← Reshaping Workflows with Dell Pro Precision and NVIDIA RTX PRO GPUs
Reshaping Workflows with Dell Pro Precision and NVIDIA RTX PRO GPUs
How AI + Dell Power Creativity and Efficiency
00:34:44
Filmmaker Ben Saltzman shares how Dell Precision and AI tools like Midjourney and Claude helped him build a game—no coding background required.
Loading summary
Transcript71 lines
- [00:05]
Ben
Welcome to Reshaping Workflows with dell Pro Max PCs and Nvidia, where innovation meets real world impact in high performance computing.
- [00:20]
Logan
Hey, everyone. Welcome back. We have another exciting episode of Reshaping Workflows with Dell Pro Max PCs and Nvidia. I. I'm Logan, your host. Hopefully I haven't worn out my welcome with you yet. This episode's gonna be a little bit different. In some of the previous episodes, you've learned a lot about Delpro Max. You've also learned about the new Nvidia kind of RTX Pro Blackwell architecture, the new GPUs that are launching at GTC. But today we're actually getting into a workflow. So I'm gonna have Ben here in just a second, you know, introduce himself from Aurora Picture Company, who I've been kind of working with on Glimmer, which is his AI video game. So with Ben, without further ado, can you go ahead and introduce yourself? 30 seconds, tell us a little bit about yourself and give us a high level of what you do.
- [01:09]
Ben
So I'm a filmmaker, but I got into AI in a big way and I got. Somehow we connected and I got a super powerful Dell computer and I've been developing an AI video game locally and it's called Glimmer. And basically the premise of Glimmer is that it takes place in New York City with a bunch of supernatural creatures and they're walking around the city talking to one another, and they're also talking to the. The player. So that is everything in a nutshell.
- [01:37]
Logan
I mean, you did, you really did sum it up in like 30 seconds. Like, most people would drag out for like 10 minutes, but. All right, so we did meet. We met on AI on the lot because you're, you know, you're based in la. Let me ask you a quick question. You know, coming from the media entertainment space, you know, with your background, you've been in LA for a while. How is AI kind of perceived in the media and entertainment space overall? I know it's a loaded question, but how is it perceived, let's say a year ago when you were at AI on the lot versus now?
- [02:05]
Ben
It's very hard to tell, like, person to person. Huge, huge varying opinions. But ultimately, what I think about AI is that it supercharges whatever you're doing, so it doesn't replace what you're doing, but it definitely allows you to get further with your ideas. And so I, you know, I mean, in a nutshell, like, I love the AI because I can explore so Many more ideas and get further than I would have without A.I.
- [02:29]
Logan
Couldn'T agree more. I mean, it's really a, A supercharger. It's that accelerant on the fire. Right. That really, like, enhances and accelerates the creative process. You're at AI on the lot. I was as well. You know, kind of some great ideas coming out. But I gotta ask the jump from I gotta. And this is a question we get kind of all the time, and I think you're a perfect person to answer it is that people are always wondering and customers or individuals are always wondering, how can I get started on our. On your AI journey? Like, and a lot of times it's like kind of with mine personally was I just kind of had to take a leap of faith. I spent a lot of time on Reddit. I spent a lot of time just learning different things to be able to do what I can do now. And it was just a time investment. So tell everyone a little bit about, like, I guess the question I'm, I'm asking is how did you get started? You went from kind of media entertainment filmmaker, all things kind of L. A to doing, you know, making an AI kind of non, you know, non player character game. How did you bridge that gap to be able to do that?
- [03:33]
Ben
Yeah, I mean, from a skill set.
- [03:35]
Logan
Skill set standpoint, not a hardware standpoint.
- [03:37]
Ben
Yeah, there's a lot I could say, I mean, like, like during, I think when I got into AI, there was a lot of film work going on, so there was some downtime and, and that doesn't, like, work for me because I, I like to make creative projects.
