Inside Joyrider: Craft, Speed, And AI

What happens when you’re asked to keep an iconic music video exactly as fans remember it except the singer must perform a different song, in perfect sync, for four minutes straight? 

We brought Joyrider Films founder and executive producer Spencer Friend into the studio to unpack the creative calls, the technical stack, and the leadership habits that make a moment like that possible.

TLDR / At a Glance

• why Joyrider built in-house production and post
• what defines fast, visual, cinematic storytelling
• how to choose between practical, CG, virtual, and AI
• where consumer AI breaks and pro pipelines start
• the Coldplay and Idols “Grace” lip sync workflow
• training, slap comps, and iterative refinement
• ethical guardrails and likeness consent in AI
• testing methods before committing to a pipeline
• why storytelling and continuity still lead hiring
• skills young directors need to thrive

Spencer walks us through Joyrider’s hybrid approach across commercials, entertainment, and music videos: when to shoot practically for authentic performance, when CG delivers the impossible, and where AI or virtual production creates real value rather than novelty. 

The Coldplay x Idols “Grace” project becomes our living lab: a model trained on Chris Martin’s face, an iterative cycle of training and slap comps, and the fine-grain compositing choices that made an overcranked, front-facing close-up hold up under scrutiny. 

We explore why consumer AI tools fall down on continuity and how a nodal workflow like ComfyUI supports professional control over look, lip sync, and consistency at scale.

We also get practical about responsible AI. Consent-driven likeness capture, clear data provenance, and ethical standards keep talent protected while improving results. 

Spencer argues AI won’t erase jobs; it expands them, demanding directors who think in shots and systems, VFX artists fluent in light and latency, and AI artists who can code while protecting performance. 

Story remains the north star: every tool decision ladders up to narrative clarity, emotional truth, and deliverable quality under deadline pressure.

If you care about how cutting-edge production actually ships—across branded content, music videos, and beyond this conversation maps the choices, pitfalls, and wins. 

Subscribe for more deep dives, share with a producer or creative who needs a modern toolkit, and leave a review with one question you want us to tackle next.

Chapters

• 0:00 Meet Spencer And Joyrider
• 1:06 Early In-House Production And VFX
• 2:41 Choosing Practical, CG, Or Virtual
• 3:16 AI’s Promise And Its Limits
• 4:06 The Coldplay x Idols “Grace” Challenge
• 6:15 Training The Face And Compositing
• 8:19 Standards, Testing, And Scale
• 10:10 Jobs, Skills, And Responsible AI
• 11:59 Story First And Consistency Problems
• 14:55 Where To Find Joyrider

Transcript

Kieran Gilmurray: 0:03

Welcome to the show. Today we meet Spencer Friend, executive producer and founder of Joyrider Films. Since 2004, Joy Riders told stories across commercials, entertainments, and music videos with a craft that blends action, CGI, and virtual and AI-enabled workflows. Spencer builds teams that move fast, stay honest about the creative risks, and deliver work that feels both cinematic and contemporary. In this episode, we'll unpack how he decides when to shoot practically, when to lean on CG, and when to use virtual production. We'll dive into some recent examples and choices that have made all of this work. We'll also talk about the responsible use of AI and what leaders need to do to keep high quality output when timelines are tight. For those who do not know you, Spencer, would you mind telling us why you set up Joyrider in 2004? What was the problem or the solution you were trying to bring to the market, and how has the company scaled since then?

Spencer Friend: 1:02

Sure. Thank you very much for time. Great to be here today. So even in 2004, it was an exciting time to break the mold of how things are made in branded content, music videos, moving image work. Even that time, we were able to find ways to instead of splitting out production and post-production to another company, we started doing things in-house. So we'd have traditional shooting as most production companies do, but we also then started doing animation and visual effects and using other CG tools in-house. And what it allowed us to do was provide a turnkey solution for our clients.

Kieran Gilmurray: 1:49

So what defines the Joy Rider project in terms of pace, look, performance, and production method?

Spencer Friend: 1:56

I'd say what we're known for is um quite visual storytelling, which I know sounds like a bit of sales puffery, but when I say that, I mean work that visually stands out, whether that's from a photographic aspect or whether that's more mixed media or computer generated. So I'd say that's what we're pretty well known for, and that's what the kind of work we like to do, because it allows us to use various skill sets to deliver what our clients need.

Kieran Gilmurray: 2:25

So projects today, I'm assuming, are very complex in the media because of every uh particular type or tool or technology you can use. So, how do you go about managing and organizing teams and technology for what is now arguably a very complex industry?

