How to Use NEW Midjourney V6 (Full Midjourney v6 Tutorial) (Dalle -3 KILLER)

With the release of Midjourney version 6, this video is going to give you a full breakdown of exactly why this is so cool and exactly how this compares to Dall E3's revolutionary Image 2 text. So that being said, don't forget to watch the video until the end so you can skip around and see which parts of Mid Journey you find interesting, including the comparisons between ChatGPT, all of the secret new tricks, and anything that you might have missed regarding tips and tricks for Midjourney version 6:6. So the first thing is first, of course Mid Journey has been released. In order to use version 6, all you need to do is just select V6 from the drop down menu under Settings or you just want to type V6 after your prompt. What this essentially means is that if you just select V6 in the settings, that just means that you won't need to type in after every single prompt. But if you do just want to test it out because you are good with version 5.2, you can just put in version 6 after your prompt and then you're going to get that new Bijini version 6 prompt. So what I'm going to firstly go through is essentially the Cliff Notes versions of everything that's new. Then I'm going to show you guys some of the cool tips and tricks because there are a lot of things that you do miss. So one of the things that most people are going to miss about the model is the fact that this is actually an alpha test. So that means that the mid journey version 6 model, although that is actually really, really good right now, this is alpha. Now, alpha is essentially the first iteration of the model. Not the first iteration, but the first first release because you do have Alpha, then you have a beta, then you usually have a full release. So there could be two more versions, two different kind of versions of Mid Journey that we do get. So I would say that whilst Mid Journey is good, even if there are some things that aren't great right now, you know, some specific instances, they will improve. So it also says do not rely on this exact model being available in the future. It will change significantly as we take V6 to full release. And of course speech image quality, coherence, prompt following and, and text accuracy should improve over the next few weeks. So like I said, it's really good at certain things I'm about to show you. But it's going to be even more good over the coming weeks because they just rolled this out in alpha as Christmas rolled around. Now, of course, V6 is a bit slower and more expensive than V5, but it's going to be getting faster as they optimized. And they do have relaxed mode available also. So then you can see here it actually talks about what exactly is new with the version 6 based model because Midjourney is already a pretty comprehensive AI tool. It's really, really good at generating these lifelike images which. But what's really new with the version 6? So it says here much more accurate prompt following as well as longer prompts. I'm going to talk about this later on in the video in supreme detail because the way how you can actually prompt with Mid Journey has changed a lot. So this is, you know, a complete rehaul, but it's really much simpler. So even though you have to learn how to prompt, it's much simpler and makes much more sense. Of course, improved coherence and model knowledge, improved image prompting and remix. Remix is really cool. Going to show you some amazing things with that. And of course minor text drawing ability. You have to put your text in quotations and also Style Raw and Stylize are actually pretty good as well. And of course there's improved upscaler. So let's get into the really, really cool stuff. Okay. And if you are going to be using V6 and you want to know which things are supported, there's AR chaos, weird tile, stylized star, Raw, very, very strong remix and blend and of course describe. But I want to show you guys how crazy it is. So for the prompting of V6 just before we get into some of the images. Okay, prompting is like I said, you need to relearn it. But V6, you know, it's no longer about those junk words where it's like. So before I decide to get into some of the more rules and guidelines, I really just want to show you guys some of the really amazing side by side examples of Mid Journey version 6 vs Dall E3 from Chat GPT OpenAI because once you see them, you'll understand exactly which you're going to need for what and. And the key differences so that you become a more effective creative or someone that just likes playing with these tools to get whichever desired output you want. So I decided to put in here a lime green McLaren 650s outside a luxury store in London. You can see on Mid Journey how it imagines that. And then of course the same thing in Chat GPT. Now you can instantly see that the realism on this side in terms of what we're looking at is extremely real. Now I didn't specify anything for the text here. So I'm sure that if I did want text up here, that, that or text on the license plate, I'm pretty sure I could have done that. But I kept this one really basic to just show you how these systems initially interpret things, because that's what they're based on. Now, of