- [03:49]
Logan
Right.
- [03:50]
Ben
And so I think that was one of the reasons I wanted to continue to explore the various scripts that I had written, the various story ideas I had written. And so that's how I got into it. And I got in, I think, like just jumping in, just talking to ChatGPT, talking to Claude, and coming up with an idea and just going with it. I think that's the cool part is you can just come up with an idea on one day, and then the next day you type in your idea to the AI and the AI doesn't know any better. And it starts coming up with a plan for you, and the plan's ultimately going to be way too complex because for some reason, AI still come up with really complex solutions that could be really simple. And then you. But once you know that, you can sort of start to like, take what you want from the AI and just implement it immediately.
- [04:32]
Logan
I love that. Now, next question. And obviously selfish. Cause I do work for Dell Technologies. Tell us a little bit about where you started. Not from a, you know, a skill set standpoint, but kind of a hardware standpoint, where you started and before you and I started working together. Tell us about, like, the difference between where you started and where you are now, from a technology standpoint.
- [04:54]
Ben
Yeah, for sure. Well, I think, like, a lot of the creative experimentation, I was working completely online. And so because I have a billion ideas, I was just going through credits nonstop. And even if those credits are really cheap, the way that I was going through those credits was just not. I was running out of money. So I think the. The way of the future, obviously, Is to have AIs locally so that you have control over your data, you have control over the AIs. And so that's very. You know, I'm appreciative that I got the Dell and that I got to do work on things locally.
- [05:28]
Logan
Let me kind of ask about a little bit into that, like, from a. From a. Let's say not. Not a technology standpoint, but let's say from a usability standpoint, someone who is working mostly online, mostly in the cloud. How was the change from that to the local PC, like, in terms of, like, how long did it take to adapt? Was it quick? Was it easy? Were there some gotchas?
- [05:53]
Ben
Well, I mean, I'll be honest. I'm a. I was traditionally a Mac person, but what I was.
- [05:58]
Logan
Oh, oh. Oh, my God.
- [06:01]
Ben
But what I was seeing with all these new AI tools coming out is that they were mostly being developed on PCs. And so I was like, how dare. How dare them? I have to, you know, switch over to PCs. And so the transition was literally just like, all right, here's my Mac workflow. I have to change over to a PC workflow. But that's it. So I literally, you know, uploaded Glimmer, the video game, to the cloud, to Google Drive, downloaded Glimmer from Google Drive to the Dell PC. And then I had to, like, mess around a little bit, figure it out, but maybe like a day that was the transition.
- [06:31]
Logan
Okay, I love that. So, I mean, we've talked a little bit about Glimmer, but before we get into that, I know there's some other projects that you worked on, and. And I think it's important not because, you know, they're not finished or anything like that, but I think an important thing to take away from you, which I think you kind of exemplify. And it is a good example for the audience is that AI is kind of a journey and where you kind of start, you're going to learn some lessons and you might fail and you might realize that X, Y, Z doesn't work, but you're going to, if you stay on that journey, you're going to kind of find ultimately something that does work for you, for your company, for your project, you know, whatever you're doing. And I would love to hear kind of a little bit about that. And you don't have to get into the specifics of it, but just like kind of where you started and where you initially thought and then the lessons you kind of learned along the way.
