Ep. 55: Nodes Are Eating the Creative World And You’re Already Late

0:00:01"Tell your dog walker to subscribe”

Play from 0:00:01 8:88:88 0:00:01 What's up everybody Midjourney Fast Hours podcast episode 55 the week after Halloween were back Rory Flynn Drew Brucker. Go ahead and let's let's get this out of the way up front. Go ahead and like and subscribe and tell your your mom, your friends, your grandma, your dog, cat. You know any any weird any weird anyone or anything?

Play from 0:00:28 8:88:88 0:00:28 How about that? Just cover the bases. if you have any of those. Someone that you see on the subway or on the train or on just riding your bike. Just be like, yeah, Midjourney Fast Hours. That's us. You guys should check this thing out. 55 episodes in, Rory Flynn. I mean, I was just talking to uh my wife about it last night. was like, I can't believe 55 episodes in.

Play from 0:00:55 8:88:88 0:00:55 I mean, we've just, we've been doing this for a year and a half. It's been a long journey. It's flown by at the same time. It feels like we've been doing this much longer than that. And I feel like we've talked about so, so many things. Obviously, like if you guys have joined us recently, right, like we kind of keep you up to date with Midjourney. We talk through sort of like the changes, the updates, the things that are coming.

0:01:0855 episodes in: what Midjourney Fast Hours is really about

Play from 0:01:24 8:88:88 0:01:24 And then we also couple that with sort of like the broader picture of things happening in the AI visual space. But just a reminder, there are some super, super deep dives. in the earlier episodes where we go into like the thick of it, where we go into levels that not a lot of people go into in these specific, you know, all the parameters, how to use each one, micro decisions that we make, know, prompt theory and structure, like so many good things.

Play from 0:01:54 8:88:88 0:01:54 So just, I just want to throw that out there. Reminder that those things are still out there, still very, very relevant. So go ahead. We should also say that when we started doing this, there wasn't a new major release from every tool every five days. It was like we could go a couple weeks without something really happening and we'd like touch on it. But that was when we were that's when we were deep diving. And it's like now this is this is like real time what it's like with Drew and I trying to keep up with everything while we do this professionally.

Play from 0:02:24 8:88:88 0:02:24 And then also like continue to bring content here because there's so much to talk about. it's like we used to just be like, we might go six weeks without something major. Now there's no, you don't go six weeks, not anything, you don't go six days without something major. But, and everything got good. Like that was the thing. was like other stuff was, everything was mediocre. Now it's good. A lot more people talking about it. mean, like at first it was just a select few people talking about it.

Play from 0:02:52 8:88:88 0:02:52 mean, all those things were true. I think, uh you know, as in tradition, we always like to lean into Midjourney first, you know, so maybe we start there. There's not a lot of changes or updates since the last time we talked. um Rory, something, and maybe we just touch on kind of like what happened with office hours and then what's coming next week. Cause they did say there were some things coming next week. And then. We kind of zoom out.

Play from 0:03:17 8:88:88 0:03:17 We can talk about some of the other things that are happening in the mix so. ah let's do it. Let's see what's going on Midjourney verse. Let's do it. So office hours this week, little bit different. So in case you guys ever want to check out office hours, this happens every week on Wednesday. I think it's noon Pacific Time 3 Eastern Standard Time. I don't know where in the world you are, so just kind of adjust accordingly.

0:03:25Midjourney Office Hours recap: dev updates, bugs, and feature requests

Play from 0:03:44 8:88:88 0:03:44 But uh that's available and open to anyone. If you want to get in there and access that, they're streaming that both on X through the Midjourney account, and they're also doing it in Discord. for the OGs. So you can access it either way. This week was a little bit different though, because they invited like their developers on and we're leading it. And so it was a little bit different of a format, a little bit different questions coming in, a lot of people asking and just kind of suggesting bug fixes or do this, could you do that, blah, blah, blah.

Play from 0:04:15 8:88:88 0:04:15 One of mine that I've been thinking about a lot lately is just like an easy way to like, photos in the create screen. So like if I'm just creating something. You can right click and like, but honestly, this is going to sound very like complaining of me. That's too much work. I don't want to right click and then do that.

Play from 0:04:42 8:88:88 0:04:42 Like I just want to be able to just click, like click, like boom, boom, boom, boom, boom. Cause a lot of times you'll find yourself in this sort of screen here. And I love that they, I love when they added this, you know, subtle and strong variation here. That was awesome. But I think, you know, just adding that would be a huge, huge fix, but there were a lot of like things like that. People asking about, could you do this? Could you do that? The developers were making notes of those things. uh There's really not much else to say besides that.

0:05:02Multiple feature drops teased for next week

Play from 0:05:10 8:88:88 0:05:10 I would say maybe the one thing Rory that was interesting for us and the audience is this one right here where we're just talking about active development of web profiles in the new style tool creator system. No immediate releases expected this week. So was on Wednesday. Multiple feature drops and updates anticipated for next week. Okay, so next week sounds like a week where we might have something we'll be able to talk to talk about it and share with you guys next week on the podcast So that's kind of where we're at there.

Play from 0:05:44 8:88:88 0:05:44 I mean like very very short um The office hours just was very you know question voice oriented very conversational with the dev team so it wasn't your typical uh Sort of office hours that you might get with with the depth and with David and sort of like the more I would say like interesting, interesting topics. was more or less like what's on your mind? What have you been wanting? What's not working well? Blah, blah, blah, blah.

Play from 0:06:10 8:88:88 0:06:10 That's good because they've been, I feel like the last release was really like Style Creator. mean, or Style um Explorer. It feels like a while ago. It was probably like a week ago. But that's, um you know, they were on like a rapid pace after V7 and it was something every week and they were being talked about all the time. So I think it's probably time that they kick that back up because it's got to be, they've been working on something, right?

0:06:39v7 release timing + hopes for a true creative studio UI

Play from 0:06:39 8:88:88 0:06:39 Have to at least, because it's been a while since when was V7 released? I don't even remember. It was probably over the summer at some point. So June, something like that? Maybe? No. it only been June? I feel like it's been longer than that. When was Midjourney V7 released? I don't know if you're using any of the AI browsers.

Play from 0:07:06 8:88:88 0:07:06 I really like Comet now for April 3rd, 25. Okay, so was April. Okay. Yeah, I mean, so I mean, if the if they don't do seven one and they go to VA by the time the eight guy would be basically 10 months. Yeah. I mean, they've added a bunch of stuff. the way they were moving before. They've added a bunch of stuff, right?

Play from 0:07:31 8:88:88 0:07:31 Like it is what those micro updates that kept us happy for a long time. You know, there's a few things. I feel like I'm confused about what comes next. I'm probably the most interested outside of V8 in the new UX UI. Like how much is that going to change? Am I going to have to relearn a lot of stuff or is it going to be relatively simple?

Play from 0:07:57 8:88:88 0:07:57 They've never had a problem with how they've done it, but... just curious of what that's actually going to look like. Cause there is a lot of stuff now. It says this does feel like a, like there's other tools where it's just like you prompt and hit enter. And this one is like, it feels like a whole creative studio in here now, which isn't a bad thing. I just, I just, I'm just curious how they're going to organize it. Yep. Yeah, I think I'm super, super curious about the UI stuff. I mean, because, you know, a lot's changed in terms of like a workflow with how people are generating images now, even videos, right?

