World's First Ever 80 Minute AI Film | Road to Cannes Ep. 1

0:00 We're presenting a full feature film in can in less than a month.

0:03 80 minutes long, fully AI generated.

0:06 Now, the issue is that we don't have the film yet.

0:09 What we do have is 10 million credits, a team of 15 people,

0:13 14 days start to finish, and no clue if we'll make it.

0:17 Welcome to episode 1.

0:21 Hi, I'm Adil.

0:22 I'm one of the people generating this film and I'm telling you right now,

0:26 this is either the future of cinema

0:28 or the worst decision we've ever made on camera.

0:30 Here's what's happening.

0:31 We're making an 80-minute film, fully AI generated.

0:34 Again, no cameras, no sets, no actors.

0:37 Well, except for me.

0:39 It premieres at Khan in less than a month,

0:41 and we only have 14 days to finish the whole film.

0:43 And with this film, we want to prove that AI can actually produce cinema.

0:47 What does that mean?

0:48 A viewer watches 80 minutes and forgets they're watching AI.

0:51 The film holds together as a story.

0:53 Beginning, middle, end, payoffs that work.

0:57 Characters feel like characters, not generator figures.

1:00 The emotional moments actually land.

1:02 And people feel something they didn't expect to feel.

1:04 If those things happen, we proved it.

1:07 If they don't, we learned what's still missing.

1:09 We have 15 people working on this, including me.

1:11 The same people that worked on our Kickfield original series like Arena Zero,

1:17 Hell Grind, Zephyr and some of whom you've already met previously.

1:40 14 professional directors, do obs, editors,

1:43 and one seli are joining to make this happen.

1:46 Needless to say, this has never been done before.

1:49 And honestly, I don't know if we can, but we're doing it anyway.

1:55 Every couple of days, I'm going to drop a new episode.

1:58 Same place, same time, same chaos.

2:01 You'll see the breeze, you'll see the prompts, and you'll see the credits burn.

2:04 And most importantly, you'll see what breaks.

2:07 That'll probably include me.

2:08 This is Hell Grind, the continuation of episode 1,

2:11 and it's the first AI feature film in history, if we make it.

2:15 So, quick context, we didn't start from zero.

2:17 Hell Grind episode 1, the one that's already out, is 22 minutes long.

2:21 You can check it out right here.

2:23 Here's what it actually took.

2:24 253 videos made the final cut.

2:27 Out of 16,181 generations on CS 2.0, 107 images made it in out of 10,71.

2:36 We generated total credit spend, brace yourself, 1,152,95 credits.

2:43 That's about $69,000 for 22 minutes.

2:47 And that's not waste.

2:48 That's the craft.

2:49 Every generation that didn't make it taught us the prompts for the next one.

2:53 That's what I want you to understand before we start.

2:56 Masterpieces aren't single prompts.

2:58 They're the 26,000 you don't see.

3:01 So, here's where it gets interesting.

3:03 22 minutes, almost 27,000 generations.

3:07 Now, we have to make 60 more minutes in 14 days with 10 million credits.

3:12 Now, in traditional production, to make the first episode,

3:14 it would have cost millions of dollars,

3:16 month of pre-production, and years from script to screen.

3:20 But don't take just my word for it.

3:22 This is Chuck Russell.

3:23 He directed The Mask, The Racer, The Scorpion King.

3:26 He's been making blockbusters in Hollywood for over 30 years.

3:30 If anyone's going to give me a real number on what

3:32 episode 1 would have cost in the traditional system, it's him.

3:35 Hi, I'm Chuck Russell.

3:36 I'm a writer, producer, and director.

3:38 And I was just asked for my first

3:40 reaction uh watching Higsfield new production of Hellgrind.

3:45 And uh I was pleasantly surprised because I was entertained.

3:49 That was my first reaction.

3:50 While I would imagining looking at any AI production with a more critical eye,

3:55 I got into the characters.

3:56 I got into the story.

3:57 It's about a bunch of edgy street kids who get superpowers.

