Ep. 7: Midjourney Parameters: The Forgotten Cheat Codes (& Best Ways To Use Them)

0:00:00Intro & Midjourney's New "Editor" Feature

Play from 0:00:00 8:88:88 0:00:00 What's up everybody? Rory Flynn, Drew Brocker, we're back in the building. It must be a Friday cuz we are recording and man, it's always a great time just being in the booth, talking shop. Midjourney, of course, dropped another big update in the late eve hours. So, I think we're just kind of getting our feet wet in that with with editor, but man, that's a huge piece of the puzzle that you and I have been talking about for a long time, even on this podcast.

Play from 0:00:28 8:88:88 0:00:28 So, pretty convinced at this point that they are listening in. So, we'll cover that, show you guys what's changed, the new abilities that that gives you to stay in tool, which I think is like a fascinating component of why that's so effective. And then the big meat of today is going to be getting into parameters at a deep deep level. We're going to go down the list of parameters. Rory and I kind of find that a lot of people either don't know of all the different parameters that exist, they don't use all of them, they don't know what each one does. And so, we're here to sort of bust some of those myths, break those down, visually show you how we use each one, and how you might think about using each one as well. So, welcome to the show. Rory, take it away, man.

Play from 0:01:06 8:88:88 0:01:06 What's up, dude? You know, again, Midjourney just drops this at I think it was like 9:30 last night on Twitter. It's like, "Hey guys, new feature. See you later." It's like, "What else is new?" Threw off my whole Friday, threw off our whole game plan on what we were going to do on this, but, you know, in typical fashion. But, I'm I'm amped about it cuz like you said, keeping me on platform there, you know, do have some thoughts and feelings on this. And it's good. It's great.

Play from 0:01:32 8:88:88 0:01:32 Don't get me wrong. I think, you know, generative fill from Photoshop still has a little bit of an edge on certain things, but this is definitely going to keep me there for a lot more. Super needed and it's odd enough that we maybe called out about seven of these things in one of our podcasts and like kind of all of them came to fruition. So, I don't know how crazy. Like, it didn't seem like this was a on the update list or on the office hours for a while and then out of nowhere it's like, "Bang, we're going to do this. Maybe it was an easier update than we thought, I don't know. But it's so much more dynamic, so much more than it was. Repaint and reframe have like totally been changed. So, this is cool and thank you, Midjourney, for doing that if you are listening.

Play from 0:02:11 8:88:88 0:02:11 Long overdue, and please, I hope this is like that last little evolution of combining all these things with reframe and custom zoom and vary region and inpainting and repaint and change AR and remix is gone, remix is back. Like let's let's settle down now. And now we've got this new UI for this tool called editor, which we're going to visually walk you through too and just kind of show you these changes. But to your point, yeah, total game changer. Total game changer because now keeps you in platform for a lot more instances. You have a lot more flexibility with making the selections that you want to make.

Play from 0:02:50 8:88:88 0:02:50 Lot more flexibility with your outcome in terms of the the aspect ratio itself. So, amazingly powerful tool to an an existing feature that we've just kind of been pining for on this podcast is like, "Yo, let's let's upgrade some of the things that we already have." And Rory, you're right. We like we're either like the luckiest guys in the world or somebody's listening. Somebody's listening to this.

Play from 0:03:17 8:88:88 0:03:17 Maybe someone's listening. That's a you know, it's our pipe dream over here, guys. We would really like them to be listening to us cuz we are nerds and we'd use this thing every day. But it's the one thing that I wish they would also now get rid of if you guys are listening, please God, just like the 10 images on Discord to get into the website. Like who cares at this point? Like let's go. 10 images? I had to I had to explain to someone the other day, and I just realized how complicated it was. They're like, "How do I actually get on the website? Cuz I'm trying to set this whole thing up." Like all right, well, you got to go to create a Discord account, go to Discord, do this, create your own server, then go in there, generate 10 images, then go back, and then did connect to Discord on, you know, on the website, then upload the way I'm like come on it's it's I was like I was writing an email. I'm like, you know what? Let me find a YouTube video that's doing this and just send it to you. It's so convoluted it doesn't make any sense for new users, right? For

Play from 0:04:04 8:88:88 0:04:04 us it's like riding a bike. Whatever we can tell someone how to do it, but someone who's never used Discord before It's like why why It's inevitable. It's inevitable that they're moving to the site. So, why wait so long to sunset I guess the the other component of, you know, the Discord? And I And again, I I like maybe it's not sunset. They're going to leave that up. Yeah. But But maybe they do need to like sunset it once once these things are over because I don't think you want to have necessarily two different environments. And on office hours and things like that, I just I think they're a little hard on the website. I know they're not they're not happy yet with how many people are using the website versus Discord. Like that split isn't ideal yet, but like to me the website is great. Especially for like this early of

Play from 0:04:49 8:88:88 0:04:49 a product this different. Like we're still in the first couple years of this. I mean that website's pretty damn good and it just got up it just got put up eight or nine months ago. So, I find it very usable. I find it very usable and super simple. Especially since we've seen it grow like every single step of the way. I think they're worried about people like losing people on Discord. Like at this point I think the people that are using Discord are using Discord. Like the community's going to stay there. They're not going to just migrate to the website. They like it there they're going to use it.

Play from 0:05:17 8:88:88 0:05:17 You know, the new users who you're trying to attract who haven't been the diehards who haven't jumped on, they're probably going to want to use the website because it's easier than using Discord. So, you know, again, I don't know what that split looks like of who's on Discord who's on website. You're going to have your diehards. You're not going to lose your data from, you know, the people on Discord. They're going to show up in the in the rooms. They're going to show up on the, you know, on the challenges, things like that. But that's just that's just to me I think maybe overanalyzing it cuz if you haven't moved off Discord by now and you've been with Midjourney for two or three years at this point. Like I don't think you're going anywhere. Like I still think you pop in for the community versus just like I I'm purely generating here.

Play from 0:05:56 8:88:88 0:05:56 If that makes any sense. you're I think you're onto something there because yeah, it's it just seems like if you're already comfortable with that tool and you've got maybe other instances where you use Discord on a daily basis, other communities, other tools, which there are, you know, a lot of those AI tools have Discord channels, too. You're probably there, right? But you still do have that obstacle that you're putting up for all these new people that you're trying to attract. And so, do they know Discord? You know, the new users still don't know Discord.

Play from 0:06:21 8:88:88 0:06:21 There's a reason why they probably haven't jumped on at this point. They're they're late adopters to game-changing tech, which means that they're the likelihood of them, you know, loving Discord or having familiarity with it only go down. I think that was probably a higher rate when it was earlier on because you're getting those early tech adopters who don't have problems learning new tools. Welcome it, you know, it's not as big of a hurdle to overcome. So, I think that's a really good point.

Play from 0:06:47 8:88:88 0:06:47 Yeah, like I said, for this one, you know, at this point I just don't think it's necessary anymore. I understand they want to grow their their, you know, numbers, whatever that might be. Or maybe they're not even concerned about it. But at at the same time, it's just the website is such a nice experience and this new tool, the editor feature, just becomes so much easier with fine-tuning. Now, before we get into it, definitely I have some I don't want to say criticisms, but a little bit of annoyances that that, you know, cuz that that's who I am. And uh you know, uh but it I definitely want to know what you're what you're thinking about it right off the bat. What are you liking the most so far about the new editor thing?

Play from 0:07:27 8:88:88 0:07:27 editor, uh for me, it's really you know, we talked a lot about the the ability to crop. That's never been easier to do. I love this ability that we're going to show you here where you can essentially take an image, drag it in either direction. You could scale the image all the way down like So, we're going to pop through all these different tabs and sliders and so forth, but that scale functionality where you could take that all the way down to like a 10%.

Play from 0:07:56 8:88:88 0:07:56 Yeah, the scale is outstanding. You know what I mean? Like I I'm sure and and just in the the few that I've run testing this, you're going to get some distortions similar to like probably what you might get with those pan functionalities or if you zoom too many times in a row, but this tackles almost everything that we talked about. Crop, the zoom out more than 2x, right? Because if you scale it all the way down to 10%, that's more than a than a 2x. We talked about the free object selection.

Play from 0:08:26 8:88:88 0:08:26 Like we were thinking like smart object, but this is just the brush, which I think is freaking awesome. And then, you know, like the ability to not necessarily be trapped in this compositional game where do I want to place this at the start or the center or the end? Because we always talk about how everything's symmetrical by default. There's never been an easier way to break free of that now. And so, I don't know if it's like one thing. It's just a combination of all those things for me, but I'll let you chime in on your end.

Play from 0:08:55 8:88:88 0:08:55 All right. Yeah, man. I mean like like this is really interesting, right? They found a smart way to do this fairly quickly to encapsulate a lot of those things without having to make like individual tool changes, which I thought was pretty, you know, from a conceptual standpoint, pretty impressive. Right? Cuz all the things you said were a lot We were asking for a lot of things within this, but you know, cuz I think one of the things you could do with Discord was basically that custom zoom feature, where you could, you know, you could do a lot of different, let's call it like remixing, changing the changing the zoom functionality within it, also changing the aspect ratio. You could do all that within the custom zoom feature in Discord, but then you couldn't really do it on the website. So, combining all that, I love it. One of the little annoyances I found is, you know, basically when I want to change smaller

Play from 0:09:41 8:88:88 0:09:41 details, they haven't been doing exactly what I want, right? So, let's just say I wanted to erase this smoke and I wanted to change it to red smoke. So, I get like two-tone, two-color smoke. When I've been doing that, it hasn't been doing that. But, if I'm doing bigger things like, let's, you know, add a dress to this person or add some clothes, that's been working. So, it's just one of those things that you know, again, you can't have everything. So, I will acquiesce cuz there's so many good things here. But, one thing to note here is that what I've also found is smaller details, like if I just wanted to change this, right? That has been not that accurate, either, if I'm going and changing this in the prompt and saying like, you know, red eyeball. That's not going and changing it that much. So, think there is some little things that have to get updated, but really I think

Play from 0:10:26 8:88:88 0:10:26 one of the nice things here now is that you can scale this brush down to this size. You don't have to use the square select anymore or the lasso select, which again, you have you run into that problem where it was essentially, you haven't selected enough pixels, like select more, which, you know, you have that hack for selecting the edges, which works, but this also just gives us a little bit of a, you know, frame in terms of what you can actually highlight. So, it's nice you can stay small, you can go big here, which is awesome. I'm working through this tool kind of left by right here. When we did erase, right? This is what erase does.