Play from 0:08:32 8:88:88 0:08:32 So how does Midjourney adapt to that? And typically when they touch something, they'll give you like a V1, but then they won't come back to it for a while, if ever. You know, so I hope it's good enough. right to accommodate for more of like the workflow efficiency connecting the dots without you know jumping over to different tabs or screens and this and that and the other or even tools and you know just kind of pushes pushes things forward in that next phase for us.

0:09:02Wishlist: multi-character control, angles, expressions, and seed editing

Play from 0:09:03 8:88:88 0:09:03 So here's what's interesting, right? What has Rory been complaining about for the last couple of weeks on this podcast with Midjourney is that number one, we just need something that's like Nano Banana. If you look at this leaderboard here, how many of those problems are solved by a tool like that? have gray multiple characters. up in one of the questions and that was one of my questions the previous week on office hours too. So I really hope.

Play from 0:09:29 8:88:88 0:09:29 That that is the case, because you're right, right? Look at this. angles of a character, pose of a character, coherent or improve as a multiple consistent characters, more advanced facial expression controls, multiple consistent characters at the same time, all things that nano, seed edit, reave, ChatGPT even which not as quality but like flux context, like all these things that they can do already.

Play from 0:09:54 8:88:88 0:09:54 So that's what everyone wants. So just build it. If anyone's listening, not just... Now, like I'm saying, just go build it in five seconds. It's just like, that's what the market wants. That's what we all want. It's very, it's a very simple, cause it's not like, oh, it doesn't have to be like the anti-marketing thing. Like we don't want marketing and businesses in here. It's just like, people want to tell stories. They want to be able to play with their characters and they want to play with their environments and they want to be able to do it.

Play from 0:10:22 8:88:88 0:10:22 Not jerry-rigged. Yeah. Yeah. So. totally control that. I forget this is in here. I need to go back in and start rating some of these things again. It's been a long time. um This is in tasks. Anybody can access this. Scroll down, community rate ideas. on. I was going to say that's a world building in there. I saw a little thing that said world building because I was thinking about that.

Play from 0:10:47 8:88:88 0:10:47 Remember, remember that release? I forget even forget when it was like their little was like the forget what they even called it. It was like patchwork or something, maybe. Yeah, like it's like that just disappeared. That was like a quick. Yeah, that was a quick flash in the pan, but it was on the same track. It was on the right track, right? Like that's where everyone's ending up now, right? Is that sort of like one step to the next to the next in a visual space or a canvas, right?

Play from 0:11:15 8:88:88 0:11:15 Like that's sort of like the idea. We'll touch on that later. Yeah, I think like some of this stuff too is I feel like. Right, like some of this stuff is already doable. It's just the problem is it's not very fun and easy or accessible to do, right? um Input of scene images to create images in that scene. Sounds like a nano thing.

Play from 0:11:41 8:88:88 0:11:41 Add more events. mean, you can get great facial expressions if you actually do it. Most people don't do it. ah I just did a whole post on that. People just don't do facial expressions. I never see it. But they can be done. I mean, it's not perfect. I will say that there are some that just don't work that you would hope would work. But there are plenty that do work.

Play from 0:12:08 8:88:88 0:12:08 uh You know, so I don't know if I can maybe show this really quickly, but that's always one that I'm focused on, which is, is that in here or here? I think it's in this one. Expression, there you go. can get these all day, right? Like it's just people don't do it.

0:12:29Prompting real facial expressions (test simple → build complexity)

Play from 0:12:34 8:88:88 0:12:34 People don't do it. And if you don't have these ideas, just go, you know, a couple prompts with an LLM. You can get plenty of ideas to test and then test which ones work right? Like keep the prompt maybe simple at first just to see if it's going to get it. And then you can that gives you then the the freedom or the sort of accessibility to be like, OK, now I'm going to like work this into a longer prompt because I do know that it.

Play from 0:13:00 8:88:88 0:13:00 So that would be my advice when you're testing any of this stuff is like start with a very simple prompt coupled with what you're testing. See if it works first, then you can sort of play with the fact that it can be included in a longer, more detailed, or more complex prompt. Totally. 100%. I think that's, like I said, like if you look at all these things that people want also, it's just like, I want to be able to save an image that like, not save in terms of like download.

Play from 0:13:30 8:88:88 0:13:30 Like if I like an image, but it didn't have the facial expression, right? Like I really like it. Like I just want to be able to edit that exact image. I don't want to, like I don't want to go and try to recreate it and get it and run Gen over and Gen over and Gen over to get the same thing. It's like, I just want to do what I do with Nano or, you know. SeedEdit or any those other ones, Seedream. Just like, do something with this exact thing. Don't reinvent the wheel.

Play from 0:13:55 8:88:88 0:13:55 I don't want to reinvent the wheel. Just want to change the facial expression of this person. That's it. Change the color of this car. That's it. know, that's... It sounds very snotty of me. Yeah, it is. remember we were maybe we were force feeding things in remember we would go like into you know, you'd go into like an image and you know, you'd be like, okay, you know, let me let me edit this and you know, image reference in a red jacket and then like yeah gonna say like image reference it in or whatever you got to do text reference plus the image right like Wait it, you know and play around with the different weights.

Play from 0:14:40 8:88:88 0:14:40 I mean, it's just like it's time man. It's time, you know I like the precision control. I like that we have all the parameters and all the things. I just want for certain things, where it's just like, make that guy smile instead of... uh You know what I mean? fly adjustments. I would say that's gotta be the number one, right? That's gotta be the number one thing you would want right now.

Play from 0:15:07 8:88:88 0:15:07 Yeah. like, why do I need to go anywhere else? Right? Like that's it. Just take this, do this with it. That's all. I also will say this. If you need more Fast Hours, don't sleep on this. mean, go into tasks. I would say you could rank these new style grids or images. Do like 50 to 100 and you're going to get a free hour.

0:15:13Pro tip: community rating = free Fast Hours learning

Play from 0:15:35 8:88:88 0:15:35 Easy. mean, how simple is that? You know what I mean? Don't sleep on that either. It's just like, boom, okay, go through and do that. If you're forgetting these things, it's just important that a lot of people stop doing this. So you don't really need to do that many in order to get these free hours. So maybe just something to keep in mind there too. um Mike, my complaining is done for today, by the way, that was just me.

Play from 0:16:01 8:88:88 0:16:01 I had to I had to get it out. know what? I'm not complaining too much because after that last week and figuring out that I can use my personalization codes from V6 on V7, I haven't looked back, brother. I haven't looked back. Ooh. looking for it. I like it. That was a good one. That was a good unlock. And I have been using my old ones again. And I'm like, thank you.

0:16:02Hack: using v6 personalization codes inside v7

Play from 0:16:27 8:88:88 0:16:27 Thank you for this wonderful gift that you've provided to me. Well, you want to zoom out a little bit too? I don't know if there's much else. Is there anything else that you want to talk about? You know, maybe really, really specifically into Midjourney remind anybody of? I mean, there's nothing really new here at. Hmm, no, but I think let's zoom out a little bit. like that phrasing.

