A (2:45)

So we've been partners now, I think, Annie, for what, you know, about a little over a year. You know, we met through a mutual connection. Actually. You guys, I think you've got a few Nomen alumni, so that's, you know, I work really closely with Nomen, and I think that's how the connection forms. But you guys do some amazing work. What I really appreciated was the fact that you guys were. Were all in with AI. And I think that's. That was hard to find a year ago because, you know, no one wanted to talk about it. Everybody was using it. And so I kind of want to dig into that a little bit and understand how you're implementing AI into your pipeline, what exciting projects you've got coming up and kind of get into some of the AI stuff. That's what I'm most excited about.

B (3:29)

I think I'd love for Alex to kind of lead it off. One of the things that we loved about Alex, and like you said, he's been with us about a year. He's one of these guys that has really jumped in, but he's a very generous person and he's got a YouTube channel and he's a teacher and he really loves sharing the knowledge, you know, and is really good at doing it. So, Alex, maybe do you want to talk about your experience with diving into these tools?

D (3:57)

It's not a straight line, that's for sure. It's quite an adventure because we don't. Nobody really knows what the goal post is, so it just keeps moving and shifting all the time. And like Andy was saying, I've always liked to teach something, so it just felt like the perfect fit because we're all trying to teach each other. I just need to find more people that think about not hiding the resources to learn from board more so sharing them. And that's what I found in Make Make. We're all trying to figure it out, right? We're all trying to build the boat and sail at the same time. And it's. Yeah, it's. It's just constant research. It's like I always thought when I was. When I was young, it's like, oh man, it's like, doctors have to learn, and then they just keep learning forever. It's like, oh, man, it sounds so like, not me. And here I am, and it's like, you know, Monday morning, it's like, oh, my God, I have to do a lot of research, see what happened from Friday here. So it's. Yeah, it's quite exciting, though. It's. I know there's a lot of fear, but you can only be afraid of something if you fully understand it. And I think that's how we're trying to. That's what we're aiming towards. So the more information we have, the better decisions we can make for ourselves, for our team, for the industry. And I guess that's the approach we're taking.

C (5:22)

I love that. I mean, I think it's a great answer.

B (5:24)

Right?

C (5:24)

I mean, I think I'm probably going to steal the whole. You can't be scared of something unless you fully understand it. Because I totally agree. Right. Like, let's. Let me kind of start with this. Is that, you know, when it comes to. I mean, we could probably talk hours, because I love this kind of stuff and I'll nerd out. But, like, let's not get into technical specifics. But where yet. But from kind of a video generation kind of model standpoint, right? Like, let's just say purely, I'm prompting to get out video. Why do you have better success using, like, a reference image than prompting to the video output? Or do you feel that you get a better output just being creative and prompting and then generating a video or a clip?

B (6:08)

It depends. Like, sometimes. So here, here's the approach that, you know, I believe that we take at make make, which is a bit unique, is that collectively, everybody in the building probably has 100, 200 years of, you know, collective knowledge of creating really compelling storytelling. Right. So it starts from. From that. Right. You know, it's. And there's a lot out there that's just a slot machine where you put something in and you just pull the handle. And that's not really the way that we work. We see it as a tool. We see it as a piece of what we're doing. Sometimes it's final pixels and sometimes it's components. So what we'll do is we can prompt something and then that'll get us somewhere. And then maybe we use that image then to modify it and then to generate a few more images or a bunch of different videos. And then a lot of times then we go back to our traditional VFX roots and we'd be like, oh, I like this part of this video and that part of this video. And then, oh, I need to use a traditional element or something that we shot put in here. And that's where the tools of the craft come in and the experience of the VFX artist that's doing it. And it's really, I see it as a force multiplier for an artist and not necessarily enabling them to do more versus stealing someone's job. And I think that's where a lot of the fear comes in.

D (7:41)

The misconception is always like, oh, man, it's like, it's so easy now, right? Just put a prompt and, you know, I don't know, it's a soft drink commercial, please. And then it comes out 30 seconds chopped up and it's like, ready? And the reality is that what we're trying to do in learning all of these new techniques, services, models, whatever you want to call it, is we're trying to find how to control them, right? Because at the end of the day, we don't want the model to give us the image it thinks it wants to give us. We want to steer it in a direction. We want our vision to be what makes it to the screen and not just like, oh, you know, just whatever the slot machine gives you. Like Andy's saying, we're trying to harness and control these models so that we get the output we're after in a be in a more efficient or better way. Whatever you want to say, whatever you want to call it. But that to me, that's the most important thing. It's like controlling the outputs is key. It's like our vision leveraged, made easier by these new tools.

