Ep. 26: To Midjourney v7 and Beyond: Personalization, Moodboards, Roadmap & More

0:00Podcast intro and news

Play from 0:00 8:88 0:00 What's up, Midjourney nerds? We are back. Midjourney Fast Hours Podcast. Drew Brucker, Rory Flynn. Back in the building. January. [00:00:11] Ten. [00:00:12] Ten. [00:00:13] Ten. Ten. You sure? [00:00:15] Somehow. [00:00:15] No one knows [00:00:15] the date of this. No one knows. Well, it doesn't matter anymore. We're living in the singularity, dude. So true. [00:00:24] So, so true. Yeah, who [00:00:26] cares?

Play from 0:27 8:88 0:27 [00:00:27] Who cares? That's it. I don't know. It [00:00:30] all feels the same at a certain point. I think now that there's like a There's definitely got to be like some sort of parent time warp effect that goes into play as soon as you have a child because it's just like May to May to January who [00:00:45] knows I don't remember anything. [00:00:46] I don't know how it got to here I don't know where I've been who I've been The last the days are [00:00:53] long. The weeks are fast.

Play from 0:55 8:88 0:55 [00:00:55] That's a great. That's a great question That's a good way to put it a very very poetic of you [00:01:00] [00:01:00] Yeah, these are long weeks fast how it is as a parent [00:01:04] and the weekends feel like the weekdays [00:01:06] they do that's right and I mentioned this before I mean it's sort of like a good little trade off because you get five days of a Weekend [00:01:15] versus two but at the same time your weekend now is considered working So right, you know take that back.

Play from 1:23 8:88 1:23 [00:01:22] It's worth [00:01:23] big facts I heard, I heard you're doing some snow angels today down in Atlanta. [00:01:29] [00:01:30] Yeah. So absolutely snowing in Atlanta, which, you know, you could say snowing in Atlanta and, you know, people get excited that, you know, you might have a little bit of flurries that hit the, hit the ground and melt. [00:01:42] This is not the case today. [00:01:45] Uh, school has been called off, which is not a normal, even if there's not any snow. We've actually had, School be called off just because of like rain before, uh, so that's, that's actually happened, but we've got legit snow. I [00:02:00] would say there's probably like a legit two to three inches already on the ground, not melting right away.

1:25Snow day antics

Play from 2:06 8:88 2:06 [00:02:06] You know, so the kids are stoked. This is really like, I think the first. Time the kids have like seen Snow there was like one winter where [00:02:15] when they were babies it snowed, but it melted right when it hit the ground So it was like cool to show them but never really got to play in it this time We got all dressed up. [00:02:25] We went outside We did the snow angels Um, we built a little [00:02:30] snowman out there And you know the kids are packing up the snow, grab the carrot through the carrot in the snowman's nose. We're looking for like something to use as the eyeballs. You know, I grabbed a pair of like sunglasses inside through [00:02:45] the snowman, you know, perfect.

Play from 2:48 8:88 2:48 [00:02:48] But then my wife's like, no, like, I don't really like the, the, the sunglass. I'm like, are you serious? Takes the sunglasses off, starts to kind of like, you know, grab some. Some stuff off the ground, you know, like [00:03:00] mulch stuff off the ground. Woodchip eyes. Yeah, woodchip eyes. [00:03:04] All right. And, you know. Now we're getting all And I'm like, yo, this, [00:03:07] this looks like shit, you know, like. [00:03:09] Ha ha. And literally, it was shit. You don't have any clothes sitting [00:03:15] around? No, hold on. She had grabbed a frozen turd off the ground. Oh. And placed it as the eyeball. And before I realized that, I had also grabbed said frozen turd and rearranged [00:03:30] frozen turd on the snowman.

Play from 3:33 8:88 3:33 [00:03:33] Was that from your own dogs or someone else's? [00:03:35] Well, that, does it really matter though? [00:03:38] Does it really matter? It's like, yeah, everyone likes the smell of their, you know, their own whatever, if you want to call it that. Little bit different, I [00:03:45] feel like if it's my dogs versus someone else's, I don't even know what it is. It's the same thing. We win. [00:03:49] Full circle full circle and you know immediately, you know, even though it's frozen turd It's like yeah, i'm going in wash my hands like this is over with the fun's over, you know, like we're done.

Play from 3:59 8:88 3:59 [00:03:59] So [00:04:00] Whoops, that was that was it man. But but actually before we did like finish up The neighbors from across the street, they've, they've got a young son. I think he's like maybe five or six brings over one of those like saucer sleds, you know, like the good ones, [00:04:15] the OG ones. And he's like, yo, you want to give this a ride? [00:04:18] Like talking to the kids, kids, like, I don't know. I was like, yeah, I don't know. Like, let me give it a test first. Cause my, my driveway goes. pretty much straight down. We're a little bit on a hill, and then on the other side of the [00:04:30] street, there's a neighbor, and they're even further down a hill. So, like, their driveway's a hill, but they go up.

4:01Sledding and childhood memories

Play from 4:37 8:88 4:37 [00:04:37] My driveway's a hill, I go down, if that makes sense. So I get on this saucer. And take off, you know, I, I go [00:04:45] face first, you know, on my belly saucer. And before I went, I was like, guys, like, based on the way this snow looks, I give this like maybe a two out of 10 chance that, you know, like this is going to work well. [00:04:56] It actually did work really well. So I got a running start, [00:05:00] flew down. You know, jumped on the saucer, went down the driveway, ramped up, went through the street, and then, like, went fast enough to where I hit the neighbor's driveway, and then went down again, and then ran right into their garage, so. [00:05:15] [00:05:15] Nothing makes you feel like a kid more than just doing dumb stuff.

Play from 5:18 8:88 5:18 [00:05:18] You know, you gotta go ahead first, because why not? Let's, let's test the, uh, insurance company's capabilities. Let's do the whole thing. Yeah. Well, I mean, if you're gonna do it, go, go a hundred percent. And there is something to be said about the, the saucer sleds [00:05:30] coming from the northeast. You really can't beat them. [00:05:32] Like, it, you can try. You know, back, we also used to just grab garbage can tops and just do that. You know, that was like the, that was like legit stuff. Do what you gotta do. You know, next time you'll build a little ramp. When they're like, [00:05:45] if the snow gets hard packed enough, get a little, get a little jump into that neighbor's driveway.

Play from 5:50 8:88 5:50 [00:05:50] It was, it was, [00:05:51] it was a lot of fun. It was a lot of fun, minus the dog. Frozen dog turds, you know, there's a lot of. [00:05:55] Yes. And if you also have a pool tube, another great thing, you know, if you [00:06:00] have like a, like a, like a noodle, like, yeah, well, no, I don't, maybe, maybe you could talk about pool too. Like an inflatable tube. [00:06:07] Oh yeah. Those things like we've definitely used one of those flamingos before. I wouldn't have thought about that. As a snow slide. I like that. They [00:06:15] glide like a mofo. They're awesome. Ooh. Okay. Yeah. Very, very slick. They get wet. So they just right over the top. Amazing. Amazing. It's a winner. Amazing. You pick up a lot of speed with those things.