Play from 0:16:57 8:88:88 0:16:57 You need a sound effect for that. Yeah, man, where you want to start? Because I know that we were talking about a few things before we jumped on here. know, maybe we want to start with the based phenomena. Let's talk about the node revolution as to what's been going on. think it's, I think, you know, going back. the emancipation proclamation of the node based phenomenons going on?

0:17:12The Node Revolution begins — why nodes fix creative workflow pain

Play from 0:17:26 8:88:88 0:17:26 It's sort of like what I've always wanted, but never knew how to ask for it. Because it's just, I can use everything all at once in the same space. And I just see it. I don't know. I like mid journeys. In regards to other platforms, Midjourney versus others, I like creating in Midjourney the best. From just a pure usability standpoint.

Play from 0:17:51 8:88:88 0:17:51 If I'm doing one thing, is images and videos. Just because it's easy, it's fast, I know it really well. um Nodes is like, I can see it all, which is what I like to. I don't have to scroll through to find stuff. It's all there. It's like, I'm putting all my ingredients on the table and then making a meal with it. That's why I personally like it, but also it's just the branching realities of things, plus sort of history tracking, if that makes sense, going back to things and it's not...

Play from 0:18:25 8:88:88 0:18:25 If anyone's familiar with designing in Photoshop, right? Like you could design one image, right? And you do work on one image at a time. And then when you had to like go, you wanted to go test something, you had to like, you know, wrap up all your design layers and do a folder, duplicate the folder, know, un- you know, un-visualize it and then, you know, work on the new thing. And if you didn't like, you had to unclick that and go. It's just like a really clunky process to like, to test different things.

Play from 0:18:53 8:88:88 0:18:53 So yeah, I like dragging some nodes. and saying like, go this way, this way, this way, this way, this way. If they all suck, I can just delete them and go back and start from that image again. Right. So I think it's just like people, think it's very familiar to people, like in terms of like how they actually create, like visually, it's like, here's where I started. Here's where I go. Here's where I keep going. let me try this. No, that didn't work. Like that's, that's sort of like the creation process in general.

Play from 0:19:20 8:88:88 0:19:20 never just go like, here's my paragraph, enter, go. That's it. So. is, I mean, this is something that we were, we've been talking about a little bit now. We've seen this mass like rush to implement this into, you know, existing tools, right? Like runway in there, you know, like, backing up, right? This was comfy UI by itself or seemingly by itself for a long time, but you know, it was just so intimidating and so complex, probably overkill for a lot of people, very like intimidating in the sense that it wasn't really early adopted by this larger base yet.

Play from 0:20:05 8:88:88 0:20:05 So it just kind of felt like this niche thing and a different way to do it, but didn't feel necessary at the time, at least from my perspective. I'm just going to speak about my perspective. And then you started to get these other locally too. That was the other thing was early days needed to run it locally. So you also had to have the hardware to do it. So it wasn't like, you know, cloud-based or like you had to have like a decent like a, Mac could not handle it back in the day. So my, my MacBook pro.

Play from 0:20:31 8:88:88 0:20:31 Yeah. had right the floor as the we've ease right that then I felt like made it way more accessible, less intimidating, lowered the learning curve a little bit like still learning curve, but like enough to get more people in there and experimenting and trying right? Because there I think there is like this barrier that if it's too high, people just you got to be in this right mindset.

Play from 0:20:58 8:88:88 0:20:58 to go over and knock down the barrier. And if it's too high, people aren't gonna come back. You know what I mean? And so like, feel like with Flora and Weevy and like tools like that, they did a really good job of making that accessible. And then they had people that were on it early like you that were really like just breaking it down, sharing how you can sort of like make this a workflow and what that actually means in terms of like scaling and efficiency and blah, blah, blah, right?

Play from 0:21:24 8:88:88 0:21:24 Not having to have, again, like to maybe like CREA's credit, the idea that you could then aggregate all these tools in one place and you don't have to have all these separate subscriptions, right? And this can be a very costly thing. like there's been this whole evolution. Now you've got, like I mentioned, Runway's got playing this game now. Adobe, we talked about last week, playing the game now. FreePic is playing the game now, which I love FreePic.

Play from 0:21:51 8:88:88 0:21:51 uh Then you've got CREA. uh in beta right now for this. And Rory, this could be fun maybe because I haven't been in Korea yet, but I do have access to the beta. So maybe like, maybe I could share as we walk through this and then you can sort of like narrate uh some of what you've seen so far or maybe even share what you've got. I have some I have some good stuff. It's, you know, also just to like, before we get into this, I wanted to talk to I just want to say something like overall general, I feel like the AI it was like this is like an organic sort of like way for everything to come together if that makes sense, right?

0:22:12Krea Nodes deep dive: blueprints, drag-to-wire, product-swap demo

Play from 0:22:30 8:88:88 0:22:30 Like where early on, there was a few tools that we were like, sort of that were like fighting for position, which was okay, if you like image generation, it was like Midjourney or stable diffusion or Dolly, right? So you had to like buy the subscription. And then it was like, everyone went to that model, right? And everyone's like, okay, we need subscription subscription. I have to have one to runway and Midjourney. Now I have to have one to runway Midjourney.

Play from 0:22:55 8:88:88 0:22:55 And, you know, Photoshop for Jen Phil, right, like going back to those days. Then it just started to explode. And there was so many tools, and they were all good at doing different things that I think probably this is just a this is my guess you tell me how you think about this from a business standpoint, that Everyone's subscription started to shrink because they were spreading out to a more broader, like it wasn't like the subscription numbers, daily active users, kind of things like that. It gets broader.

Play from 0:23:21 8:88:88 0:23:21 So then these tools, what do they do? They start releasing APIs so that you can get like the residuals like, oh, if I'm not going to get subscription, then like, wherever people are using the APIs, I'll get the API hit money. Then everyone had an API. So then you see like the rise of tools like CREA, FreePick, Higgs Field. all these tools to like aggregators that people don't have to have a subscription to every tool, they can have one subscription and use them all.

Play from 0:23:46 8:88:88 0:23:46 Right. Then like it just became like, what's the easiest way to use every single tool? And it feels like that birthed like, okay, nodes, use them all in one space, you can use every single one, you just do it the way you want to do it. And like, it doesn't have to be some prepackaged things. So that's why, to me, it's like this natural progression of how it gets there. And it's all sort of based on our problems, like Too many subscriptions, too many tools.

Play from 0:24:12 8:88:88 0:24:12 Right? How do you solve that? You put them all in one tool and then you make them all accessible in the same space without having to make it complicated. that I just wanted to, I was thinking about this the other day as to how we got there. So yeah. think I don't really have much to add, but I think you pretty much nailed it. The thing is that I'm also thinking about is you don't have to learn all these different website UI UXs, which is also a problem.

Play from 0:24:39 8:88:88 0:24:39 I mean, for me, it's like, I use tech every day, I'm in new tool, but that's also an exhausting process to figure out how things work on each different platform. And then if I haven't been there for a while, have things changed and moved around? Like, that's a problem. So this solves that problem. You talked about this, we've already, you you and I have already talked about the subscription costs, the tab hopping. This also then, I think allows now, if you don't have this, you're sacrificing uh memorability and stickiness with the product.

