WEBVTT 00:00:00.000 --> 00:00:02.200 So the frustration is real. I think a lot of 00:00:02.200 --> 00:00:04.299 people feel this. You find yourself writing these, 00:00:04.360 --> 00:00:07.820 I don't know, 500 word novels for an AI video 00:00:07.820 --> 00:00:11.839 tool. And what you get back is still blurry or 00:00:11.839 --> 00:00:14.619 inconsistent. It just looks like amateur garbage. 00:00:14.679 --> 00:00:16.559 You're spending hours on it. You're trying to 00:00:16.559 --> 00:00:19.000 describe the lighting, the mood, the exact movement. 00:00:19.339 --> 00:00:21.960 And the model just seems to ignore the most important 00:00:21.960 --> 00:00:25.500 parts. It's exhausting. Stop wasting that time. 00:00:25.940 --> 00:00:28.780 I think we've figured this out. The problem isn't 00:00:28.780 --> 00:00:30.719 that you need more words. It's that you need 00:00:30.719 --> 00:00:33.079 a stronger foundational framework. The real figure 00:00:33.079 --> 00:00:36.899 here is mastering seven, just seven simple prompt 00:00:36.899 --> 00:00:39.960 styles. Welcome to the deep dive. And yeah, that 00:00:39.960 --> 00:00:41.479 novel you're trying to feed the AI, it's not 00:00:41.479 --> 00:00:43.140 designed for that. It's looking for structural 00:00:43.140 --> 00:00:45.920 cues, not, you know, beautiful prose. We've basically 00:00:45.920 --> 00:00:48.100 boiled down the key workflows from all the top 00:00:48.100 --> 00:00:50.579 AI filmmakers. And the main insight is this. 00:00:51.020 --> 00:00:53.579 The best results, they come from frameworks that 00:00:53.579 --> 00:00:56.280 give you directorial control. And these aren't 00:00:56.280 --> 00:00:58.799 unique to one tool. These seven styles work across, 00:00:58.979 --> 00:01:02.399 well, everything. Veo, Kling, Runway, Sora 2, 00:01:02.719 --> 00:01:05.299 Pika. It's about moving from just describing 00:01:05.299 --> 00:01:08.439 a scene to actually directing it. So our mission 00:01:08.439 --> 00:01:10.519 today is to really distill this craft, to give 00:01:10.519 --> 00:01:13.200 you the architecture to stop guessing and start 00:01:13.200 --> 00:01:16.340 creating high -fidelity, consistent work. Let's 00:01:16.340 --> 00:01:18.760 just jump in. framework one is all about getting 00:01:18.760 --> 00:01:21.379 that directorial control over the space in your 00:01:21.379 --> 00:01:24.480 scene we call this one prompt style number one 00:01:24.480 --> 00:01:27.700 cinematic prompts this is where you use the language 00:01:27.700 --> 00:01:31.040 of film to tell the ai what to do it's not just 00:01:31.040 --> 00:01:33.379 what's in the scene but how the virtual camera 00:01:33.379 --> 00:01:36.340 is seeing it a tiny tweak to the camera movement 00:01:36.340 --> 00:01:38.700 can completely change the whole psychological 00:01:38.700 --> 00:01:41.319 impact of the clip yeah think about it like passive 00:01:41.319 --> 00:01:44.200 versus active observation if your prompt is just 00:01:44.200 --> 00:01:48.260 uh an artist in a studio, the model picks the 00:01:48.260 --> 00:01:50.439 perspective. It's almost always flat and boring. 00:01:50.579 --> 00:01:52.480 Right. But if you start using intentional camera 00:01:52.480 --> 00:01:54.620 language, you change the whole experience. Yeah. 00:01:54.780 --> 00:01:57.040 It's contrast to static shot where the camera's 00:01:57.040 --> 00:01:59.019 just locked down. That gives you a feeling of 00:01:59.019 --> 00:02:01.760 stillness, maybe contemplation. With a rotating 00:02:01.760 --> 00:02:04.659 zoom in. Exactly. That rotating zoom instantly 00:02:04.659 --> 00:02:07.439 tells you something's changing, right? It builds 00:02:07.439 --> 00:02:09.800 suspense or maybe a really intimate emotional 00:02:09.800 --> 00:02:12.960 connection. The model knows what these moves 00:02:12.960 --> 00:02:15.569 mean. And for dynamic movement, you have to be 00:02:15.569 --> 00:02:18.629 really specific. A tracking shot that runs alongside 00:02:18.629 --> 00:02:21.069 someone, that creates a sense of flow, of progression. 