Play from 0:10:57 8:88:88 0:10:57 So, it's very similar to repaint. You can go and like select this if you want to, you know, to get rid of it. You can also restore, which is basically just erasing your eraser. So, erasing the eraser. You like that? I do like that. So, that's pretty simple. I mean, this is very easy to go and utilize now the undo, so you can undo what you've painted. You can just restart from the entire thing, which is great. Again, the brush size as we just went over, but this scale feature that Drew was mentioning is really just game-changer.

Play from 0:11:27 8:88:88 0:11:27 Now, you used to have to, if you wanted to be this small, it would have to be, you know, you zoom out seven times, eight times. Now, you can scale this within the frame, which is exceptional. Now, the other thing that's great is it's very much so, you know, dynamic. Now, you can move this here, you can move this to the left. You can actually scale the image within the frame, so you can cut out and crop some of the sides if you want to. If you wanted to, let's even say, you know, now over here you can switch this aspect ratio. Say it's 9 16. We wanted to scale this thing and let's move this over here. And we just wanted to have a little bit of this in the middle. Now, you can go and generate all around that.

Play from 0:12:07 8:88:88 0:12:07 So, it's way more dynamic, very much so like Photoshop generative fill. But, let's reset this entire thing so we can see this again. I mean, if you're a creator, this is where you can get lost in the sauce, dude. There's so many different ways to manipulate that. I didn't even just think about the way that you just slid off half the image outside of the frame, right? Like that. Okay, like Different mediums, right? And and again, going back to that point, too. This is now the ultimate symmetry killer right here. This is how you avoid it.

Play from 0:12:38 8:88:88 0:12:38 Perfect perfect segue because I think, you know, one of the things that, you know, if you're doing this stuff for organic social, you're doing this stuff for your website, whatever, it's never going to be like, oh, I generated in 3 2. Like, that's sometimes not what you need. You need some maybe you need a website banner. You know, now we can go to 3 1 and we can put this in here. But, most of the time you either have to worry about do you want to do do it at the end or in the center? Right? Like, that was the only options with this. So, you'd have to generate all this stuff to the side. But, now we can just kind of like put it where we want it. Let's say we want it a little bit more like space on the side here and then some space for text, we can generate around it, which I think is just awesome. And even with this, too, cuz now we can scale it. This is basically like having a little bit of Photoshop {slash} Canva baked in here with what

Play from 0:13:24 8:88:88 0:13:24 you're able to do. Again, even if you wanted to scale and crop, which I think is just outstanding that you can do that. Like even starting from the side over here, it's like a great way to just play around. So, this is where I found it the most interesting. Now, like I said, going to show you an example. If this generates great, if it doesn't, you know, we're we're running a lot of running a lot of power here on uh between recording and doing this. So, Don't crash. Don't crash it right now.

Play from 0:13:50 8:88:88 0:13:50 Don't crash. Don't crash. So, let's just say I wanted to, you know, add some finer details. Like most of the time, maybe I'm switching. Maybe I wanted to do this for something that's more social heavy. Let's say it's going to go in a reel or it's going to be a cover for something on mobile. Let's just make this 3:4. Now, we can generate some space up top and down the bottom. If I go and maybe drag it down a little bit and scale, maybe move it up here.

Play from 0:14:16 8:88:88 0:14:16 So, this is like super dynamic and I love this. But, let's just say we erase. Let's go. Make this a little bit bigger. Let's add a Let's try to make this like her wearing like a red dress. So, I'll go and I'll erase this bottom here. And then you can go into the edit prompt, and you can do spiritual hero healer wearing a red dress. This is what the edit that I made before. So, somehow it just showed up in here again. But, now when I hit generate, it's going to generate all around this plus add some details. So, when I hit submit, we'll be able to run through this stuff and hopefully this is what it would come out like.

Play from 0:14:50 8:88:88 0:14:50 I don't have to make you guys wait. So, it adds the red dress. It adds a little bit of the smoke down to the bottom here, which I find to be super helpful. Because you can also then go into editor and let's just say you needed it to be wider. You can pull this either, you know, in, out. We can make this, you know, 16:9. We can make this a square. We can do whatever the hell we want. We don't even know what that aspect ratio is there. But, you can hit submit on this, too, and then add some side details. Super helpful with this stuff. Again, love this functionality, and I think there's so much that you can do within it and even, you know, here this actually went and I always love some of these pans when they generate like this one. We got another random guy with, you know, smoke hair, which is great. This one added a actually a nice little touch here with

Play from 0:15:36 8:88:88 0:15:36 this bowl, but again, more dynamic cuz you can just keep going with this for days. Then you can make this 1 by 2, you can make this 2 by 1. Let's drag this out. Again, so you can build you can start and I think this is a good way to think about this, too. It's like when you have to build something, let's just say it's a website banner, right? And we're going super wide here. You know, if you want to start with the person, you get all the fine detail. Then you can add. Again, think about this in steps. Instead of trying to generate her in 2 1 to start with, I'm generating her in 3 2, so we get the close-up sharpness, and then we can just keep adding the background details cuz it'll just keep picking up on that. So, let's scale this and then we'll we'll pause on this stuff for a second once this is done. It's just like the way you're thinking about utilizing it and going with it cuz again, even if

Play from 0:16:21 8:88:88 0:16:21 we like this, we can go and edit the bowl. Maybe we want something else now. It's really like a different way of thinking about prompting. And I just love that this is now available cuz we just went from this close-up image of the face with the smoke to now having like this person standing in a cinematic style like stream in the background. This just makes it so much more dynamic. It was much needed and I don't have to go to Adobe Photoshop generative fill as much. A couple things that I'll just add on to what you just One, like the idea that you could create an image that's up close, right? That macro shot, that close-up, something that's relatively close to the screen, especially with a person. It now seems that that that's almost now even more important because this this kind of lends itself into

Play from 0:17:07 8:88:88 0:17:07 composition doesn't necessarily have to be prompted for anymore. With maybe with the exception of like when we've talked about stuff like leading lines and stuff. Like sure, if you want some like elements to drive focus or storytelling in that capacity, sure, but like in terms of medium full shot and that ending at somebody's waist, you know, or something like that, you don't need to you don't need to know those terms anymore. I mean, you just kind of need to know what you're looking for. So, that changes the process completely in terms of maybe you don't necessarily need to be as specific with prompting with composition. You probably are prompting maybe even a little bit less and and then tailoring a specific image. So, so to your point, having a good enough starting spot and then making all these tweaks now that

Play from 0:17:55 8:88:88 0:17:55 you can just build on and add versus like prompting a thousand times to find something that you like and then, you know, finally, you know, getting lucky to get it, right? So, I think like all these different controls can can really just kind of find their way right into the user's hands and make the the experience a little bit easier and more efficient to get there. I think so cuz now now I'm just thinking about it. I haven't even tested this yet cuz it's still so fresh, but you think about how we started with the woman just on a black background. Right? Normally, I'd have to prompt for the background for certain things, especially if I wanted it to be a very specific style.

Play from 0:18:30 8:88:88 0:18:30 But, if you start up that like that close, you can really start to expand it into the different environments. And then that's where once you get the body down, then you can expand the side, change the prompt, and then, you know, change the environment. So, you can switch up the environments if you want, which I think is going to lend itself to for, you know, maybe some different characters as well. Like, characters in different places doing different things. You can get really creative with it, which is what I love about it now is that it's not really, you know, it's not really limited by the functionality. You can kind of go and do what you want. So, it's cool that this is all, you know, the symmetry portion portion of it, bye-bye. I think there's probably some other things. I don't know if you've noticed this, but like I never really never really had luck with pan when you were going like pan down or pan up. To the side tended to work, but like up or down if I you

Play from 0:19:16 8:88:88 0:19:16 know, let's talk about like a medium like a medium shot starting at the waist. If I wanted to get like a full body from there and like panning down, it always seemed like the legs just like got longer and longer or like the bottom half didn't match the top half. So, very very interesting to see where that goes. I will also say I tried some really small scale stuff. So, putting that down, you know, that scale down to like 10% and generating around it, you get some funky stuff, so just be aware of that. Maybe do it a little bit more in steps.

Play from 0:19:44 8:88:88 0:19:44 That's a really great point. So, I I just want to show one other thing with that because I was curious about this scale 10%. So, I took this image. I scaled it all the way to 10%, which then gave me these, right? Mhm. And the first thing I notice is, okay, like pretty decent scene generation, but oh, we got a little bit of Hey. NSFW there, but like this is this is probably the part that they're going to have to clean up. Zooming in on her, she lost her detail.