Play from 0:25:12 8:88:88 0:25:12 You almost have to have this now to be a sticky product because... with the velocity of the changes that come through here, I mean, you'll be forgotten instantly if you don't have this, because this is going to be that way again, to just access all the tools without the UI, know, like white labeled tools essentially in the same place. You don't have to log in. What's my password here? Do I have it saved?

Play from 0:25:37 8:88:88 0:25:37 Oh my God, like I need to up my credits, blah, blah, blah. Did something move? I can't find it. Like you don't have to worry about any of that shit. You know what I mean? So this is... this is really interesting. then I think um just the last piece of it, Just everybody knows this, but it's like every tool is good at something different. This solves that problem, right? So, yeah. It makes us tool agnostic. That's sort of where I think it goes is right. You don't have to. It isn't like I, what should I choose between this tool A or tool B and like which subscription, which one's better for what I needed for it's like no more of that.

Play from 0:26:13 8:88:88 0:26:13 You just use whatever you have to use to get it done. Yeah. Yeah. So like, and CREA to their credit was onto this really early. They were just like, okay, we have our real time editor and a few other things. Then it's like, well, let's just start bringing everything else in. And they made it easy. And it was like, to me, they sort of like started the snowball effect here with the aggregator sort of sense. But I'm, really liking, uh, I'm really liking their nodes.

Play from 0:26:40 8:88:88 0:26:40 One thing to know about Korea nodes, um, that I will say is that they can do batch processing, which I think is a differentiator now in the node space. If you don't know what I'm talking about, meaning like, and maybe I'm not calling it the right thing, I just call it batch processing. Like if when you run in nodes, something like things like Flora and things like FreePick, um they're super pretty, usable, easy.

Play from 0:27:05 8:88:88 0:27:05 The one thing they can't do is batch processing, meaning I give it one prompt, it creates 10 more prompts, and then goes and creates 10 separate images. With things like Flora and FreePick, you can give it... one, you can only give it one prompt and then it can go create 10 images of that same prompt. can't do variation. So, cause there's no splitting or like splitting function to actually like, if you had an LLM, for example, like create 10 prompts, you can't like go into that LLM box and then create 10 different.

Play from 0:27:40 8:88:88 0:27:40 mean, you could technically in theory go and have something that goes and creates a prompt and then makes them all different. But like, there's no sort of cohesiveness in terms of like the LLM. complicated. You'll see we'll go through I'll show you what I mean. Sorry for that, that complex rant but um it's good man. This is good because I think like you know we it's time we did a deep dive on nodes. I think this episode is going to be.

Play from 0:28:06 8:88:88 0:28:06 Really depth focused on this node piece, right? So backing up to this screen there are uh codes to get into the beta, so I would say maybe go on X. Follow Korea. I think they were throwing around a bunch of codes. I was just looking. ah So I think like if you're looking for something, maybe do that. And then just like Weevy did, just like FreePic did, which we'll show too, you've already got some nice little, they're calling them blueprints or workflows that are already kind of set up for you.

Play from 0:28:44 8:88:88 0:28:44 So ah my experience in learning this, couple two things. Of course I reached out to my guy Rory who spent uh like an hour with me kind of like walking me through the basics. But one of the other great ways to learn this stuff, well, I'll give you a couple different ways to learn this stuff. But one of the best ways to learn this stuff is to click into these and just like reverse engineer them in your mind, right?

Play from 0:29:09 8:88:88 0:29:09 Like actually break down what's going on here, right? So like, let me zoom in on this. Okay, so we've got like an original product photo. They just added this photo in. Now let's like see what they're doing inside of here. so, you know, they're basically taking this, well, they created this photo. It looks like in CREA. ah Here's the prompt for that, right?

Play from 0:29:37 8:88:88 0:29:37 And they did this in, you know, in nodes here. So right here, ah this is my first time looking at it. It looks like you've got some other settings here. know, aspect ratio, style, blah, blah, blah. ah Here's the option to run the node. So you can run that. Yeah, you can also run the whole thing, which basically connects all of those dots. But then they generated this, right? Then they basically opened up another thing.

Play from 0:30:03 8:88:88 0:30:03 This is Nano Banana. You can see all the sort of like models here. And it's taking this, replace the sneakers with the teal colored ones, right? So then it took this original image, connected that over here, connected this here. So it's taking the pose, the background, everything about it, and it's just swapping out the shoes. Then it's connecting this one again, add a white title in a thin old school serif font saying run with style, just connecting the dots.

Play from 0:30:33 8:88:88 0:30:33 All it is is this dragging here to here. So that's all that's happening here. And then you have these other settings available here. This does look pretty sharp. I'm excited to like jump into this a little bit further. And then they did this nice little before and after thing, which is pretty cool. So they did. right here for the, for the upscalers. I like, I like the slider in the node.

Play from 0:30:59 8:88:88 0:30:59 That's good. It's good. little, good little UX UI sort of a piece right there. It's just not available in any other tool. So that's a cool little one. mean, that's pretty sweet. then like, let me see if I can, how do I, here we go. Okay, so then we go here. What do we do here? We took this, isolate the sneakers, separate them, make them look like they're floating in the air.

Play from 0:31:25 8:88:88 0:31:25 Okay, so this is like text prompt coupled with these shoes. They took this, they said, uh elegant product, they have for two sneakers floating in mid air, sneakers should fly into the air. Okay, so this is a video. So we can click this. Boom, right? And you're just connected. Like it's just a simple connection, right? So if you haven't used this yet, you're just dragging this here, you're going into settings, you're selecting the duration, blah, blah, blah.

0:31:39Image-to-video inside Krea: turning stills into motion

Play from 0:31:53 8:88:88 0:31:53 And I would say even here, like I just right clicked, this is where you would add, you know, what you're gonna use for this. So it looks like they've broken this out by image, video, et cetera, right? So they just did this, or maybe that's enhanced. So that's not actually it. That's for upscaling. Let me see. You can also search through the nodes. You can just type in there too, up at the top, whatever you're looking for. Good one, easy. we got Hicksfield in here for example, maybe.

Play from 0:32:21 8:88:88 0:32:21 No. Nope, they don't have Higgs field. Okay, they do have, yeah. Okay, so I mean like you can select any of these, they have SOAR 2 in here and blah, blah. You would select that, that's gonna give you this box and then you're gonna have the same option, right, with the settings. Based on the model you choose, it might have slightly different settings. But essentially you're just connecting these things. And then Rory, you know, like what makes this so great, and I know you can speak to this, which is just the ability to then change something very quickly and have it affect.

Play from 0:32:53 8:88:88 0:32:53 You know the entire workflow for scalability and efficiency and maybe this isn't the exact best uh Example here, right? But it's essentially like hey Maybe I just want to swap out these shoes at the beginning and maybe it does 80 % of the work for me I might have to go in and adjust like a few quick things Maybe it's I want a different color for example or whatever on the background, but then it's like I can I can do all that What's really cool though are the deeper examples which you and I have talked about, which is like, you're getting into something where you're talking about all the different camera angles too, right?

