (Link) (Shot) (Dash Dash). I am going to circle back fairly soon (once the stupid style guide is done), and I'll add that in! Thanks so much for the question!!
@@TheoreticallyMedia Thank you for the clarification! I appreciate it.
love your content subscribed and liked
just a though have you thought about creating a pdf of the prompt templates and sell them as another revenue stream?
Thanks so much, Dwayne! I do need to do that at some point, but I'd like to approach it more from more of a tutorial perspective-- less of a "formula" (which, ironically I use in the title of this video) and more as a perspective.
I have been putting up PDFs of later videos and offering them for free/Donation-- and I'm very happy to say that the donation response has been super generous. Can't thank those people enough!
Style guide? Example refernece prompt? Anything?
Awesome!!!! Thank You!!!
Thank you for checking it out!!
Please, when you say the words, like you did on 9:09, for those who don't have english as first language.
Happy to help! But do you have the correct time stamp? At 9:09 I am apparently reading a prompt and the text appears on screen. Happy to clarify anything you need though!
I’m in - thanks.
MIDJOURNEY CINEMATIC STYLE GUIDE link?
Oh man, still working on it. Sorry-- It turned into a bit of a monster (my fault). I wanted to kick it up a notch, and it ended up being a bit more than I could chew. I'll get it there soon though!
@@TheoreticallyMedia hahaah no worries just wondering :)
@@dm.b7560 Ha, there was also the problem that while I was working on it, v5 came out! Haha, I need a week where progress stops on all AI!!
@@TheoreticallyMedia hahaha v6 will be out before you finish :P
good stuff
Hey Tim, and all. I was not sure where to post this so thought I would do so on this video since Tim is talking about Cinematic prompts here. And this is a real business application where I am building a business around AI Character-driven book trailers. Here is one I did - "The High Priestess". I used Midjourney to generate the images, then morphed the faces to get consistent characters with Patreon, then used Photoshop for final clean up (eye and hair color for the most part), and then pulled into TensorPix to upres some of the final stills. For the second to last shot I used Runway to generate the video (from a Midjourney generated still). All then edited in Adobe Premier Pro. Enjoy: th-cam.com/video/UMLpGROtrxQ/w-d-xo.html
Still no style guide?
Not yet. I dug myself into a bit of a content hole...too much stuff going on. I'll try to get to it soon. Likely to revamp the whole video with 5.1...
Cool 😎 Video
Thanks so much! This one was a lot of fun to make!
@@TheoreticallyMedia I still have a lot to learn
@@ryanhowell4492 To be honest, we all do! This is such new technology that we're all learning on the fly! New stuff is popping up every day, leading to new techniques and approaches. And we get to be a part of it all! It's a pretty exciting time!
Pretty sure the 'Star Wars' look was triggered by 'Lightsabres' not Akira Kuwosawa in space.
John Carpenter did make an Alien movie (kind of) before Ridley Scott: Dark Star
Perhaps try "firing at a lovecraftian monster" rather "shooting at a lovecraftian monster". I think Midjourney will associate the word shooting with a camera especially if you've specified a camera or lens type.
Yup! I've got a revision to the Cinematic Prompts thing coming up soon, since this was made before v5-- I'll make sure to include that in it!
谢谢你，i mean thank you
不客气 -- I hope my translation software worked!
Soul left my body at Star Wars 😂
Haha, it so true! And watching Ashoka so far, it’s clear that they’re leaning into the Kurosawa aspects in that show. I know a lot of people are complaining the show is “boring” and I’m like: no, dude- it’s a 60s Samurai movie!
I mean…in space. With space wizards. But still!!
TM, do you/could you create images for me? For one of my short films I would like to put on TH-cam. Please advise. best, GG
I 100% agree, I like to use clear and concise direction, you might get something you like from spamming the prompts, but you're just playing the odds at that point hoping to get something cool. For more control, less is more. Also, wanna let you know that I'm going to try out your "non-Star Wars Star Wars prompt" in GEN 2 lol I might change it up as GEN 2 isn't midjourney level, but I gotta say I've been getting pretty decent stuff. Mostly through proper prompts and getting the settings just right. Let's see how I do, I'll post it for you in this comment. Thank for another great video, you videos have been a great help, so much appreciated!
th-cam.com/video/uiFcBBhX8io/w-d-xo.html
Here are my attempts hah @theoreticallymedia
as usual, I see folks uing MJ and getting cool results. But I want to see someone getting results that were intended or changing the prompt to get the right results
How's so? Like, specific poses/facial expressions and the like?
