WEBVTT 00:00:00.000 --> 00:00:03.020 You know that specific kind of exhaustion? The 00:00:03.020 --> 00:00:05.480 one that comes from editing video. Oh, the Sunday 00:00:05.480 --> 00:00:07.740 night dread. I know it well. Exactly. The Sunday 00:00:07.740 --> 00:00:09.960 night dread. You spend an entire weekend, you're 00:00:09.960 --> 00:00:12.939 fighting with clips trying to match audio. And 00:00:12.939 --> 00:00:14.640 you end up with what, like a minute and a half 00:00:14.640 --> 00:00:17.519 of usable footage? If you're lucky, it's just, 00:00:17.620 --> 00:00:20.739 it's heavy. It's the friction. It is. It's the 00:00:20.739 --> 00:00:22.620 friction of the tools getting in the way of the 00:00:22.620 --> 00:00:25.039 actual idea. And that friction is, you know, 00:00:25.039 --> 00:00:27.219 what usually kills the idea before it even gets 00:00:27.219 --> 00:00:29.760 a chance to breathe. But I've been reading about 00:00:29.760 --> 00:00:31.940 this concept. It's called the speed gap. Right. 00:00:32.060 --> 00:00:35.460 It's this massive divide between that old manual 00:00:35.460 --> 00:00:39.219 way. The agonizing weekend edit and this new 00:00:39.219 --> 00:00:41.880 standard that's, well, it's quietly taking over. 00:00:42.000 --> 00:00:43.700 And we're not talking about just making things 00:00:43.700 --> 00:00:46.359 a little bit faster. No, not 10 % faster. We're 00:00:46.359 --> 00:00:49.200 talking about automating 50 consistent scenes 00:00:49.200 --> 00:00:52.439 in just minutes. It's a total collapse of the 00:00:52.439 --> 00:00:55.299 production timeline. And the craziest part about 00:00:55.299 --> 00:00:57.619 the speed gap. Go on. The bridge to cross it 00:00:57.619 --> 00:01:01.929 costs exactly $0. That's the hook. And that's 00:01:01.929 --> 00:01:04.250 what we're doing today on The Deep Dive. We are 00:01:04.250 --> 00:01:06.629 unpacking a roadmap for what's being called the 00:01:06.629 --> 00:01:12.829 2026 Free Stack. It's a guide on building a professional 00:01:12.829 --> 00:01:16.209 AI video production system, an actual assembly 00:01:16.209 --> 00:01:19.609 line from just a text prompt to a finished movie. 00:01:19.950 --> 00:01:21.930 And we should be really clear here. This is not 00:01:21.930 --> 00:01:24.469 about just typing make me a video into a chat 00:01:24.469 --> 00:01:26.689 bot. No, that doesn't work. It never works. This 00:01:26.689 --> 00:01:29.269 is about chaining together five very specific 00:01:29.269 --> 00:01:33.250 free tools. We're talking ChatGPT, a Chrome extension 00:01:33.250 --> 00:01:37.090 called AutoWhisk, Google AI Studio, Grok, and 00:01:37.090 --> 00:01:39.150 CapCut. It sounds like a Frankenstein's monster. 00:01:39.450 --> 00:01:41.390 It is a bit of a Frankenstein. But when you see 00:01:41.390 --> 00:01:43.230 how they all lock together, you realize, you 00:01:43.230 --> 00:01:45.030 know, this is the difference between an amateur 00:01:45.030 --> 00:01:47.569 messing with tech and a producer building an 00:01:47.569 --> 00:01:49.790 actual workflow. Okay. So let's walk through 00:01:49.790 --> 00:01:51.609 this factory floor. It's a seven step process 00:01:51.609 --> 00:01:53.829 and it starts where every movie starts. The script. 00:01:53.969 --> 00:01:56.450 The script. And usually for me, when I try to 00:01:56.450 --> 00:02:00.209 get AI to write a script, it just feels so generic. 