From brief to New York Subway in 2 weeks
We sat down with Chad Barrigan, Co-founder and Creative Director of Diagram Learning Studio to discuss how Hedra has helped them launch the Lunos campaign featured at the New York Subway station.
Chad's Favorite Things About Hedra
One Canvas, One Workflow
Chad really enjoys that everything is in one place and that he doesn’t have to switch between tools or hunt for files.
Creative Collaborator
What Chad loved most was being able to just talk to the agent and the natural back-and-forth that actually delivers.
_________________________________________________________________________________________________
Full Interview
Can you tell us a bit more about your role?
I am the creative director of a company called Diagram Learning Studio. We are a UK-based learning design company, but we also make marketing content. Our core business is learning consultancy. So whether that's e-learning design, workshop design, leadership training design, but we also make our own product, where we create innovative leadership scenario experiences that use a lot of AI video. I have a background in animation, and we did a launch video for Lunos, where I used Hedra. I have gone firmly down a rabbit hole of the type of things we could make. Having AI video just elevates everything, as long as it's done right.
How did you first hear about Hedra?
I first heard about Hedra through Lunos, actually. While I was working on the initial project that I ended up using Hedra for, we tried a few things and then Alex, the Chief Marketing Officer at Lunos, was like, "You need to speak to my friend Max because I think this is going to work really well for you." And by this stage I tried iterating a few different ideas for the Lunos project, and then chatted with Max, when he showed it to me, I was like, "Oh, I get this now. This'll really work. This is brilliant. I'm going to use that."
What was that first thing you created with Hedra?
As soon as I got access, I started making strange things that I can't even remember.
It was like, “oh let's see what happens”. But the first real thing I made was actually the ad that we ended up using for the Lunos campaign that appeared on the New York subway. I had already generated some other images and then I was like “okay, now give it a pink-like hue”. And then it did that. I was like, okay now put this cowboy in it. And it just put it in. I like it because it's working on the canvas as well and everything is in one place. I don't have to download anything, then forget what I've called it, try to upload it to another tool, to hope that it’ll work, and then compare it in Photoshop. It's all here. It's all in one place. I don't have to run around and do any extra steps. As soon as I saw that canvas, I knew this works.
Was there a specific moment early on where you thought, okay this is the “thing I want to use”?
The canvas was the first thing, but not all canvases are created equal. What I liked is how you could then start changing things on the canvas. It's not just a place to put things, it's a workspace as well. So when I could start changing things in the canvas and then use the bar on the left hand side to rearrange the canvas, I thought this can be a home for all the assets where you can iterate quickly and try out different things. It might sound like a small thing, but there's a huge difference between just telling it 'put a green shirt on this person' and watching all the variations appear, versus the old workflow where you're dragging in reference images, wrestling with prompts, and iterating just to get something close. All I had to do was click on the image, a little dialogue box pops up asking what I want, I type 'make that shirt green’ and it just does it. It saves so much time.
Can you share a bit more about the Lunos campaign?
Yes, because I know those guys really well, the brief was, "You've got two weeks, we want some ads for the New York subway to advertise Lunos, off you go." Part of the integral to their brand is the idea of a cowboy. I was just throwing ideas around, thinking about what we could do with it. I've played video games my whole life, so it clicked, if he's in the subway, why not lean into that? A classic '80s, '90s arcade game aesthetic with the Lunos cowboy right there in the subway.
That's basically where my mind ended up going. With that level of creative freedom, there comes a part where you have to be like, okay let's see what the client actually wants, it's not just me. I'm not the final stakeholder and can sign-off on this piece. I was able to iterate quite quickly, it moved very quickly from the idea I had. When I saw that I could use different models through Hedra, I got excited. I can't remember exactly which model I used for the 15-second generation, but since it was the only one that did it, I didn't dive too deep into the others. Every time the agent recommended a model, it worked, so I never really felt the need to change.
When some video generation doesn't hit the mark on the first try, what is your process of tweaking or fixing it?
It depends on what mood I'm in, because sometimes you just end up getting incredibly frustrated. But having the agent and that bar on the side to actually talk to was really really helpful because it's very easy to just start shouting at the generation like "No, I wanted it red”. But when you're able to explain it and actually have the back and forth with the agent, I found that really useful.
With AI video generation, there's always a risk, it's amazing, but the fidelity and customization don't always quite get you where you want to be. What made the difference for me was just being able to talk to the agent directly. Like, 'I want it to do this, can we make sure it's doing that?' With other generative models, you'd have to go consult a separate AI just to get your prompt right. Having that conversation natively in Hedra was really, really useful.
Where do you see AI video going for marketing teams?
It's an interesting one because marketing is something we've done but it's not our main focus, but if we're looking at the ability to make videos quickly at a very high standard, it means that more things can be made. There is always the risk with lowering the bar to entry where the quality can take a hit. But I think someone who knows what they're doing will be able to get fantastic results. I know there's some controversy around the use of AI but for us we wouldn't be able to do these things without AI. To reference back to the leadership scenarios we do, we don't have a Hollywood budget to have live action and all these things, it just wouldn't exist without AI. It's opening up possibilities that previously weren't available to people. If it's in the right hands, you can create some really innovative things.
Do you see any potential limitations with AI video as an industry?
One of the biggest limitations isn't really technical, it's perception. You hear people talk about AI slop, and honestly, some of that criticism is valid. There is bad AI content out there. But slop is slop whether a human or an AI made it. The risk is that people write off the whole thing because of the worst examples, when some of it is genuinely really good. That's almost a challenge I enjoy, proving that it can be done well. As for technical limitations, every time a new model is released, something that wasn't possible before suddenly is. So for us, the real limitation isn't on the tools side, it's appetite. It's whether customers are ready to embrace it.
When can we expect your next piece of content?
That is a very good question. I hope it's around September or October this year. But anything AI video, I will be using Hedra for. So if another job comes in, then it'll be sooner than that.
Any idea yet what this next piece of content will be?
We're going to work on a new leadership scenario, which will be a cinematic visual. I’m not sure that it’ll be ready for October. But there's potential for some other client work coming in. It's using a lot of credits for generating these 60 videos and it's a pretty heavy thing, but Hedra will make it a lot easier.