course, there's probably going to be a Dali 4, which is could be hyper realistic. But the point is, so far, what we've seen From Mid Journey vs Chat GPT, I want to just give you guys, the main takeaway from the two systems is that Mid Journey is super, super realism with the Star Raw with stuff like that, with ChatGPT, what we do have is, is something that's much more creative, much more playful, and something that doesn't look as realistic. But it isn't bad, it's good. It just doesn't look as realistic. Like, if you see an Image From Dall? E 3, you're going to know it's from Dall? E 3. But if you see an image from Midjourney, you're going to have to look at it twice. For example, the license plate, it's smaller details, maybe the guy in the back in order to be able to tell exactly what's what, because I'm not going to lie, that is definitely much more realistic than that on the right hand side. Now, there were some other examples, and like I said, this is an alpha because I did typed in prompt the Batman punching a criminal in the dark alley because I wanted to see if it was able to do some kind of action shots. And although Mid Journeys did look a lot more realistic, I didn't really get the exact thing. I think this one right here was probably really cool. And trust me, guys, I'm a super Mid Journey fan. I really, really do like the software. But you can see that Dall E3 does capture this one a bit better. But I do want to say, before you guys take away saying, oh, Dall E3 still better than Midjourney versus 6. This isn't the truth, guys. This is just honestly one very small, small example. And like I said, when you watch the video and you see tons of numerous examples between the two softwares and just between version 5 and the version 6, you're going to see that this is truly, truly a huge upgrade. Because these ones right here, these ones are like comic book illustrations. I kept this one super basic, but like I said, I'm going to show you an example that someone else did and why it's pretty crazy. So someone Else did this and they said, okay, Dall E3 is still more accurate at generating text than Midjourney version 6. But if we look at the aesthetics, Midjourney just blows it away. Which is completely true, because comparing these two images together, you can see that one of them looks much more realistic and one of them you could definitely use for like a realistic campaign or form, just more realism. This one is more of like a 3D kind of arty style. And then this one's like much more realistic. And of course, the prompt that they had here was Fresh blend. I do like Midjourney's typography because I'm not going to lie, that level of typography and design and coherence is really, really hard. Like someone who did graphic design before and understands that, I can definitely tell you that that is a. Is really hard. So, yeah, it's really, really good. Because I'm not going to lie. Realism combined with great text is definitely something that we can look forward to. And of course, it's something that's really cool. Now, what you might also want to know is some other stuff, because there's some other stuff I do want to show you before we look at the version 5 and version 6 differences. So for the prompting, because you could see there that my prompts are really small, like I said, prompting is going to be completely different. So essentially V6 is much more sensitive to your prompt. So what you need to do is you need to avoid junk like award rating for photorealistic 4K 8K. So previously what we did was we like stuffed up these prompts with like four photorealistic 8K super high quality HD, you know, like HDR, whatever image, like all of that stuff that we thought would make it better doesn't make images any better. And of course, you know, you do need to be explicit about what you want. It's going to be less vibey, but you are explicit. It's now much better at understanding you because essentially what they've done is they've made midjourney version 6 achieve natural language. So it understands natural language, it can process it, and that means it's going to give you a better output. So I'm going to show you guys some side by side examples of why that is also really cool. Now, if you want something more photographic and less opinionated and more literal, which just makes it look more photorealistic, you just want to put dash, dash, style, raw, and it's going to make those images look super photorealistic. So it also says lower Values of stylized, which the default is 100, may have better prompt understanding, while higher values of up to a thousand may have better aesthetic. So, yeah, it's up to you whether you want prompt understanding or better aesthetics. You can change the stylized feature. Just put dash, dash and then put whichever number you want it to do. Because the default is at 100 and you can put it up to 1000 or you can put it lower than that. So that is going to be that then, of course. Okay, we do have the version 5 to version 6. So this was one example I saw on Twitter. This is version 5. This is version 6. Definitely a huge improvement. I did use version 5 to generate some photos of Elon Musk. They were really good, but these ones were just. These version 6 ones are just a level