00:02:21.449 --> 00:02:24.449 But if you swap that for a handheld drift. Oh, 00:02:24.469 --> 00:02:26.509 that little wobble. Yeah, that slight instability. 00:02:26.550 --> 00:02:29.330 It injects this human tension, this realism, 00:02:29.490 --> 00:02:33.129 that a perfect, smooth tracking shot. just doesn't 00:02:33.129 --> 00:02:35.050 have. And we've seen people get really good with 00:02:35.050 --> 00:02:37.449 the vertical element, too. A vertical tilt up, 00:02:37.509 --> 00:02:40.370 it can suggest something huge, ambitious, maybe 00:02:40.370 --> 00:02:43.270 even overwhelming. A tilt down, on the other 00:02:43.270 --> 00:02:46.090 hand, it immediately suggests fatigue or someone 00:02:46.090 --> 00:02:48.430 searching for something. It's psychological control 00:02:48.430 --> 00:02:51.090 with just two words. And the pro move, I think, 00:02:51.090 --> 00:02:52.729 is combining these. You can actually string them 00:02:52.729 --> 00:02:54.870 together. You can prompt. The camera starts with 00:02:54.870 --> 00:02:57.189 a slow tracking shot from behind. Then it smoothly 00:02:57.189 --> 00:02:59.349 tilts up to their face and then it circles around 00:02:59.349 --> 00:03:02.099 to reveal the city below. That's a whole whole 00:03:02.099 --> 00:03:05.039 dynamic sequence built into one cohesive prompt. 00:03:05.319 --> 00:03:08.020 So how does controlling the camera angle fundamentally 00:03:08.020 --> 00:03:11.240 change the meaning? It just changes how the audience 00:03:11.240 --> 00:03:14.759 experiences the scene completely. So we've got 00:03:14.759 --> 00:03:17.360 control over space. Now we need to talk about 00:03:17.360 --> 00:03:20.580 control over time. And that brings us to style 00:03:20.580 --> 00:03:24.740 number two, timestamp prompts. This is so important. 00:03:24.879 --> 00:03:27.419 I've tried to do this without timestamps. If 00:03:27.419 --> 00:03:31.060 you try to script a sequence, like a character 00:03:31.060 --> 00:03:33.840 walks in, finds a key, then reacts to a sound. 00:03:34.560 --> 00:03:36.780 If you put that in one paragraph. It's a mess. 00:03:36.840 --> 00:03:39.000 The AI just dumbles the timing. It might show 00:03:39.000 --> 00:03:41.159 the reaction before the key is even found. Or 00:03:41.159 --> 00:03:43.840 it merges everything into this one chaotic simultaneous 00:03:43.840 --> 00:03:46.400 moment. So timestamp prompts, they basically 00:03:46.400 --> 00:03:48.680 turn you into an editor. You're dictating the 00:03:48.680 --> 00:03:50.580 timeline second by second, and you segment your 00:03:50.580 --> 00:03:52.979 video into these exact blocks, and you force 00:03:52.979 --> 00:03:55.020 the model to follow a precise sequence. It's 00:03:55.020 --> 00:03:57.259 how you choreograph something complex. Let's 00:03:57.259 --> 00:03:59.039 use an example, like an eight -second video of 00:03:59.039 --> 00:04:00.919 a barista. Right. You wouldn't write one block 00:04:00.919 --> 00:04:03.300 of text. You'd break it down. Zero three seconds. 00:04:03.500 --> 00:04:06.039 Slow pan across the empty coffee shop, stopping 00:04:06.039 --> 00:04:08.680 exactly on the barista. Then you force the shift. 00:04:08.819 --> 00:04:12.000 Three to five seconds. Camera pushes in close 00:04:12.000 --> 00:04:15.349 as the barista pours steaming milk. Very focused. 00:04:15.629 --> 00:04:19.310 And the final beat. Five to eight seconds. Camera 00:04:19.310 --> 00:04:21.689 tilts up to the barista's face as they pause, 00:04:21.870 --> 00:04:24.569 exhale, and look out a rain -streaked window. 00:04:24.910 --> 00:04:27.649 Does this method prevent that chaotic randomness 00:04:27.649 --> 00:04:30.430 we were talking about? Yes, it dictates the timeline 00:04:30.430 --> 00:04:33.029 and the sequence precisely. No more guesswork. 