- [07:22]
Ben
See if this crashing burns or not. But one of the projects that I started to work on was Quill Quest. And so Quill Quest was like real time image generation as you told your story. So I could be talking to you right now and images would just automatically pop up onto the screen without any prompting. Because like I, I was a big mid journey prompter. I probably have like 20,000 mid journey prompts and I'd do these very careful prompts where it'd be like, you know, like a retro spaceship in the desert in the style of Stanley Kubrick with 8K resolution, et cetera, et cetera, et cetera. And you know, I'd get some cool results, but like, you know, I wanted that to be all automatic. And so that was the idea behind Quill Quest, is that we'd just be talking and images would pop up on the screen like, like if you're sitting around the fire just telling stories. So I tried to launch that and I had some success in the sense that I was able to actually launch it, but didn't get a lot of traction with quillquest. But at the same time, it taught me all about AWS. It taught me about APIs, how to connect your AIs together, taught me about building a, a platform, taught me about money. I had to like, you know, start charging for credits. You know, I didn't have limitless credits. So, you know, I was trying to start a business in that sense. But even though Quill Quest is pretty much shelved at this point, I mean, I couldn't have transitioned to glimmer, this video game or other projects without having that solid attempt. And I think that's what's so cool about AI is that you can, you can try these things really quickly. I mean, Quill Quest at the time, it didn't seem like it was quick, it seemed like it was a lot of work, but still I mean, I feel like I got through that and now I'm on to the next project, which I love.
- [09:05]
Logan
So we've talked about. About, you know, Glimmer quite a bit. So let's kind of start. Give kind of a high level of Glimmer. And then I kind of want to go into, like, the individual aspects and, you know, get kind of more into the technical details and stuff like that. But let's do high level, you know, give us kind of an overview of Glimmer.
- [09:26]
Ben
Yeah, so Glimmer started as an idea for a pilot, like a TV show, back when I was going to nyu. And so I've always wanted to make this pilot with my friend Brendan. But again, you know, it was during a time when I wasn't working very much and I was like, you know what, let's just not worry about the pilot. Let's make this video game to sort of explore the world of Glimmer. And so that sort of brings me to another thing. It's like, with me, at least with ideas, it's like you can. You can make a pilot, you make a short film, you can make a feature film, you can make a video game. If it's like an interesting idea, it doesn't really matter to me in the sense that, like, I like to explore all of them together. So I started designing the characters in Mid Journey. I started writing the through line for the video game with Brendan. And then I started. We also started creating databases for the characters. So we, you know, we came up with like, what is a vampire in this world? What is a werewolf in this world? And in these databases, because databases are key for AI to keep continuity, it's basically you design the database of your character and then you connect up the LLM and. And then you have character continuity.
- [10:30]
Logan
So let's kind of start breaking down. For the folks that aren't kind of super technical, is first bit is. And you'll see here, you know, throughout this conversation, you'll see bits of Glimmer kind of come up on your screen. But, you know, one. One bit I want to talk about, which I thought was really striking, is the. The image generation of like, you know, kind of the trailer video when you're in it. So kind of talk us through the process of how did you create kind of the character images and the stills? Like, did you use flux? Did you use stable diffusion? Like, which workflows did you actually use?
- [11:04]
Ben
For now, we're just. I just used midjourney, and I created a database of characters. So I Probably I selected like 10 different, you know, five males, five female vampire characters, et cetera. But the cool thing, or at least I think is really cool, is a lot of these NPCs, they don't really exist until you run into them. So in the interface of Glimmer, you see all these dots moving about the map. And I can't really like, even with this superpowered Dell computer, I can't have hundreds of AIs, like all talking at the same time and walking around. And so basically what happens is if you actually run into one of these dots, which is a character, it looks at the database and it randomly picks one of the characters and then it creates their backstory and then it also looks at the current through line, like the main storyline, and then it opens up a dialogue window and then you start talking to your characters.
- [11:55]
Logan
I love that. So, you know, and that's one of the things I think was, was really interesting is about the, the character consistency. And I think that, you know, I do some work with. We won't say which one, but you'll. Everyone will find out surely who it is. And we talk about character consistency and on model and listen, like, I am not an artistic person. Like, I just, I am not. I don't see the things that artistic people do. But it is very important to have the consistency because that's what kind of ties the story together. So, you know, how did kind of tell me about the process? I mean, you kind of explained but like which LLM you're using. Obviously you're creating kind of a database or a rag or a vector database of what you're doing. I kind of talk through that process on, you know, how did you iterate to test and refine to ensure that that was happening with each character that you bumped?