Play from 0:20:18 8:88:88 0:20:18 Oh, all right. You see this? And also I'm seeing this little box here. I didn't notice that either, which is odd. I don't know if that's on the other images or not, but like I am kind of surprised that it didn't maintain that resolution backing all the way out. Having said that, 10% is a long Yeah. way to go. Right? Like that's that's the equivalent of doing what? Like an 8x zoom out. So, I mean, that's probably understandable, but just something to just something to look out for because I just noticed that. You brought that up that that scale thing, and I wanted to look back and just kind of zoom in just to see. And so, I'm super zoomed into this, but you know, that's something that you just have to watch out for. It's it's not perfect. So, that's where maybe something like one of

Play from 0:21:04 8:88:88 0:21:04 those tools we talked about last week like a Korea Magnifique can really kind of clean that part of it up. I'm also seeing some distortions here in like the hands and stuff, too. I mean, that doesn't even look like the same the same image. So, pretty impressed that it matched the environment that well for 10%. Mine was not that great. But, the other interesting thing to note about that, I think, is I haven't tested this. So, if anyone wants to go do this, I wonder if you can when you zoom out like that far, if you can also add, let's just say like a style reference code and then change the entire environment. So, I mean, you can get like super creative. If that if that works, that might be interesting cuz you cuz you can edit the prompt and if you add like a style code into it. So, I'm thinking like maybe it's like a realism

Play from 0:21:50 8:88:88 0:21:50 type portrait of like my face. And then if I scaled that down to 30%, then edited the prompt, put in a style code that's maybe like illustrated, if it blends like the illustration into the realism side of things. Cuz it could be you could create some really funky stuff that way. I'm guessing Like that's interesting. Because like I wonder if you could do that from like an image, too. Like you drop in an image to almost guide it with the the Gen Fill.

Play from 0:22:15 8:88:88 0:22:15 Now, these are good these are good things. We we haven't tested this. We don't we don't know if that's a possibility. But like that could yeah, if if either of those things are true, both of them are true, that's that's a whole different rabbit hole to go down. But look, as we sort of like wrap this up cuz I know we're going to dive into parameters, like I feel like this is a really good drop. This is a needed upgrade. Not necessarily from like the combining the tools part, didn't really care for that.

Play from 0:22:42 8:88:88 0:22:42 I was fine with the way that existed before. More from like the control, the dynamics, the break free of the symmetry hell, you know, that we were in. So. Also, just so you guys know, this is what when Drew and I talk to each other on the phone, this is exactly what it sounds like. Oh, maybe we should try a style reference code while we're zooming out. Yeah, this is the this is the this is the behind the scenes. You get on the phone with someone who knows what they're doing and likes to get weird and here we go. We end up with some crazy stuff, but totally totally agree with you. Love that I can just dynamically change the frame. That's probably the biggest thing for me. Specifically for things like presentations where I have like a great image, but I want to make it my cover image. Now I just go and drag it to the size of the presentation image. Done. Like awesome. I don't have to get into generative fill. Don't have to worry about the the loss of quality with generative fill cuz that's one of

Play from 0:23:29 8:88:88 0:23:29 the biggest things I think on Photoshop. Like if you're using a mid-journey image and then you take it into Photoshop and do the generative fill, often times it looks like it's generated in two different programs. One looks stellar and detailed and amazing and the other one looks kind of like eh you know, so even if that gives you know, the more of like a 90% head start. Thank you. mid-journey again. Look, like we we brought up Photoshop as one of the tools last week that you should be stacking with it. There's still the use case for that like you mentioned.

Play from 0:23:58 8:88:88 0:23:58 It's just now this is a another big piece that you can solve in-house. So, you know, like we we also brought up like color grading and all that stuff, right? Like how far that's come. You know, by the time that we got to V2 or 5.2 and then V6 and you know, how everything now kind of has a little bit of that post-production type of feel to it, like post-process type feel to it. So, I think you know, you're starting to see a lot of those needs to take it outside disappear, which is great. It's a great thing for all of us. So, yeah, let's get into the deep stuff, the parameter you know, beast of the puzzle, which I think probably doesn't get it talked about enough. I mean, we focus a lot just people in general the same sort of components, which I think are like image ref, style ref, character ref for good

0:24:26Aspect Ratio (--ar)

Play from 0:24:46 8:88:88 0:24:46 reason, too. Like and what we've what Rory and I have talked about for this particular episode is we're going to almost take those three and treat them as a separate entity for this. While they do belong in parameters, we want to focus on those more dialed-in kind of customizations that you can make at a high level, the the OG parameters, if you will. And so, there's a lot to talk about there because people tend to think everything's all about prompting, words, words, words. But, like to me, I always saw it as an A and B. Like, 1A, 1B. You have to have a good prompt, but like the parameters are so powerful that can take a bad prompt and do amazing things with it at the same time.

Play from 0:25:26 8:88:88 0:25:26 And so, let's get into that, man. Let's Let's start to kind of like dissect these and break them down one by one. Maybe we start with aspect ratio because probably for both of us, that's something we're thinking about every time we prompt. Every time. I think it's probably overlooked cuz people probably set it as their preset and just go with it. Either you do it in 1:1 or you do it in 16:9 or you do it in, you know, 3:2, whatever. It all comes down to To me, it's like, what are you trying to create? Cuz it Again, the aspect ratio makes a difference. So, human faces, things like that. If you want a close-up of a face and you do it in 2:1 aspect ratio, you're going to get the eyes and the nose. That's it. So, if you're thinking about wide, like a human face is not is not wide like that. It's long and thin naturally. So, it would be like a portrait style, which is you're going to

Play from 0:26:12 8:88:88 0:26:12 want to go with, you know, a lower width and a higher height. So, something like 3:4 or 9:16 or whatever, cuz that's going to give you more of the face. If you go with something that's 2:1, super wide, then you're just going to get the eyes. And that's just something that oftentimes used to be more, I think now with this feature that we just talked about, a little bit more important. Now, you can go and edit, but just know that when you do something like that, especially close-up of a face, like anytime you try to edit beyond like the eyes or, you know, pan down or pan up, you typically get some distortions.

Play from 0:26:45 8:88:88 0:26:45 Like, the forehead will grow like super high, and it's like, that doesn't look anatomically correct. Or the chin will be like super low. It's like that looks like, you know, Frankenstein. So, that's one thing to to definitely take note of. And then the other thing I would say, too, is depending on what you want in the background, think about that as well. So, if you have a person, an object, a thing that you're trying to focus on, if you go with four, you know, something that is again lower width by higher height, so a 34, anything along those lines, those aspect ratios, you're not going to get much background detail because there's only so much room for the subject. So, again, if you want to start, I typically like to use I don't know what you like to use a lot. I I like either This is weird, 54 or 32.

Play from 0:27:29 8:88:88 0:27:29 It's just a good starting point. It feels like it gets a lot of the image. It gets you, you know, your subject, gets you some background, gives you some things to work with. That's typically where I start. But, where do Where do you typically start? Like with your I like this. I like this discussion, man. This like yeah, our favorite aspect ratios. Okay, so number one for me, 45. It's not even a default in Midjourney, 56 and 34, so it skips 45. But, man, I love 45 for anything with a person, anything with a portrait. Yeah, anything photography that isn't like a street photography or documentary maybe even like cuz those I think lend themselves to more of like that landscape or 32 like you mentioned, but it depends on what you're prompting.

Play from 0:28:08 8:88:88 0:28:08 So, that's why I use a lot of 45. That said, I also I mean 11's great for certain things, right? So, like I use 11 for anything that is like a logo or an icon or maybe even an illustration or something that, you know, I'm just trying to create like this nice area for negative space to be included. And I do love 169. It's just not the best for for generating great detail with faces most of the time. So, usually I'll do a 45 and then convert it to a 169 or a 32. You know what I mean? So, like 90% of the time is 4 5 for me.

Play from 0:28:49 8:88:88 0:28:49 I have a tough time getting away from it. Interesting. You're 4 5, I'm 5 4. This makes This makes a lot more sense now. But there's The other thing to note on the 1 1, what you said about logos, which is great, just another thing that came to my brain. When you're trying to create symmetry with like a perfectly symmetric Maybe you want like a There you go. pattern or something like that. Symmetry is going to work best in 1 1 because it's just It's working within the bounds of like a computer, right? So it can do symmetry in 1 1. Let's call it some super minimalist photo of like three skin care bottles like laying on a flat flat lay, something like that. Where it was like product photography. It's going to be so much easier for it to do it symmetrically when it's working within a square because a square is inherently symmetrical. It's easier that way to think about it. Something super minimal and you want it to be placed like center

Play from 0:29:36 8:88:88 0:29:36 justified right in the middle. Like use a square, it's going to work much better and then just out paint. Those are two little hacks. The other one I would say too is if people are using this for design or you're using it for a specific thing like let's just say it's a LinkedIn banner, you know, 534 by 1533, whatever it might be. Um you know, or the opposite. You can just put that in. Like whatever the pixels, you can put that in specifically.

Play from 0:30:02 8:88:88 0:30:02 You Midjourney can handle all of it. It just need It'll just default to the lowest common denominator, right? So This This is why This is why I keep you around, brother. Like You know, like of course Of course it does, but like I just never even thought of like I I never that thought never crossed my mind. It's a good way to It's so easy now. Once they Once they got rid of the like, you know, you can't do these odd aspect ratios cuz you know, you'll try to do a banner like I think the LinkedIn banner is like the weirdest one of all time, especially the company one. The company page is like 1500 by 200 or something. Like nothing fits in that other than like a block of text, but it's It's It's al- It's almost as bad as the Twitter one. So, Yeah. okay. LinkedIn banner number one.

Play from 0:30:47 8:88:88 0:30:47 Yeah. Terrible like terrible dimensions. Twitter or X right there at number two. Pretty bad. They cuz why doesn't why doesn't lend itself to many things. If you want to show something visually, like super wide and thin, it just doesn't. I mean, you can do stuff, but like it's it's everything's either super small or super scrunched. Crowded. Yeah. You got it in there. And let's say it's like six the the aspect ratio is or the the size reference is 600 pixels by uh 200 pixels.

Play from 0:31:18 8:88:88 0:31:18 Midjourney will automatically default that down to 3 by 1. But it can also be six 615 by 217, and Midjourney will just do that. So, you know, it'll make it. It won't be that size specifically in pixels, but it'll make it that exact size. And then you can upscale it, do whatever you need to do with it. So. I I like that little hack, brother. That's a good one. Yeah. It's very helpful for many things. The many odd, you know, the banner ads and like that where like you're doing like a side runner for for Google, and it's like 60 wide by 450 long, and you know, like the nothing fits in this either.