00:04:33.170 --> 00:04:35.149 Okay, that makes sense. But what if you need 00:04:35.149 --> 00:04:38.269 more than just one smooth shot? What if you need 00:04:38.269 --> 00:04:41.750 dramatic visual variety? That leads us to style 00:04:41.750 --> 00:04:44.810 number three. cutscene prompts. Right. This technique 00:04:44.810 --> 00:04:47.129 lets you script actual cuts like you're in an 00:04:47.129 --> 00:04:49.889 editing bay. If timestamp prompts control when 00:04:49.889 --> 00:04:52.230 things happen, cutscene prompts dictate a hard 00:04:52.230 --> 00:04:54.430 instant change in the camera angle mid -video. 00:04:54.649 --> 00:04:56.310 And the trigger is usually super simple. Something 00:04:56.310 --> 00:04:58.939 like the phrase cut to or new shot. So you could 00:04:58.939 --> 00:05:00.660 have a wide shot of a street photographer walking, 00:05:00.740 --> 00:05:02.439 you see their whole body. And then you dictate, 00:05:02.620 --> 00:05:05.399 cut to a close -up, a macro shot, focusing on 00:05:05.399 --> 00:05:07.279 their fingers, pressing the shutter. It's that 00:05:07.279 --> 00:05:09.720 immediate shift that builds impact. But the best 00:05:09.720 --> 00:05:12.079 way, the pro workflow, is combining these two. 00:05:12.240 --> 00:05:16.500 You'd write zero threes, wide shot, three fives, 00:05:16.519 --> 00:05:19.560 cut to a close -up of the eye and the viewfinder, 00:05:19.740 --> 00:05:23.379 five eights. Cut back to a medium shot as they 00:05:23.379 --> 00:05:26.079 walk away. That's multi -angle storytelling in 00:05:26.079 --> 00:05:28.240 one prompt block. Now here's a crucial warning. 00:05:28.779 --> 00:05:31.319 Don't, and I mean don't, cut to radically different 00:05:31.319 --> 00:05:33.360 visual styles. Oh yeah, I've seen this happen. 00:05:33.600 --> 00:05:36.920 If you go from, say, a photorealistic street 00:05:36.920 --> 00:05:41.060 scene to like a 3D anime character. The AI just 00:05:41.060 --> 00:05:43.160 freaks out. It tries to smear the two styles 00:05:43.160 --> 00:05:45.519 together and you get this incoherent mess. You 00:05:45.519 --> 00:05:48.199 have to maintain visual consistency. So what's 00:05:48.199 --> 00:05:50.060 the biggest risk when using the cutscene technique? 00:05:50.589 --> 00:05:52.709 Cutting between radically different visual styles, 00:05:52.810 --> 00:05:55.389 it breaks the AI's brain. Okay, so we've directed 00:05:55.389 --> 00:05:57.529 the camera and the clock. Now let's get some 00:05:57.529 --> 00:05:59.329 help writing the script itself. This is prompt 00:05:59.329 --> 00:06:02.410 style number four, GPT prompts. We should be 00:06:02.410 --> 00:06:04.790 using AI to help us write for another AI. The 00:06:04.790 --> 00:06:07.509 concept is, well, it's pretty elegant. You create 00:06:07.509 --> 00:06:09.829 your own specialized prompt helper. You take 00:06:09.829 --> 00:06:12.389 a large language model like a custom GPT, and 00:06:12.389 --> 00:06:14.569 you feed it the official documentation for your 00:06:14.569 --> 00:06:16.589 video tool. So you give it all of Runway's rules 00:06:16.589 --> 00:06:19.800 and syntax guidelines. Exactly. And the job description 00:06:19.800 --> 00:06:22.139 you give it is simple. You are an expert prompt 00:06:22.139 --> 00:06:25.220 writer for Runway. I give you a scene. You generate 00:06:25.220 --> 00:06:27.319 an optimized prompt with all the right language. 00:06:27.560 --> 00:06:30.180 It's meant to save you hours of manual formatting. 00:06:30.459 --> 00:06:33.879 So the GBT spits out this technically perfect, 00:06:34.040 --> 00:06:36.399 beautifully structured prompt. It follows all 00:06:36.399 --> 00:06:38.759 the rules and it sounds like the perfect solution. 