- [12:49]
Ben
It's an ongoing process that I haven't settled on. I think that Anthropic is very creative, which is not a local LLM. But I think like something that I really want to experiment with in the world of Glimmer is using different sized LLMs within the video game. So what I mean by that is if you have your main characters, they're talking, you want to have a super powered LLM talking to you as the player. But maybe, you know, if you have all these other NPCs going around, you could have a smaller sized LLM that's also happening in the background. So that's definitely something that I'm experimenting with, like not just, you know, doing a combination of online LLMs, locals, they're all working in concert.
- [13:33]
Logan
Okay, how are you, like, orchestrating that? Because that's one of the things I thought was so interesting about Glimmer is where, you know, you kind of run through an example. You bump into, let's just say character A, and then you go down a little bit, you come bump into character B. There was a few times you showed me where character B actually knew that you talked to character A, which I thought was really interesting. Like, how did you orchestrate that? Yeah.
- [13:57]
Ben
And then we're back to databases again. You have to keep the databases, and I'm sure, like. So, yeah, I have no coding background whatsoever, so I had to teach myself Python as much as I possibly could. But again, AI to the rescue. And so I use Klein to do a lot of the programming. I'm. I literally just opened up VS code and I have Klein, and I'm like, hey, Klein, can you make this change? And then Klein makes the change to the code, and then I. I see what happens. And then Klein's like, uh, sort of got it. And then it. And then I give feedback, and then it. It just works. So it's very much like almost a verbal conversation at this point.
- [14:31]
Logan
That's awesome. So, you know, in terms of, like, where is the. Where. Where are you taking Glimmer? So, I mean, I know this is kind of like, let's say beyond, you know, the alpha, it's probably into Beta. But where. Where's the vision of where you want to take this?
- [14:47]
Ben
I think the. The problem with me is I have a lot of different projects going on, but there's a couple different directions that I'm trying to take it. So I'm trying to do a local simulation on the Dell computer with a lot of LLMs talking to one another. So, for example, they could go to their local grocery store and there could be a werewolf and a vampire, and they could be shopping for groceries and they could talk to one another. And this is all happening without any user input. And I just want to leave that going for a month and just see how that pans out, because right now, where I have it, it's sort of like there's not enough progression. And so one of the things I've had to learn is I had to add time into the video game. And once you add time, that sort of changes up the whole setup. And then the other way I want to go is it would be really cool to have a Steam game. And so with the Steam game, I'd probably have less LLMs. I'd probably just have one LLM that's running that game and I'd have like a more portable version where anybody could just plug in their, their own API or their own LLM and then play the video game that way. So in other words, like the Dell is used as like the testing ground, the big time simulator.
- [15:54]
Logan
Sandbox. Sandbox.
- [15:55]
Ben
Sandbox. To play around. Yeah.
- [15:57]
Logan
That's cool. So you know, in, in terms of feedback that you have on the game, I mean, have you had feedback? Like are there other tweaks or changes that you're going to make? What's been the overall reception?
- [16:10]
Ben
Yeah, I mean anybody I show it to is like this is awesome. So. And I think it's awesome.
- [16:14]
Logan
I think it's awesome.
- [16:14]
Ben
Playing around in it is awesome. Like the, I think the, the characters are way too verbose. Like they, they, they just like, they talk way too much. But that's like an easy change.
- [16:24]
Logan
Yeah, I mean that's just changing kind of the, you know, the overall tokens and context window. Like. But yeah, I mean I think that's interesting and it's, it's interesting because with AI think overall is like, you might think that they talk too much, but other people might think they talk just the right amount. Right. And like all of that is really adjustable and tunable kind of depending on, you know, the overall use case or what you're trying to do. Ultimately at the end of the day, like I got to ask, like I don't know if you've tested and I don't know the answer to this and they always say you should never ask a question that you don't know the answer to, but I'm going to ask it anyways. Is Windgamer is kind of running locally on your 7875, you know, precision workstation. My, my question is, is that, have you opened up Task Manager to see how much of the, the Nvidia RTX ADA 6000 that you're actually tapping into from a consumption standpoint? I'm very curious.