Play from 0:31:59 8:88:88 0:31:59 But we'll we'll see what we can generate. That's a brilliant little tip. And then obviously like what we just talked about with the editor now, that gives you the capability to just kind of like create these really unique aspect ratio sizes. I don't know what else you got for aspect ratio. The final note for me is really just aspect ratio should be something you're thinking about really for virtually any image. And it goes back to the point that you brought up at the very beginning, which is aspect ratios can drive the outputs. And the reason for that is tied to the training data and how they went about putting this model together. And so, you know, when I say 4 5, it's just like more of that portrait style is what you know, all those photos of people from the, you know, mid-chest to their

Play from 0:32:45 8:88:88 0:32:45 head are trained on, right? Like most of those shots are like a 4:5, 3:4, 5:6. And so, you're going to get better results if you're prompting for a portrait if you use those dimensions. And so, you could start with the 4:5 or a 3:4 and then stretch it out to a 16:9 and all those things that we just talked about. But, point being, keep aspect ratio in mind every time. It's just It's just such a killer in what it can do. And if you don't know what the best aspect ratio is for the image that you're creating, run a permutation on it. You know, nest it in those nice little brackets. Try two, three, four different aspect ratios and you'll know exactly what we're talking about because you'll start to see like like 2:1 aspect ratio in that, garbage.

Play from 0:33:27 8:88:88 0:33:27 You know, like so So, like that's the That's the key takeaway. Do you have anything else you want to add on top of that? Shot type can also be affected. So, low angle often times is probably to me the biggest one cuz if you're doing low angle, you kind of want some more of the body versus like, you know, the face. So, like low angle a lot of the times for me on 4:5, I get it, but it it's really just like from neck up. Like if I'm prompting for medium shot, it doesn't really do it cuz it's it's such a tight space. It can't get all that in there from a low angle. So, something like a low angle, you might want to go wider. For high angle, same thing. You know, so it's very very contingent on understanding like again, having some sort of vision or just when you're editing and like I'm not getting what I want. Sometimes it's it's simple thing like aspect ratio that's not doing it

Play from 0:34:14 8:88:88 0:34:14 for you. So, that's a I'll wrap it up on that one. Perfect way to end it. Aspect ratio, super important, something we're thinking about every time. Outside of aspect ratio, what else are you thinking about every single time that you're putting in a prompt? Normally people would go stylized here. I'm going chaos because to me chaos is super important in what I'm doing. Sell me. By the way, I'm prompting. If I'm going, you know, that to me there's a lot of uses for it. I tend to use it, you know, just so everyone knows what chaos does.

0:34:21Chaos (--c)

Play from 0:34:42 8:88:88 0:34:42 It varies your image grid. So, if Midjourney spits out four images, basically the higher the chaos value, the more each individual image is changed or like further away from each other. So, you get like more variance. I guess that's why they call it variety on the website. But, it's kind of like a like a math equation in my head. If I have a super broad prompt, like a super general prompt, maybe it's like Lamborghini on the street. Midjourney's going to go and create interpret that a million different ways. Now, if you want to add some chaos to it, you're going to really expand as far as it can go with its imagination because it already is a broad prompt.

Play from 0:35:20 8:88:88 0:35:20 You didn't tell it much. So, now it can really go think about that and create, you know, four radically different images. But, if you have a really long prompt, so something that's really finally detailed, and a lot of times you'll see the longer your prompt gets, the closer the images get together. They look very similar. So, that's when I like to dial up the chaos, too, to like add some flare to it. So, not going too high. I probably stay between 0 and 20 most of the time.

Play from 0:35:46 8:88:88 0:35:46 But, longer more detailed prompt, I like to add a little bit of chaos and variance to it so it's not all the same image cuz I can also be the X factor in like uh it's not there and I don't know what's wrong. Sometimes a little bit of chaos can add some some pop to it. So, I use it in that sense. And then also, I'll use it for fun because you go up in value and you really stray from the prompt. So, also to know this, the further you go up on the chaos scale, probably it starts breaking around 20. Like you get between 20 and 100 and it is total it is totally random. And from there, you know, anything in the 60, 70, 80 range won't even look like your prompt most of time.

Play from 0:36:27 8:88:88 0:36:27 Sometimes you get hidden gems in there. I found some of my best images with like chaos 70, 80, but that's just something to note that it can be utilized in that fashion to get some variance. And also, it can really just take something and run absolutely wild with it. So, I use it a lot. Probably my default is like five, like chaos five. It's just like it gives me some basis, then I'll then I'll play with it from there. But, I don't know if I'm weird like that. Maybe I am.

Play from 0:36:55 8:88:88 0:36:55 Well, you Well, you are weird. You are weird. But, you know what? I agree with all that. I don't really have too much to add. The thing to know about chaos here is you touched on it. It's a very sensitive parameter, so you don't need to go high to get variety. I think five is a great one. Eight is another one. 10. You know, like just you could just throw something in there. But, yeah, to your point, like that 70 80 thing, I I honestly hadn't run a lot of that. And then I remember you and I talked about this like, I don't know, a couple months back. And I I tried to You know, I started trying it a little bit more often, just like throwing in the occasional like chaos 70 or 80. You know, especially like when you kind of get to Well, I you actually have use cases for both longer prompts and shorter prompts in that way. But, like if it's really short, you're going to get something totally different, right?

Play from 0:37:42 8:88:88 0:37:42 So, Yeah. just like in the spirit of continuing to make this like really, really visual, I'll just show you this because I was doing a training for somebody last night. And we were just talking about something like this, right? Where it's like portrait photography, we've got the camera angle. You know, like this isn't a long prompt, but it is very specific. 30-year-old blonde woman with a bob haircut. She's wearing a simple sweater, symmetrical, looking at the camera, studio lighting, background. Right? Well, I guess for me, like it it's just that long in length, but there are a lot of like very specifics here. And so, without chaos, these are so damn close.

Play from 0:38:22 8:88:88 0:38:22 Similar. Very. I mean, there is a strategy for that and they're they're different enough to say, "Hey, I really like you know, I really like like sort of like the the crushing of the the highlights on this versus something like that, you know, or whatever. But, here's where it got interesting is, you know, when I do these trainings I like to kind of prove out the threshold of any of these parameters to show, "Hey, this is what like a five looks like versus like uh a 90." Right? And so, here's here's that with a 90 and you see how different these images like but then we got this and I was like, "Yo, that looks like like It's cool.

Play from 0:39:00 8:88:88 0:39:00 like a demon sea, you know what I mean? Like this is like this is so cool and so like we just you could just see the variety here and then we took that and kind of used that as some other examples with with some of the other things that we were doing. But, that's where you can really just kind of break your prompt and almost like explore the latent space of what you could get back within Midjourney. So, I think it's it's definitely worth playing with regardless. The low values are are fantastic, too.

Play from 0:39:27 8:88:88 0:39:27 100%. I love that I love that, you know, that one random one that you just like, "Oh, damn." You know, it's just like that. So, so that's what happened with that one. Yeah. It's It's a cool one. Like I I barely ever touch Chaos 100. I don't even know if that makes sense to utilize it. That's just if you want to if you want to live dangerously. That's That's a That's a fast hour for you probably on most on most generations. But, yeah, Chaos to me, once you get into being a little bit more controlled with what you're looking for and your vision, you know, you can use it strategically. I wouldn't say everyone has to use it on every prompt. It's just like when you know you want a little variance, a little pop, a little bit of intrigue, maybe, a little bit of a grab grab bag mystery bag type feel. That's what you're going to get with it, which

Play from 0:40:12 8:88:88 0:40:12 I I tend to love, especially when I'm not doing something very specific or when I'm trying to be creative cuz that'll take you in a different direction. You know, you can take that image and maybe reverse engineer it. Build it out again in a totally different way. Now you can use it as a style reference. There's There's a lot of things you can do when you generate cool random stuff. But that's my soliloquy. I was hoping you messed up soliloquy so we could just kind of leave that in there. It was smooth.

Play from 0:40:38 8:88:88 0:40:38 God damn. It was It was almost like too smooth. Like I feel like you might use that a lot now because it. Yeah. I I I think you do. But All right, bro. Yeah, so I mean like let's start Let's wrap up chaos. Let's move on to What's What's keep it with like the things that we're using the most often. Is there Is there another one that comes to mind? I mean we mentioned stylized. Do we want to go there yet? What else is on your mind?

0:40:50Weird (--w)

Play from 0:41:03 8:88:88 0:41:03 I can run through weird cuz I do weird a lot. Um I I had a feeling you'd bring that up. Let's do it. Yeah. Weird There's a default aesthetic for Midjourney. I think everyone knows it. If you've prompted enough on here, you know there's a very default look to it. It has this very HDR toned look unless you're prompting specifically out of it. It's very colorful. It's very rich. It's very saturated. Very textured. Weird's again that that variance. I I I don't know why they go up to 3,000 on the scale. It's between weird zero and weird 3,000.

Play from 0:41:31 8:88:88 0:41:31 Default is zero. So It's It is a little odd, right? Like Yeah. It's just make it make it 300. Like, you know, there's so many different possibilities with it. But I love it in low values, too. Again, it used to give me a very natural feel to photos which I think is interesting cuz it takes some of the polish away. You can get a little bit more realistic. If you don't know what I'm talking about, go type in, you know, Facebook photo from 2012 and, you know, shot on iPhone and then hit it with like weird 15 and you'll get something that looks like out of your yearbook. It It's very good for stuff like that. Kind of looks like some of the images you see coming out of Flux now.