Play from 6:27 8:88 6:27 [00:06:26] I miss being an idiot. Okay, [00:06:29] so [00:06:30] that's, that's me. What about you? Because you had a, you had an eventful week in Florida, sir. Since the last time we spoke. [00:06:37] Nothing, nothing, you know, we will go from nailing the foot to going to, I went to the hospital this week for, God knows [00:06:45] what. Who knows. I'm in the emergency room. [00:06:46] I'm back. I'm healthy now. But it was a, it was a pretty dodgy day or two there.

6:31Florida trip recap

Play from 6:53 8:88 6:53 Dodgy. Yeah, and we got a, you know, I like to say that I bring the worst of whatever environments weather. [00:07:00] With me when I travel went to South Florida, Naples. It's supposed to be, yeah, I think when I, with the last podcast, it was like 60, it dropped until like the forties. [00:07:08] Like when the hell is South Florida in the, in the forties? [00:07:11] No one's bringing a jacket when they're traveling to South Florida. [00:07:14] Yeah. [00:07:15] We were like going out in the morning to get coffee. I'm like, what is this? Come on. It's like, do I need sweatpants in southern Florida? It's like, this probably happens once a year.

Play from 7:26 8:88 7:26 [00:07:26] Where did you check the weather in New York while you were [00:07:28] there? Like, yeah, it was like, [00:07:29] [00:07:30] it was like negative five. Yeah. Okay. It was really, you know, trading a little bit, but you don't go to, you don't go to South. You know, I brought my bathing suits, not sweatpants. So did you touch water while you [00:07:41] were there? [00:07:42] I did go in a pool and I did [00:07:45] touch the ocean. There was one day we had about a 75 degree day. And I went, we went to the, to the beach. I stepped in the water was like, it's not warm enough for that. It's it's, it's good. Like if this is the middle of the summer, it's 85 degrees [00:08:00] out, 90 degrees out. I'd be right in, but it was, you know, just cold enough to be like, that's yeah.

Play from 8:05 8:88 8:05 [00:08:05] We'll just, we'll just ride here. We had one, one great day at the pool. Where the water was low, some relatively warm. Then they refilled it. You know, when you put the hose in the pool [00:08:15] and it becomes like 10 degrees colder immediately, that was the next day. So it was, you know, it wasn't going in there, but [00:08:21] maybe can't have it. [00:08:22] Maybe Mr. Uh, I'll be, I'll play Mr. Brightside here. Like, was it still good to get away and slow down a little [00:08:30] bit? Like get into that slow swing of things that the South brings away from New York.

Play from 8:35 8:88 8:35 [00:08:35] So that being said. I want to say Naples is just an extension of New Jersey or New York. Oh, is it really? It's, you know, there's traffic, there's, [00:08:45] you know, busy places, everything's busy. [00:08:47] Like, it's not, it's not like I went to like an island and chilled. It was like, I get in the car and it's like I'm battling people, switching lanes and stuff. I'm like, oh, damn. I go home. But, you know, regardless, we had a, we had a [00:09:00] good time. We laughed. A lot. [name] is absolutely out of her mind, so she's, uh, she provides constant entertainment, and, uh, we, you know, it's good.

Play from 9:10 8:88 9:10 [00:09:09] We just, uh, hopefully next time we go down a little bit warmer, we can actually, like, do Florida [00:09:15] things. Ah. Florida things. As you heard [name] in the background, she's, uh, she's on board with that. Heh heh heh. She's like, yeah, [00:09:22] next time. No playing around next time. Yeah. [00:09:27] Well, it did give me an opportunity to get back into [00:09:30] some of the uh, into some Midjourney when I had a couple, had a couple minutes. [00:09:34] I was using it as like my, Downtime. Which, I like when there's no, agenda. It's just like, let me go and play. Just like, I've been missing out on [00:09:45] this tool. I feel like I've been, I feel like I've missed it. You know what I mean? Oh yeah. So many other things trying to learn and keep up with, I'm just like, I just want to play on here.

9:28The magic of Midjourney

Play from 9:53 8:88 9:53 [00:09:53] I hadn't had the time to do that in a while, so it was good and it was just like cathartic. Like kind of our conversations. [00:09:59] [00:10:00] Every time. Every time. Totally agree. Yeah. Like, I get that. Because like we might be on a client call or doing a training or this or that. Like that's not. Playing in Midjourney, It's just like, I'm in the tool, but I'm not actually [00:10:15] doing it from a no agenda, purely creative standpoint, you know?

Play from 10:20 88:88 10:20 [00:10:20] Yeah. There's no like no structure. Like also our episode last week, I think got me back into it. [name] is, is really like, uh, not happy with our conversation right now, but [00:10:30] it got me back into the, um, that mindset of just like being creative, just like, instead of just trying to get something and then utilize it for video or get something for a post or whatever, it was just like, look. [00:10:39] Where can we go? Can we make the walls, clouds? Can we, you know, put Carlos Santana at the end of a, [00:10:45] you know, at the end of a stairway up to heaven? We, hell, we did and we did. You know, it was, and we did. It was like, that was like, oh, I forgot that. I can have a lot of fun here. But it also. You know, in testing all of these other tools, because we're always kind of like looking for what's new [00:11:00] photon or Korea or flux or ideogram or whatever.

Play from 11:05 88:88 11:05 [00:11:05] It's just like, I get to the point where I'm like, I've just, it's Midjourney. It just is like, you know, I look at them. That's funny that you say that. I like to go [00:11:15] back. [00:11:15] Yeah. there's just that, uh, that magical element of Midjourney. [00:11:20] You know, it's the, it's the parts that we talk about. It's the part that you talk about. It's the part of this creative interpretation and this [00:11:30] underlying magic that exists within the tool for whatever it is, as intangible as it is. It's not replicated to this point. No, no, it's, it's almost like it's, it's its own thing.

Play from 11:43 88:88 11:43 [00:11:43] You know, like [00:11:45] Midjourney is an AI image generation tool, but there are other AI image generation tools. None of them are like Midjourney. Midjourney is almost like its own. its own category, you know? Like, [00:11:57] yeah, it's really, it's just like you go back and forth [00:12:00] between like the differentiating factors. [00:12:01] Like control is awesome. Testing all these other tools and realizing that the control is there. You can get what you want. [00:12:08] It's cool, but at the same time, like it just doesn't have any. Pop and maybe I'm prompting wrong on these things, but there's just [00:12:15] sometimes you just need like that little extra look when you first use the stylized parameter in Midjourney and you like dialed it all the way up and you're like, Whoa, that's cool.

Play from 12:24 88:88 12:24 [00:12:24] That's not what I prompted for, but it's cool. You know, uh, the chaos parameter when you went all the [00:12:30] way and you're like, that's not even something I would ever expect or ever prompt, but I like to play with it because there's like this branching effect to it, you know, like we did, like we did with the I was gonna say, are we talking about micro decisions here? [00:12:44] Yeah. [00:12:44] It's like [00:12:45] the, it's like the multiverse in terms of like mapping your imagination or something like that. But yeah, it's just a, you know, feel, feel a little bit, uh, was feeling a little bit nostalgic to get back to just like going on there and doing stuff.