B (8:46)

And it's really a different kind of work. There's a little bit of blending, a little bit of hybrid. It's not like, oh, I do this in five seconds and it's done. It's just if you're making it for you and you're the only client and you could pull the handle a few times and it's done. That's great. But a lot of times, you know, our clients are very discerning and they come to us because we can make a really high quality product and change just about every aspect of the image.

D (9:14)

Yeah, and there's something quite funny if you go on X or LinkedIn and you see all these people posting things like, oh, man, it's like VFX is so done. And it's like, look what I did in just three hours. And it's like that person has clearly never had a client. Clearly. It's just they're their own client and they're happy with it. But yeah, get a client through the door and have them start giving you notes and then you talk to me about VFX being done, about you taking my job.

C (9:42)

And it's true. And I mean, I'll be the one. I played around with a lot of video generation models and I am not artistic. And you're right, like I produced something where I took a photo of a hibiscus flowering in my yard and then I made it look like it was windy and storming and, and I thought it was so dope if I gave it. I mean, be embarrassed to tell you that I made it because you would just probably rip it apart, right? Because it's not, it's not like that level of stuff. And I think there's a huge misconception of people thinking that really any AI of any sort is you can go from A to done, right? Like completely just type in, give me a 30 second, you know, soft drink commercial with this person, this person, that.

D (10:21)

And.

C (10:21)

But I have a question is that someone told me this the other day is how much when you, you said you have discerning clients, how many of them are using AI on their side to come to you, to kind of bring ideas or to prompt you before you even start your work, like where they have a more formed up idea. Do you see that a lot or not?

B (10:42)

A lot of times yes and a lot of times no. You know, so a lot of briefs from clients, they're written copy and some, you know, some basic storyboards and they're pulling reference. The great thing about AI is you've been able to get a lot more clear reference, you know, and stills and then allows us to concept a little bit more. But then what we've noticed is all that stuff starts looking the same. Like, you know, Mid Journey came out and you know, then all the pitches started looking like they came from Mid Journey and then you're losing like an artist touch. So it's. If you can put real stuff into it and augment that, that's always better than just starting directly from a model for something like, for a vision that you're creating.

A (11:29)

Going back and looking at all of the stuff you guys have worked on, you know, over the years. What do you wish you could go back and redo now that you've got these AI tools at your disposal?

B (11:39)

Well, I mean, here's the thing is that, you know, at least for me I never like anything that I do. Right. Especially when you deliver it. And then you're like, oh, this is garbage. And they're like, oh, it's great. It's fine. And then you look at it a year or so later and you're like, okay, that's not too bad. But we have a word that we don't use, at least at our company. It's called this is done. We're never really done with anything. We just ran out of time. And, you know, it's done for now. Or, you know, complete. Ish. Delivered. It's delivered.

D (12:14)

It's delivered. Yeah, that's accurate. That's accurate. I mean, the reality is that, yes, you can go back and fantasize on what. Yes, my. My life would have been so much easier if I had a lot of the tools I have now. But the reality is that we're still troubleshoot. We're still solving old problems in a more efficient way. And having done it the hard way, I think makes you a better artist too, right? Like, you grow muscle by making resistance, right. I think it's still important to have those struggles, right, for as long as we can have them, because that's what you circle back in your mind to try to solve new problems. It's like a compound effect, right? It's like, we'll have a bigger problem later, but we'll solve it with death by a thousand cuts. Like a bunch of different approaches, and we'll get it done. And I guess that's. I guess that's a curve of progress, right? Like, we're moving in the right direction, I hope.

A (13:09)

I guess as traditional filmmakers, like, how do you feel about kind of this. The rise of, like, 100%, you know, AI generated content. There's so many AI film festivals right now. And, I mean, some of the work that is being done is incredible. I mean, you wouldn't necessarily know that it wasn't, you know, produced by a huge studio with a massive budget. I mean, what are your thoughts on that?

D (13:32)

Here's a hot take. I don't think a lot of people share this, but let's say, let's talk about ComfyUI, right? ComfyUI is one of these UIs that are used to leverage all these tools. And people think of it, I guess people that don't really use it think of it as just like another one of the slot machines, right? But to me, that thing is, at this point, it's like a comp package, right? It's like you are compositing in There if you want to. Right? You're creating maths, you're putting one thing over the other, you're trying to mix different things. So it's not just an input output tool. And many of the very good people doing AI videos out there are really treating it like a VFX software, right? Because that's kind of in the direction it's going and that's why a lot of us have hopped on the, on the, on the trains, like, yes, this is going the direction we want. I think there is a, there is a misconception of all these great videos that we see out there. It's like if you really look into it, there was still a lot of work put into it. It's like I have yet to see other than, you know, funny videos out there, like the Yetis, you know, doing vlogs or whatever. It's like most of the really good stuff that's AI still has a lot of work and yet like what we have on screen right now, it's like that's not input output.