4:01 So it's a traditional genre.

4:03 Very ambitious for an AI production.

4:05 About a 20-minute production.

4:06 Uh, and it was fun.

4:08 So, this was full AI.

4:10 What we're doing in mainstream is liveaction

4:14 actors with AI enhancements for sets, visual effects.

4:18 It's a new tool in the visual effects toolbox.

4:20 For a full AI production like this, though, it's evolving quickly.

4:24 Uh, congratulations to the filmmakers.

4:26 You made me care about your characters.

4:27 Something I almost never see in AI.

4:30 Uh, so I thought it was a great job.

4:31 You have to understand this is a micro group of people,

4:33 probably a handful of people.

4:35 It's a micro budget for sure.

4:37 And so if this was a traditional studio workflow with the same kind of material,

4:42 you'd be talking about maybe a minimum of $5

4:44 million depending on cast for this 25minute piece.

4:48 So it's a whole new opportunity for independent filmmakers

4:52 and for people that are making films in the mainstream like myself.

4:57 We're still developing techniques that use our live action actors.

5:00 This all started back.

5:01 It's not that new in 2019.

5:03 My in my impression, the the Mandalorian creating endless landscapes on volume

5:08 screens was one of the first forms of AI, which to me is only smart CGI.

5:14 So, fear no AI.

5:15 It's a wonderful tool.

5:16 It reminds me a lot of when we uh did CGI

5:19 in the mask where no one knew it was brand new,

5:22 no one knew where it was going to go.

5:23 So, I think it's always going to be about the human heart.

5:26 Uh there's authenticity in these these mini productions.

5:31 Uh that's one of the biggest advantages and I saw that in Hellgride.

5:35 So congratulations folks.

5:37 Now you did the math whether it's worth it.

5:41 So here's the process.

5:42 Before any of the generation starts, we read the script.

5:45 I know it's hard to believe,

5:46 but it actually took only 3 days to finalize the entire script for the film.

5:50 And I'll show you in a second how we streamlined that entire process.

5:53 As for the script itself, I'm not going to spoil it, but I'll tell you this.

5:57 There's a moment in this film where the room went completely silent.

6:00 I even cried a little bit.

6:02 So now our mission is to convey

6:04 those emotions to the viewers through the generations.

6:07 So how did we write the script?

6:08 We built it using a custom claw skill we trained for screenwriting.

6:12 Instead of asking an LLM to write a scene,

6:15 we built a system that knows our characters,

6:17 our tone, our pacing rules, our world.

6:20 We feed it context and it outputs scenes that actually fit the film.

6:24 That's the difference between AI written content that sounds

6:27 like AI and AI written content that sounds like us.

6:30 I'll leave a link to it in the description.

6:32 Just make sure to adjust it to fit your writing style and voice first.

6:36 The easiest way to do that is to run it

6:38 once with your idea and then any parts you don't like,

6:41 you can tell Claw to change it in the skill itself.

6:43 If you want a deeper dive on this, let me know in the comments.

6:46 But let's move to the most interesting part of this video.

6:51 So once the script was logged, every asset went into one shared canvas.

6:55 Now this feature is still in development and the directors

6:58 and I actually had a lot of input

7:00 on how to make it the most useful and easy to use for you and well for us.

7:05 It'll be live on the website by the time you watch this video hopefully.

7:09 So definitely go check it out.

7:10 Here's what we used it for.

7:13 So first is the style prefer.

7:15 This will be hard baked into every single prompt

7:17 when we make the shot list in the next video.

7:19 We got the style, lighting, color, camera.

7:23 I mean, very detailed so that shot one and shot 80 stay consistent.

7:28 We got some old iterations here, but this is the final one we locked.

7:32 Next up, we got assets.

7:33 So, we have locations, characters, props,

7:37 and as you can see, it's not just the names,

7:39 but a detailed description so that we

7:41 give the model as much context as possible.

7:44 Next up, we have the characters and a bunch of different forms.