12:59Mood boards and customization

Play from 13:00 88:88 13:00 Yeah. [00:13:00] I mean, there's [00:13:00] still, there's still a lot to do. [00:13:02] Like, I mean, in all honesty, I'm still thinking about. All the things that just happened with mood boards and the personalization stuff. And it's like, yeah, we've created mood boards and we've started to [00:13:15] train, you know, other personalization things. But like, if you're like me, it's like. I've only done so much, you know, I've got a few, I'd like, I'd like to do more. [00:13:24] I would like to press the gas on how far that goes, right? I would [00:13:30] like to set up more mood boards that are more, uh, specifically focused and targeted towards some sort of like outcome that I envision either that I need now or in the future, you know, and starting to go down those paths. But it's [00:13:45] like, Yeah.

Play from 13:46 88:88 13:46 [00:13:46] There's only so much time, right? Like, so maybe we do that for an episode [00:13:49] too. Just build a mood board to collectively see what we can do with it. Like that, see how we, see how we edit it, see how we like, you know? I, I like that a lot. Yeah. [00:13:57] Maybe what, maybe we do that here in the next [00:14:00] week or two because, um, I agree, right? [00:14:02] Like that's, that to me has been one of the more fun things to play with. Lately, obviously it's a new feature, but also just trying to see how far you can push the boundary on it, which you and I both do, right? We want to see [00:14:15] what are all the different ways that we could use this. What are all the different ways this, you know, potentially has application for what we're doing, either just for personal fun or on the other side of it.

Play from 14:25 88:88 14:25 [00:14:25] Yeah, what are the rough edges? What do we need to avoid? What do we need to push? So, [00:14:30] maybe we, maybe we tee that one up, guys. We'll do that one for, uh, cause I've been waiting to play around with it a lot more. Maybe we do it live and see, you guys can see our real reactions to it. How we'd probably get around some of this stuff. [00:14:40] Now that we have this somewhat of a new format, seems like people liked it. Um, [00:14:45] People liked that a lot. And I, I mean, I, I [00:14:47] felt like you and I both thought Thought that would be the case and just so you guys know we wing we wing that one You know like that that was a hundred percent of a wing But I think like there are a lot of those [00:15:00] types of formatted episodes.

Play from 15:01 88:88 15:01 [00:15:01] We could still go further into right? It's the it's the mood board side We didn't even go, you know, into I think one of the like the things I think about like with micro decisions, which is I've got one additional idea and just the [00:15:15] variations and just like those initial controls and how far you could go with that to get something you like you brought up the mood board stuff, one of the things that Midjourney brought up on, you know, the most recent office hours is the ability to mix mood boards and personalizations, which [00:15:30] by the way, I tried again.

Play from 15:32 88:88 15:32 [00:15:31] Now I can't do it. I [00:15:32] told you you had like some special access that no, there was [00:15:35] something going on there. You had [00:15:37] a glitch in the matrix, dude. I did [00:15:39] because I went, cause I, cause I tried it again. Nothing. Then I was like, okay, did I just do this [00:15:45] wrong? Like, did I, did I maybe like overlook something? I went back. [00:15:47] I successfully mixed mood boards. Successfully makes, yeah, it's like, okay, there's a [00:15:52] selection tool too, with the dropdown selection tool. Yeah. I was like, this doesn't work. It's not working for me. Well, and like, I took [00:15:59] like two totally [00:16:00] different things to just to see, right? Like a mood board that's like this and a mood board like this, just to make sure that was actually happening and it was so sadly enough, that's not the case, but it sounds like that will be accessible here soon, [00:16:14] so [00:16:15] trajectory to, uh, or similar progression to SRF,

16:09Personalization focus

Play from 16:17 88:88 16:17 if anyone's familiar with how SRF, when it was. [00:16:19] You could do one, and then you could do two, and then you could wait. So I'm assuming that's the same way it's gonna progress, uh, from that standpoint. [00:16:27] We'd like to also, you know, [00:16:30] anecdotally, random thought, we'd also like to see the ability to run the permutations on the retexture, you know, with the, with the style weights, like, come on. [00:16:37] Good idea. That, that and if we could get on the personalization side, like on Discord, we talked about this last week, where you can find all of your old [00:16:45] personalization codes, Like, can I just have that as a drop down to make my global profile because I like my older ones. There you go. It is. Um, you know, that's just a hard fact for me.

Play from 16:54 88:88 16:54 [00:16:54] I just like them more. But I did hear something and we're definitely going to talk a lot about office hours today [00:17:00] because they, you know, they came back swaying after three weeks off. Um, you know, for the five minutes that I was on that I heard David say, we're basically going to force people to use personalization or want them to do that. [00:17:10] So I feel like that's going to be way more. Part of their [00:17:15] trajectory moving forward is like personalization really is we didn't really know where that was going I think but that seems to be like their focus moving forward is like real Personalization each individual user like less less mass appeal kind [00:17:30] of stuff.

Play from 17:31 88:88 17:31 [00:17:30] Yeah. [00:17:31] You could tell that sort of been that, that focus for them to really like get people to use it. It's not like more people are using it. Um, but it's, it's always good to, I guess, hear some of the feedback that he's got when it comes to like the [00:17:45] totality of users and how they, they use the tool, because those are reminders. [00:17:49] I think for me, probably for you too, where it's like, okay, well, still a lot of people aren't doing this stuff. Right? Like we're at the end of the day. We're kind of in our own echo chambers, whether that be in [00:18:00] our conversations, one on one or on a linkedin or on an X, you know, like we follow people and people follow us that we're in the same circle and, you know, just generally speaking, we're, we're one of the first movers in this technology.

Play from 18:14 88:88 18:14 [00:18:13] Those reminders are [00:18:15] important to kind of like ground us in terms of what. And where people are at, like what they're focused on and where they're at. So like personalization is definitely a part of that. [00:18:25] I think they're, they're trying to get everybody to use it. [00:18:28] Definitely. I think there's also just like a, [00:18:30] there is a diehard contingent that just likes Mitchell default style. I mean, I'm, I'm one of them. I still like just, I gravitated towards it. Immediately before personalization, you know, 80 to 90 percent.

Play from 18:43 88:88 18:43 [00:18:42] But for [00:18:43] personalization, it's all you knew. And it was like, Oh, this is [00:18:45] still great. [00:18:46] You know? So I still think there's people that just like it. I mean, I know me personally, I've, I was like, I don't really want to go too far on personalization cause I don't want to Stray from the aesthetic that I like and that I did go too [00:19:00] far. [00:19:00] And now my P code is like a horror film, everything I prompt with it. It's just, you don't like it no more. It's like four iterations ago. I loved it. And that's when I didn't do anything with it for about like a month or so. I'm like, I'm not going to rank any pairs.