@@TheoreticallyMedia I often see huge and probably redundant prompts people make and MJ will ignore basic requests like giving closeups instead of full body. Trying to get things closer or further away in frame. Getting MJ to put a background that YOU request. Not putting a tail on something when to ask for no tail
@@DKGCustom Oh, I get ya. It's funny, I just made a video about that after someone commented something very similar:
th-cam.com/video/j3LiGOBglCs/w-d-xo.html
Long story short, there are some tricks to get what you're looking for-- but there are other times that, yeah-- MJ is just gonna ignore you. Or give you some really terrible results!
11:48 Jordan Peterson on the left and Gigachad on the right. It's gotta be them.
Do you pay for this?
Please do
You'll want to use --no camera
I ask this in the least condescending way I can, but what about this process do people like? Why do people like assembling pixels this way, why are these even being made? I can't really warp my head around what's enjoyable about making and viewing these, for a lack of better terms, "images". Looking for a real conversation
Thank you for your question! It's great to see that you're interested in understanding the appeal behind AI-generated images, and I'd be happy to provide some insight. Here's a few reasons, I'm sure other people can add to this list.
Creative exploration: AI-generated art offers a unique opportunity to explore new and unexpected visual realms. By training AI on a diverse range of styles and visual elements, it can generate images that blend these influences in novel ways, resulting in truly unique pieces. This can be a source of inspiration for artists and art enthusiasts alike.
Collaboration between humans and machines: Many artists have embraced AI-generated images as a way to augment their own creative process. By using AI as a collaborator, artists can push the boundaries of their work, blending human intuition with machine-generated suggestions to create innovative pieces that might not have been conceived otherwise.
Entertainment and fun: Some people find enjoyment in the surprise and novelty of AI-generated images. As the AI learns from its training data, it can produce images that are unexpected, strange, or even humorous. This can be entertaining for users who appreciate the unpredictability of the results.
Customization and personalization: AI-generated images can be tailored to individual tastes and preferences, making them an appealing option for personalized art and design. For example, users can input their favorite colors, styles, or themes, and the AI will generate a piece that incorporates those elements, creating a one-of-a-kind work of art.
Practical applications: Beyond entertainment and artistic expression, AI-generated images have numerous practical applications as well. They can be used for tasks such as generating textures for video games and movies, creating advertising materials, or producing visually engaging content for social media platforms.
Accessibility: AI-generated art democratizes the art-making process, allowing people with little or no artistic skill to create visually appealing images. This can be empowering for those who might have previously felt excluded from the world of art due to their perceived lack of talent or expertise.
Research and innovation: AI-generated images are not only a form of artistic expression but also an area of active research in the field of artificial intelligence. As researchers continue to develop and refine these technologies, we can expect to see new and innovative applications emerge, pushing the boundaries of what's possible in the realm of AI-generated art.
I hope this helps you understand the various reasons people enjoy creating and viewing AI-generated images. It's a fascinating intersection of art, technology, and human creativity that has captured the imaginations of many. If you have any further questions or would like to discuss this topic more, please feel free to ask!
Hey man, thanks for the thoughtful question-- and thank you for your opening, I think that sometimes the tone of questions come off wrong in a text/comment format, so I greatly appreciate your clarification at the top!
Anyhow, I do hope you get a few different answers, because honestly, it is different for everyone. Since starting the channel, I've heard all kinds of reasons-- one person that always wanted to visualize a dream they had as a child. Many like creating these cinematic images for mood-boards, or creating images from a film they wished had been made. There are a few comic book authors (although, I still personally feel that Midjourney is the wrong tool on that front), folks that want to make books for their kids...The list goes on.
I got really into it via a fake-documentary I made using Midjourney, detailing a 1977 John Carpenter film that was Alien, before Alien. If you're interested you can check it out here: th-cam.com/video/BbeQDnBSjD8/w-d-xo.html
Ultimately, I think the real joy of it all is thinking "I wonder what X would look like?" popping in a prompt and seeing it. Once you go down the rabbit hole, you'll see that first image, and a bunch of other ideas will come to you. Before long, you're coming up with specific ideas, and that's where the process factor kicks in.
Thanks again for your comment! If you have any questions, please don't hesitate to reply!
It’s a fair question and I can answer it from a practical point of view as someone who uses Midjourney for work purposes. I make thumbnails for TH-cam and the amount of time it saves is phenomenal. A client could give me a really obscure piece of work which could lead to me spending hours and hours looking for the right assets to make the thumbnail. Looking at different assets with different backgrounds, lighting, composition, etc can be so draining if you can’t get it right. With AI tools I can prompt exactly what I need and make minor adjustments instead of spending hours just searching for assets. This allows me to spend more time perfecting the lighting, blending, etc instead of that time going on searching for assets. That’s my take on it anyway.