C (14:53)

No, I mean Comfy is awesome. Like I use it in a couple of projects and I mean I might hit you up for that workflow because even in Comfy I have found that, you know, things that like Runway are a tad bit faster and quicker. Now granted they cost you and it's cloud based, so it's not my favorite. But like putting in models into Comfy, like it's not something that you put in and just click run and then you're right, pops out. Like to generate any type of video, it's even with a 6000. Well back in the day, 6008 or 6000 Blackwell, it takes some time to generate. Like, I mean it takes some time and then that's time to kind of go back and edit. But I'm going to follow up with a question is like, I do think your hot take is, is very interesting because Comfy is way more powerful than people actually give it credit for with, you know, masking and filling and all this type of stuff that you can do to kind of edit video, you know, photos and stuff. But with the models that actually go inside Comfy, which ones are your favorite? I mean you've got Moshi, you've got ltx. Like which ones do you use locally in Comfy that you found the best results?

B (15:57)

My answer today is different than it was probably four days ago.

C (16:01)

Well, give me your answer as of today and then give me your answer as of last week.

B (16:06)

You know, WAN 2.2 is like the new hotness, right, it's delivering some very good coherence. Motion is really great. You know, prompt more control on prompting and camera. And that's from a video standpoint and to kind of retouch about, about comfy and kind of what, what we do is, you know, there's a place for cloud services, but a lot of the work we do is in a secure environment. There's a lot of IP that we have to protect around it, so we don't have the luxury of putting stuff up on the cloud because we can't control that security. So having it run local is a really important thing for us. Besides the amazing control and extra training that we can do to augment these open source models, that's a very important point for this. And then having obviously the hardware to run it is an important thing. And that's really where our relationship started last year, Cindy, was you guys were really generous to give us one of these monster workstations. And it's been the heart of our AI team and allows us to work a lot faster. Having the Pro Max and I don't know what the crazy Pro Nvidia card is.

C (17:34)

I know, man.

B (17:35)

Probably worth, it's probably worth more than my house, you know, I don't know

C (17:39)

but I mean, honestly, I will tell you. It cost more than my parents paid for their house back in 1971, if that gives you any frame of reference.

B (17:48)

It's probably Corvette when it first came out in the 60s.

C (17:51)

Probably, probably less actually. Corvette was cheaper.

B (17:54)

There's that aspect to, you know, filmmaking, you know, also to touch upon what Alex was talking about earlier with, you know, a commercial and what you were saying about, you know, the filmmakers and all these things. On one hand, I think giving people access to tell stories has always been a huge barrier in, you know, Hollywood. So giving people a platform to kind of be creative and show their vision, I think it's great. The more people creating the better because that inspires other people. Right? So I don't think we should ever kind of trample on creativity or putting yourself out there by creating something and putting it up on the web. You know, a lot of these, these posts are divisive because they're like, oh, you can't do. Then starts turning these like flame wares. A lot of that, like in the beginning was like, you're taking away from your, stealing from this and you're, you know, all this kind of, you know, stuff. But for me it's just another way that people can use a tool to Exercise their creativity. And then I had another point and I forgot. Oh, and the other point is, at least as far as on the professional side, is when we deliver a commercial, it's not just 30 second commercial that you see that people will post on LinkedIn. You're like, Hey, I made this commercial in like three hours. We work with, you know, high bit depth, anywhere from 16 to 32 bit, anywhere from HD to 8K. And what I don't think people realize is that a single commercial can have over 120 versions that we deliver. We deliver on average hundreds, 100 versions of a commercial upon delivery. You have your, you know, your one by ones, your 9 by 16s and although that's new, you know, all the social kind of stuff. But then there's different markets and regions and different languages and what we call localizations. Right. So all the legal has to go through and be changed. You know, a lot. You know, a lot of commercials have disclaimers or even languages are different. Right. You know, so there's different overlays and text that comes up on the screen. The stuff, you know, the little cards at the end that show the product. Like all that is a different version that is created for our clients so that they can play literally internationally. And it still has the same effect as it does when you watch it here, you know, in English.

C (20:21)

That's wild. Okay, I took your comment a little different direction. I thought you were saying I get the legal, the legalese, the language, all of that. That makes sense. Different sizes. I thought you meant 120 different variations of the ad. And I was like, whoa, what are we doing? Like can the client not pick one, maybe five and be done with it?