00:02:00.769 --> 00:02:03.469 Hollow. Like plastic. Yeah. It gives you that 00:02:03.469 --> 00:02:07.590 bland corporate AI voice. The guide we're looking 00:02:07.590 --> 00:02:10.550 at attacks that blank page problem in a different 00:02:10.550 --> 00:02:12.849 way. It doesn't just ask for a story. What does 00:02:12.849 --> 00:02:16.409 it do? It uses this rigorous persona switching 00:02:16.409 --> 00:02:19.969 strategy. It treats the AI like a specialized 00:02:19.969 --> 00:02:23.150 employee you can hire and fire. Okay, explain 00:02:23.150 --> 00:02:25.669 that. Because most people just dump a paragraph 00:02:25.669 --> 00:02:28.330 into ChatGPT. You got to hope for the best. And 00:02:28.330 --> 00:02:30.590 you get garbage if you do that. So step one. 00:02:31.240 --> 00:02:34.340 You don't ask for a video script. You tell ChatGPT 00:02:34.340 --> 00:02:37.039 you are a professional children's author. You 00:02:37.039 --> 00:02:40.419 give it rules. Human characters, one animal, 00:02:40.659 --> 00:02:43.439 inspirational, no dialogue. You get the story 00:02:43.439 --> 00:02:45.340 first. So you lock in the narrative structure 00:02:45.340 --> 00:02:47.099 before you even think about the visuals. That 00:02:47.099 --> 00:02:49.020 makes a lot of sense. Precisely. But here's the 00:02:49.020 --> 00:02:50.599 technical pivot. And this is where most people 00:02:50.599 --> 00:02:52.800 fail. You don't take that story and just paste 00:02:52.800 --> 00:02:54.719 it into an image generator. You have to clean 00:02:54.719 --> 00:02:57.020 the data. You have to. You go back to ChatGPT 00:02:57.020 --> 00:02:59.539 and you say, OK, now switch hats. You are an 00:02:59.539 --> 00:03:01.740 experienced animation director. I like that. 00:03:01.819 --> 00:03:03.740 You're firing the author and hiring a director 00:03:03.740 --> 00:03:06.650 for the next task. You are. And you tell it to 00:03:06.650 --> 00:03:09.229 break that story down into 20 storyboard scenes, 00:03:09.449 --> 00:03:12.729 but the prompt engineering here is very, very 00:03:12.729 --> 00:03:16.750 strict. How so? The output has to separate the 00:03:16.750 --> 00:03:18.909 narration, what the audience is going to hear, 00:03:19.069 --> 00:03:22.569 from the image prompt, what the AI needs to see. 00:03:22.650 --> 00:03:25.629 Why is that separation so critical for the system? 00:03:25.729 --> 00:03:28.469 Can't you just describe the scene? Well, if you 00:03:28.469 --> 00:03:30.729 mix them, the image generator gets confused. 00:03:31.520 --> 00:03:33.860 It sees the emotional language of the story and 00:03:33.860 --> 00:03:35.900 doesn't know what to do. You need the image prompt 00:03:35.900 --> 00:03:39.500 to be cold, descriptive data like Disney Pixar 00:03:39.500 --> 00:03:42.860 style, wide shot, warm lighting. I see. Completely 00:03:42.860 --> 00:03:45.139 distinct from the narration. Totally. And then 00:03:45.139 --> 00:03:46.939 there's a third little step in the scripting 00:03:46.939 --> 00:03:49.860 phase, the extraction. Right. This is pure data 00:03:49.860 --> 00:03:52.199 prep. You tell the AI to strip away everything 00:03:52.199 --> 00:03:55.080 else, see numbers, headers, all of it, and just 00:03:55.080 --> 00:03:57.599 give you the raw visual descriptions, each one 00:03:57.599 --> 00:04:00.150 separated by a blank line. It feels less like 00:04:00.150 --> 00:04:02.110 writing at that point and more like, I don't 00:04:02.110 --> 00:04:03.949 know, coding. You're just preparing the raw material. 00:04:04.229 --> 00:04:06.810 That's exactly what it is. So why is that formatting, 00:04:06.990 --> 00:04:08.710 that separation of the prompts, so critical? 00:04:09.009 --> 00:04:12.009 Clean data allows the AutoWhisk tool to read 00:04:12.009 --> 00:04:14.629 distinct instructions without any manual tagging. 