@@James777x Thanks for the deep and thoughtful response! Entertainment and fun, I get that. I'm sure it's fun to see all these things come to life almost instantaneously.
I guess where I get tripped up is the whole 'art' thing. You say "it" a lot and not "me" or "you". The idea of a person taking credit for this as art based on typing a few words into, essentially an intake form, is a little bit unsettling to me.
While I'm not Wes Anderson, although I hope to be at that level some day, I don't think I'd appreciate someone saying "make me a Wed Anderson style photo" having it generate instantly then going to take credit for it as art. I would argue that had nothing do to with the "artist" and 100% to do with the machine biting his style.
I think creative exploration is more about thinking and playing and working it out, which in turn will inevitably force the artist to put their own spin on it, not just waiting of AI to do all the work for you.
I have hope that society will reject it as "art", but that really isn't up to me and we'll just have to wait and see I guess!
@@TheoreticallyMedia thanks for responding! I threw some thoughts on the comment above
Jesus, it´s always creepy to see those images containing humans. There is some kind of horror, knowing that nobody real created this person.
Oh, it’s crossed my mind more than once. The really weird part is when you run across an AI character that looks like someone you know in real life.
Like when there is a background character and you’re like “geez, that looks like Jeff in accounting!” That’s weird.
It also might be Jeff’s fault that he’s so bland looking. For god sakes, Jeff, get a personality!!
Very helpful video thanks. I've been trying this whole AI thing over the last few days (video below). But irs really the prompts that get me confused. It seems the order is very important. The first item mentioned is a big deal apparently. I was trying to recreate a Moses character and wanted a wide angle shot but with a bad order Moses appeared with a DSLR to his eye! 😂
th-cam.com/video/YEYkcwJrNrA/w-d-xo.html
Ha! Yeah, the camera thing happens pretty often. You can try dropping specifics on the camera model if it keeps happening, or try something like “style by DSLR” and see if that helps!
Oh, and order is 100% clutch!
What do you want to do when you grow up?
Ow, you know, talk on TH-cam about prompts for AI images generaters.
You wanna do what?..
Haha, yeah-- life takes you in weird directions. Funny thing: This isn't even the strangest job I've had. But it is the most fun!
@@TheoreticallyMedia curious about the strangest job now ;)
a little diversity like people of color or women would be a nice addition to your example
Oh, I can’t say I disagree with you. There were obviously women in the example images, but they were mostly Caucasian. But I think that has more to do with the directors chosen in the image prompts.
There aren’t many people of color in Tim Burton or David Fincher films, right? And Midjourney is basically taking the median of character that appear in those films and creating a person who might be cast.
I will say that the entire back half of the video contains examples from Akira Kurosawa films, so there is a pretty good amount of Asian representation in this tutorial.
But, you bring up a point that I’ve been mulling over for the last week or so, and I think your comment finally pushed me over the edge to make a video about it. There’s a problem with both body type and representation in AI imagery, and we should start having a meaningful discussion about it.
Thanks very much for your comment.
thank you
This is absolutely inherent in midjourney, it’s amazing. You HAVE to specify a specific race to get anything but white people. Systemic biases illustrated for all to see! 💔
@@TheoreticallyMedia well said ❤
What about doing your own Midjourney stuff and keep your vertue signaling and "anti"racist rants for yourself ? He could have shown only whites or blacks or whatever, what's the deal ? You get a free video with lots of cool tips, that doesn't talk a single second about race, and this is the only comment you could think of ? What about thanks for the content ?
What about your own style?
Just rips offs I guess, that's what Ai is all about, right?
Yeah, I'm not wading too deeply into those waters. I'll say this though: David Bowie had a quote that I always liked (I'm paraphrasing it here): "There's no such thing as an original artist, just very tasty thieves"
AI is not very tasty. It takes human input to tell it what to create. AI is very good at taking two or more styles and blending them together to create a new look, but it still takes a person to decide what those things are.
The thing that I'm always pushing is: We have these tools now that will allow you to create anything you want, go MAKE something with them. Make a short film, make a music video, make a comic book, make a video game.
To me, that's where a creative person's style will evolve from all this.
haha, people with zero skill still struggling to make good images with AI is hilarious
How do you figure? If someone is new to creating images, are they expected to step in and create the most masterful prompt on their first go around? You might be thinking of babies. Babies have zero skill and when they try to do something it is hilarious.
useless