B (20:42)

It's usually one main commercial and then it's modified 120 different times for to be delivered to all these different markets.

A (20:52)

We have a lot of partnerships like with the ISVs. What other softwares are you guys using at the studio? And then are you leveraging some of the AI capabilities within those softwares? Like, you know, you've got Adobe substance and the text to texture kind of stuff. Is that helping? Is that, you know, I feel like that was a really safe place for creatives to kind of start dipping their toe into AI was use some of the AI features of their softwares and that way, you know, no one could say they weren't using it because you know, it was in Adobe Premiere and it was in a lot of these software. So what is it that you guys are using and are you using, utilizing

D (21:29)

those Features, we'll take what we can get. Anything we can get our hands on. The whole lineup of VFX tools that has been used for a long time, that's obviously in the list. Like you said, more and more of those are getting more AI or really machine learning features. And for sure, it's like I, for me, I started as a compositor, so Nuke is my, you know, my bread and butter. And yeah, just since Copycat was added a few years ago, it's like, it's amazing. Honestly, it's amazing. You can't. There's so many things that you can do with it that would take ages or a whole team and now can just be handled smartly just with a bit of organization and planning. And then again with a Nuke, it's like the whole category system, which is just a lot of these models, be it for cleanup or roto or, you know, you name it. Depth generation. Yes. And I think they're getting more and more exposure and hopefully, at least that's what I tried to get people to use them more because I think they're so, so useful and can just, you know, save you so much time. And it's not just about making the tasks faster, it's about being able to iterate more. Right. So to me, if I, if in a day I can just have a one version that looks good, it's like, yeah, that feels good. But if I can do 10, and from those 10, I can select the one that I feel better about. To me, that's, that's where this technology comes in. It's like, let me do more and in the same amount of time and, and that way we'll get better outputs, hopefully. Sorry, not to deviate, but yes, anything, I guess. And internally, Andy, I don't know if you want to, if you want to talk about some of what we're doing.

B (23:06)

Yeah, the Adobe suite is really nice. Generative fill in Photoshop has been great. You know what's nice about that is it works on larger images and bit depth. You're not generating an image that's really small, which is a lot of times is what you get in a bunch of the image generation models and then go through an opera process. Like the uprest tools that are out there, like from Topaz, that run, that run locally on, on our machines. They're beasts, they're amazing. Like, that's like the chef's kiss. You like the little, you know, salt down the arm. Yeah, exactly. At the end of, you know, a lot of our Projects like, that's the final thing that we run it through Topaz. We get that little. You know, it just tightens it up just a tiny little bit. And I can't say enough good stuff about their products. I think just about everybody's using some sort of chatgpt variant. You know, I think I forgot how to write emails. I just kind of, like, mash in a couple sentences, and I have a GPT that's called make me sound smart. And I just copy that into that, and it just, like, gives me a coherent email, and then I pass it along and they're like, oh, Andy, you're so smart.

A (24:12)

I'm like, so I have to nerd out for a second. I just rewatched, like, the full seasons 1 through 8 of game of Thrones, and I hadn't watched it, obviously, since the. The final season had aired. I'm still completely blown away by the opening sequence. And that was you guys, right? The amount of detail that went into that. And then it kind of evolved as the seasons progressed. And I never really noticed that. And I was, you know, now, since I had a very vested interest and you was. I was going to be, you know, on the podcast with you guys, I started kind of paying attention. Just walk us through that. Because that was game changing. And I feel like there was now a lot of shows that came after that that kind of followed that same sort of. I don't even know how to describe what kind of opening sequence that was, but they were starting to look a little similar. And I feel like you guys were like, the first ones to do it.

B (25:04)

Elastic, which is one of our, you know, Make. Make is the parent company. Elastic handles all of our main title work. And, you know, we've done stuff from Game of Thrones to Last of Us and Shogun the Crown, like, you know, a bunch of, like, the original Daredevil on Netflix. The creatives do an amazing job of coming up with new ideas for these main titles. That particular one has a lot of blood, sweat and tears from Kirk Shintani. He drove the CG side and, you know, later on was. Was art directing some of those, along with Angus Wall, the owner of our company. And I wasn't here or I wasn't with the company when we did those initial ones. I was, you know, for the last couple seasons, I was around. And it just comes down to, I guess in the beginning, it was probably just seizing the moment, you know, and understanding that they wanted to do something different. And there was such a cool vision that was executed and just a lot of sleepless nights, a lot of love, you know, and detail into something that is special. And the best jobs in this industry are the ones that are the most painful, you know, because you're. You're living there, you're putting your guts into it, and it comes out on the screen. So, like, you know, hats off to Kirk and the team and Angus, you know, for really coming up with a unique vision and creating a really new genre. And Angus has been doing that for years. Pretty sure did the opening for the main titles for seven. He was involved in that. There's a pedigree there as far as the creativity.