Play from 0:42:12 8:88:88 0:42:12 Uh but you know, going really high on the scale, there's diminishing returns. I find going beyond about 1,500 to 2,000, you don't really get much different. I did one with a did one with a chameleon. Let me actually see if I can pull this up real quick. This actually something I ran a while back. This might even be 5.2. But just so you guys can see what I'm talking about, right? Like I did this prompt, basically this is a macro photography of a green thorny-tailed iguana on a rock, National Geographic style photography, right? You run this normally, you're just going to get really good mid-journey looking stuff. Like this is all very mid-journey.

Play from 0:42:51 8:88:88 0:42:51 Color saturation, super deep super deep textures and detail. It looks very much like something mid-journey would produce. Bro, you got to you got to rerun this on 6.1 now. I know. Now it's a this is a good This is what I mean, there's so much you forget on this stuff. But when we go and we start dialing up the the weirdness, right? Like Let's go to Yeah, I mean, like that. Weird even 100. Like that to me, I love that color scheme. It might not be perfect, but I can edit it, right? Like some of these stuff that's a frog. It gets weird. But you start getting here, like this looks like a real photo. You know, like this doesn't look like something mid-journey would produce. The colors are desaturated. It's just like very interesting.

Play from 0:43:31 8:88:88 0:43:31 You know, we get to something like this now looks like a dinosaur. But as you go up, you'll start to get more dynamic. Like here, like again, I wouldn't even know how to prompt that. Dude, that's cool. I can post-process that. You know, start getting some funky stuff with them on like on a wall. You start getting some really really unique thing. I don't know if you know what the hell that is. But you know, we're up into the weird, you know, 1,100s. This stuff is just so hard to get if you're just looking for something new.

Play from 0:44:00 8:88:88 0:44:00 Like and I love this one. This is weird 2,800. Like this one to me looked like something I would I would take on like my phone. You know, like a maybe if I had just a quick shot, that's what it would look like. You know, you're just looking for things that are different. And that's sometimes if you don't know how to prompt certain things, you wouldn't know how to actually type this out like a split kind of level shot that you're looking for with an iguana on top of a white, you know, of a white cinder block with a navy black background. This is why I like weird, because you just get stuff that you can't prompt for. When you prompt Midjourney like this, this is what you get. You know, you normally that's what you're going to get. It's going to be as highly stylized as possible. But when you go and add some weirdness to it, you might get something that's very different and looks like this.

Play from 0:44:42 8:88:88 0:44:42 How often would you say you use that? I probably like 30% of the time. A lot of times in lower values. When I'm doing stuff that's more art focused, too, that's more like fine art, I'll definitely throw it in there. This is where I find it to be most intriguing is in like when you're doing more like art focused stuff. Cuz I'll throw I'll throw weird into stuff like this so you can get some really, you know, interesting types of imagery. Now, I'm throwing a bunch of style references in here and this is a low level of weird, but if I go back and find, you know, something like this where like I'm doing like some trippy mushrooms or whatnot. Uh I'll throw I'll throw weird into things like this.

Play from 0:45:22 8:88:88 0:45:22 Stuff where I'm trying to blend like art and reality most of the time. Like that's where this can be really, really fun. It just it can provide some pop. If you want to break the monotony, you want to break the default of Midjourney, introduce a little bit of weird. Again, I would say very similar to chaos in terms of the sensitivity. You don't need to go up very high and I would in fact say that in a lot of cases you're not going to want to go above 100. You know what I mean? Like but it's worth exploring. That's another great like parameter type of test or permutation. I'm sorry. Type of test where you just did. Like run run something like a three, something like a, you know, 50, run something like a 100, maybe something all the way up just so you can kind of see what that is yourself. So the only thing I'm going to

Play from 0:46:08 8:88:88 0:46:08 add to that is just reiterating the sensitivity there the uniqueness to break the default and then just maybe like a general overview point. I think the thresholds and the values especially if we add any more parameters at this point maybe we need to bring this to a universal scale of 0 to 100. I dig it we're there you know because I I just think like unless you're a power user like us or you've just been in here for a long long time.

Play from 0:46:36 8:88:88 0:46:36 It it's tough unless you put in the reps to just like know all these different values and how they affect each different parameter and have you used them enough to really know how to master them. So if they were to not dumb it down but just like create some universal consistency with the parameters 0 to 100 for example we'd be in a lot better place. Yeah weird for me can go down to 100 you know chaos can stay stylized whatever. I think I I kind of like the the scalability in stylized which we're probably going to talk about next but I don't see a need for 3,000 different options with weird but it's cool. I mean it's probably great for like you know just developing the Midjourney brain because people just go a lot of weird places with it.

Play from 0:47:18 8:88:88 0:47:18 Yeah let's let's tackle stylized that's just sort of another one for me besides aspect ratio every time no matter what I feel like I'm doing myself a disservice if I don't run permutations on stylized because I'm just like what if I miss out on something great. You know that's where like stylized can have a similar effect to chaos on some level too because it's like the higher the value of stylized the more it kind of veers away from the prompt too you know and so there are you know some similarities in that aspect but back to the function of stylized. We talked about this in a previous episode but I think it's worth reiterating because it gets confusing to a lot of people, but stylize is essentially how much of mid journeys

0:47:19Stylize (--s)

Play from 0:48:05 8:88:88 0:48:05 aesthetic you want to pump into your image. Uh another component of that is the higher the value, the more it veers away from the prompt, the text prompt. So, you know, you do need to strike a balance. That's why I always try to to run the permutations. It's just like my own uh preference there, but like stylize to me, you and I tend to stay somewhere between like 50 to 250, I think, 90% of the time, I would guess, but yeah, you know, that's because a lot of what we do is, for the most part, photorealism. If you're getting into stuff like logos, illustrations, like that's where it's like that stylize 05, less than 50 can be pretty damn awesome. Like maybe the higher stylize values, you just kind of

Play from 0:48:50 8:88:88 0:48:50 see and test how far away it veers to add in some of that chaos. And it's not like every time you run a high value, either, it gets to be too much detail. I don't find that to always be the case, but you run that risk, too. So, like it's it's just definitely worth running that value threshold from top to bottom in some capacity and run multiple stops when you generate an image. 100%. To the point, illustration, also, like you said, going lower levels, that between like zero and 50, you know, keeps things flat. Once you start getting like if you try to do an illustrated style, you dial it up to like 700, 800, it starts to get 3D, and if that's not what you're looking for, just know that's probably a thing. But the one thing I find interesting with stylize is, again,

Play from 0:49:37 8:88:88 0:49:37 there's not only do you stray further from the prompt the higher you go up, but this is like with people and portraits, I don't know if you get this, either. Like the higher you go, the closer someone gets to the frame. So, it's like it's like if you did stylize like 50, it's like people are all the way back here. When you do stylize a thousand, they're like right up in the frame. It's It's something I've noticed. I don't know. I think it's probably just focusing on the details way more, and that's why it's bringing it forward. But, it also does have this like very polished AI bot look to it. I've changed my default to 200. I just like 200 to start and see where things go. You know, 100's great. Typically, I'm using at least 115 to start. I'm like, you know, I don't I don't want to be default.

Play from 0:50:17 8:88:88 0:50:17 I don't I don't want to be default. it up. I feel like you bumped up a little bit, brother. I feel like I feel like you're at like that 90 100 125 sweet spot, which I think is awesome, too. Like, I My default's still 100, but I don't really run 100. Like, I probably run two more than one. So, I probably should change it. But, I think like for me, going for photorealism, I'm almost always running permutation something along the lines of like a 125 150 somewhere around a two maybe something like a 250, and then I'll always run something way high just to just to see. Just to see what happens.

Play from 0:50:54 8:88:88 0:50:54 What happens, you know? Like, more or less just like to scratch that itch. But, yeah. I mean, that's that's cool, though. I mean, we we we tend to stick in that same range. For you, whoever's listening to this, maybe totally different, right? Just Again, we do a lot of photorealistic stuff. Those numbers between 100 to Let's just call it 250 300 work uh amazing for photorealism. Totally. I think sometimes, you know, using the personalize, which we'll probably get into also here, I tend to go up on that for some reason.

0:51:15Personalize (--p)

Play from 0:51:23 8:88:88 0:51:23 Maybe I just I've because of like the repetition of like the higher it gets, the more personalized it gets to me, which is why I like it more. That's also a great point worth like we should actually elaborate on that. I think I'm glad you brought that up because when personal personalization There we go. What's up with my zations? Um But, you know what I mean? Like, that added a whole 'nother level or layer of how you think about stylized because then it's like oh well it's not necessarily stylized on Midjourney's aesthetic now it's stylized as a part of my personalization so you know like does that sweet spot change so I think that's a great point dive into that a little bit please cuz I'm glad you brought that up.

Play from 0:52:06 8:88:88 0:52:06 Yeah so it's pretty interesting you know I still haven't cracked the nut on exactly how it works but when you stylized in combination with your personalization code so your dash dash p you know stylized it basically just tailors it more to your style or exaggerates your style more instead of like a general stylized setting. Your style is like mine mine is like high contrast like super super light elements super dark elements it just gets more expressed as you go up in scale.

Play from 0:52:36 8:88:88 0:52:36 So I've tend to be utilizing more in the 200 range to 300 range cuz it just feels like oh it's like exactly what I like let me push it further. I I probably stop around 650 is where I hit my tipping point as to where things get a little bit too much but it's interesting cuz I think the other thing to to think about where does chaos play in with stylized with your personalization cuz the more personalized you get the more it's going to look like your style but you also want some variants so then you can utilize chaos. It's like this sliding scale of what you're looking for that's where you know using something like chaos weird can just add the variants to it that you're looking for it to break maybe even if you're not totally happy with your personalized code maybe like it half the time not the other half. The chaos combination there

Play from 0:53:21 8:88:88 0:53:21 can help stray a little bit and get you more maybe back into the Midjourney aesthetic cuz I know some people have told me they don't really like their personalized code which mine just happens to be so I don't know how it's so tailored to me at this point but I don't even have to you know too. Yeah yeah. I was thinking back to when personalization dropped. Remember when we did I don't know if it was a podcast. I think this was a LinkedIn live that we did. It was some sort of group event. Yeah, I'm pretty sure, but we were talking about the strategy now with personalization. And right when it dropped, you and I were like, "Yo, run that all the way up stylized now because this has changed the game." You almost want to see what your personalization looks like. So, how do you do that at the max level? You need to run stylized

Play from 0:54:06 8:88:88 0:54:06 up, you know. Now, again, diminishing returns because then how much of that is it going to be away from the prompt? So, Yeah. I say all that to say if you are newer into using personalization and understanding what that does for you and how you're going to leverage it, start by running values all over the place with stylized and maybe even more so higher up just to get some initial indication of what it's going to give you back. That's both what we did at first. It was run those high, just trying to get some nuggets or breadcrumbs of oh, okay, like this is what it thinks my personalization looks like running that value high.