00:06:39.160 --> 00:06:42.300 But, and I'll admit this, I still get tripped 00:06:42.300 --> 00:06:44.759 up here. I wrestle with a prompt drift myself 00:06:44.759 --> 00:06:47.769 when I lean on these helpers too much. There's 00:06:47.769 --> 00:06:50.290 a deadly blind spot. And that blind spot is the 00:06:50.290 --> 00:06:53.310 visual rendering capability of the actual video 00:06:53.310 --> 00:06:55.870 model. The GPT knows language, but it doesn't 00:06:55.870 --> 00:06:58.149 know what the video AI is currently, you know, 00:06:58.149 --> 00:07:00.550 bad at rendering. Like crowds. If you ask the 00:07:00.550 --> 00:07:03.709 GPT for an angry mob of a thousand people shouting. 00:07:03.870 --> 00:07:05.730 It'll write you a beautiful detailed prompt. 00:07:05.990 --> 00:07:08.110 But when you feed that to the video AI. Blobby 00:07:08.110 --> 00:07:10.550 robots every time. Because rendering hundreds 00:07:10.550 --> 00:07:13.149 of distinct moving individuals is one of the 00:07:13.149 --> 00:07:15.149 hardest things for these models right now. The 00:07:15.149 --> 00:07:18.129 human fix is so important here. You have to simplify 00:07:18.129 --> 00:07:20.709 the action but keep the emotion. So instead of 00:07:20.709 --> 00:07:23.550 a shouting mob, you change it to silent, still 00:07:23.550 --> 00:07:26.970 townsfolk watching with growing, ominous unease. 00:07:26.990 --> 00:07:29.569 It simplifies the geometry but keeps the tension. 00:07:29.889 --> 00:07:32.850 So why is human judgment still necessary, even 00:07:32.850 --> 00:07:35.810 with a prompt helper? AI generates technically 00:07:35.810 --> 00:07:39.009 perfect prompts, but lacks awareness of what 00:07:39.009 --> 00:07:41.089 it can render visually. Okay, that makes total 00:07:41.089 --> 00:07:44.089 sense. Now let's talk about making sure our scenes 00:07:44.089 --> 00:07:46.389 stay the way we designed them. Which brings us 00:07:46.389 --> 00:07:48.790 to prompt style number five, anchor prompts. 00:07:49.810 --> 00:07:53.389 This is your consistency insurance. AI loves 00:07:53.389 --> 00:07:56.230 to hallucinate away details. The things you care 00:07:56.230 --> 00:07:58.750 about most, like a unique scar on your hero's 00:07:58.750 --> 00:08:01.550 face. Or the color of a prop. It can just disappear. 00:08:01.889 --> 00:08:04.810 Or change from one shot to the next. Your protagonist 00:08:04.810 --> 00:08:06.910 suddenly looks like a different person. Anchors 00:08:06.910 --> 00:08:09.680 are the fix. they are specific repetitive phrases 00:08:09.680 --> 00:08:12.279 you use to lock down those critical details physical 00:08:12.279 --> 00:08:15.060 appearance like scars or wrinkles or even more 00:08:15.060 --> 00:08:17.839 important spatial relationships there was a great 00:08:17.839 --> 00:08:20.120 example of this an orc warrior riding a dire 00:08:20.120 --> 00:08:22.259 wolf right without an anchor you get the shot 00:08:22.259 --> 00:08:25.259 back and the orc is just floating like three 00:08:25.259 --> 00:08:27.560 inches above the wolf's back as it runs it just 00:08:27.560 --> 00:08:30.019 breaks the whole illusion so the anchor is a 00:08:30.019 --> 00:08:32.759 phrase you add and repeat the orc is securely 00:08:32.759 --> 00:08:35.149 seated on the back of the dire wolf Locked in 00:08:35.149 --> 00:08:37.970 position. You're just forcing the AI to maintain 00:08:37.970 --> 00:08:40.289 that connection across the whole movement. And 00:08:40.289 --> 00:08:42.730 they're also great for off -screen details. If 00:08:42.730 --> 00:08:45.190 your warrior has armor only on their left shoulder, 00:08:45.350 --> 00:08:47.889 you anchor that detail. So when they turn and 00:08:47.889 --> 00:08:49.490 you see them from the other side in the next 00:08:49.490 --> 00:08:51.950 shot, the AI remembers the armor is still there. 00:08:52.210 --> 00:08:54.470 So are anchor prompts essentially just forcing 00:08:54.470 --> 00:08:57.009 visual consistency? They are an insurance policy 00:08:57.009 --> 00:08:59.909 against the AI assuming details disappear. Perfect. 