00:04:14.729 --> 00:04:16.870 Okay, which brings us to the engine room. Step 00:04:16.870 --> 00:04:20.850 two, bulk velocity. This is where we stop making 00:04:20.850 --> 00:04:24.029 images one by one. Yeah, this is where AutoWhisk 00:04:24.029 --> 00:04:26.389 comes in. Tell me about this tool. This is what 00:04:26.389 --> 00:04:28.149 really creates that speed gap we were talking 00:04:28.149 --> 00:04:31.050 about. auto whisk is a chrome extension and it 00:04:31.050 --> 00:04:33.209 just sits right on top of google whisk and google 00:04:33.209 --> 00:04:35.910 whisk is the actual image generation engine that's 00:04:35.910 --> 00:04:38.550 right and it's free currently the quality is 00:04:38.550 --> 00:04:40.750 surprisingly high too if you use the settings 00:04:40.750 --> 00:04:44.329 from the guide english version 7 .6 .0 aspect 00:04:44.329 --> 00:04:47.189 ratio 16 .9 pretty standard stuff very standard 00:04:47.189 --> 00:04:50.389 but the extension is the absolute game changer 00:04:50.389 --> 00:04:52.920 because it automates all the clicking it automates 00:04:52.920 --> 00:04:56.019 the entire batch you take those 20 clean prompts 00:04:56.019 --> 00:04:58.339 you extracted you paste them all into the extension 00:04:58.339 --> 00:05:00.699 at once and you just hit start and it does the 00:05:00.699 --> 00:05:03.759 rest the extension sees the line breaks it feeds 00:05:03.759 --> 00:05:06.459 them into the engine one by one generates the 00:05:06.459 --> 00:05:08.240 image and then downloads it straight to your 00:05:08.240 --> 00:05:10.319 hard drive i actually laughed when i read the 00:05:10.319 --> 00:05:12.160 warning in the source material for this step 00:05:12.160 --> 00:05:16.180 it says very strictly Don't touch your mouse. 00:05:16.379 --> 00:05:19.279 It's serious. The extension is literally simulating 00:05:19.279 --> 00:05:21.680 you clicking and typing. It's kind of hijacking 00:05:21.680 --> 00:05:24.639 your cursor. So if you tab away to check an email 00:05:24.639 --> 00:05:26.480 or something. You break the loop. You have to 00:05:26.480 --> 00:05:28.480 just sit there and watch the little file count 00:05:28.480 --> 00:05:30.899 go up in your download folder. It's kind of mesmerizing. 00:05:31.180 --> 00:05:33.560 It's a funny image. Surrendering control of your 00:05:33.560 --> 00:05:36.199 computer to gain all this speed. But this is 00:05:36.199 --> 00:05:39.220 where the guide introduces a huge problem. A 00:05:39.220 --> 00:05:41.300 critical problem. You have a speed. You've got 00:05:41.300 --> 00:05:45.769 20 beautiful images. But they don't look like 00:05:45.769 --> 00:05:47.709 they belong in the same movie. The consistency 00:05:47.709 --> 00:05:51.889 problem. This is the bane of AI video. Right. 00:05:51.930 --> 00:05:54.350 In scene one, your main character is a boy in 00:05:54.350 --> 00:05:57.089 a blue hoodie. Scene two, the AI decides he's 00:05:57.089 --> 00:05:59.490 wearing a red jacket. Scene three, he's suddenly 00:05:59.490 --> 00:06:02.189 Asian. Scene four, he's a cartoon. It's just 00:06:02.189 --> 00:06:04.930 chaos. It's the hallmark of what people call 00:06:04.930 --> 00:06:07.509 AI slop. It looks like a fever dream. It does. 00:06:07.689 --> 00:06:11.269 The AI has no object permanence. It has no memory 00:06:11.269 --> 00:06:13.629 of who the character was five seconds ago. So 00:06:13.629 --> 00:06:17.029 speed without control creates chaos. What is 00:06:17.029 --> 00:06:19.329 the missing variable here? A reference image. 00:06:19.529 --> 00:06:22.389 Without it, the AI hallucinates a new protagonist 00:06:22.389 --> 00:06:26.290 every single time. And this brings us to what 00:06:26.290 --> 00:06:29.310 the guide calls the secret sauce. It's step three. 