of photorealism that is just. I mean, you know, part of me today was just looking at these images and thinking how far we've come is just honestly incredible. But, you know, we don't have time to dilly dally. Let's take a look at just how far these images have come because it's truly incredible. So you can see right here we have a corner bar with a neon sign that says Open late. We got version 5.2 on the left and version 6 on the right. You can see, of course, there's no text there because, you know, version five can't do that. Then of course, you know, on the right hand side you can see it says Open late. And it's really, really, really nice, like how realistic that is. I mean, maybe you're a brand owner, maybe you're someone who wants to just theorize what your bar could look like when you make it. You wanted to put your text there, you put it there. You can get like a quick mock up. That is something that a lot of businesses do want, a lot of businesses do need. So being able to use the software is really cool. Of course, when you look at these further examples, you're going to see just how crazy it is. Version 5 to version 6. So 1960s street style photo of a young woman sitting on a sailboat wearing a green silk dress and a pearl necklace. The sun is setting over the ocean shirt shot on Afgha Vista 200. So you can see here that these are two different styles. V6 is the top and version 5.2 is the bottom. And version 6 is absolutely incredible. Like, I mean, this one really looks like you can see the pearl necklace, the, you know, the sailboat. You can see like the sun setting in the bottom that's what version 6 is truly all about. The term like the accuracy of this model is insane. And I think like Mid Journey, the main thing that they're doing now is they're just making the model more and more accurate because quality, we've already surpassed that. Like this image here is high quality. Okay, it rated quality. The only thing we need is more accuracy. Like when sometimes you input a couple of words, it doesn't get that. But version 6 is pretty much hitting the nail on the head. Take a look at this example 2 1970s medium close up ectochrome fill of a still of a 25 year old woman with curly red hair and freckles standing in front of a vintage car in downtown Los Angeles. She's wearing a green silk jumpsuit with a plunging neckline. Golden hour. So you can see that this is of course version 5.2 and, and of course version 6. So this is crazy because you can see that the freckles are much more pronounced. I mean, nothing about this image on the right or the left looks AI generated at all. So like I said, the quality is just unbelievable. And like I said, I was just looking at these images today thinking how on earth is it that we've come this far with AI? If you would have told me like 3 years ago these are AI generated images, I would have said, you know, just stop lying because that isn't possible. And I knew that AI generated images were coming. But to get to this level, and there's no other company that does it as well as midjourney, they definitely have some super, super secret sauce. Then of course we have a 1980s Mysteries film. Low angle shot of an evil French eyed butler sporting a black suit and grasping a candle in the hallway. I actually don't know which is which here, but I do think that the bottom one is version six, just in terms of the quality. But like I said, either way it's crazy. I'm pretty sure the bottom one is version six. And I mean if you would have told me that that looks like it's AI generated, I would have said no way. Because there's just some things that are just too good. Like you AI system would just struggle with it, but the realism is just unmatched to the point where I'm starting to get a little bit afraid of what AI can do. Then we have documentary photography of lion family in the African savannah at sunset. The scene includes a male lion, a lioness and their cubs. The background features acacia trees setting in the sun. So this is of course Another one that looks really good. The only problem with the version 5.2 is that it didn't really capture the, I guess you could say male lion and the lioness, but I'm guessing this one as well only really had the lion cubs, so. So, like I said, there's small nuances where sometimes it doesn't work. But at the end of the day, if I would have seen this image and someone told me it's from like a nature documentary, I would have said, sure. Like, I would have believed it because, I mean, it looks like it. Like there's nothing about the image that explicitly says AI generated. We don't have the weird hands, we don't have that. And then if you want to get into more realism, you can take a look at this property stuff right here, which is just crazy. And it says editorial style. Photo, eye level, modern living room fireplace. Leather and wood built in shelves, neutral with pops of blue. West Elm. Natural light, New York City. The afternoon cozy. Art deco. Version 6 is top and version 5.2 is the bottom. I mean, they both look really, really good and really, really realistic. But yeah, I mean, I really can't tell the difference. I mean, version six, I mean, it does look a little bit better. I mean, here we do have a little bit