Play from 0:33:30 8:88:88 0:33:30 Where you're getting these new shots and blah, blah, blah, blah, blah. So. The one and then one thing to think about with like workflow building, right? When you look at this one, this workflow is very specific to this individual use case with these pair of shoes. So like if you want to go and do this just to be like project wise, like I think there's there's two ways to think about this, like individual projects and then like building workflows.

Play from 0:33:55 8:88:88 0:33:55 Like this is an individual project, right? You can't just drop in a different pair of sneakers and the whole thing runs how it should. Like a workflow would be like built a little bit more generic, if that makes sense. So I drop in any pair of shoes and then it does this, these steps, regardless of what the shoe looks like. So, you know, where they're saying like teal background and run with style, like those are all individual to this specific image. So if you're building a tool, you'd have to think about it a little bit differently, like a workflow to be utilized over and over again, if that makes sense, because it can't be so specific.

Play from 0:34:28 8:88:88 0:34:28 It has to be generic, right? like flexible and versatile enough. So instead of teal shoes, it's just like, you're just saying shoes, right? And then maybe you've got another node box where it's just, maybe that's where you would change the color, right? Or whatever it is. Or maybe that goes back all the way to maybe the system prompt that you have at the beginning that has it describe everything that's there based on that particular.

Play from 0:34:57 8:88:88 0:34:57 output or outcome that you're looking for. And then that makes the rest of the workflow a little bit more automated, right? Like that's what I've seen you do in the past too. so that's that's where we get into and I can show I built I went nuts on Korea for like 25 minutes last night built just one the one that's are just the Korea examples. Let me stop sharing here. Yeah, this was this was me like the first thing I went to with Korea. I was like, let me let me see if I can do batch processing, which I can.

0:35:04Batch-generation magic: LLM → 10 prompts → parallel image runs

Play from 0:35:27 8:88:88 0:35:27 So I was pretty amped about that. Delete this and this here. But like this is me like, I can take one image and then just create whatever from it. Right. I wanted to just I wanted to auto prompt it. And then I have actually this is a very unique feature for the video side of things, which I was liking to. But essentially, because you have this, it's called an LLM call, meaning you can just use an LLM.

Play from 0:35:55 8:88:88 0:35:55 ah But as you'll see in the LLM call, basically going to generate, it's going to use a custom GPT. I'm giving it a prompt, I'm giving it the image input, and then I have the system prompt, which is basically saying, analyze the image, try to create the image as close as possible, make it as detailed as possible with a... 200, 300 word range written in natural language paragraph.

Play from 0:36:20 8:88:88 0:36:20 So up here, once I run this, it'll create 10 prompts for this car. See how generic it is? I'm just saying 10 prompts for this image in action situations from different compositions. I could use this for any car now, because I'm not saying which specific car it is. I'm just using it as car situations, compositions, and then it's coming up here, like low angle tracking shot.

Play from 0:36:45 8:88:88 0:36:45 you know, Twilight, Tuge Drift, 100%. That makes it really scalable, right? So that's like the, that's the whole, what I'm talking about, about making it generic. But this is the, this is the key piece, right? This is what Weeby has, this line splitter, which essentially what it's saying is like every time there's a line break, like every time, okay, we have prompt one. And then let me see if I can expand this, can I?

Play from 0:37:13 8:88:88 0:37:13 I don't think I can. uh So like after prompt one, you see there's a paragraph and then there's a line break, right? So that's two. So that's how it knows with the line splitter what to split. So it's like, okay, prompt one, then, all right, break. And then prompt two is a different one. So, you know, I have this running to 10 different prompts. And then I actually ran these. I don't know if you've played around with, I really like Creo one. It's not like it's not going to translate this one to one, but I do like some of the Creo one aesthetic.

Play from 0:37:44 8:88:88 0:37:44 So I just ran it to Korea 1 just because I wanted to check, know, just want to play with Korea 1 for a little bit to give it like basically image prompting, nothing like not like using the same subject here. So these nine were Korea. And then I used SeedDream to actually place the car in different situations, just running these and getting different examples. Right. So all of these are attached.

Play from 0:38:11 8:88:88 0:38:11 Just I can I can do this all in one click now if I wanted to just. gonna say, you mind running, show these people too, if they've never done this, how do you then run all these images again? How do you create new variations of these really quickly? Yeah, so if I wanted to run this whole prompt again, I just go and select the whole thing.

Play from 0:38:37 8:88:88 0:38:37 And then see how it just says run nodes up here. Run nodes and hopefully this works when we're on live. I don't know if it will, but let's see. like those all of those images now that he's got there are generating or they should be. That's the thought and it's possible. I think you're doing that right. I would be my guess, but that's essentially what we're doing here. mean this is so going back to that really important point that you mentioned at the beginning, which is like making you're going to have maybe specific use cases to you, but if you make things a little bit versatile upfront.

Play from 0:39:14 8:88:88 0:39:14 Now all of sudden, don't, know, Rory doesn't have to go back and look at every single box, you know, and update text and this and that. Now we know that it's versatile and flexible enough and specific enough, yet vague enough to then just run this. And then hypothetically, right, I don't know how many images are here. Let me see, six, 24 images maybe. If I can still do math, which I was never good at math, but here we are. And you're not gonna get 24 new images, right?

Play from 0:39:41 8:88:88 0:39:41 And... You don't lose the other images that you had. They're in there, they're scrollable. like Rory zoom into one. So you see five of five, those are all things he's already run in that. So you can try and test, hey, can I get a better image than this? And blah, blah. You don't lose what you've already created there. So it's there. So this is like me running.

Play from 0:40:07 8:88:88 0:40:07 So this is like mid-journey running permutations, right? Almost, but like at scale with much more differentiation, right? So like that's sort of like the idea of here, but this is the, this is the cool part. This is again, another, I think you can probably do this in Comfy, but you can't do it at most of the other tools. So here it's like, I'm using the same sort of system prompt, except I'm saying create video prompts. And then I'm attaching the images for, you know, it can add up to five images here.

Play from 0:40:33 8:88:88 0:40:33 So I attached four of them. So it's taking those four images individually, writing four prompts for them, and then creating videos automated, right, based on the image itself and the system prompt. So system prompts basically saying, look at the image, write the prompt. that's basically it, right? So, you these are automated, so you never know what you're going to get. I didn't look at any of them before I ran it, but you can see, like, some of them, not bad.

Play from 0:41:01 8:88:88 0:41:01 Like this one's pretty... Is this a van? Is this a vans? Style is that what I'm picking up on these vans? Okay. I just picked this. I saw this car on Pinterest. I was like, that's cool. I like the way that looks. Let me just go play with it. But know, too, one last point on that when you were showing the videos, you know, it said that you could only connect five there, right? But like, you could then just go back and clone, duplicate it and connect five more, right guys?

Play from 0:41:32 8:88:88 0:41:32 So like that's how this stuff works. You're not, you don't have to then go and change all those again. You just copy, clone that whole little section, which Rory has shown you already where you can just drag. clone it and then just reconnect the new nodes, right, to the other five that you want. And all of a sudden you went from five videos to 10 videos, right? Like all of this is sort of built out. So this looks intimidating, right? Like this is very intimidating, this whole node situation.