00:06:29.750 --> 00:06:32.610 And honestly, this feels like the most vital 00:06:32.610 --> 00:06:35.149 part of the entire workflow. It really is. This 00:06:35.149 --> 00:06:38.689 is the barrier between amateur slop. and professional 00:06:38.689 --> 00:06:41.529 storytelling. So before you run that bulk batch, 00:06:41.750 --> 00:06:44.029 you have to create a reference anchor. You do. 00:06:44.149 --> 00:06:46.529 And the fix is actually pretty clever. You go 00:06:46.529 --> 00:06:49.389 back to ChatGPT again. Okay. And you ask it to 00:06:49.389 --> 00:06:51.470 generate a character prompt, but specifically 00:06:51.470 --> 00:06:53.689 on a white background. White background. Why? 00:06:53.870 --> 00:06:56.610 It isolates the features. It tells the AI, focus 00:06:56.610 --> 00:06:59.209 only on the face, the clothes, the identity of 00:06:59.209 --> 00:07:01.470 this character. You generate just one good image 00:07:01.470 --> 00:07:04.649 of, say, Mila or the brown dog. So you're creating 00:07:04.649 --> 00:07:06.930 your cast's headshots, essentially. That's a 00:07:06.930 --> 00:07:08.649 perfect... analogy you download this now you 00:07:08.649 --> 00:07:10.829 go back to the auto whisk extension but this 00:07:10.829 --> 00:07:13.430 time before you paste in your 20 scene prompts 00:07:13.430 --> 00:07:16.490 let me guess there's a button there's a reference 00:07:16.490 --> 00:07:19.089 image option you click it you upload that file 00:07:19.089 --> 00:07:21.250 of mila you're telling the system this is mila 00:07:21.250 --> 00:07:24.370 then you run the bulk generation so you are anchoring 00:07:24.370 --> 00:07:27.410 the ai's imagination you're saying paint whatever 00:07:27.410 --> 00:07:29.730 scene you want but the person in the middle has 00:07:29.730 --> 00:07:32.870 to look like this precisely The AI forces every 00:07:32.870 --> 00:07:36.110 new scene to match that uploaded face and visual 00:07:36.110 --> 00:07:39.269 identity. It connects the dots for you. It feels 00:07:39.269 --> 00:07:41.910 like anchoring the AI's imagination. How much 00:07:41.910 --> 00:07:45.029 time does this step add? It adds about 15 minutes, 00:07:45.089 --> 00:07:46.730 but it's the difference between amateur slop 00:07:46.730 --> 00:07:49.410 and professional storytelling. 15 minutes to 00:07:49.410 --> 00:07:51.709 say the soul of the story. I think that's a trade 00:07:51.709 --> 00:07:54.329 -off most of us would take. Okay, so visuals 00:07:54.329 --> 00:07:58.720 are locked, but video is 50 % audio. And there 00:07:58.720 --> 00:08:02.240 is nothing worse than that robotic, glitchy AI 00:08:02.240 --> 00:08:04.339 voice. Or the one that sounds like a GPS trying 00:08:04.339 --> 00:08:06.879 to read a bedtime story. It's awful. The guide 00:08:06.879 --> 00:08:10.060 pivots here to Google AI Studio, specifically 00:08:10.060 --> 00:08:12.600 using their Gemini model. Okay, why this tool? 00:08:12.639 --> 00:08:13.819 I mean, there are a million voice generators 00:08:13.819 --> 00:08:17.610 out there. A few reasons. First, it's free. No 00:08:17.610 --> 00:08:19.310 credit limits, which is huge when you're just 00:08:19.310 --> 00:08:22.269 trying things out. But technically, the big advantage 00:08:22.269 --> 00:08:25.149 is it supports long -form text. So you can paste 00:08:25.149 --> 00:08:27.170 the whole story in at once. The entire thing. 00:08:27.310 --> 00:08:30.439 And the quality is surprisingly high. The guide 00:08:30.439 --> 00:08:34.460 recommends the Gemini 2 .5 Flash Preview TTS 00:08:34.460 --> 00:08:37.899 model and a voice called Enceladus. Enceladus. 