of weirdness. I mean, that's only like if you're nitpicking. But yeah, I mean, it's incredible. Like, honestly, it is absolutely insane. I mean, just look at the window. Like the. Just the way how the window is just like you can kind of see like a building outside. I mean, it's just. It's just too good. Like, it's. It's just. I mean, part of this video is going to be me being amazed at the kind of stuff that we have here. And of course, we have reflection of a sunset in a rain puddle, version 5.2 on the left and version 6 on the right. I don't think there's that much difference here apart from the fact that version 6 just seems to be a little bit more detailed in terms of the quality. And of course, this one I really do like. This one I really do like the old styles as well, because it shows us that not only is Mid Journey able to capture new things in a variety of angles, it also shows us that it's able to get, you know, different genres and different styles really, really correct. So you can see that this one we have detective interviewing female witness at crime scene. And we can see the females taking notes here. But this one doesn't really look that good because the way he's holding a paper in his hand, but this one definitely looks like a shot out of a movie. And that sentence that I just said will pay not dividends, but pay attention because at the end of the video I'm going to show you guys something and if you want to skip to the end, you can, but it's really going to surprise you. And I don't know how midjourney are going to deal with that problem. But anyways, now that you've seen the images, how do you deal with text properly? Because you might be thinking, okay, you told us you need to prompt, how on earth do you deal with prompting properly? So what you need to do is essentially you can add text to your image. So essentially, if you want to add text, all you need to do is just add it inside quotation marks. So all you need to do, like I said, just add it in quotation marks, use your words and of course use the other words that describe exactly how that text is. So for example, you can say printed on, entitled, inscribed, labeled as, marked with brand, as, engraved with, stamped with, adorned with, scripted with, lettered with. And if you don't know what any of these words mean, just have a Google, like just, you know, printed, embossed because you're going to want to see exactly what these are so that you can gain an understanding of how things work. And of course you can also print text on things such as speech bubble, post it, note, book cover, poster, sign T shirt, mug, billboard newspaper, just pretty much anything that you can think of, you can put text on. So essentially that is how you use text. You just type it in, put it inside quotation marks, just, you know, you pull it in like a red car with the text, awesome stuff in quotation marks. And then it's going to be able to do that. And of course we've got positioning and this is, like I said, just mid journey's main focus. So what you want to do is make sure that you learn a natural language so that you understand how to control the composition. So you can see here they've highlighted stuff like the basket in the middle, the basket on the left, the basket on the right and in the background. So if you really want to make sure your images are good, make sure that you state where things, things are. If something's on the left, state that this is on the left, something on the right states on the right, state what's in the background, state what's in the foreground. The foreground just means what's in the front. The background just means on the backs. The left is of course your left and your right. And of course, yeah, this basically a simple thing, but it's able to understand that a lot more regularly. Then of course we have only important words are kept. You can see here that like I said before when we used to stuff these entire things, you can see all of these different text pieces right here. All of these words don't matter. The only words that do matter are the ones that determine the output of the image. And here you can see he's wearing red glasses. And you can see right here that the glasses are in fact red. The point that they're trying to say is that only important words are kept so you know, clear details. All of this, you know, cinematic lighting, yada yada, all that stuff, you don't need that stuff anymore. Like having 500 words is not something that's going to be done in version 6. They're making it really, really simple. Kind of like a chatgpt style where you can just literally say, you know, a man with red glasses doing X rather than hyper realistic 8k hyper realism octane render. That stuff that you're not, you don't need. So you can see here you need visually descriptive adverbs, visually descriptive adjectives. So like crowded, empty, textured, intricate. Intricate just means like, you know, like if you've ever seen like the inside of like some kind of like it's kind of hard to describe, but just imagine like a bunch of small things like that are like thin and just like you know, like a clock, like it was really, really cool. Mechanical clocks that will have those thin, intricate pieces and of course