18:50Personalization strategies

Play from 19:13 88:88 19:13 I love the way it is right now. And [00:19:15] then I was like, you know, for the purposes of. [00:19:17] And talking about the stuff. I'm just going to go further. I got it to like, I went from like level four to like level eight or something like or level five or six, whatever. I jumped up a few levels and now it's just like, it's a pure horror [00:19:30] show. Like it's, there's like blood and like weird stuff and I don't prompt anything or anything. [00:19:36] I don't, I don't like anything. That's I don't rank any pairs that are horror. I don't do anything that's in like that sci fi mix. It's like what, what [00:19:45] personalized this? What did you, what did you [00:19:46] answer on those surveys?

Play from 19:48 88:88 19:48 [00:19:47] I don't know. They, they, maybe they know something about me that I don't. I think [00:19:51] that's a great point though. [00:19:52] Well, that's what, okay, so that's what I did too. I, I, I think we're in the same boat. I really liked my personalization for a long time. I still [00:20:00] do my original personalization. but what I started to do is once they released that, uh, I started training a brand new one 20, you know, I just called it 2025 personalization and I'm about maybe 600 images [00:20:15] deep on ranking that, you know, and so I'm just trying to like basically use that as, Hey, like I want to use, I want to have a secondary option that's trained.

Play from 20:24 88:88 20:24 [00:20:25] You know, I'm probably not going to get to like the 18 or 19,000 that I had on the other one, but like a [00:20:30] strong model That's maybe a little bit more updated and maybe I have a different mindset going into how i'm going to rate those Because I feel like you because right to to our point Intention going in when we got personalization [00:20:45] the first time things had changed we did, you know This is what personalization is. [00:20:49] Okay, we knew We knew A or B part of the equation, you know, then, you know, weeks later, maybe we know a little bit more about personalization, our works, and then a little bit more. And [00:21:00] maybe at first we're starting to rank these things based off, which image do I like better? And then maybe halfway through it, we're just starting to like images that is this something I would prompt for or not, right?

Play from 21:10 88:88 21:10 [00:21:09] Like, so those are two different intentions altogether. And so, you know, I'm trying to [00:21:15] take something like this new personalization into account of like one singular mindset. Is this something that I would maybe produce and is a, is a good quality image in my style, [00:21:30] meaning like if it's, you know, an illustration style that I like and something I could use, maybe, yeah, I'm going to, I'm going to rank that even though maybe I do a lot more photo realism versus something that maybe is a good image that maybe it's something I never prompt for, like, [00:21:45] those are strategies and decisions that you.

Play from 21:48 88:88 21:48 [00:21:48] Yourself have to make, but I think maybe the idea that you could have one singular focus into how you're going to decide what you're going to rank could actually help. So it's just something to think about something I'm thinking about right now. [00:22:00] [00:22:00] I was going to ask you now that we have the ability to train in like a different personalization code, cause I've taken a different approach to it. [00:22:05] Also. Now, like the last, my, my global one was only like, I want this very specific look and that's how I want it to be. Cause I want it to be my, my aesthetic right now that [00:22:15] I can go a bunch of different ways. Like I found, like my P codes are very gritty, textured, super contrasty, looks like everything is Q2 from the start.

Play from 22:25 88:88 22:25 [00:22:25] Like it's just like that's what I ranked the entire time. I would skip like 40 images until I found ones that I [00:22:30] liked, which is probably why I never got to really high levels. But it was then like these new ones, like now I'm like, okay, well now I, I rarely, I rarely prompt for softer looking images, right? [00:22:41] Like, it's just not my, that's just not my preference regardless. I still like it [00:22:45] every once in a while when I want it. So now I'm like training one that's specifically on a softer aesthetic. And then now I have another one where it's like, I'm just liking everything that I like. Like, I don't care what style it is.

Play from 22:56 88:88 22:56 [00:22:55] I just want to see like sort of the variance. When I use them because the softer [00:23:00] profile that I've created works really well when I want something to be softer, right? Like it's now it's like, okay, that's, that's a very useful tool, like depending on what you need it for, right? Like if you understand sort of like there's going to be a weight attached to [00:23:15] whatever you You know, start to rank and they're going to, it's going to like heavily influence your outputs, like, then you can start to do it more strategically in terms of like what you're looking for.

Play from 23:25 88:88 23:25 [00:23:25] You know, maybe if you're looking to do more artistic or fine art focus stuff, [00:23:30] or you're trying to do more animated things, or maybe you just work a lot with Niji, right, go and rank accordingly. Like that way, maybe create one that's specifically for that. Cause now it doesn't have to, now it's not like. [00:23:43] Everything you do has to be [00:23:45] attached to this one per P code. It can be a bunch. So let's try to find those little like intricacies.

23:51Workflow optimization

Play from 23:51 88:88 23:51 [00:23:51] A hundred percent. Like what I'm doing now too, is I I'll take basically, I would say most of the things I'm prompting for experientially or personally at [00:24:00] this point, I'll run through both of those personalization codes that I've, that I've trained the then I'll just, I'll also randomly throw in. [00:24:12] What would this look like with this mood board? What would it look [00:24:15] like with this mood board? So like, oftentimes I'm running that same prompt across two personalizations and three mood boards, you know, just to see what comes out of it and how, you know, how that interprets what I'm putting in and the ability to get [00:24:30] unique, but great results from that.

Play from 24:34 88:88 24:34 [00:24:34] Has been massive. and so there's, I, I, that's what I mean, like going back to sort of how we, we started this conversation. I would like to spend a lot more time with mood [00:24:45] boards and personalization a lot more time, you know, because I, I do think that's where, This can really be streamlined. It can be very strategic. [00:24:55] And, just going back to personalized versus mood board, maybe one of the way you could think [00:25:00] about it, personalize, maybe personalize. And that profile that you're creating from that standpoint is more of a generalized approach of, do I like this image versus this image, or is this something I would prompt for period the end doesn't have to be specific [00:25:15] to an illustration or.

Play from 25:17 88:88 25:17 [00:25:17] a photo realism type of thing, but just like, yeah, I like this image or I like this image over this one because this one has more color and saturation versus this sort of flat interpretation of it. Right. Like [00:25:30] things like that versus a mood board where that I think is maybe the more specific nature of it, which we talked about too, where it's like keeping it tight is the word that you use, right? [00:25:40] Like keeping that mood board a little bit more buttoned [00:25:45] up tight and specific. To an idea that you have, whereas the personalization can maybe broad and span to different things that you like and don't, don't like.

Play from 25:56 88:88 25:56 [00:25:55] There's, there's two things that I would love from Midjourney in [00:26:00] this, in this personalization sort of quality of life updates, quality of life. [00:26:05] Yeah. Good. Number one, I'd love to be able to rename the codes. So coming back to like, you know, custom options set in discord. So I [00:26:15] don't have to like. Cycle through or select or whatever if I could just type like dash dash Rory one, whatever Yeah, like that would that would be great. I agree very simple for me totally make it a lot easier because number two I mean I can like [00:26:30] look at them in my feed, but I don't even know which one it is You know what?