00:08:37.980 --> 00:08:40.720 Yeah, it's described as warm and friendly, but 00:08:40.720 --> 00:08:43.220 the real trick is the style instruction. You 00:08:43.220 --> 00:08:45.720 actually type into the prompt, read the story 00:08:45.720 --> 00:08:48.600 aloud using a warm, gentle, and engaging storytelling 00:08:48.600 --> 00:08:50.940 voice appropriate for children. You're directing 00:08:50.940 --> 00:08:53.580 the actor, not just the software. Exactly, and 00:08:53.580 --> 00:08:55.799 because you generate it all in one go, you get 00:08:55.799 --> 00:08:58.399 a single audio track. Why is generating the full 00:08:58.399 --> 00:09:00.860 narrative at once? better than scene -by -scene 00:09:00.860 --> 00:09:03.480 audio. It maintains natural pacing and emotional 00:09:03.480 --> 00:09:06.639 continuity, avoiding that disjointed, choppy 00:09:06.639 --> 00:09:09.000 AI sound. Okay, let's just unpack where we are. 00:09:09.059 --> 00:09:11.179 We have consistent images, we have a warm, flowing 00:09:11.179 --> 00:09:14.899 voiceover, but we still have basically a slideshow. 00:09:14.980 --> 00:09:17.860 Right. It's a series of still images, and static 00:09:17.860 --> 00:09:21.279 images are boring. To compete in 2026, you need 00:09:21.279 --> 00:09:24.120 motion. For sure. This is where we bring in Grok, 00:09:24.220 --> 00:09:27.220 specifically their Imagine feature, to add that 00:09:27.220 --> 00:09:30.019 life. this is step six in the guide breathing 00:09:30.019 --> 00:09:33.009 life yeah but it's not just hitting a button 00:09:33.009 --> 00:09:36.009 that says animate, is it? No. And that is a really 00:09:36.009 --> 00:09:37.769 important distinction. If you just let the AI 00:09:37.769 --> 00:09:40.889 guess, you get weird warping or these random 00:09:40.889 --> 00:09:44.009 nauseating zooms. So you need control. You need 00:09:44.009 --> 00:09:46.470 control. The system here relies on a control 00:09:46.470 --> 00:09:48.929 mechanism. The source mentions having a text 00:09:48.929 --> 00:09:52.090 file with about 38 specific cinematic camera 00:09:52.090 --> 00:09:55.070 techniques. Like pan left, dolly zoom, that kind 00:09:55.070 --> 00:09:57.629 of thing. Exactly. So you go back to ChatGPT 00:09:57.629 --> 00:10:00.169 briefly. You ask it to look at your script and 00:10:00.169 --> 00:10:02.649 assign a camera movement to each scene based 00:10:02.649 --> 00:10:05.750 on the emotion. So a sad scene gets a slow zoom, 00:10:05.950 --> 00:10:08.330 an action scene gets a quick pan. You got it. 00:10:08.549 --> 00:10:10.549 Then you take those instructions over to Grok, 00:10:10.629 --> 00:10:13.269 but, and this is a big but, the guide points 00:10:13.269 --> 00:10:15.389 out a very specific setting you have to change 00:10:15.389 --> 00:10:17.649 first. What is it? You have to go to your profile, 00:10:17.850 --> 00:10:20.210 then settings, then behavior, and you have to 00:10:20.210 --> 00:10:24.230 turn off enable automatic video generation. Wait, 00:10:24.250 --> 00:10:25.809 why would you turn off the automation? Isn't 00:10:25.809 --> 00:10:28.879 that the whole point? That's the pro move. By 00:10:28.879 --> 00:10:31.279 turning off the automation, you regain the ability 00:10:31.279 --> 00:10:34.480 to paste in your specific command. You drag in 00:10:34.480 --> 00:10:37.320 your image, you paste the movement prompt, slow 00:10:37.320 --> 00:10:40.059 zoom in on the character's face, and then you 00:10:40.059 --> 00:10:43.100 generate. We are injecting human intent back 00:10:43.100 --> 00:10:45.200 into the machine here. What is the result? It 00:10:45.200 --> 00:10:47.500 stops looking like a slideshow and starts looking 00:10:47.500 --> 00:10:50.440 like a movie with intentional direction. So we 00:10:50.440 --> 00:10:53.139 have all the pieces, the animated clips, the 00:10:53.139 --> 00:10:56.559 audio. Now comes the assembly line. Steps 5 and 00:10:56.559 --> 00:10:58.840 7 in the guide kind of merge here in the editing 00:10:58.840 --> 00:11:00.779 workflow. Yeah, this is where it all comes together 00:11:00.779 --> 00:11:03.940 in CapCut. And the workflow is, again, designed 00:11:03.940 --> 00:11:06.860 for speed. It uses a two -pass system. A