Play from 0:54:42 8:88:88 0:54:42 Totally. If you're going to talk about what probably most people are using right now, it's going to be your personalization code and stylized together. That's really if you're going to put any parameters in there, obviously aspect ratio, but also those two just seem to be from what I'm seeing out there, very similar, you know, we're going to get into style or S ref and C ref in a later episode, same thing with image weight. But, those are it's important. I think it's just not understanding how to leverage it cuz I feel like most of the time if you can't get what you want, tinker too much and you know, the parameters can sometimes be the benefit, but also the problem.

Play from 0:55:16 8:88:88 0:55:16 Sometimes you're like, "Why won't, you know, I prompted him with a girl, you know, raising a fist in the air and you have really high stylized cuz you like the way it looks, but her fist isn't in the air because it's not listening to you, you know, so." That's that can be the issue sometimes and then sometimes you need to generate at a lower level, you know, at a lower at a lower stylized to actually get her fist in the air and then you can probably take that and either remix it or you know, take it and use it as an image prompt or style reference, whatever that might be. We'll get into all this stuff at at a later episode, but you know, that's that's kind of where you got to finagle the system a little bit.

Play from 0:55:50 8:88:88 0:55:50 Bro, I I I love that you brought that up because, you know, it's easy to to go back to what you typed in the text prompt and think, "Oh, it's not listening to me." What are your parameters set at? You maybe need to look at that, too, right? And especially if you you go back, you try to change the text prompt, but it still doesn't work. Yeah. That you need to audit, you know, like you need to look at okay, am I like am I am I leveraging any parameter that could conflict with this? Stylize value, chaos, weird, aspect ratio, right? Like we talked about that. Like that's what you need to do to troubleshoot anything that you can't achieve that you're trying to basically get Midjourney to output for you. So, something to keep in mind, it's not all just based on the text prompt itself.

Play from 0:56:32 8:88:88 0:56:32 It's just not. Yeah, quick clarification on what I was just saying. For this is probably for anyone who's like a real diehard. But if you are trying to do super complex prompts with complex motion or complex, let's just say details, and you want to go the high stylize route, and you like the way that image looks, use that as an Sref. So, go all the way up on the stylize, 800, 900, 1,000, and then, you know, drop the stylize back down to get the action or the control that you need on the image, and then just use the higher stylize value as a as a Sref, cuz that'll get you there. And it'll kind of keep the pose, and it'll kind of keep a lot of the details that you're looking for without having to really dial up the stylize. This is all from problems of trying to get things done and can't figure it out, and then it's like, well, now we have Sref. That can be Love that. Love that one, dude. Hadn't

Play from 0:57:18 8:88:88 0:57:18 Hadn't even thought about that. I mean, you know what I mean? Like that's that's what I think it's it's I love Midjourney because of that. Yeah. It just seems like there's always these little things that are yet to be discovered in different ways to manipulate the model. While it's great to have something that's super transparent clear, it makes it a little bit of a gamified thing. I I love I love that part of it to me. 100%. You find these little unlocks, you know, where it's I think I did something where it was like I did this low stylized value. The The image was jacked, right? Like guy had like four arms, you know, but it was like this really cool style of what I wanted. But I also wanted like a little bit more firepower in terms of the control, so it's just okay, let's do the

Play from 0:58:04 8:88:88 0:58:04 same prompt, but let's use that image from stylized zero as the style reference. And then bang, like it came out perfect. I'm like, uh-huh. Didn't even It was by mistake, but I couldn't get stylized zero to work for it. It just wouldn't It just wouldn't listen to the prompt. Remember what we were talking about last episode? We said, you have to almost look for problems to know that how to solve it, right? So, I think that's like a perfect example. You could have just gave up. Your curiosity got the best of you. You figured something out. I just like No one I've never heard that before. So, like these are the things that's just like you don't get those opportunities to know the intricacies unless you experience the problems and are willing to work through it. So, love that. Are you good with moving on to the next one? Okay.

0:58:46Quality (--q)

Play from 0:58:47 8:88:88 0:58:47 Why don't we touch on quality, man? Because this is sort of like something that's become, I think, relevant again with 6.1 and talking about things like the update which introduced a lot better texture, micro contrast, the details, the fine details, even. Like, you know, for a long period of time, quality was a parameter that existed and was used a lot when I think I was first introduced to Midjourney. I remember Q coming up a lot. Something you had to almost think about as one of those things every time you put in a prompt, disappeared. Like I I Correct me if I'm wrong, but like Overnight.

Play from 0:59:24 8:88:88 0:59:24 It still worked. It still worked. Like you could still use it, but like no one was using it. Or maybe even the effectiveness wasn't there. But they made it a point to like put that out there that Q2 now matters a lot. It will take a little bit longer to generate the image, may veer off a little bit, but you're going to get you're going to be rewarded with those fine details, micro contrast, all those little things that just kind of take it to the next level.

Play from 0:59:50 8:88:88 0:59:50 So the Q threshold, correct me if I'm wrong, Rory, but I is two the highest you can go now? Two is the highest. Yes. And that that sits between, you know, zero and two in terms of the value. So the parameter for that {dash} {dash} Q, what what have you noticed maybe with now that 6.1 is launched outside of maybe what I mentioned or like how do you think about introducing Q into your parameter strategy now that it's more relevant again?

Play from 1:00:19 8:88:88 1:00:19 Like used to be way bigger of an aspect. I remember being like you dial it up to quality four back in I think that was like Midjourney either four or five, but then it like disappeared overnight. And now it just plays such a huge role, I think. I'm using it on just about everything. Tend to find that it works better with certain images, right? Like when I my style cuz it's very harsh, like the lighting's always harsh. So I like this harsh look with the very direct light, like very harsh skin tone.

Play from 1:00:46 8:88:88 1:00:46 I'm prompting a lot for texture. So sometimes it overdoes it and I'll get skin texture that looks like you dialed up the sharpness way too high. So I have to take it out every once in a while, but I do love it for textures. If you're doing macros, close-ups, things like that, like it works very well. I think it's mostly cuz of the personal personalization. Now I now I have that bug. It's It's um Sorry, brother. I know. I know now it's in my head. That in tandem with personalized can sometimes just be a little bit tough, but I really like it because, you know, the one thing that it switched overnight to also was you could do like quality {point} seven. And that used to work.

Play from 1:01:22 8:88:88 1:01:22 Now I think it's just and I could be wrong about this. I'd have to double-check cuz I haven't done it in a while. I think it's just like .25, .5, .75, 1, and then 2. Like it jumps to 2. So, you can drop it because it's good for some things. What quality is the time it spends rendering the image. So, sometimes minimalism can be better with a lower quality. Sometimes with illustration also a lower quality. Um if you want more fine detail, it just goes up in that sense. So, you're looking at 1 or 2 at a minimum.

Play from 1:01:52 8:88:88 1:01:52 Yeah, like so run it at 2. Like especially if you're going for some of those shots that you mentioned, macro, texture, fine detail, things that really matter to tell a story. And you nailed it. Less GPU time to generate the image, which then as a result, less detail to it. So, actually I I've never really used it as a way to test and experiment. But we can make that argument right here. Like that could be another way to throttle and play, you know, devil's advocate with something like stylize.

Play from 1:02:25 8:88:88 1:02:25 Because like what if you run a high stylize number, but a low quality? What does that do? You know, like Yeah. I I don't know. Like I actually haven't done that before, you know, so that I think is probably worth experimenting with just as another way to get maybe more of that flat component or more of these like micro detail components. But yeah, super relevant again, so I'm glad we're we're talking about it. Yeah, like I found it I used to run this um used to run this sort of experiment with like a tattoo design. And most of the time quality like I think it was .75 worked really well because if you did quality one, like the tattoo was just so unrealistic. Like it was just like you could never have it shaded like that or never have those lines would not make sense. But like the lower you got, the more it looked like a tattoo where it was like flat and defined and the

Play from 1:03:11 8:88:88 1:03:11 shading was correct. And that was one thing I mostly saw it with with art-focused prompts being more useful like if you want to trim down on like the super fine details, give it a little bit less firepower. Same thing with stylized, you know, you're working with that in a style a lower stylized could be really nice. It's like knowing where these little levers go up and down and like how you can manipulate them to find the right balance cuz often times it's just, you know, sometimes that's what your prompt needs. Sometimes your prompt is tattoo of a dragon, you know, unless you can actually fill in all the specific details that you want, you know, sometimes it's that's as much as Midjourney can do. So, the parameters are what's going to balance it out and get you what you need to get to where you need to go.

Play from 1:03:55 8:88:88 1:03:55 Love that. Yeah. Love that. Yeah, I don't think there's like a ton else to to really like dive into there, but like my takeaway I'm going to start I want to experiment with that lower quality high stylized low stylized high quality. Yeah. just like what what does that variance look like and how does that affect the output? So, yeah, super interesting that like what do you think about moving on to what about repeat? Just like a small one, but I think also like something that could be super helpful. So, {dash} {dash} R, I know you use this.