Spencer Friend: 2:40

I think every project is different. So every project has its own challenge, uh challenges and has its own requirements as to how you make it. Obviously, it's really, you know, it's it's exciting times now because there's many more tools to use and we've got the know-how to do that. So whether it's shooting live action, whether it's using CG elements, whether it's compositing, whether it's using AI tools, we've we've been cutting edge on the approach to using all different sources.

Kieran Gilmurray: 3:16

Well, you mentioned AI there, and AI is taking on an increasing role in media. Why is that?

Spencer Friend: 3:25

It's much more accessible. So um anyone can go into an AI generative platform and generate something still moving image. You know, it's it's quite easy. However, that's a cut I would say at a consumer level, to create a standalone piece where the shots match and the and the let's call them characters or the actors look the same throughout, that's a different ballgame, completely different ball game. And it requires expert know-how to do this.

Kieran Gilmurray: 3:59

Well, as a company, then you mix, you know, live action, CG, and increasingly AI or virtual production. Have you got an example you could walk us through, you know, a recent one from pitch to delivery? Where did AI or virtual tools create the value versus the nice to have experiments versus, you know, dare I say old school classic methodologies?

Spencer Friend: 4:22

Um, sure. We so we did a really cool project last year. It was um for Idols and Coldplay, and that combined pretty much some old school shooting, photography, some old school compositing, and then much more newer thinking with the AI route. And it was quite an interesting project because basically Chris Martin from Coldplay and and Joel from Idols are good mates, and um Chris was singing, uh guest singing an idols track, and the track was called Grace, and they basically came up with this plan where they would take the yellow video from 25 years ago and have the lip sync completely changed so that Chris would sing Grace, even though the video looks the same as it did 25 years ago, and that created huge challenges. The most the biggest one, of course, making Chris Martin look like he did 25 years ago, otherwise, we we'd just shoot Chris and then try and find a way to composite his face. So this video required an AI toolkit. The the most amazing thing about it was basically developing a pipeline and skill sets to do this, which fortunately we've got some very smart people at Joyrider, and we managed to do it.

Kieran Gilmurray: 5:43

I was gonna say, because those aren't just uh, you know, when you talk about film grade or music video grade skills, you can't walk into Marks and Spencer's and get them. So where did you get them from and what were the decision-making processes between when to lean on CG, when to use AI, when to use virtual production, when to use people?

Spencer Friend: 6:05

Yeah, so a lot of these things are driven by, let's say, whoever the technical lead on it is. So with the Grace Music video, it was Jonathan Irwin, one of the Joyrider directors, who's um he's a really smart guy from Ireland. Um and when I when the video commissioner brought us the project, because it was an official project between the two record labels, uh, we talked, Jonathan and I talked about how we could possibly approach it and do some initial tests. And the key with that video was to basically recreate or generate an AI model of Chris Martin's face singing the new track. So it took millions of iterations of training the model using stable diffusion and and other AI tools to get him to sing that. And then once that was working, we actually use some multiple compositing tools to do to slap comp the face back on the rest of the video. And the video is tricky because it's an overcranked uh the original video. Sorry, the original video is tricky because it's overcranked, so it's shot at I think it's at least double speed, and you've got Chris Martin front and center for nearly four minutes. So there's no hiding the lip sync of what he's singing, which made it very uh interesting uh to achieve the results we managed to pull off. No, very, very difficult, very, very difficult because it's under a lot of scrutiny to make Chris look obviously as as good as he does, but also to make the lip sync perfect.

Kieran Gilmurray: 7:48

Z looking back at that, would you change anything next time?

Spencer Friend: 7:52

Um, no, I think I think our approach was right. You know, the original tests um showed potential, but this was also a year ago in AI. So things have evolved a lot in that time. What we found was we had to, as a as a kind of pipelines generally, create the face, do the training on the face, get it to a level that we thought it was half decent, do a slap comp, so it's rough comp back onto the original video, take that, retrain it again, and round and round and we went until we could get it to a level where it was almost perfect. And we had a lot we were very fortunate because we had a lot of compliments that it looked better than you know, they did some AI work with uh Mark Hamill in Star Wars, and we were told what we had done was even better than that. So it's a really nice compliment.

Kieran Gilmurray: 8:46

Yeah, I was gonna say I'd want that in my gravestone.

Spencer Friend: 8:50

Yeah, it was uh it's really high level that we managed to pull off on a music video.

Kieran Gilmurray: 8:55

So so what capabilities must a production company have today to stay competitive? As brand work is streaming, short form platforms are going, everything's moving at a rate of knots. You know, what do you need to do? And are you doing anything in the in that space?