00:09:00.049 --> 00:09:02.750 So if anchors handle consistency, then style 00:09:02.750 --> 00:09:05.269 number six, image prompts, is all about defining 00:09:05.269 --> 00:09:08.049 the style of that consistency. This is so key. 00:09:08.210 --> 00:09:10.750 Image to video workflows are absolutely essential 00:09:10.750 --> 00:09:13.470 for professional output. Text describes motion. 00:09:14.110 --> 00:09:17.149 The image defines the style. Trying to describe 00:09:17.149 --> 00:09:20.450 a really specific visual look with a just text 00:09:20.450 --> 00:09:23.490 -like hyper -detailed cinematic photorealism 00:09:23.490 --> 00:09:27.309 with a 1980s neon color palette is so hard to 00:09:27.309 --> 00:09:29.309 do consistently. Yeah, you'll get 10 different 00:09:29.309 --> 00:09:31.169 versions of that. The professional workflow is 00:09:31.169 --> 00:09:34.090 this. You generate a high -quality base image 00:09:34.090 --> 00:09:37.549 first. Use something like Midjourney or Dully. 00:09:37.889 --> 00:09:41.029 That image establishes the exact composition, 00:09:41.409 --> 00:09:43.870 the lighting, the style you want. Then you just 00:09:43.870 --> 00:09:46.370 feed that image into your video tool, like Clang 00:09:46.370 --> 00:09:49.870 or Runway, and use a simple text prompt to describe 00:09:49.870 --> 00:09:52.710 the motion. Something like, the character slowly 00:09:52.710 --> 00:09:55.840 raises their head. The model keeps the style 00:09:55.840 --> 00:09:58.419 from the image, and the text just handles the 00:09:58.419 --> 00:10:00.460 action. And this is the secret for consistent 00:10:00.460 --> 00:10:03.200 characters. You don't try to generate a 30 -second 00:10:03.200 --> 00:10:05.080 video of your character walking and talking. 00:10:05.240 --> 00:10:07.340 You generate a perfect, high -fidelity -based 00:10:07.340 --> 00:10:09.940 image of them first. Then you use an image tool 00:10:09.940 --> 00:10:12.480 to generate variations, a side profile of them 00:10:12.480 --> 00:10:15.220 sitting, an action pose. And then you animate 00:10:15.220 --> 00:10:17.019 each of those images separately, so you end up 00:10:17.019 --> 00:10:18.919 with five different perfectly consistent shots 00:10:18.919 --> 00:10:21.120 of the same person. So which is more critical 00:10:21.120 --> 00:10:24.269 for setting the visual tone? The image or the 00:10:24.269 --> 00:10:27.250 text? The image defines the style, and the text 00:10:27.250 --> 00:10:29.990 describes the motion. Simple as that. Okay, last 00:10:29.990 --> 00:10:33.470 one. Prompt style number seven, negative plumps. 00:10:34.000 --> 00:10:36.820 This is often the easiest and fastest tool for 00:10:36.820 --> 00:10:39.519 a quick surgical correction. Yeah, instead of 00:10:39.519 --> 00:10:41.279 adding more instructions, you just tell the AI 00:10:41.279 --> 00:10:43.700 what you don't want. There's super targeted corrections 00:10:43.700 --> 00:10:46.879 for when the AI makes some weird, unwanted assumption. 00:10:47.259 --> 00:10:49.580 Like if you're generating a futuristic city and 00:10:49.580 --> 00:10:52.360 the AI keeps adding flying cars and drones because 00:10:52.360 --> 00:10:54.679 that's what it thinks future means. Right, you 00:10:54.679 --> 00:10:56.860 just add a negative prompt, no flying vehicles, 00:10:57.120 --> 00:10:59.519 no drones, no traffic in the sky, and it just 00:10:59.519 --> 00:11:01.500 cleans the scene right up. They're also really 00:11:01.500 --> 00:11:03.639 good for sound control. A lot of these models 00:11:03.639 --> 00:11:05.840 are starting to add their own sound design now, 00:11:06.000 --> 00:11:08.220 which is terrible if you want to do professional 00:11:08.220 --> 00:11:10.580 audio in post. Exactly. So if you're generating 00:11:10.580 --> 00:11:14.620 a meditating monk and the AI adds, I don't know, 00:11:14.679 --> 00:11:17.240 dramatic chanting. You just add total silence, 00:11:17.379 --> 00:11:20.179 no chanting, no music, no ambient sound, and 00:11:20.179 --> 00:11:22.000 you get a clean slate for your own sound design. 