serene, atmospheric. I. It's going to be good to just go and Google and just Google a bunch of these words just to see what they kind of want so that you know exactly how to create those images. Then of course you can reference artists, you can reference art movements, design movements, you can reference photography styles, you can reference ethnicities, cultures, photography. And of course you can use words like vividly, sharply, dimly, brightly, subtly, boldly, hazily, seamlessly. And some of these words I'm not entirely familiar with, but like I said, Google is your friend. Just Google them. See images that perfectly describe that, then use those words and then of course use that to your advantage. Then of course you can see here are example words that don't, you know, insane, extreme, you know, somewhat. These are words that just don't give you much. Like 8k ultra before, since these words just really don't help you. So if you want to screenshot this, you can, of course, if you just go into mid journey v6, if you go into the prompt area, that's where all of this stuff is. I'm just doing it so it's easier for you guys. Then, of course, we have some more examples from Twitter. You can see that this is version six. They wanted to make a coat with the text Ventura on it. And like I said, if someone were to show me this, I wouldn't think that this is AI generated, because the way the text is, it's too seamlessly imprinted on the jacket. So it just. I mean, it's just. It's just too good. And then we've got another one here. And the way that the text just deforms that. I mean, I don't know, guys. Like, it's just. It's just truly, truly crazy how this software works. And then I want to show you guys something funny, because I found this really cool. So this guy was sharing stuff, some Star wars stuff. So he did. So you can see right here we have mid Journey version 6. And essentially with the Star wars stuff, right, you can see that his prompt here was a stormtrooper with a text loser on it. I mean, it looks absolutely crazy. The detail, the dirt. Just everything looks like. I mean, there's no way that looks AI generated. But. Okay, what's funny about this is that I did try this with ChatGPT, and guess what? Okay, they said I'm unable to generate images that don't align with our content policy. And this includes creating images that may portray characters or figures and demeaning or negative lights. Just adding the word loser to a stormtrooper helmet. If you have another phrase, I'm happy to do that. Now, the worst thing is I tried to put the name awesome on the helmet, and it still declined because it said copyright. Okay, but you need to take a look at Midjourney's copyright because they seriously probably have maybe another lawsuit coming. I mean, they got all the cash in the world to fight it. But it's going to be pretty crazy because I need to show you guys that in a moment. But anyways, like I said, we're going to take a look at some of the. More images. This one was really cool. I found this one really cool. Yoda eats frogs. And we had this one right here, which is Luke's dad. Like I said, really cool. Although the text on this one does look a bit. A bit weird, if I'm being honest. And then someone else in the replies did it again and it did look a lot better. So text has improved. Now, of course, we do have remixing, which is really cool. Remixing Mid journey has improved. V6. It lets you imagine an image with an image while carrying over the same exact form and basically the same image composition. So what this guy did was he used remixing to turn this woman into a robot. And you can see right here that once on running variations, open the new prompts box that let you adjust and change your prompt. And of course you can just change it so you can remix it into whatever you want. It's really cool. And like I said, V6 seems to be far more accurate than before. And yes, it's crazy. Now, this is the main thing that most people do want to know. Multiple subjects. How do you do multiple subjects in Mid Journey? So of course, for prompts where it makes sense to do so, set up the scene in generic terms using archetypes. In the first statement, there's a sweet spot for specific. So essentially this is showing you the good example. So good is three friends sitting on a park bench. Better is three different friends sitting on a park bench. And without different midjourney gets to decide their general appearance. And they may appear similar. And then of course, course, be specific. So three different friends sitting close together on a park bench. And without best friends and sitting close together, we get a more generic vibe. So it's always best to just, you know, describe the image as you would, as if you were talking to a friend and you might be thinking, wow, to do that. I'm going to show you in a moment, but with multiple subjects, things that you don't want to do. So, for example, you know, what they're saying here is that if you want to do multiple subjects, you need to understand callbacks. So basically what this means is, is that you can see here, this is this initial setup. So pay attention to this words right here. Three different best friends sitting close together on a park bench. Callbacks