Play from 0:42:00 8:88:88 0:42:00 I built it in Weevy and it can be very, you know, it'll be much stripped down, right? Like this is the same workflow, same system prompt, same everything. What's going on here is you have the Image, Your Prompt, System Prompt, same way it is here as Image, Prompt, System Prompt, right? Once it runs, as you can see, like in the LLM here, same thing, it's just going to create 10 prompts for us.

Play from 0:42:27 8:88:88 0:42:27 Then this goes to the array, which is, you know, split each one of them based on, you know, individual prompts. They also have this thing called a text iterator in Weeby, which is like, instead of doing it by the list format, you don't have to go and break out like into each individual. Like if I wanted to do like list selector, like if I want to do this, I don't have to go and, you know, create 10 of these to have 10 different prompts. You can just run it into one of you can just run it into one node and the well generate.

Play from 0:42:55 8:88:88 0:42:55 Right. So if I just go, you know, hit run, we'll start to generate the 10 prompts that are already here. So this can be stripped back a lot. Right. Which is which I think is. maybe it doesn't look so overwhelming. Yeah. because then this is like if I'm building an app, right, like there's something to share with someone that they're just going to use it this way. Right. Like this is so much easier now than than doing it that way, because I can let me just lock this up here.

0:43:07Weavy “app view” — simplified node interface for creators

Play from 0:43:24 8:88:88 0:43:24 We'll make this app a little bit more visually aesthetic for everyone. So like this is what you can do on Weeby that the other things, you other tools you can use on comfy. But like, again, I just hit run. And then you're going to see it here. 20 are just going to happen at the same time, just based on sort of what I asked for. And we'll get decent stuff that comes out of this basis. The same exact, same exact prompts from, uh, from what's it called from Korea.

Play from 0:43:53 8:88:88 0:43:53 It's just a, it's, it's looks at it now. Now feel like this is what it comes down to, right? It's like who makes things look the best and who makes things the simplest? Cause if everyone's got everything and you can do every tool in on node-based create, then it's like, who makes the system the easiest and the cleanest? The easiest, like the most simple to use. So this is what I love about this thing. You know, it's just, it's just awesome.

Play from 0:44:19 8:88:88 0:44:19 But that's also radically different than FreePick, right? Like FreePick is also different. So have you been, have you tried FreePick's nodes yet? Or spaces? All right, let's... I've got I've got it up. Looks like let's get into it. um Here's me with cowboy hat. I just I had to show you that you know I felt like I felt like you needed to see it. These are fun by the way, dude.

Play from 0:44:44 8:88:88 0:44:44 I mean like you can create any image you want in here. I got even take. I'm even taking some images that I'm doing for like clients and then I'm like hey like. uh What is the problem for this? know, it's like more contrast saturation, more analog film look, right? Like harsh flash type, you know, like, so, I mean, this is possibly something that you could scale in spaces too, or Weevy for that, or Creea for that matter.

Play from 0:45:12 8:88:88 0:45:12 It doesn't matter, because they all have Nano Banana, right? But like, when you, like, this is a very big need too that I see from clients. It's like, hey, look, like this picture looks too perfect. I want it to like, feel like somebody took it. Yeah. what mean? Like that looks significantly different than maybe what I had in there. But let's go to FreePix piece here. um I don't know the best way to get there other than going to all tools down here at the bottom spaces.

Play from 0:45:41 8:88:88 0:45:41 And then I opened it up in this tab. today. The cowboy hat drew might make its way into the thumbnail now that I'm thinking about it. Get my cowboy boots on, I've been rocking the boots lately. It's cowboy boots season, uh Very impressive. Going back to FreePick, hey, loving the templates, loving the templates. I haven't clicked into these, but again, same deal, right? Like we were just talking about the nodes, the workflows, know, the blueprints or whatever Crayo was calling them.

0:45:58Freepik Spaces walkthrough: collaborative canvas + node workflows

Play from 0:46:11 8:88:88 0:46:11 I mean, you got the same thing here. So it's like, hey, is there something here that feels maybe relevant to what you would work on right now? And then again, like going in here and taking a look at how it was built. There's a good template because this is also the differentiator now for a free pick, right? Like you can do a lot more in here, like a canvas style. Like it's not just nodes and creating.

Play from 0:46:39 8:88:88 0:46:39 Like this is more like I can, you know, add text and like design it a little bit. And this can be like a collaborative space, which is also good from the free pick side. Smart feature on their part, smart design for this. this is my first time seeing it. So this is cool. All right. This is, yeah, this just, mean, this is really interesting. Okay, so you've got the nodes here as you zoom in. It's also got some information here.

Play from 0:47:08 8:88:88 0:47:08 Like you could share this as like a, you can share this. can be collaborative. Yeah. This is what I have a feeling. If you're going to try to interpret what Weeby and Figma looks like, this is probably going to be how it looks and feels in a very similar way. Yeah. and connect. Boom, boom, boom. Yeah, it's just, man, I just find it so fascinating to see how each of them kind of put their own little spin on it.

Play from 0:47:40 8:88:88 0:47:40 Super clean. I'm excited to play around with this. I think I made a case for this last week, but I like what FreePic's doing. I mean, I really do. I think like. You know, the image generator thing is, it's nice. know, like the, can go, like I'm on a plan. I think I pay 40 a month and I'm paying month to month.

Play from 0:48:10 8:88:88 0:48:10 I think I'm on this one. So like, obviously it's cheaper if you, if you were to just do an annual upfront payment, 40 a month. And I get unlimited image generation on Nano Banana, for example, or yeah. I didn't know that piece. like so that speeds up the process. Also, when you're using Nano Banana.

0:48:25Quick win: “4 on demand” + unlimited Nano runs in Freepik

Play from 0:48:37 8:88:88 0:48:37 So let's go here. Let me just show something really quickly, but you can also just like go ahead and click four on demand. So every time I'm like. Hey, make this you know instruction. I'm prompting for some iteration I generate. It's going to generate all four. ah Which is also nice, right? Because I get unlimited so I could just hit that a couple times and it's like, OK, if I hit if I got eight iterations of that and it still didn't nail it, I probably need to tweak my prompt.

Play from 0:49:04 8:88:88 0:49:04 I didn't get something right. I need to do something different, right? Also, I mean this is we're kind of getting a little bit tangent here, but like. Let's circle back to this Nano Banana 2 I've been hearing. Rumblings coming this month. So. I'm very excited. I'm wondering if it's going to be like what seed seed. Like I've been really liking Seadream for this stuff. It's like really I think Nano is better at nailing the product consistency and like object consistency and like person consistency.

0:49:18Rumor mill: Nano Banana 2 incoming

Play from 0:49:37 8:88:88 0:49:37 I feel like Seadream is better at like changing angles and like being a little bit more dynamic in terms of like visual aesthetic. But it also can produce in 4K, which is great. So, you know, a lot of times it's really good. But there's definitely some differences. I'm sure Google's right, you know, got something up their sleeve as usual. I mean, Nano Banana 2, probably VO4, maybe end of the year, before end of the year?

Play from 0:50:06 8:88:88 0:50:06 I don't know, I'm just guessing. But that would be crazy velocity, what Google's doing. This shift for that was that's massive. If I could release that that quick, I'm sure they all got it. I'm sure they got nano 2.5 in the pipeline. There's you know, that 0304. So when mid journeys probably working on 8.5 or nine. just different, you know, scenes of the same shoes with a different prompt, right?