00:11:22.299 --> 00:11:24.940 So when should negative comps be your first instinct? 00:11:25.220 --> 00:11:28.179 When the AI's default assumptions conflict sharply 00:11:28.179 --> 00:11:30.490 with your vision. Okay, that covers all seven 00:11:30.490 --> 00:11:33.070 styles. But these aren't just individual tricks. 00:11:33.330 --> 00:11:36.110 The real power is when you see how they all stack 00:11:36.110 --> 00:11:38.549 together. Yeah, let's do a quick recap of the 00:11:38.549 --> 00:11:41.429 function for each one. Cinematic. That's controlling 00:11:41.429 --> 00:11:43.929 the camera. I'm stamp rag. Choreographing your 00:11:43.929 --> 00:11:46.529 sequences. Cutscene. That's for visual variety. 00:11:46.909 --> 00:11:49.529 GP key prompts. Speeding up the writing, but 00:11:49.529 --> 00:11:52.110 with human oversight. Anchor prompts. Locking 00:11:52.110 --> 00:11:55.350 in all your details. Image prompts. That establishes 00:11:55.350 --> 00:11:58.519 your style and consistency. And finally... Negative 00:11:58.519 --> 00:12:01.179 prompts. That's for surgical removal. And when 00:12:01.179 --> 00:12:03.779 you look at how a 30 -second short film is built, 00:12:03.879 --> 00:12:07.019 you see this integrated approach. You don't start 00:12:07.019 --> 00:12:09.080 from scratch. You start with control. Right. 00:12:09.240 --> 00:12:11.899 A single shot might begin with a strong image 00:12:11.899 --> 00:12:15.120 prompt to set the aesthetic. Then you layer an 00:12:15.120 --> 00:12:18.059 anchor prompt on top to lock down a crucial detail 00:12:18.059 --> 00:12:21.080 like ash on the character's armor. And you finish 00:12:21.080 --> 00:12:23.580 it with a negative prompt to block any distracting 00:12:23.580 --> 00:12:26.059 music that AI might want to add. And then you 00:12:26.059 --> 00:12:27.940 move to the next shot where you dictate the timing 00:12:27.940 --> 00:12:30.399 with a timestamp and you integrate a cutscene 00:12:30.399 --> 00:12:32.799 halfway through to jump to a close -up for dramatic 00:12:32.799 --> 00:12:36.289 pacing. This whole framework just moves you completely 00:12:36.289 --> 00:12:40.009 beyond guessing. It's repeatable creative control. 00:12:40.230 --> 00:12:43.309 This is what separates hobbyist work from professional 00:12:43.309 --> 00:12:46.230 asset creation. And when you think about that, 00:12:46.330 --> 00:12:50.190 whoa. I mean, imagine scaling that level of precise 00:12:50.190 --> 00:12:54.169 directorial control across hundreds of assets. 00:12:54.330 --> 00:12:56.529 For a full -length project, the efficiency gain 00:12:56.529 --> 00:12:59.649 is just massive. That's the main takeaway here. 00:12:59.830 --> 00:13:02.190 Prompting is a craft. It needs a directorial 00:13:02.190 --> 00:13:05.159 structure. And the future of creative video production, 00:13:05.340 --> 00:13:07.820 it's going to depend entirely on people who understand 00:13:07.820 --> 00:13:10.639 how to direct the AI, not just use it like a 00:13:10.639 --> 00:13:13.440 magic button. These seven styles are the foundational 00:13:13.440 --> 00:13:16.039 techniques that every pro -AI filmmaker is using 00:13:16.039 --> 00:13:18.419 right now. Master these, and you will fundamentally 00:13:18.419 --> 00:13:20.399 change the quality and consistency of everything 00:13:20.399 --> 00:13:22.840 you make. So now that you know how to truly direct, 00:13:23.139 --> 00:13:25.580 what is the first impossible shot you were going 00:13:25.580 --> 00:13:27.679 to create, something that was visually out of 00:13:27.679 --> 00:13:29.659 your reach before this? We look forward to seeing 00:13:29.659 --> 00:13:30.120 what you build.