Play from 26:33 88:88 26:33 [00:26:33] I mean? Like it's TX 45 That's right, like a football play, right? And when you, and when you're talking about the [00:26:40] feed, you're talking about, Hey, I've created this now and let me bring up that [00:26:45] same prompt. And it says the profile dash dash profile. And then I've got a list of letters and numbers. Which one is that even associated to without looking at the create and organize screen, right? [00:26:55] Because that's [00:26:55] my next feeds. And it is a [00:26:57] huge pain in the ass, especially if you're doing it on mobile, [00:27:00] right? Because then you really can't tell [00:27:02] my next problem, which is, I think, a relatively small one. Which is like, can we override if our defaults are selected on the profile codes? Cause like I was using, like, I'll use one of my old P codes, right?

Play from 27:15 88:88 27:15 [00:27:14] And it's like [00:27:15] in the prompt, I go reuse prompt, it pops up in there. And even if I have the default set with a mood board and a profile code, hit enter, it still doesn't, it doesn't override. Great point. I got to like go delete that, redo it. Cause it's like, you can't like for prompting, I don't wanna say prompting for [00:27:30] speed, but I don't know how many times I've run either permutations on something like that and have like 15 generations and I'm like, this doesn't look like the mood board I attached. [00:27:38] It's like a hundred percent, dude. I'm glad that you [00:27:40] brought that up.

Play from 27:42 88:88 27:42 [00:27:41] Yeah. Simple one. [00:27:42] These are, okay. These are things that you're only thinking [00:27:45] about if you're playing with this stuff a lot, because we, this has been on my mind since the last time we spoke. It's just the. Just that, right? Because going back to my point of I'm trying different personalizations, different mood boards with the same [00:28:00] prompts, and I want to do it quickly, but also you can't permutate profiles and mood boards, can you?

27:43UI pain points

Play from 28:07 88:88 28:07 [00:28:07] Can you? No. [00:28:09] You can, you can on, on, uh, I've tried it for profiles and [00:28:15] for mood boards. You can do that. [00:28:17] You can do permutations on profiles and mood boards? [00:28:19] See, the mood board, like, I don't understand, so here's the thing, like, when you use a mood board, this is, I think I might get this wrong, so if anyone fact checks me, you're probably right, but I'm pretty sure if you use your [00:28:30] global profile and a mood board, and you push, you know, you hit enter, then it's like a totally different code, rather than if you just put mood board and no personalization code, like, it's a, it gets confusing as to what, what I actually did with it.

Play from 28:44 88:88 28:44 [00:28:44] Do you know what I [00:28:45] mean? Like what? Oh, yeah. Yeah. No, I filed. Yeah, it's just the mood board Not saying that this has to be ironed out or like it's a big Factor for them or like me getting upset with it Just like some of these things because you just want to if you're we're gonna focus on personalization You [00:29:00] know, like I kind of want to do things that are you know, personalized especially like naming my own stuff Like the mood boards great. [00:29:06] You can name it Like, you know, awesome. But still is that bike coding sort of backend there where it's like 75 32 [00:29:15] GXW 43 set hike, you know, like that's kind of how, um, yeah, it, I would really like that to be, I don't know how to envision it. And every time I do say something like this, [00:29:30] they, they come up with something that's way

Play from 29:31 88:88 29:31 easier than I could ever envision a solution, but at least when you're At least when, you know, maybe having like the, if your defaults are set, like prompt override on the profile codes. [00:29:41] Prompt override. Yeah, maybe that's one. [00:29:45] I want to show you this. Maybe it's good to show this. [00:29:47] Well, cause this is what I'm, this is, and I would love to just show you what I'm doing to make sure. Is that calculus? So, I saw this random equation somewhere on X. I don't [00:30:00] even remember the post.

29:45Creative experimentation

Play from 30:01 88:88 30:01 [00:30:01] It looks sick though. [00:30:02] And it generated some sick stuff. Sick stuff. I like that. but like, let's use this as an example, right? Like what I'll first do is to like, make sure when I'm selecting, right. I'll go [00:30:15] here. Okay. So like, let me run this and like, let's just run these on fast. Let's do all this good stuff. So let's, uh, let's just do AR one, one. [00:30:25] So, okay. That's my, that's my default anyway. Okay. I'll run the default. What I have is a [00:30:30] stylization. Let's just move this to three. so I don't have to put in a bunch of stuff. Okay. So now that I've got that, let's go ahead and hit enter. Now, let's just say I want to switch the profile or the mood board.

Play from 30:40 88:88 30:40 [00:30:40] What I'm doing is I'm going and I'm selecting the new [00:30:45] one, unselecting this one, then I'm clicking off, then I'm clicking on. Okay. [00:30:51] Why? [00:30:52] Well, because. I don't want to have to copy and paste the code in there. So, for example, I just click [00:31:00] this, now I've already got this selected. [00:31:03] That's because your D well, you're also, your, your aesthetics are selected at default, like the aesthetics, the stylization, correct.

Play from 31:12 88:88 31:12 [00:31:12] Correct. If I'm doing like Q2, [00:31:15] you know, S uh, you know, uh, image, weight, whatever, what, you know, you have to like copy those and you've reused the prompts and then that other profile code comes in there and then you got to delete it. Yeah. Yeah. [00:31:24] You have a hundred percent. Yeah. Like I'm just showing this like just for speed here, but yes, like you would [00:31:30] have to like delete the profile. [00:31:31] You're like. But I guess I'm just showing you this part, which is now I got to go in. Okay. I'm going to select this. I'm running that now. If I do this, and then I'm going to go ahead and select this again, [00:31:45] like to your point, if you're just pressing the up command, which is the shortcut to get that prompt to reenter, it's going to throw that profile or that mood board code right in there.

Play from 31:55 88:88 31:55 [00:31:54] So you're going to have to delete it out. So I'm just saying from this example, like I'm just taking that. [00:32:00] Now I've got to turn this off, turn this on with the selection so it doesn't confuse that and add that back in. That's all I'm saying. Like, so I'm just pretend like I'm not doing Q2 for example, right? [00:32:10] Yeah. I'm running 300 stylized, I'm running 1 1. [00:32:15] So, then I'm, you know, then I'm running that. Then, You know, I'll say, okay, now I'm going to do this. okay, let's run this one, this one. I'm like hitting off and on every time because I don't want that code to just populate.

Play from 32:30 88:88 32:30 That's [00:32:30] like my shortcut of how I'm doing it without having to delete all this stuff. [00:32:33] But it's still a pain in the ass. It's a huge pain in the ass, you know, because I'll often go back and I'm like, to your point, I'll look and I'm like, oh, well that just ran it under my global profile, you know, and then I've [00:32:45] wasted, you know, maybe a fast hour or something like that. Like trying to do that, that kind of stuff. [00:32:48] But, this is a little cool little, uh, little prompt here. I mean, you get some really neat stuff out of here, you know, [00:32:55] I'm digging that. I like this. I like this idea. So again, that's something you can't put into another [00:33:00] generator and get something cool out of it. Like that, that exact prompt, right?

Play from 33:03 88:88 33:03 [00:33:03] There you [00:33:03] go. [00:33:04] And, you know, maybe while we're talking about that, like another cool thing that I really love about Midjourney is that you could throw in a, an emoji and get cool shit, you could [00:33:15] throw in lyrics from a song. Or a one liner from a song or a poem or whatever it is. And you get really cool, unique, creative stuff. [00:33:26] That's part of the magic, right? That's part of the magic.