just means that you need to use the same words you did before. So they used three different best friends sitting close together on a park bench. So when you're using a callback, the next part of the sentence should be the friend in the middle is a cheerful blonde Caucasian woman wearing jeans and a green tank top. So you can see how they said the friend in the middle. Okay, so three different best friends, okay? And then with this one, it's just like Jennifer is in the middle, a blonde Caucasian woman that doesn't have anything. Do it then it says in the middle. That's not good because it doesn't say the friend in the middle. So if you've got three best friends. So for example, if you've got three cars, you wouldn't say the red one, you'd say the car in the middle. So essentially it just means use the kind of same linguistic terms if that does make any sense to you at all. Hopefully it does. But like I said, you can screenshot it, you can test it out and then figure out exactly what works. And this is exactly something that I was saying before. The backdrop, the background does make sense to use before. So of course it's something that you're going to need to do. Of course, again with multiple subjects as well. Of course, you know, not greatest photo, it's better color photography or just use the word photography. And then of course, you know, if you do want like specific styles and stuff, something that I did find useful is you can also, you know, use ChatGPT to figure out which styles you do want. Just ask it, can I get some ideas? And it will literally give you those because it knows a vast, vast amount of different photography styles that you could then go ahead and research. So I would advise you to do that because once I did that, my Mid Journey game just completely changed. So that is a secret tip. And also, okay, like it says, don't use art, use digital art. And of course, I guess you could say Mid Journey does require you to have a decent understanding of certain styles because you can't just ask it do something. I mean, it can, but it's a lot better when you understand all the different styles because if you just take like a 15 minute look to look around the different art styles, because midjourney was trained on tons and millions of images, you're going to really get a lot of better results in terms of what you actually want to see. So this is where we say talk to it like ChatGPT because the example that they showed us was, and I'm sure this might be from Twitter. So I will, like I said, leave links down to everything in the description. But you can see a raven purchase on a dead branch in the style of a gothic charcoal sketch on antique paper. Then on ChatGPT style, this is the same version, create a charcoal sketch of a raven. The sketch should look like it's on antique paper. Don't include any leaves on the branch because it's a dead branch. Make it look gothic. You can see right here that this is what most people are going to be able to do. And this right here is something that looks a bit advanced. So like I said, talk to it like it's chat GPT and you're going to be doing far better off in terms of your prompting. So this is the new way to prompt and this is what is far more effective based on what Mid Journey have told us. And then you can see right here, I actually did find this one really funny because you can see right here that if you do have the right punctuation, you can get the wrong thing. So you can see a panda eats, shoots, then leaves. And of course, we've got a panda with a gun. That's not what we want to have. So because of the comma right here, a panda eats, then shoots, then leaves. But a panda eats, shoots and leaves. You know, it's two completely different senses. So your grammar and punctuation is a bit important. Much more in V6. So that is something you do want to take part of. Now, one of the craziest things that we did see with Mid Journey, and I know most people won't get to this part of the video, but for those of you who do, okay, I really want to know what your opinions are on this, because Mid Journey, we know, was trained on millions and millions of images. But there was a tweet that essentially said that there was some data leakage from Mid Journey. So people were trying to create screenshots from the movies. Okay, just like inspired by. But they got pretty much exact replicas from the movie outputted from Mid Journey. So I think this is a big problem because if someone wants to do a lawsuit for midjourney, I'm not sure the status of any legal cases that they do have. But you can see right here it says, looks like the last example of data leakage that I shared wasn't at all hard to replicate. And you can see the Mid Journey prompt. Thanos Infinity War 2018, screenshot from a movie, scene 4K, Blu Ray, yada yada yada. Even though they're not supposed to use this prompt, you can see how similar that stuff is. So, I mean, it's pretty clear that these are just images from the training data and someone else was able to do this. I'm definitely going to include another part about this in the video. But if you did enjoy this Mid Journey full breakdown, don't forget to leave, like, comments below all that good stuff. And I'll see you guys video in next. The.