1:04:15Repeat (--r)

Play from 1:04:25 8:88:88 1:04:25 Before I turn it over to you, just a couple things to keep in mind. You can't run repeat in relax mode. You need to be in faster turbo mode to run it. And then if I remember correctly with repeat it's different values based on the subscription that you have. Yeah. So, for example, if you're if you're a basic user, I want to say it's somewhere between two and four uh ish like repeat and then I think that jumps up to 10 for standard and then I think after that it goes up to 40 for pro and mega.

Play from 1:04:56 8:88:88 1:04:56 Yeah. Yeah. Yeah. Yeah. Right now, repeat use it all the time cuz a lot of times I just like three different I'll just rather run three different generations and hit rerun. Oh okay. I was going to say we we actually haven't told anybody what it does just in case you haven't used it, but it's essentially just running the same prompt over again. So, now that we've said that, go for it, Rory. Yeah. So, you know, you type in {dash} {dash} r and then two, it'll generate the same prompt twice automatically. If you do a {dash} {dash} r 10, you'll get 10 generations. So, sometimes easier when you're in like a flow of something or you know you have a good variance and you want to just like keep running it.

Play from 1:05:33 8:88:88 1:05:33 This is where I find it to be the most helpful is running style random. If you do style random, if everyone's familiar or anyone's familiar with the style reference codes that you can run, you know, you have to type in style random to generate a code. But, if you do {dash} you know, style style random or sref random, and then {dash} {dash} 10, you'll get 10 style codes. Now, you can do that if you have run it 40 times if you have the pro subscription, you get 40 style codes.

Play from 1:06:01 8:88:88 1:06:01 So, sometimes it's just easier that way to like explore something like that. That's just like a little hack, but I use it all the time. This is like one of my favorites. Yeah. One of the one of the things that bugs me with with repeat, but I I understand why why it has to happen, is just another thing to keep in mind, you cannot run repeat with permutations. Yeah. So, you know, when we use permutations a lot, right, with with stylized different values, you know, you can't use any interchangeable parts or dynamic parts of your prompt in addition to repeat.

Play from 1:06:31 8:88:88 1:06:31 So, it's just got to be whatever the prompt is, all your specific values, one value only, and then repeat, right? So, for example, I can't have a permutation of stylized 125, 250, and 500 and run repeat three. I can't do it. I have to run stylized 125 repeat three, right? So, just something to keep in mind, but um yeah, repeat repeat's great. I think that's a great point with the style codes. That's one of the the ways that I used it the most was just like unlocking things at scale. If you really want to just go down a rabbit hole with a particular subject or style, too, like that could be great to to do more of like a deep dive exploration into something specific, You know, just to kind of see what the latent space is to like what are all the kind of results I could get. You know, if I were to run this multiple times.

Play from 1:07:15 8:88:88 1:07:15 100%. I mean, it's just simple for prompting for speed. There you go. That's one of Yep. Big fan of repeat, always have been. All right, where you Where do you want to go next? Tile. Tile. Let's do tile. All right, tell me why you like tile and tell people what tile is for those of you who don't know. Forgot for mobile. There's a the really quick If you've ever designed patterns, they suck. So, like this is to me this was like the most ridiculous feature. You just type in {dash} {dash} tile, you know, after the end of your at the end of your prompt, and it'll generate an image that's lock and key. So, essentially you generate one image, then you can place it on top, to the left, to the right, on the bottom, and it'll seamlessly integrate. So, if you have a

1:07:22Tile (--tile)

Play from 1:08:01 8:88:88 1:08:01 pattern, you can make the pattern will expand into infinity. If you've ever tried to make a pattern on Illustrator or something like that, it takes forever. There's so much that goes into it. This doing that automatically is just insane to me. So, if you want to like a you know, by like a red red chili pepper pattern on a yellow background, you just prompt for that {dash} {dash} tile, it'll give you that exact one. You export it, and then you can basically stack the boxes and they'll all be one seamless pattern. Anytime you're trying to do design, Packaging.

Play from 1:08:32 8:88:88 1:08:32 packaging, things like that. If you want to go, you know, going back to one of our other episodes, side hustles, you want to go make fabrics or something like that. Wall wallpaper, some Yeah, Anything. Get your Get your home and garden on. Like you could I don't I don't really use it that much, but I do think it's it's a super awesome parameter, having said that. And I think that's just a matter of what I prompt. But like for sure, and then you throw that into the idea of what we're talking about with something like editor now where you could scale that and position it and just like create a a totally different resolution and keep expanding that. Now it's easier than ever.

Play from 1:09:09 8:88:88 1:09:09 It also ends up with some really cool stuff. Every once in a while I'll prompt something really random and throw tile in there. And you end up like I did leopard print like leopard print whatever. I don't even remember what I was doing, but it ended up with like a bunch of like cheetah bodies like stacked on top of each other with just like the cheetah's head like with the sharpness of the eyes sticking out right in the middle. This is pretty damn cool. Like way way to think about that if you're not going for just patterns if you want to try to really be creative and just like you know there's I know some people just in here to push some damn buttons and get some really awesome stuff. That can be like a little bit of an unlock sometimes for some really crazy looking material.

Play from 1:09:49 8:88:88 1:09:49 Just a little tip there. Love that one. Anything else you want to touch on with tile? I mean like I think the use cases are the one thing that we could spend a lot more time on just thinking through those. So if you haven't used tile before or you could explore what that repeating pattern or seamless pattern can do for you, but it's like it's super cool. It's a super cool parameter that I think deserves its its mention. So we got a couple more. So let's dive through these. We've got the no parameter which I don't think you know it is super complicated in any way, but I think there are a few things to keep in mind. If you use dash dash no in your parameter section, essentially what you're telling Midjourney is don't include this in the results.

1:10:09No (--no)

Play from 1:10:28 8:88:88 1:10:28 So if it's a picture of me and I'm playing basketball and I put no headband in the parameter, you know there shouldn't be a headband. The reason for this is like if you prompt no headband in the prompt itself, it's not reading that is no headband in the prompt, you're likely to actually get a headband. Whereas the parameter controls that a little bit. It's not perfect like anybody that's used it could tell you that. There are times where where a little bit inconsistent, but I'll give you like a really relevant example. Um you know, I've got a client that I do work with that is in the metal recycling space.

Play from 1:11:06 8:88:88 1:11:06 And with a lot of scrap yards when you're trying to create images, it really makes those as dirty as possible. And you know, all the metal as rusted, oxidized as possible. You know, like even if you are creating uh an image of a person next you know, standing next to a pile of scrap. It starts to put all this dirt on them and mud. You know, it makes sense, right? From like a training data model perspective, but that's a perfect example of why that came in is like I started to notice that and then all of a sudden now, you know, you can use like parameters like no dirt, mud, filth, oxidation, rust. You know, like so I'll I'll incorporate those things as well as indicate like the type of metal I want being like

Play from 1:11:52 8:88:88 1:11:52 lustrous is a good one, meaning like shiny. So like you do something like that and you have sort of this counter balance with the no rust or oxidation and you get what you're looking for. So it's got special use cases. That's the one that that hits home for me, but what about you with with the no parameter? Tattoos. I I know Midjourney loves their tattoos. So I've I always try to negative prompt, especially when I'm not looking for it. No tattoos.

Play from 1:12:19 8:88:88 1:12:19 Hey Hey, quick note. Their tattoo game could be a little bit better. Could be. It could be most of the time. He's talking Yeah, all right. It don't make sense. You know, but it's it's one of those things where a lot of times that and jewelry. I mean, they overdo jewelry for me all the time where it's like Like you'll have two or three fingers together and they got a ring on each finger or one that's like taking up two knuckles and you're just like what the hell is that? Seven necklaces. I'm like no one wears that many necklaces, you know, dangling earrings always. A lot of times I'll I'll negative prompt that stuff out. And also if you want to take multiple things out, just to know you when you write dash dash no you type whatever it is, you know, maybe maybe it's tattoos. Then you got to use a comma and then put jewelry and then a comma.

Play from 1:13:01 8:88:88 1:13:01 Yeah, I'm glad you brought that up. Yeah. Yeah. Yeah, so I don't know if you've experienced this like where you've you've prompted multiple things with the no parameter, but do you feel like it is desensitized the the more no's that you start to include there? Have you seen that or is that just maybe my sort of speculation on what I've seen? Kind of I don't think it's as big of a you know, as big of a factor as it is in like something like stable diffusion where you have a lot more negative prompting going on. This one, you know, to me most of the time it's really like subtle things that I don't want. You know, it's most of the time it's it's tattoos, it's jewelry, it's I don't want like extra extra things that it likes to put in there.

Play from 1:13:41 8:88:88 1:13:41 There was a time when heavy stylize in I think it was maybe 5.2 like you got to have the stylize 700 800 and they would automatically add flowers. I'm like, why is there flowers? You know, it's just I remember when you told me about that, dude. Yeah. Yeah, because I mean, you just got to think like their training days got so much with flowers, but you're right. You I mean, we talked about that so long ago and that was I I I just started opening my eyes to that.

Play from 1:14:07 8:88:88 1:14:07 I'm like, yeah, dude, flowers are just why? I love flowers, but why? Yeah, so it's just no flowers and it would still be like, wait, cannot compute. Need flowers. Yeah. Yeah. That was That one you know, I use it for sometimes it's again to play with this stuff. Guys, like this is like really random. Like I just have these ideas sometimes where it's like, oh, what if I write, you know, a prompt for an illustration and it's like super crazy and then I just type in no illustration, you know, no some of it just to see what it'll do. That's where you get some really creative stuff is like trying to break the system to see where they like it'll in its armor. You get some you end up with some really cool stuff.