0:50:25Seedream vs Nano: angle agility vs object consistency

Play from 0:50:33 8:88:88 0:50:33 So you could, yeah, it's kind of interesting connecting those things. This is like a social media flow, create social media content, not super complicated, but these are just three different prompts, right? Leveraging this as an image reference essentially. And what is this? So they've got copywriting instructions. Yeah. Going into an LLM to then write copy based on the images that are in there.

Play from 0:51:00 8:88:88 0:51:00 So it'll probably write you three pieces of copy based on the instructions analyzing the image. from? That right there. They're just writing it. They're just giving the... yeah. okay. Okay. Okay. I see now. So this is yeah, this is what they're pulling in. But they didn't but this isn't run through an LLM though, is it like this they could have copied and pasted this in here and then That's the nice part is that you can do stuff like this, where it's a little bit more living, breathing sort of design space than just using the AI tools.

Play from 0:51:37 8:88:88 0:51:37 So here, for example, this can stay the same on every single piece of... This can be like a workflow piece. This is the finishing piece, where it stays the same. You just import your copy guidelines, you run it to the LLM, you also run the three images. to the LLM, it'll be like, okay, I'll read the three images or look at the three images and write three pieces of copy. Right? So this is, this is how like we have to start thinking about this stuff because you can do it.

Play from 0:52:08 8:88:88 0:52:08 of it Rory that you want to spend a little bit extra time on to make sure you got something that's a banger, right? Because if you're going to use it time and time again, nothing against this. This is structured really, really well, but it's like I want to give it more than that for the hook. You know, I want to talk a little bit more about what I'm looking for with the hook, you know, and provide a little bit more like specific constraint.

Play from 0:52:35 8:88:88 0:52:35 and direction, but I like visually the way that this is displayed. cause again, we're just comparing, right? Like we're not picking on it. We love all these tools, but like if you go in here, you know, like it's just harder. It's just harder. You know, like same with. with Weevy going to Figma. It's going to be something more that looks like, like Free Pick.

Play from 0:53:02 8:88:88 0:53:02 I would assume Free Pick. You guys win on design so far. display of it is really, really interesting. um This feels like, this whole thing feels like a creative brief, the way that it's laid out, which I think is interesting, it's smart. Can you go into that LLM box that over there on the assistant? just want to see if you can write the last box over there.

Play from 0:53:29 8:88:88 0:53:29 Can you switch the model? Let's see. I'm just interested if it's like running on a. I think I have to change that let me let me bring this Like, that be Claude instead of ChatGPT, or is it just GPT that we're using here? I do want to do your tutorial, but not right now. OK. So let's do. um Let's see.

Play from 0:53:54 8:88:88 0:53:54 of poke around here. Prompt, result, copy, export result, text, settings. Yep, see. you go. Oh, you can use GPT-5. Okay. And you can do a system prompt, though, which is good. isn't that is that surprising to you? Kind of, um, you know, that's, that's also like the benefit of tools like, like Weeby and, you know, they have all of them.

Play from 0:54:22 8:88:88 0:54:22 You can go and it, which is, you know, for copywriting, like Claude to me is my, is my go-to for, for, and if I want something that sounds more human, less, you know, step into timeless style and effortless comfort, which is so ChatGPT that I can't even like, you you see it and you're just like ChatGPT wrote. um But that's, you know, like it's really pretty.

Play from 0:54:48 8:88:88 0:54:48 Let's put it this way. Like it's really well designed. It's there's a lot of different things that mood board one that you were looking at. Like that's all that's all really, really legit. Like I like how you can go in there and make it like a design piece and like share that. And I'm pretty sure it's collaborative in terms of workspace. I don't know if it is or not. That's the guess. But like I can share that. I don't even have to go and externally design that.

Play from 0:55:15 8:88:88 0:55:15 That looks good. That's pretty done. And all that stuff's being added after the fact. That's not just all generated. That text is added individually. Some of those boxes are added individually. um I like this. So, free pick. Well done on this one. I'll show you. Hold on. You want to see our t-shirts that we did last night in about two seconds?

0:55:27Merch detour: Fast Hours T-shirt mockups built with nodes

Play from 0:55:40 8:88:88 0:55:40 please, please. Guys, think we're getting close to merch. I think we're getting close. I mean, I think like we're onto some things. We've got little cute stickers, cool. But I think like the wearables component ah is really what I have in mind when I'm thinking merch.

Play from 0:56:07 8:88:88 0:56:07 And obviously like we looked at Rory's thing last week with. Dart and Scatabo. You know and just kind of putting that on a shirt, bringing it into real life, which was which was dope and then Rory was messing around yesterday with some with some fast hour shirts. So I thought. two seconds to respond to Drew's comment on X. And I was like, let's go back to the beginning here.

Play from 0:56:34 8:88:88 0:56:34 bro, what are the, is? Well, this was, I was like, let's get the base of the shirt, right? This is the other thing that I want to mention about FreePick, which I really like, is that this sounds like a little tiny UX thing, but you can prompt within the node, right? Which is great for just like doing one thing at a time, or you can prompt outside the node and bring the text in, right? So it's, you can do both. That's not something that's available on most, because if you're doing that, right?

Play from 0:57:02 8:88:88 0:57:02 Like it's just, it's a little thing, but it makes a difference. So this was. This was, let's see, that one? No, this one was it. Let's go back one. There we go. So I was like, you know, Drew said, where's our t-shirt? you know, in terms of our design and I was like, screw it. I'll make it in five seconds. So he's using that out here, taking our two pictures, making it like a rap album cover.

Play from 0:57:27 8:88:88 0:57:27 should have put parental advisory on here. Now that I'm looking at it with a low angle. now drop the t-shirt example from as well. And then we had this, and then Drew responded, hey, can we just use our characters, like our caricatures? And I was like, all right, well, let's just run the whole thing over again, the same exact prompts. Right. I know they're cool. I like this.

Play from 0:57:53 8:88:88 0:57:53 This one was funny to me. throw me dude, you throw an explicit content parental advisor on there? It's over. I need to. I should throw it on this t-shirt, right? And then we're like, all right, cool. So we can run this stuff relatively easy. Then I actually had it put it on us. So dumb. But this one, again, I really like some of the things in here.

Play from 0:58:22 8:88:88 0:58:22 I like, again, if I'm using image tools, I'm able to... sort of branch reality here, which I like doing. You know, I have like my standard image. I can, you know, you see that it to get the close up. We can remove the person and then we can get sort of key frames. Right. So we can always have like, like this is what it's easy about doing stuff like this, which is nice because there's oftentimes too many things that are that are missing here.

Play from 0:58:51 8:88:88 0:58:51 Like this is a perfect example. Right. Like taking the image of the car here, but I want the car pulling up. Right. So you remove the car with CData and then have it go in. But it's just so using key frames and like it's so simple in tools like this to do that instead of downloading both uploading to another platform. Like I can also just swap this if I don't like it. know, key frames, same thing here. Like to get like the perfect sort of like still shot of whatever.