Play from 33:30 88:88 33:30 [00:33:29] [00:33:30] One word prompting stuff like that. You know, [00:33:32] things [00:33:34] like that. I love just starting to like it more now that I try to like. Skew from control every once in a while just like yeah, let's just go down this weird route See what it comes up with cuz and that's [00:33:45] that's like my way That's like my idea inspiration now at this point It's like let's just let it do the thinking in terms of like creativity and I'll run with it from that totally No, everyone totally so did it's it's good I think that was just a definitely wanted to hit on the personalization stuff of the office hours because I feel [00:34:00] like you know That was to me.

Play from 34:02 88:88 34:02 [00:34:01] Let's let's go into that listen Yeah, [00:34:03] I like that because, okay. And then that was the idea that we had going into this too, is where we're at with Midjourney, the updates, we're getting close to a lot of things here. Yes, that [00:34:15] gets tossed around a lot, but, but I think that's the truth here. As we get close to v7, [00:34:20] you get close to v7 one, which shouldn't be very far behind v7. And the reason because is v7 is something they've been trying to push out for a [00:34:30] while, but they also realize that they're not gonna be able to shove everything into v7. So they're going to have that phased approach of a v7, which will probably have.

34:02V7 model expectations

Play from 34:38 88:88 34:38 [00:34:38] 50 to 75 percent of what you would normally have, right? There might be still a few features that we can't access in [00:34:45] V7 that we do have access to now. Then you get the V7 one, which has the extension of the other pieces of it, but V7 couple of weeks away, touching on the notes. Maybe I'll share these two cause you and I both. [00:34:59] Um, we're [00:35:00] on at least part of the call. And then we also have gone through the notes here. Let's pull these up so everybody else can see them. And we'll just kind of talk through these together. First piece being the seven. So new model, I would take this back to like a month or two ago.

Play from 35:13 88:88 35:13 [00:35:13] They said January. [00:35:15] The reality of that now, to me, looks like late January, mid February, with mid February, early to mid February feeling like the most likely. Because, them talking about this idea that we've got, we're now in this [00:35:30] testing phase. And that can take anywhere from a week to a month. Plus there's a two week final tuning phase, [00:35:37] and then the alpha period follows the final tuning. And if you're not familiar, alpha is typically that access that's [00:35:45] given to folks that have generated. A certain number of image, there's some sort of threshold. They've generated a certain number of images. They've been on, they've been a Midjourney user for a while, or they have a yearly plan, whatever it is.

Play from 35:57 88:88 35:57 [00:35:56] Right. So then that will be the case. Not everybody will have [00:36:00] access, but you know, it looks like February for that. And I think the thing that I was the most surprised about with this piece of it was David V7 is totally [00:36:15] different. than v6 And I was not expecting that because even if, if I take that back to one of those office hours that occurred before the holidays, because we had a couple of weeks off there, I think we might've even talked about this.

Play from 36:27 88:88 36:27 [00:36:27] I was a little concerned because it didn't really sound like [00:36:30] V7 was going to be much of a. Jumper a difference at all. And now I heard this right. New architecture, different data sets, revised language processing. So different prompting is going to work. Now it's going to be a whole different coherence.[00:36:45] [00:36:45] They, they even talked about the jump from V five to V six again, [00:36:49] which [00:36:50] you and I know quite well, because we remember when 5. 2 was over and then all of a sudden six was there and it was like, Oh, these images are totally different. And there's. A lot of the stuff that I [00:37:00] can't just copy and paste over, I got to make some tweaks to it.

Play from 37:03 88:88 37:03 [00:37:04] Because it was like V5. 2 was like, and V5 essentially was like broader concept prompting, which was going to work really well. Like if you're just using keywords and like large sort of placeholders, like you're going to get something creative [00:37:15] and good. And then V6 sort of like went back to that, you know, super coherence model where you need to like, really define more blank canvas style, what you're looking for. [00:37:23] Um, But, you know, now that once they started adding the things like S ref and the S ref [00:37:30] codes and the editor and the retexture and, you know, all these other different variables and play and personalize and mood boards and whatever, like you can kind of take it in a million different directions. So it's less, you know, less of a deal than it was when it was first V six, right.

Play from 37:43 88:88 37:43 [00:37:43] When you just like jumped in there and you had [00:37:45] to kind of text an image prompt and that was like where you were going with stuff. Um, so yeah, I mean, like, I, I feel like. We think it hasn't, to me, it's ambiguous, right? Because when we look back, we thought V5 was crazy in [00:38:00] addition to V4, right? [00:38:01] And like, in reality, it probably wasn't, it wasn't like crazy big, but then like V6 for V5 was another massive jump, but we were just able to kind of pick it up much quicker because of, you know, older prompting architecture from like [00:38:15] V4 kind of stuff. I have a feeling this one's probably going to be pretty, uh, the one thing I'm See, now the one thing.

Play from 38:22 88:88 38:22 [00:38:22] This is, this is interesting for V7 that I'm hoping is that now after testing some of these like super coherent models, I [00:38:30] hope it's not down that route because it's awesome to control everything, but if you can't control everything and you don't have a vision for everything, Then like, you're just going to have blanks, you know, kind of average image, right? [00:38:43] Like you just, that's how it will be. [00:38:45] You don't have that like creative flair that Midjourney we've talked about so many times. We're like, let it's imagination also drive some of the conversation to let's clip [00:38:52] that. Let's clip that.

Play from 38:55 88:88 38:55 [00:38:54] That's, [00:38:55] that's, that is it, because do you want to change the things that make you [00:39:00] different? [00:39:00] And if you do, how certain are you that those are the right things to change? [00:39:05] Yeah. [00:39:06] that's it, dude. You know, like I still love having fun with the ability to take some of these tokens that you and I talk about, or [00:39:15] this equation that I just brought up and you throw that, you know, like that's fun, that's the magic.

Play from 39:22 88:88 39:22 [00:39:22] And to your point, there is value in the coherence and the clear and matter of fact [00:39:30] way. It's not a lot of fun. No, it's not a lot of fun to do it [00:39:33] that way. No, it's a lot of rules. . [00:39:35] You know, [00:39:36] like, it's a lot of rules. Who likes rules here? None of you. Because you're listening to this. That's right. [00:39:40] That's why. So, that's right, dude. It's, that's the one thing. It's just like, you know, I, [00:39:45] I, I'm being a hypocrite. 'cause I would like the ability to do both. Look, I would like the ability for some reason. Well, why can't you [00:39:49] do both? You know? I think that's, that's a, that's a common problem that a lot of people run into.