C (26:51)

Andy, I know we talked about before where I know you've been working with Cindy for, you know, a little over the year, over the last year, but we actually met at AIM a lot and you showed a video that. I know I came rushing down. Well, I wasn't the only one, but a lot of people came rushing down like, oh, my God, how did you do. This is the most amazing thing I've ever seen. And you've agreed to show it, so can you. Before we show the clip, can you kind of tee up kind of the video, kind of what it's about, the background, and then, well, whenever you're ready, we'll start showing it.

B (27:21)

Make make is kind of a collective campus of a number of different companies. There's RPS for editorial, a 52 for commercials, elastic for, like, main titles, and a lot of our own internal creative. And this is just an amalgamation of all of that work, from feature film editorial to television commercials to special venues. And, you know, it just shows the amazing amount of work the artists that work at make make have have been doing for the last 30 years.

C (27:53)

All right, with that, let's roll it.

B (27:58)

We all want to be problemless, to fix ourselves. We look for some magic solution to make us all better. But none of us really know what we're doing. And why is that so bad? That's not what humans can do. Guess try.

C (28:31)

Hope.

B (28:58)

Star shining bright above you Birds singing in the sycamore trees

A (29:12)

what?

C (29:41)

Finish.

B (30:09)

Ra. I'm sorry. Yes.

C (30:39)

I don't understand.

B (30:40)

Which part?

C (30:50)

So that was a great video, Andy. Alex, that was just super impressive. I know Cindy's been chomping at the bit to ask a question.

A (30:58)

So, Cindy, what's next? You know, where do we go from here?

B (31:01)

Alex, what do you think?

D (31:03)

Wow, that's. Like I said, it's a moving goal post, so it's hard to pin it down. What's next? I Think we keep going on the path that we're going. And to me, well, this is something I live by. It's like you just have to stay curious. If you stay curious and you just keep digging and you just keep trying to figure out what's going out there and staying on top of things and we'll have the answer next time we come. But for now, we'll keep curious and busy.

B (31:33)

For me, it's myself, Alex. Everybody at the company is. We're storytellers. Angus is a storyteller. He loves telling stories. So we're going to use crayons, pencils, markers, AI film, video. We're going to use whatever it takes to tell a really good, compelling story.

D (31:56)

And Dell workstations and Absolutely.

C (31:59)

Dell Pro Max with Nvidia RTX GPUs.

B (32:03)

Yeah.

C (32:03)

Don't forget.

B (32:04)

Absolutely. In all seriousness, like, it's really kind of the heart of what our innovation is. Like we couldn't innovate and test a lot of these tools without a Pro Max because it's just a beast.

D (32:19)

It's always busy, isn't it? Like, people are lining up. It's like, I know.

C (32:24)

And I can't wait to hear about the results from the Blackwell card as well. So that's going to be very interesting. Doubling.

B (32:29)

Yeah, we're. We're chomping at the bit, waiting for the FedEx guy.

D (32:32)

Every day.

C (32:33)

It's keys. They're coming, they're coming. I promise. It's in the mail. It's in the mail. Make, Make. I promise. So with that, Andy, Alex, really appreciate it. We're going to go ahead and wrap it up. Before we do, we're starting with Alex. Tell everyone where they can find you, you know, on social media, LinkedIn. And Andy, you do the same.

D (32:51)

Sure. So, yeah, like I said, I'm Alex Villabon. You can Find me on YouTube. I've been a bit busier than before, so less, Less going on there, but you can find me under Alex Villabon. And same on LinkedIn. That's. That's kind of between the two. Where I live.

B (33:06)

Alex is really modest.

D (33:07)

His.

B (33:08)

His, his YouTube channel is great. He's very generous and teaches some really amazing things. So, you know, subscribe. And smash that, like, button for Alex.

C (33:18)

And turn on notifications.

B (33:20)

And turn on notifications.

D (33:21)

Smash the bell. Right?

C (33:23)

Smash the bell. Really appreciate Cindy. It's always a pleasure hanging out with you, even though we probably do it more than we should. So with that, we're. We'll wrap it up till the next time of reshaping workflows with Nvidia, RTX, GPUs and Delpro. Max, I'm Logan, your host. Keep your AI workflows running locally and we'll see you on the next one.

B (34:01)

This podcast was produced in partnership with Amaze Media Labs.