7:48 Uh in the base form, as you can see,

7:50 we have the prop sheets baked into the character sheets.

7:54 That way, we don't have to upload multiple images at once.

7:56 And Cance actually does a pretty good job

7:59 at rendering such small objects throughout all shots.

8:03 And generating them in different forms is very important,

8:05 especially when working in a team.

8:07 Say I'm generating shot 78.

8:09 And in shot 13, Frogo received a lot of injuries.

8:13 If I just use his base form, the viewers might not necessarily notice,

8:17 but they will know something is off subconsciously.

8:19 We've locked everything we could as of right now.

8:22 As we work on some of these scenes and create more assets,

8:25 they will be immediately added to this canvas.

8:27 Actually, I'll quickly show you how to create

8:29 the characters just so you have an idea.

8:31 For this, I'm heading over to Cinema Studio 3.5.

8:34 Here, I have AI cast selected, and I want a Tibetan monk.

8:39 Uh, let's do 40 years old.

8:41 Uh, and let's make him bold.

8:44 Now, I'm going to generate four images, but I'm going to do one at a time.

8:47 That way, every single one of them is completely different.

8:53 All right, we got four nice options.

8:55 I like him right here the most.

8:57 I think he would fit our film perfectly.

9:00 This one's not bad, but he's not bold like we wanted to.

9:03 So, let's switch to GPT image 2.

9:07 Drag him in.

9:08 And let's make him completely bold.

9:18 And there we go.

9:19 The same actor, completely bold.

9:21 We got four options.

9:23 All of these look nice.

9:26 I think this one is the best one for comparison.

9:28 Let's see what it look like before.

9:32 All right.

9:32 And the after.

9:33 I mean, same face, same clothing.

9:35 Perfect.

9:36 Now the same workflow was used to create all these characters.

9:39 But the key here is iteration.

9:41 We used Soul Cinema to create the characters and then GPD image to edit them.

9:45 So for Jack for example, we had nearly 600 generations in Soul Cinema and then

9:50 a couple hundred more in GPD image to get the perfect outfit.

9:53 This step is very important because these images will be used for all

9:56 of the CE 2.0 generations and will affect the output a lot.

10:01 So make sure to spend your time here.

10:02 And lastly, we have the locations also from a bunch

10:06 of different angles to keep everything consistent across shots.

10:09 And this canvas is what holds the whole world together.

10:14 We split the film into chunks.

10:16 Beginning for sequence, the battle, etc.

10:19 Each chunk goes to a different team.

10:21 Each team has its own director.

10:23 We're going to use this to create a short list with a prompts in the next video.

10:27 Now, the plan is that every team generates

10:30 2 and 1/2 minutes of usable footage every day.

10:32 At the end of week one, we need to have the full 80 minutes completed so

10:36 that we can spend week two on refining key moments.

10:39 So, at the end of each day,

10:41 we submit edited 2 and 1/2 minutes to the lead editor who put

10:44 everything together and give us feedback on what to refine for week two.

10:48 And it's crucial for us to iterate fast.

10:50 If anyone falls behind by a minute on say day three,

10:54 they have to make up for it on day four.

10:56 That's the plan anyway.

11:00 So, here's where we are.

11:01 Script done, canvas built, shot list split, teams assigned.

11:08 Day one of 14 starting now.

11:10 And in the next episode, I'm going to take you inside the first generation date.

11:14 I'll show you how to create a short

11:15 list with prompts for the entire film using Claw.

11:19 You'll see me try to land my first shot, and I'm going to show you the breakdown

11:22 of where we are in terms of credit spend.

11:24 We've got the script.

11:26 We've got the assets.

11:27 The only thing we don't have yet is the film.

11:30 Day one starts now.

11:31 Let me know in the comments whether you think we'll make it in time,

11:34 how many credits we'll spend in total,

11:36 and what your expectations are from this film.

11:38 As always, if you found this video helpful, hit the like button,

11:42 subscribe, and I'll see you guys in the next one.