Play from 39:54 88:88 39:54 [00:39:53] It's like it's gotta be one way or the other. It's one thing or the other. No. Both things can be true. Why can't we [00:40:00] have a more coherent model and a more abstract magical model? Why not? Why? I don't know. Like, I'm not a technical nerd. I'm not building these programs. So I don't know the ins and outs that would be super [00:40:15] valuable. [00:40:15] No one else is doing that. [00:40:18] That's why I like, you know, that's why having something like stylize is so great because stylize is basically like. You know, lower levels. It's what you said, higher levels. It's like, let the creativity run wild. Awesome. [00:40:30] Um, you know, there's not a lot of, I just feel like that's like, people are cheapening the idea of like the, the creativity portion of the tool itself, because you know,

Play from 40:40 88:88 40:40 that only can get better with the immense amount of data that it's trained on. [00:40:43] You know, it's only gonna get bigger, bigger, [00:40:45] bigger, the more creative it gets, but if you put too many guardrails around it, then, you You like limit what that output can be. And then we go back to, you know, version five when everyone's stuff looked exactly the same and super symmetrical and it was the exact same face and it was the exact same setting. [00:40:59] [00:41:00] And it was all, you know, neon lights in the background, looking straight at the camera, [00:41:04] you know, Hey man, that was, that was cool for a couple of minutes though. For a couple of minutes. That was cool.

Play from 41:10 88:88 41:10 [00:41:09] Yeah. I like the, so I'm like, I'm very much so looking forward to [00:41:15] see what this actually means. You know, like what this all means as improved prompt understanding coherence and flexible styling, you know, from there are some tokens that you can go that you go through that don't necessarily pick up, right? [00:41:27] It doesn't pick up on the tokens as [00:41:30] some of it does. Like we've talked about a lot of the power tokens that just do work, right? Then there's something that, you know, it's kind of like, I don't even know what that is. So I'm guessing like expanding data sets in terms of like Prompt understanding and coherence to like, it just, you just have to keep like adding [00:41:45] stuff to that, into the tagging mechanism to get somewhere.

Play from 41:48 88:88 41:48 [00:41:47] But also like, you know, I'd love to have, if we want to have like super control, I'd love to have more color control. Right. Like more like control, like something like that. Yeah. Cause there's a way to codify that, you know, [00:42:00] like there is a standard Pantone set with, or RGB set, like, you know, CMYK set where you can do that. [00:42:05] No, man. So I want [00:42:06] Sherwin Williams colors, please. Well, [00:42:10] it'd be a great licensing deal if anyone wanted to pull that [00:42:12] one off. You know? Yeah, no, I mean, I [00:42:15] agree. Right. Like style references and you know, style codes and you know, all that. Yeah. It's it's. Definitely very, very valuable, but. To fine tune that at this point, it feels like the next step.

Play from 42:27 88:88 42:27 [00:42:27] I mean, it feels like it's coming up and [00:42:30] inevitably the, the other piece of V7, maybe just talking about that model as a whole is character consistency, which I know you love multiple characters long overdue, right? Like I even think when we, when [00:42:45] we got C ref. Do you remember when that was? But whenever it was July, we thought this is the V one V two's coming shortly, you know, and, and it hasn't happened yet.[00:43:00] [00:43:00] It's been waiting for this. So like that's long overdue because now you've got all these other tools, right? That have sort of got, you know, leaped Midjourney in terms of the character consistency, part of it.

42:31Character consistency

Play from 43:12 88:88 43:12 [00:43:11] Midjourney [00:43:12] brings that into the fold. Watch out. I think that's, [00:43:15] that's, that is going to be one of the key pieces of V7. [00:43:18] They've said that. So if that's something that you've been waiting for, and you're listening to this, you're going to love that. I feel pretty confident in saying that because they've stressed that from the beginning, [00:43:30] but they've also talked about that, ability to bring multiple characters into the scene, which as good as.

Play from 43:37 88:88 43:37 [00:43:37] Your walkthrough and loophole was [00:43:41] hard [00:43:42] fucking sucks that you have to go through all those [00:43:45] hoops to do it You know what? I mean? Like we got to make that part easier So they have [00:43:50] to for video, right? Like that's the thing is like there's they have to realize that that if they're gonna do video People don't want to use a different character in every scene. [00:43:57] They want to use the same they want to tell stories So like [00:44:00] they have a storytelling team they have, you know, this new video sort of initiative. They have patchwork They have all the stuff to get People want to tell stories with the tools, whether that be still or video. They want to use the same characters, and they want to use the same aesthetics.

Play from 44:13 88:88 44:13 [00:44:12] Aesthetics things, now, nail down. [00:44:15] Like, they don't even have to, to me, they don't even have to do much more there. Like, I don't even care what else they do. I have enough control, aesthetically, to do what I want. Right? It's now, it's just, now it's the finer details. It's the characters, objects, places. Things like [00:44:30] that. [00:44:30] So [00:44:31] now, Hey, this is an interesting question and I know we're going to like continue to like push through this, but I wonder, cause I was just on it, I was just doing a Training the other day, character consistency, people using that as like, you know, [00:44:45] this, this quasi object reference, which we've talked about to like using an object as a character reference to try to, you know, see what juice is there.

Play from 44:55 88:88 44:55 [00:44:55] Um, it's kind of hard. Sometimes it works. Sometimes it doesn't. Do we get an object reference [00:45:00] at some point, [00:45:01] I have conspiracy theories on this. conspiracy theories. I think it honestly, I think it's like a legal thing. [00:45:08] I don't think they want to touch it because I don't think like they've already had their, their, you know, toes dipped in hot water with a few, [00:45:15] you know, legal situations with, you know, IP infringement and stuff like that. I don't think they want people like uploading random, you know, intellectual property into the, into the product and then like that being replicated in some way, shape or form, which is almost impossible, but at the same time, [00:45:30] like I, I just have a feeling like there is some sort of like They've been so quiet on it and it was something that people really wanted, you know, for a long time It was high up on that on the ranking list of the things that people wanted

Play from 45:42 88:88 45:42 So yeah that if I'm thinking like them, you know, it's [00:45:45] something all right. [00:45:46] We already have like enough noise Around like that's not something that's necessary There's other tools for it like let people go in that direction with other tools. So, you know granted I would love it I have a feeling I'm not going to see it. [00:46:00] That's just how, how I feel about it. But I think if they're going to, if they're going to give in some areas, right, like not push their, their luck, like, I think it might be that one.

Play from 46:11 88:88 46:11 [00:46:11] Um, cause everyone's talked about it for a long time. Like, [00:46:15] can we just get an object reference? Yes. That'd be great. But. We haven't seen it or a development about it or even a mention of it very true No, I think it's intentional that it's not there. [00:46:25] Yeah. Let's go through the rest of this with v7 than what we know I guess there's [00:46:30] The large batch modes, which should be a part of it. [00:46:33] Right. So we're thinking about larger than just getting your four grid back without, without using permutations. So, you know, I know we talked about like that batch of [00:46:45] eight. Does that expand to bigger? I don't know. They haven't talked about that, but maybe that's something that could be like one of those things that just kind of jumps in there.

46:22Future video features

Play from 46:52 88:88 46:52 [00:46:52] Is there anything else that we miss with that? Like what's draft mode. [00:46:57] Don't remember that from [00:46:58] the [00:46:58] call. I didn't remember [00:47:00] reading anything about draft modes, especially as offshoots at varying speed and quality. I don't know. What, what is draft mode? Well, I know this [00:47:07] just like, they've talked about the idea of like, do we have different models? [00:47:11] Like if you think about runway runways, got turbo, for [00:47:15] example, maybe you sacrifice some quality. To get something faster, but I don't know, I have no idea what draft mode means now with video and all that haven't heard how that [00:47:30] attaches to V7 yet. I don't think that's like my speculation.