- [17:17]
Ben
I, I have not. I mean I, I wanted to sort of stay away from the tech, from the technical for the most part, but I've been able to run at least 10 AIs at the same time, like no problem, which is pretty awesome. So. And then I've also had, I think the only time that the computer started where you actually started to hear the fans going on the computer was when I started adding like I think I was trying to get up to a hundred different AIs running around and they weren't talking to one another. But what I was trying to do is I was trying to do pathfinding, so I was trying to get them to explore the map. And so I was. I was trying to maximize that and see how many I could simply have in the game at any one time. And that's the only time that the Dell has actually started to, like, been like, all right, now you're all right, right?
- [18:01]
Logan
I'm getting hot, I'm getting hot. Please turn on the fan.
- [18:04]
Ben
And also, like, the other thing is currently Glimmer is very. I'm keeping it sort of 2D based for the most part, for. Because I want to just explore the story, I want to explore the characters, I want to explore the AI aspects of things. So I haven't gone into Unreal at all whatsoever because I've worked with Unreal a little bit in the past. And once you do that, it's like a whole nother level of complexity. So I think at some point it'd be very, very cool to make this into a 3D game, to take these 2D buildings and extrude them upwards. But at this point in time, I'm not doing that yet. And maybe I never will. Maybe I'll just keep it really simple and just, you know, sort of work with what I have.
- [18:40]
Logan
I mean, that makes perfect sense. I mean, I think. And that's. I've never used personally Unreal. I know that it's fantastic. And the. A bunch of people, you know, in M and E use Unreal, like day in and day out. Like, what would that process look like if you were to take Glimmer into Unreal? Like, I. I feel like it would be mega complex.
- [19:00]
Ben
Yeah, well, I think I just need, like more expertise, like. Yeah, like I need more people working on the project.
- [19:06]
Logan
Yeah, that makes total sense. So, you know, getting kind of back to the technology piece a little bit. And we talked about you have a Precision 7875 workstation, like, and we're not going to say the computer you used before, maybe it's a Mac, maybe it's not. We're not going to get into it. But how would Glimmer work on Mac if you were trying to do it?
- [19:28]
Ben
Well, I would have to use the smaller. Yeah. So to. I think Glimmer is going to work on both Macs and PCs. But the difference being on my current Mac, I'd have to use this, probably the Steam version of Glimmer, meaning only one LLM working at any one time. So you know, whereas with the Dell computer, I can have, like, I can sort of go to town and have a bunch of LLMs talking to one another or NPC characters talking to one another at the same time. I think that's why I'm trying to have two different versions of Glimmer, because some computers just simply can't use the all the LLMs.
- [20:02]
Logan
That's true. I mean, from. Not necessarily like a deployment standpoint, but a development standpoint on the Mac.
- [20:08]
Ben
Yeah, I mean, I'm just not developing on the Mac anymore, so. So there you go.
- [20:14]
Logan
Well, I got my point there. So let's talk about possibilities. Right. We've kind of talked about Glimmer in this podcast. You know, at the bottom you'll hopefully see a link. We'll have a link to Glimmer. Please go check it out. But tell us a little bit, Tell me a little bit about what do you want to do, not necessarily with Glimmer related, but like, from an AI software standpoint, like, what's kind of next? What are some of the ideas floating around in your head?