two 00:11:06.860 --> 00:11:08.799 -pass system. How does that work? First pass, 00:11:08.980 --> 00:11:10.480 you don't even wait for the videos to finish 00:11:10.480 --> 00:11:12.480 generating. You just drag your audio track and 00:11:12.480 --> 00:11:14.700 all your still images into the timeline. Okay. 00:11:14.779 --> 00:11:17.659 You sync them up. You listen to the narration, 00:11:18.000 --> 00:11:19.960 find the end of a sentence, and you snap the 00:11:19.960 --> 00:11:21.960 next image to it. You build the whole rhythm 00:11:21.960 --> 00:11:23.879 of the video with just the stills. That makes 00:11:23.879 --> 00:11:26.279 sense. It's much faster to edit photos than video 00:11:26.279 --> 00:11:30.059 files. So much faster. Then, once your grok videos 00:11:30.059 --> 00:11:33.059 are ready, you do the swap. The swap. You just 00:11:33.059 --> 00:11:35.039 right -click the still image in your timeline. 00:11:35.179 --> 00:11:38.759 You choose Replace Clip, and you select the animated 00:11:38.759 --> 00:11:41.200 video file. It keeps the timing. It keeps the 00:11:41.200 --> 00:11:43.940 transitions. But it just upgrades the visual 00:11:43.940 --> 00:11:47.220 from a photo to a movie. You know, I have to 00:11:47.220 --> 00:11:49.960 admit something here. Yeah. Reading through this 00:11:49.960 --> 00:11:52.860 whole process, I thought, this is amazing. But 00:11:52.860 --> 00:11:55.679 I also know myself. I know that if I were doing 00:11:55.679 --> 00:11:58.220 this, I'd get lazy. I'd see a generated clip 00:11:58.220 --> 00:12:02.840 that was just okay. Maybe the character's eye 00:12:02.840 --> 00:12:04.899 is a little wonky and I'd be tempted to just 00:12:04.899 --> 00:12:08.399 leave it. I still wrestled with that prompt drift 00:12:08.399 --> 00:12:10.620 myself. You and everyone else. And the source 00:12:10.620 --> 00:12:12.299 actually calls this out in the quality control 00:12:12.299 --> 00:12:14.679 section. It's like the vulnerable admission of 00:12:14.679 --> 00:12:16.799 the whole system. Right. It says even with all 00:12:16.799 --> 00:12:18.980 this automation, if you skip the manual review, 00:12:19.159 --> 00:12:21.500 you break the spell. It says specifically do 00:12:21.500 --> 00:12:24.539 not trust the system blindly. Exactly. If an 00:12:24.539 --> 00:12:26.899 image is off, you have to go back and regenerate 00:12:26.899 --> 00:12:29.120 just that one. That's the discipline, isn't it? 00:12:29.159 --> 00:12:31.799 The tools remove the manual labor, but they can't 00:12:31.799 --> 00:12:34.049 remove the need for good taste. They can't. You 00:12:34.049 --> 00:12:36.149 have to be the final curator. Because if you 00:12:36.149 --> 00:12:38.909 let a glitchy face slide, the audience immediately 00:12:38.909 --> 00:12:41.750 clocks it. They know it's low effort junk. So 00:12:41.750 --> 00:12:44.250 the human role shifts from maker to editor. What 00:12:44.250 --> 00:12:46.490 happens if you skip the review? You lose trust. 00:12:46.929 --> 00:12:49.570 Small glitches compound and the audience immediately 00:12:49.570 --> 00:12:52.110 senses it's low effort junk. We are going to 00:12:52.110 --> 00:12:54.009 take a quick break, but when we come back, we 00:12:54.009 --> 00:12:56.490 are going to look at the big picture. What this 00:12:56.490 --> 00:13:00.049 whole speed gap really means for the future of 00:13:00.049 --> 00:13:03.289 creativity. Mid -roll, sponsor, placeholder. 00:13:03.629 --> 00:13:06.389 Okay, let's recap this stack because it is a 00:13:06.389 --> 00:13:08.950 lot of moving parts, but they do fit together 00:13:08.950 --> 00:13:11.529 so beautifully. They really do. Think of it like 00:13:11.529 --> 00:13:14.990 a relay race. First, ChatGPT handles the structure. 