Play from 0:59:18 8:88:88 0:59:18 Yeah, that's so. Hey, I really, I really love the uh color grading on that, that classroom school day, dude, that's all Mystic. That's what we've been... But I'm prompting for that, right? So here's a fun one for everyone. I've been using sort of sketches as storyboard images, like prompting sketches and getting this sort of shading and coloring, sort of how I want it in there, and then taking it into Mystic and then using structure reference to then animate it.

0:59:26Sketch-to-cinema using Mystic (Magnific) for structure-reference

Play from 0:59:54 8:88:88 0:59:54 But I'm also... totally working on the color grading, like, consistently sort of doing that. Okay. don't, like you were telling me about that before we jumped on. I don't know. I mean, I'm sure there's somebody maybe doing that. I just, hadn't heard of that. When you're, when you're prompting for sketches, are you doing that and are you using Midjourney for that? I guess that's my first question.

Play from 1:00:20 8:88:88 1:00:20 Now let's, go to, I'm actually using... would love for you to just kind of show that process. I think that's an interesting way because the way that you described it to me was... you were able to get some things maybe mechanically or physically that felt like maybe it would be much harder to do just by prompting for photorealism. yeah, just maybe talk through that. I'm curious, talk me through that.

Play from 1:00:46 8:88:88 1:00:46 now. you guys, whenever you guys have, if you guys have seen my WeeVee workflows before, you see the cleaned up version. This is the non cleaned up version. This is the chaos, right? This is where I'm doing like my workspace in here. So like, for example, wow, this is just a, you guys all, you guys are good to see all the chaos. What I'm doing here for these prompts, right? I have, I'm doing like a stranger things aesthetic.

Play from 1:01:12 8:88:88 1:01:12 So I wanted to sort of get that to come through, but I'm doing. Storyboard sketch, cinematic composition, Stranger Things aesthetic 90, nostalgia film tone, soft graphite shading with selective color wash, dramatic lighting, film grain, whatever. This is what I'm using on every single prompt to create eh the sketch tone, right? And then it's just basically adding the prompt, like whatever's going on in there, right?

Play from 1:01:39 8:88:88 1:01:39 So it's sort of like, that's the one, and then I'm running it for the exact same. you know, over and over and over again. But it's giving like this very consistent sort of like structure and style to the whole thing, which is what I like. Because this is, you know, if I go through here, you'll see it's all pretty much exactly the same. And the character stays fairly consistent with the sketch, like the car stays fairly consistent with the sketch.

Play from 1:02:07 8:88:88 1:02:07 But like this one, I there's a few in here that I really like in terms of the car. If I get through, you know, like that one, like the sketch is good because it provides good structure. Like the light is coming from the right direction. have sort of the, if I was doing this in black and white, I might not get sort of this, you know, the halation and the refraction, right? And like the reflection of the light on the smoke. So I was thinking about this. I'm like, Oh, this provides like, instead of me going to like actually like recreate this image with the same prompt, let me go and use structure reference with a Magnific or Mystic.

Play from 1:02:43 8:88:88 1:02:43 So here's like a few uh tests that I ran with the other tools where I'm like, let's use Reeve. What I'm doing here is, you know, where's the prompt that I'm using this one? Keep the image construction and characters. You know, we want to use, we're to use a specific camera. We want to color grade it with the contrast. We want the color scheme, want the lighting. Yeah, this is a little bit more detailed, but in running those, right?

Play from 1:03:10 8:88:88 1:03:10 Like I sent it to Reeve, which actually didn't do a bad job. Right? Like went there, but it wasn't like visually memorable. That looks like way too desaturated for me. Seed edit or seed dream sort of brought it to life. And I was like, all right, cool. Not, not exactly photo realistic. Gemini didn't really do it that well, but then I ran it to Mystic and I was like, you, that just totally, totally turned it. And these are sort of like the, the parameters I was working on.

Play from 1:03:38 8:88:88 1:03:38 So you have to, this is, if you're using just or Magnific specifically on their website used the reference image of the sketch as a control image or the structure reference. And then was using, you know, control image strength, adherence, HDR, resolution. Like these are the sort of the settings if you guys want to screenshot these, but model realism, Magnific Sharpie, creative detailing around 33. But it was working for just about everything, which I was amped about because I would do Like this one's a great example.

Play from 1:04:12 8:88:88 1:04:12 This is one of my favorites. These two together. Like that sketch to that right there, like picking up on the light, picking up on sort of the, again, like the atmospheric haze and the smoke and then picking that up off the ground. Yeah. And it's really clean. Like I actually, uh, I'll show you this. Where am I in? This is a good, good way to visualize how good this structure is.

Play from 1:04:37 8:88:88 1:04:37 Cause I wanted to just see it. was like, all right, let's see how it goes. Like when it, oops. You can see it just transform. like, all right, that's really good. Like from a structure reference standpoint, it's basically nailing it and just changing the lighting. Right? So that's, I've been looking for something like this. guess I've just been sort of missing on Mystic for structure reference and things, but yeah, dude, there's so much going on.

Play from 1:05:07 8:88:88 1:05:07 I mean, you could, have you tried that in Midjourney? I mean, I know you can't do it in the node based, but. I did it wasn't it wasn't the same like it wasn't really it wasn't as clean as that like it was a like overly saturated or like overly green and like everything was green like it wasn't sort of a rest of the retexture would change the face the retexture would change the clothes so it was kind of like but uh yeah but we just ripped on a bunch of stuff there that was a lot of yeah up. But uh yeah, man, I enjoyed this episode.

1:05:38Wrap-up: what’s next for nodes and upcoming Midjourney updates

Play from 1:05:45 8:88:88 1:05:45 Hopefully this was valuable for you guys too. I know a lot more people are starting to dip their toe in the water with the node based stuff. So, you know, we even got a comment last week on one of the videos asking for it. ah You know, if this is helpful too, I think there's a lot more we can go into here. And I think if this continues to kind of be that thing, right? Like we probably should, we probably should go much deeper in on this.

Play from 1:06:13 8:88:88 1:06:13 So let us know if you agree. In the meantime, Rory, I'm hoping maybe we get a little bit of sprinkle updates from Midjourney between now and next Friday that we can play and experiment with and share with you guys on the next one. Totally. What's everyone, you know, stay sane with your you know, with the introduction of everything that's going to happen over the next two months. Who knows, it's just going to be like rapid fire.

Play from 1:06:41 8:88:88 1:06:41 So we'll try to keep it real here. We'll try to like put in the hard work for you guys. So you don't have to go crazy. We'll try to figure this stuff out for you. But if you're looking at any specific things like that, especially what we talked about today, drop in the comments, right? Node based stuff questions if you want more of that, if you don't want more of that Midjourney stuff, whatever your questions are, always leave them on the If you're still here, if you have questions, just leave it on there. We see it. We respond to it. And we'll answer it in the show, too.

Play from 1:07:07 8:88:88 1:07:07 know, don't don't be scared. Like and comment. Yeah. Definitely. sometime November. We should do a live sesh. think AMA would be good. We've got a lot of different requests, a lot of different questions. So maybe that's something we do too. All right. Well, hey, good to see you, man. Good to see you always. I'm excited for next week and we'll keep things rolling.

1:07:08Tease: live AMA event coming in November

Play from 1:07:33 8:88:88 1:07:33 Thank you guys for joining us. We'll see you guys in the next one. See ya.