- [20:38]
Ben
Well, let me just share my screen, see what happens. So here we go. Here's the. The Glimmer page. We can always do that. But I'm also trying to launch this idea platform and you can vote on the ideas. And so there's a lot to come here. But for example, any one of these ideas could be a video game. Any one of these ideas could be a film. And then. So I think that's going to be a really fun idea to explore. The other thing that I'm working on is speaking of, you know, 3D characters, here's like one of the random short films I'm working on that uses Wonder Dynamics, which uses, like 3D characters. So in that way, I am exploring, you know, using 3D characters in that use case, or in this case, I'm using 3D characters plus live action characters as well. That's kind of where I'm going is like, I have a whole bunch of ideas that I want to explore, and they could be a short film, they could be a feature film, they could be a video game. That's the future.
- [21:35]
Logan
So you basically have your own personal AI coming up with ideas that you're going to go tackle, more or less.
- [21:40]
Ben
Well, that's the thing. Like, I think most of these ideas are my ideas. Like, I don't need an AI to come up with ideas. I want to come up with the ideas, but then I can have the AI run with those ideas.
- [21:50]
Logan
Got It. That makes perfect sense. That makes perfect sense. Well, kind of another couple of questions and then we will kind of wrap it up here. Cause I think we've had kind of a great conversation, like, tell me. And I, and I really want to hammer this home, this point home with the audience. Right. Kind of a couple points. First point being, and I know you've touched on earlier, but I definitely want to hammer it home is AI for you, is an accelerator. It's not taking your job, it's making you, it's accelerating the time for you to do A to X. It's allowing you not to necessarily be more creative because you're a creative guy, but it's bringing those creative ideas to life.
- [22:28]
Ben
Yeah, I mean, like, at the end of the day, if we're talking about the film industry, I would have maybe had an opportunity to direct one film in my lifetime, maybe two. But with AI, why not do 100 of them? And you know, I, I'm very aware you can have lots of ideas and they're just ideas. You have to think, take any one of those ideas and then spend a lot of time on it and maybe actually work with real people, with actors, and go out and make a film or go make a video game. But like you said, like, you, it allows you to explore all the ideas.
- [22:58]
Logan
And I love that. So next, you know, from a technical standpoint that I want to hammer home is that, I mean, I think you said Klein. I mean, there's a lot of different, you know, there's deep sea, there's so many different Python coding or code enhancement kind of LLMs and things out there. But have you in your journey? I know that you've said you've spent time and you've worked on it and like all of that, but like, did you think that it was possible to do it? Meaning, like without kind of the assistance of AI? Like, was it easy to get started? Do you know what I mean? Like, like 10 years ago I would not have been able to do what I'm capable of doing, like, because AI wasn't there. Do you know what I mean? So I'm curious to hear your experience and kind of drive that home with the audience.
- [23:46]
Ben
Well, I think there's like this, this, at least for me, it was like mid journey making images, too many images to even talk about. Like 20,000 or more images.
- [23:55]
Logan
Oh my.
- [23:55]
Ben
And I, you know, I think everyone, like, like I, I've run into this a lot. Like everyone sort of goes into this. Oh my God, AI can create anything. A Phase. And so you spend a good amount of time just generating, like, silly things, or maybe not so silly things, but that was sort of like the journey. I started off creating, like, lots of ideas with images and videos. And then from there I was like, hold on a second. AI can do other things other than creating, you know, images and videos. And so that was when I was like, all right, so then how do I do this? How do I create a video game? Okay, well then, like, let's learn some python, or at least learn enough python such that I can talk to Klein. And then from there I was able to work on a video game. I've been working on stories that. The never ending story concept, where I basically have a narrator that just like, it's like. It's just like constantly creating scenes for itself to tell a story that just never ends. I've been experimenting with like, creating like a. You know, I'm. I'm again, I'm not a engineer, but I've was thinking it would be interesting to create an AI brain in the sense that you have, like a logical side of the brain, you have a creative side of the brain, basically connecting a bunch of LLMs together to sort of see how that would be different from just talking to one LLM. So I think the future is just like, do I have a week of time where I can work on a project without too many interruptions? And how far does a week of time take me? How far does a month of time take me? So having these creative sprints where you can just be like, here's an idea.