Spencer Friend: 9:12

I think the the key to it is knowing what tools to use when. So whether it is AI or whether it's something more computer generated, I'd always recommend trying to shoot as much in camera as possible. Um, but as soon as you as soon as things aren't possible, for example, I can't go into the name of the project we're working on or we're about to deliver, but it's it's epic in scale, absolutely epic. Um, it's got Hollywood type action, uh motorbike chasing, all sorts of things going on uh for quite a high profile uh musical, I'd say, or high profile music musical acts, which would cost millions to make. And we've basically done the whole thing in AI, using all the new AI that's out there, bringing it into a nodal-based system like Comfy UI, and creating something that is on a very, very high professional level.

Kieran Gilmurray: 10:17

You'll have to tell me that story off camera then. So is there a risk that what we're describing here, like is there a risk that human talent disappears if AI takes on more roles in the industry? Or am I reading that wrong? Is this a moment to add more tools to the creative tool set of the existing talent?

Spencer Friend: 10:36

Or I don't I I know a lot of people fear for their jobs. I personally think AI tools would could actually bring us bring more jobs and more job vacancies because we need directors and VFX artists and AI artists that know how to do a combination of coding, understand camera lighting, um, and more traditional photographic techniques in an AI environment. So I think AI tools are gonna are definitely here to stay. Uh, but I would always combine them if I can with compositing and shooting elements. Even even when we do a video that is predominantly AI, we still need to shoot reference elements of uh faces and actors, which is another, which will lead us on to another question, probably about how uh AI can stay in a much more safer legal domain as well.

Kieran Gilmurray: 11:32

Yeah, well that that is a question that's coming up a lot because again, interestingly, in the in the last little period, there was a presidential election in Ireland, and someone used AI to uh fake a resignation speech, which was actually then picked up by mainstream media, was published online, went viral until they realized that it was a deep fake. So, what does responsible AI look like when you're training or applying AI and AI models in advertising videos or entertainment contexts?

Spencer Friend: 12:06

The whole deep fake thing is, you know, we're not going to be able to get rid of that, that's for sure. But I think um the responsible aspect for us is these are real projects that stakeholders and clients bring to us and expect. So we would always try and um take real imagery, uh photographic imagery and pull that into AI, which means actors, spokespeople, they have to agree to that and they have to agree to let us take pictures then to use in AI. And I think in some respects that that will safeguard um the use of actors in films as well.

Kieran Gilmurray: 12:47

Well, safeguard them and their careers because someone has to be responsible here as well and set standards. So, what what changes do you expect to see as as AI tools mature? Will production companies be able to survive if they don't use AI Spencer?

Spencer Friend: 13:03

I think, well, I think uh today's production company, like Joyrider, I think to be successful, you need to have the technical abilities and know-how which we have to use certain tools for the right projects, whether it's AI, whether it's compositing, whether it's a grander chute, whether it's something like virtual production. I think the testing phase of projects is is crucial and not just rushing in and then realizing it's not the right approach. And we we've got quite a lot of experience in doing that now.

Kieran Gilmurray: 13:38

Yeah. Well, see for young professionals or young directors, what skills matter most now? Is it AI? Is it storytelling? You know, are we actually hiring young talent or uh is it like other commercial industries where everybody thinks AI will be able to do everything?

Spencer Friend: 13:56

Yeah, I don't I'm not sure AI is definitely you know the panacea that everyone claims it it is. Um you know, the one of the biggest problems in you know average AI platforms and models that are available is you could create a lovely shot, but it won't match the next one and it won't match uh the actor in the previous shot. So uh you need professional and technical level understanding to realize how to pull that off beautifully. Um, for people coming into the industry, I would say it's always going to be about storytelling. Always. You know, you've got to not have laying the groundwork to a good script or a good concept before you go into the technical aspect is always going to be the most important job.

Kieran Gilmurray: 14:43

Yeah, well, it's a theme that I repeat week in, week out that humans are core at now and they will be core ahead. Talent is talent, that doesn't change. AI is an additional tool we should be using, but it's interesting to see in yet another industry that AI is not the only tool in our toolkit. Uh, Spencer, if people want to find out a little bit more about you or Joyrider, where do they connect with you? How do they go about finding you?

Spencer Friend: 15:09

Uh they can find me at joyriderfilms.com, uh on Instagram, Joyriderfilms, my LinkedIn uh is just Spencerfriend. By all means, please do get in touch. Um, we love a good technical challenge.

Kieran Gilmurray: 15:23

Yeah, well, if I can have my way, they will, and I'll pop some of those links and some of those socials into the into the comments uh which will accompany all of these videos, all of these stories and everything else. Spencer, thank you so much indeed for sharing the story. What a fascinating space. I think there's going to be more in the media space coming soon, and I'm delighted to see you and the company at the head of the curve.

Spencer Friend: 15:47

Thank you. Thank you very much for your time.