Play from 47:35 88:88 47:35 [00:47:35] That's not going to be a part of V7. Definitely like I would feel very strongly. That's not gonna be part of the seven will be a part of the seven one. [00:47:45] That's the question that I would have, right? When does that come out? Is it attached to just v7 at that point? Because there was that initial conversation. [00:47:53] If you remember, like, do we have a training model that's just built off the six? they did start to talk about video on the call. [00:48:00] They talked about they're working.

Play from 48:02 88:88 48:02 They've got like four options here. They've got one model they're building themselves. They've got two models that they're talking to potential partners, potential partners, collaborators with. [00:48:12] I have no idea who that is. I have no idea [00:48:15] whether that pans out. Probably stuff. Right. [00:48:17] You know? [00:48:17] And so, you know, like there's a lot of question marks still on the video side, but I thought there was some really great points, you know, by David. That were made. And I think, like, I think [00:48:30] about this and I know you relate to it, but like, I know I thought about this too, because I recently like, you know, actually labored to put together a video that was a few things in a row, right?

48:23The state of AI video

Play from 48:42 88:88 48:42 [00:48:42] Like more of a vision, more of a story. [00:48:45] And as much as I like doing it. It's a lot of work. It's a lot of fucking work and the way that David put it I think was was was correct. You can build passable things with painful [00:49:00] workflows Yes, great, great line, passable things with painful. That's exactly the state that we're in, right? [00:49:06] It's not a fun, it's not as fun as it's going to be. It's not as fun as it should be right now because of where we're at [00:49:15] with the tech, And that's not a knock on the tech. The tech is going to keep evolving and growing up and becoming more mature and unlocking things. But that's kind of where we're at right now, you know, like you do a lot of video Rory, I know, with me doing some of [00:49:30] that, like, how much that

Play from 49:32 88:88 49:32 goes into that. And it's kind of a pain in the ass. It's kind of a pain in the ass. And it's when it's that it takes away from the fun, which is part of why we always talk about mids kind of why this [00:49:45] podcast is about Midjourney. It's that magical, fun part of creating right. [00:49:48] And so when video unlocks that and these things become more consistent, more high quality. More predictable, less painful in terms of like connecting the dots between five or [00:50:00] six tools. That's going to be that inflection point for video. And does that come in 2025 or does that come in 2026?

Play from 50:10 88:88 50:10 [00:50:10] Good question. [00:50:11] I think they're probably going to partner API at some point. [00:50:15] Like, I don't know who it's going to be with. It sounds like they're looking for something like that. Maybe who knows, maybe they acquire someone like minimax. Like, I don't know, like everybody might be just an easier way for them to get to the. [00:50:24] To get there quicker and cheaper is to buy an existing product, you know, to buy something that's already [00:50:30] available They definitely have the cash flow. I wouldn't be surprised. I don't think they're they're underwater by any means considering You know again the way that they operate so it's it's interesting because I'm not sure what they're gonna do and how they're gonna make it different because video is [00:50:45] so It's so hard.

Play from 50:47 88:88 50:47 [00:50:47] I mean, you know, technically if they had if they had character consistency You Right. Like what video does is really just, you know, 20, a lot of them are 2430 frames per second, you know, really what they're doing is just generating small variations [00:51:00] between point a point B. So like every second you get 24 images, right? [00:51:03] So this larger batch to me is like another step in the way of how they want to do things, because it's just like, basically you're, you're doing. The way that I understand it is if there's, you know, 10 second [00:51:15] video, you have 24 frames per second, it's generating 240 frames in between, you know, each little step of the way, and then sort of like need the model to interpret, you know, interpolate them to like blend them all together.

Play from 51:26 88:88 51:26 [00:51:25] So yeah, they have the, the images, right? The images are there, but they can [00:51:30] produce all that. They just need the consistency, like style, aesthetically works. It's just like the characters and things like that. And then you have to, you know, from my understanding, again, Of like how they're training models. [00:51:41] It's mostly on motion. So that's the differentiators. You have to have enough, [00:51:45] you know, motion data, whether that's human, animal interactions of light and things like that, which is like a ton of work, which I can see why they're like, it's going to cost millions upon millions to do this, [00:52:00] and it might be easier for them to buy someone, I don't know.

Play from 52:02 88:88 52:02 [00:52:01] I think there's probably, they probably have their pulse on, you know, a lot more of these smaller, Sort of products that are kind of developing and things of that nature. And maybe that's a, maybe that's a way for them to acquire, [00:52:13] maybe two things to leave you [00:52:15] guys with, as we wrap this up, one, going back to personalization, David really stressed that you mentioned this early on, make sure you guys are going through that process personalizing, this is where. [00:52:26] I think the secret sauce lies in getting things that you [00:52:30] want to create with less effort, so going back to the idea, you got to create these long, very specific, very strategic prompts to get something you like. Yes, there are places for that. Yes, there are times for that. But it's mostly unnecessary in a lot of ways when you set up things like [00:52:45] personalization.

52:14Final tips and conclusion

Play from 52:46 88:88 52:46 [00:52:45] So that's number one, go through that process. And maybe the last thing like we should mention and bring up Rory is just the fact that like now is a great time You To use Midjourney on relax mode because they off, [00:53:00] they have offered up their servers to basically run relaxed mode at a fast mode speed right now. [00:53:05] I can't remember when the cutoff is for that, but it's coming soon. So I think you have at least maybe by the time this comes out, maybe another few days, if not a week. [00:53:15] To run that. So this is great. If you want to explore style codes or run prompts in different mood boards or different personalizations, or just whatever you want to do without burning Fast Hours, now's the time to do it.

Play from 53:30 88:88 53:30 [00:53:29] [00:53:30] Meaning you can basically have unlimited generations if you want it. So there you go. That's it. That's your, that's your new year's gift. [00:53:38] It's been a hell of a gift. Like I would say like when that, when they first announced that I was running so many damn style codes just in relaxed [00:53:45] mode. [00:53:45] Perfect for that. [00:53:46] R40. Enter. I'll see you in 10 minutes. Exactly. Exactly. Well, gotta wrap this one up, Drew and I. Let's wrap it up, dude. Tight schedules here. We are. We are on tight schedules. Unfortunately.

Play from 54:00 88:88 54:00 [00:53:59] [00:54:00] But, we're gonna be back next week with a good one. Rory, good to see you, brother. Glad you're safe. Don't bring any cold weather down to your Atlanta. [00:54:06] We're good. We already got the snow. So [00:54:09] just wait, I might bring something else, but you know, with that being said, everyone knows the deal here, like, and subscribe because that's, [00:54:15] that's what we have to say. But again, dude, awesome talking. what's ripped through this next week. We'll go through some, hopefully we'll have some v7 coming up in the next two to three episodes. [00:54:24] And then, uh, I think we'll just go from there. Let's get this thing going, man. Good to see you, brother. Good to see you, dude. Have fun. You guys [00:54:29] [00:54:30] catch you in the next one.