- [25:29]
Logan
Here'S a week, let's go for it makes sense. So, you know, I mean, it makes sense. And that's kind of the same thing with me is that, you know, when working on a project or whatever, there's sometimes, I mean, not recently, because I've been extremely busy, but now there'll be time where I can carve out, I can get a lot done. But it's really about. And that's a really good point that you made. It's about, you know, as you're going, you know, kind of through this AI journey, it's like making time to do it or having. Having a little bit of time, which I would encourage everyone, no matter what is to. To take the time, to spend the time to try it out. Because ultimately, at the end of the day, you know, AI is kind of here to stay, and it is going to become a bigger and bigger part of our lives as we move. You know, what are we in 2025 now. You know, like, things are changing faster than I can even really keep up with them. So, you know, AI is going to be. AI will continue and implore people and ask people to, hey, give it a shot. Like, go out and try something. Go out and try, you know, Nvidia's workbench. Go out and try to install Comfy ui. Get yourself a Dell Pro Max with an Nvidia RTX gpu. Like, give yourself a shot to go, try it, because you'll be amazed what you can actually do.
- [26:41]
Ben
I mean, my core thing with AI is like, it gets rid of the busy work. Let's get rid of all the busy work so that we can actually work on what we want to work on.
- [26:49]
Logan
Yeah, no, and agreed. And to be honest, like, and it's a different, a little bit of a different use case for me. It's because you have the creative vision. I don't. I use AI in a sense where it's like, hey, I can kind of create the workflow and stuff like that, but I don't necessarily know if I'm trying to create. I'm just making stuff up completely. But like a 1950s, you know, short, like 10 second reel that looks like life in Americana in 1950. Like, I am not that creative to say, like the, the Stanley Kubrick prompt. Right. Like, you know, I'm not. That's not how my mind works. My mind is more of like, okay, I kind of see the end to end, but I don't. I need it to help me be a little bit more creative or give me a direction I can go. And then I can refine it where yours is like, hey, I'm already creative, but I just need help creating the artwork. Do you know what I mean? And it's like, that's what I love.
- [27:42]
Ben
But also at the same time, there's a lot of creative in designing the workflows too. I mean, I think that's super creative in coming up with those workflows as well. Because, like, that's not just like, always simple.
- [27:53]
Logan
No, I mean, it's been a labor of love on that animation project because it is, you know, I mean, I. This is completely off topic, but I'll say it, because it's just kind of funny is that we kind of have a running joke that we'll do some work and then I'll say, how close is that? Give me a percentage. Like, and we started, you know, a while back, we were at like 25%. And I said, I think I feel if, like, if we could get to 80, we would be pretty dang good. And I think the last check in we had, which was a couple weeks ago, was about 75. And so it's like fluctuating a little bit. Yeah, so we're going up, we're going up. But there's been a lot of labor of love and a lot of things that I've learned kind of over the course of the journey, you know, and one of, one of the things, the many things that we're doing has been around trying to. And this animation company, they have, you know, done all types of stuff you've seen on Netflix. You might not know their name, but you've definitely seen one of their. They're animated shows and they want to, they have a very specific style and 2D animation. And they were like, hey, we want to be able to take all these different episodes that we've done, kind of create aura to then if we have a new concept and we want to go start show abc, how do we, if we can take a sketch, can we kind of create a control net that can like transfer the style of what we want to a sketch where we can then give that to, you know, one of our animators that's 80% done and accurate, where they can just tweak and make it better and add in detail and stuff like that. And you know, there was a whole process of learning starting like, you know, in stable diffusion, like training stable diffusion 1.5. Then I was like, eh, probably need to move to stable diffuse in Excel, then realize like kind of the limitations with that and then move to Flux. And then so I've kind of had this whole journey where I've learned a lot of things. Much like, you know, we were talking earlier about how started with one project, but then that led to something else and starting. And a lot of it is just not being scared to try because you're going to fail. Like you're going to fail, you're going to have to learn things. But like from that you're going to learn that hey man, Flux is way, I mean probably unpopular opinion, but Flux is way better than stable diffusion. That's just a fact. In my humble opinion. For the control of the prompts, being able to train the laws to impart, like we're basically, you know, character consistency wise we're up 75% which is pretty amazing. So yeah, I mean, long story short is like depending on whether it's you know, kind of idea generation or if it's like taking an idea and Then accelerating that through kind of the art design or anywhere in between. You know, AI can really fit within the media and entertainment space.