Play from 1:14:45 8:88:88 1:14:45 Like it'll be like a mixture of animation and real you know, realism, which I tend to find to be awesome when it combines those two things seamlessly. Especially with things like Niji. That's another one that we can probably get into. We don't use it that much, but we'll save that one for another episode. Go into that specifically. Niji's Niji's cool, man. I just forget about it. I I forget about it, obviously. I mean, there's just so many things that you want to want to create. But, if you guys aren't familiar with Niji, it's just like a How would you describe? More like an anime type style photo. So, it's not realism, per se. It's more like fantasy anime based.

Play from 1:15:18 8:88:88 1:15:18 Is that it? Another hot tip. Minimalism looks really cool with Niji. Like, if you go and do some minimalism stuff with Niji. That's just a little I'll just throw that one tidbit in there. Well, I got a whole list of things I want to test after this episode. These are These are all like random things that I was trying to help people with. And then it ends up being like, "Well, what if we try Niji to make things more minimalistic?" And it's like, "Oh, damn. That works." It's like, you know, we're trying to do some like bonsai tree in a Japanese garden, but make it look really soft.

Play from 1:15:46 8:88:88 1:15:46 Okay. Hey, man. I I There was a nice little segue in there, and I want to take advantage of it. You said soft. Let's go to like stop the stop feature. I don't think a lot of people use it. It's not the greatest parameter, but I do think like it's got some pretty cool use cases. Like, one of the things that we were talking about is softening an image. So, like this could come into play Essentially, what you do is you type {dash} {dash} stop, and you're going to put a value between what? 0 and 100, right?

1:15:50Stop (--stop)

Play from 1:16:14 8:88:88 1:16:14 Mhm. And I don't even think 100 you would use because that's the full image. So, essentially what it's doing is when you generate an image in Midjourney, you see it building and creating in real time. The that that noise starts to develop into an image. Stop, if you put a value in, let's just call it 90, it's going to stop at essentially 90% of that generated image. And what you could use that for is like to create like sort of the softness or blur or, you know, like finesse to a photo. So like maybe if you're introducing some sort of motion, motion blur, maybe you're going for a little bit something more fantasy or just like feminine or sensitive. Like this is this is something that I think is is a really cool thing to play with.

Play from 1:16:57 8:88:88 1:16:57 Like I going back to our our lovely friend flowers for a second. Like if you have like a rose, right? And the wind is sort of blowing the rose. Like one way to do that is to prompt that in itself, but you could prompt that into a stop of like 95 or a 90. And I bet it's going to be a damn good photo. You're going to blur it just enough where it it feels like more of an artistic interpretation. I love that. I don't use it a whole lot. With that said, like I just again, it's not something that comes to mind a whole lot, but stop is super cool. Do you use it at all? Have you used it at all?

Play from 1:17:33 8:88:88 1:17:33 I do use it. It's just it's it's case by case. I mean, I really don't go anything beyond 90 to 100. That's really it. I don't go anything lower than that. It looks cool with water a lot of times, too. Especially if you got some flow, some long exposures that you're trying to really enhance. It can be nice for that. Sometimes a stop like 95 can be really good for softening the water even more to give it that like ethereal type like dreamy liquid.

Play from 1:17:58 8:88:88 1:17:58 Ah, there you go. That's That's what I was trying to say. Thank you for that. Yeah. This stuff is great and I think we just got to do a whole episode on style reference because something like this again, now that you can take any image and kind of like transfer that style, using stop on these certain things to get that effect and then using that as a building block to another image. That's where to me these parameters now they're like the ones that we use less and less and less. That's where they come in as like developing stuff, you know, around style reference cuz you know, style reference too, high chaos levels, going back to that. Use the crazy stuff. You just take that and turn that into your own like little style reference. It's or or weird or high stylized. Right? So, to me it's all like little building blocks that work on each other. Stop to me is just another another way to kind of play

Play from 1:18:45 8:88:88 1:18:45 around if you have the Fast Hours to burn. I really like that way that you described that. I think that's that's perfect. Why don't we move on to I want to say the last one we're going to cover today, which is seed. Yeah. Before we get into seed, we have talked about style before and not to be confused with stylized, but like the standard versus raw. Mhm. I mean, I I just think there's a little bit of ambiguity there to like really touch on too much. Play with that.

1:18:50Seed (--seed)

Play from 1:19:12 8:88:88 1:19:12 If you're going for photorealism in any aspect, raw is a great way to go. I just tend to leave style raw on as a default for me. I'm probably guessing that you might do that. I don't I don't actually I don't know. You think you might use standard more, no? I'm standard. Yeah. standard. I never really know the difference, so. I just So, that's what that's that's what I'm saying, man. I think like that that is just so minor of a thing now that it's sort of come together where it's not as big of a difference as it as it used to be in our eyes. So, let's talk about seed. I think this is like another one where just a lot of people get confused. I've been so confused by seed in the past. Let's just kind of talk about what it is and if and how you use it.

Play from 1:19:53 8:88:88 1:19:53 Yeah. I think you know, it's really to me the most I use it for is comparison purposes. What seed is is basically if you have 1 billion random generation points like of where this where your image is going to start from, you type in a prompt, Midjourney basically selects like this random little point in its brain where it's like, "Oh, I'm going to start from here." That's what the seed is. And then, you know, if you ever want to go back and use that random starting point again, that's essentially what your seed is. So, a lot of times people think it it kind of affects, you know, make your images similar, make your images more consistent. Most of the time it's just a starting point, so you'll get the same four images in your grid. And then you can build off that.

Play from 1:20:36 8:88:88 1:20:36 So, I find it good for comparison purposes. Like, oh, I want to compare this image with, you know, higher stylized. So, I'll just use the seed and then dial up the stylized or dial up the chaos to see how different it gets cuz most of the time it'll stay exactly the same. But, if you run, for example, if you run the same prompt and the same seed twice, you're going to get the same four images. That's that's basically how it works. From there, it's like it whatever little adjustments you make, that's how you can keep it fairly similar to that initial starting point.

Play from 1:21:03 8:88:88 1:21:03 That one to be is often times the best way to do it. I also like it for small changes. I did it recently with an image of a guy, like a close-up of a man like holding a bunch of tomatoes. And I was like, oh, let me just use the seed to change those tomatoes to cherries. So, I got sort of the same thing. You can use style reference or image prompting for that, but seed sometimes can be a little bit more fine-tuned in that initial structure. But, that's uh yeah, really comparison purposes. I don't know if you're using it at all.

Play from 1:21:32 8:88:88 1:21:32 no, not really. Same thing. Comparison. Let's just show how you would go about getting it. I'll just quickly share my screen because it's not the most obvious thing. So, when you use seed, you know, you have to do {dash} {dash} seed. And then you have to specify what that number is that Rory was talking about. So, in order to get that, you know, I would need to go into any given image. And like Rory mentioned, if I use that same exact seed, I'm going to get these four images again if I don't change anything. But, how you get it is you're just going to click into any of that any image in that grid, go up to this options tab, copy seed. So, then now that I've got it copied, boom.

Play from 1:22:15 8:88:88 1:22:15 Right? So, just wanted to show that because, you know, that's that's always kind of been something that's been a little bit hard to find, but that's how you That's how you access it. Yeah. And if you're using Discord, you have to hit the little envelope key when you go into the open the Yeah. So, then it'll send it to you in your Discord bot, then you got to copy it. It's just you know, I think if it was more important they'd make it more accessible. I think a seed is more important in a lot of other image generators or video generators, personally, but it's to me it's very very case-by-case and it's great for when we're showing examples. I mean, that's really Well, I actually really like the the just the use case you brought up with something like even stylize, like you mentioned. I mean, that's a great way to get apples to apples comparison. Here's what this exact grid looks like just

Play from 1:23:02 8:88:88 1:23:02 with changing stylize, right? Not the same prompt and changing stylize, but that specifically. So, I I think that's uh that's kind of like a brilliant, I think, example of how you could sort of test that, right? It doesn't have to be stylize. Are there better examples of people using seed out there? I'm sure. If you're one of them and we didn't mention anything that didn't let us know. Let us know. We're all we're always interested to learn more, too. So, Rory, man, we have gone for a while.

1:23:26Wrap-Up

Play from 1:23:31 8:88:88 1:23:31 Yes. I'm just curious, like is there anything else parameter you want to cover? The big three, image ref, style ref, character ref, I think we hold off on. We do a whole 'nother episode on that. I mean, shoot, we could do a whole 'nother episode just on style ref if we wanted to. Think so. I think we're at a really good place to sort of wrap this up. I mean, we dropped a bunch. Like we probably could have went in even more detail, but we're trying to like really give you a consolidated, but also detailed enough perspective of how all these parameters come into play, what they do, how we use them, how we think about them, and how you might test and experiment with these things. My conclusion on parameters in general are one, it drives uniqueness, two, it drives creativity in your images, three,

Play from 1:24:18 8:88:88 1:24:18 tailoring it, right? Anything that you want to wrap up with on your end. Like or subscribe if you appreciated this. That's all. That's all I got to say. Hey, if you made it this far, you appreciate this. So We appreciate you, too. We do. Thank you for staying with us. You know, we probably could have done an episode on each one of these. I think it's good to have an over overview of everything cuz there is so much to do here if you love this stuff like we do. Knowing this is just way more way more useful in your in your creation journey. But, yeah, man. I you know, I'm I'm I'm glad we kind of ran through some of this stuff, brought up some old ideas that I totally forgot about. So new ones, some new ones I hadn't even heard before, so I love it, dude. I mean, it's always great getting into the weeds on this kind of stuff with you too because I know you're just as deep and you do

Play from 1:25:04 8:88:88 1:25:04 different stuff, and so I know I'm going to learn something that I haven't tried before. So, thank you for that. Uh as we wrap up, guys, thanks for joining us. Another episode of Midjourney Fast Hours. We'll be back next week. We're going to drop the heat. We'll figure out what we're going to do then. Maybe it's the the iteration on the the other part of this equation with the trifecta of the references. We'll figure that out, but in the meantime, stay up, have fun. Thanks for joining us. Keep it real. See you.