- [30:46]
Ben
Yeah, I mean it's like a lifestyle. Like, I think like in the future we're going to be able to make a feature film in a day, like no problem. But it'll be a certain style of feature film or a certain style of TV show. And it sounds like what you're working on is like training, training a world so that, you know, an individual can go into that world and create a TV episode is really empowering to that individual or individuals working on it.
- [31:10]
Logan
Yep, exactly.
- [31:12]
Ben
So, but you also, you also can go, and you can go make a, a live action feature film with actors and so like. But I feel like that's going to be the difference. You can do, you can do both. It's not like you have to do one or the other.
- [31:24]
Logan
I absolutely agree. And I mean, and there is, you know, some AI being used. I mean it's just not super widely talked about. But you know, in some big films that have been used and then we'll continue to see that kind of growing over time. I don't think at the end of the day that's going to replace scriptwriters. I don't think that's going to replace, you know, actors or anything like that. But like, if you can, you know, have a good idea for a scene and instead of spending 27 hours like trying to design that scene and get you 80% of the way there and then you can touch it up and storyboard it out, it makes total sense.
- [31:56]
Ben
And I will say this, for a lot of these, since I've made so many images, I think that, yeah, like these models are trained on, you know, depending on the model on, on various pieces of data. But for me, like when I'm trying to do like an astronaut floating on the International Space Station eating pizza, it doesn't always work because there's, there hasn't been a lot of pictures of an people eating pizza on the International Space Station.
- [32:20]
Logan
That's right.
- [32:21]
Ben
So like there, there is a use for. What I'm saying is we need artists to make original images because some of the generations are not going to give you original images.
- [32:32]
Logan
I agree 100%. We've been talking for a little over 30 minutes. Ben, this has been an absolute pleasure. Why don't you take a couple seconds, go ahead and tell everyone where they can find you, give you any parting thoughts and then I'll close this down.
- [32:47]
Ben
The parting thoughts are so, so Difficult. Yeah, just go to aurorapicturecompany.com and there's tons of ideas. Vote on the ideas and you can contact me or us through aurorapicture company. Com and my game plan for the future and keep on working on Glimmer locally on my Dell PC and see how far I can take those simulations.
- [33:09]
Logan
I love that. Well, this kind of wraps up our episode. So one, kind of hope this was informative and that you learned a lot. And then two, you know, if you have ideas for episodes, I mean, I'm, I'm all about it. Would love to hear kind of your feedback. And you know, the whole purpose of this show is to be able to talk about not just AI. I mean, obviously this episode was very AI focused, but being able to show kind of the power of, you know, dell Pro Max PCs powered by, you know, Nvidia RTX GPUs. Whether it is AI, whether it is reality capture, whether it's AR VR. The world is changing based on the power of the GPU and overall compute power. And there's a lot of great things happening. That's the whole purpose of this show, is to try to teach you something you didn't know, show you an opportunity, maybe show you a workflow that you can insert kind of into your day to day life. But with that, we'll go ahead and, you know, appreciate you watching. Next episode's coming soon. And Ben, appreciate you coming on, buddy. It's always a pleasure to chat with you, of course. All right, with that, we'll call again.
- [34:36]
Ben
This podcast was produced in partnership with Amaze Media Labs.