00:13:15.090 --> 00:13:17.710 It gives you the story and those clean, separated 00:13:17.710 --> 00:13:20.049 prompts. Right, the raw material. Then the baton 00:13:20.049 --> 00:13:22.269 goes to AutoWhisk and GoogleWhisk. They handle 00:13:22.269 --> 00:13:24.389 the bulk visuals, and you use those reference 00:13:24.389 --> 00:13:26.610 images to keep the characters consistent. Got 00:13:26.610 --> 00:13:30.250 it. Then audio. Third is Gemini AI Studio, which 00:13:30.250 --> 00:13:33.850 gives us that warm single track voiceover. Fourth, 00:13:34.090 --> 00:13:37.269 Grok takes those still images and adds that controlled 00:13:37.269 --> 00:13:40.549 cinematic motion. And then finally, CapCut. And 00:13:40.549 --> 00:13:43.129 finally, CapCut is where you assemble it all 00:13:43.129 --> 00:13:46.330 and do that quick swap workflow. It's an impressive 00:13:46.330 --> 00:13:49.710 system. Yeah. But I want to zoom out to the big 00:13:49.710 --> 00:13:52.230 idea here. The guide starts by talking about 00:13:52.230 --> 00:13:54.919 the speed gap. It's a powerful concept. It is. 00:13:54.980 --> 00:13:58.080 The idea is that the barrier to entry for making 00:13:58.080 --> 00:14:00.480 video has basically dropped to zero dollars. 00:14:00.980 --> 00:14:03.679 Anyone can get these tools. Right. But the barrier 00:14:03.679 --> 00:14:06.500 to quality has shifted. It's no longer about 00:14:06.500 --> 00:14:08.980 who has the most expensive camera or software. 00:14:09.340 --> 00:14:11.860 It's about who has the best process, the best 00:14:11.860 --> 00:14:13.899 system. Exactly. The winners aren't going to 00:14:13.899 --> 00:14:16.120 be the ones who just use AI. Everybody's going 00:14:16.120 --> 00:14:18.080 to use AI. The winners are the ones who build 00:14:18.080 --> 00:14:20.879 a system like this that allows them to fail faster 00:14:20.879 --> 00:14:23.549 and succeed more often. What do you mean by that? 00:14:23.669 --> 00:14:25.669 Well, if you can make a pretty good video in 00:14:25.669 --> 00:14:28.350 15 minutes, you can afford to make five bad ones 00:14:28.350 --> 00:14:31.269 to find that one great one. That is a luxury 00:14:31.269 --> 00:14:33.690 you just don't have when one video takes you 00:14:33.690 --> 00:14:36.690 an entire week. It changes the economics of creativity 00:14:36.690 --> 00:14:39.690 itself. You're not so precious about the output 00:14:39.690 --> 00:14:41.809 anymore. No, you're focused on the pipeline, 00:14:41.990 --> 00:14:43.809 and that pipeline is what lets you actually tell 00:14:43.809 --> 00:14:46.309 stories instead of just, you know, managing files 00:14:46.309 --> 00:14:48.990 all day. So here is the challenge for you listening. 00:14:50.000 --> 00:14:52.000 Don't try to build the whole Hollywood studio 00:14:52.000 --> 00:14:54.679 today. No, start small. Just install the Autowisk 00:14:54.679 --> 00:14:57.759 extension. That's it. Generate one story script 00:14:57.759 --> 00:15:00.759 using that specific three -prompt structure we 00:15:00.759 --> 00:15:03.100 talked about. Don't overthink it. Just start 00:15:03.100 --> 00:15:05.919 the system. Watch the files download. See that 00:15:05.919 --> 00:15:08.480 magic moment for yourself. And here's a thought 00:15:08.480 --> 00:15:12.659 to leave you with. If you can produce 10 professional 00:15:12.659 --> 00:15:14.799 -looking videos in the time it used to take to 00:15:14.799 --> 00:15:18.440 make just one, what happens to the value of the 00:15:18.440 --> 00:15:22.100 video itself? Does scarcity even matter anymore 00:15:22.100 --> 00:15:25.019 or is it now finally all about the story? That 00:15:25.019 --> 00:15:27.019 is the question. Thanks for listening to The 00:15:27.019 --> 00:15:28.419 Deep Dive. We'll see you in the next one.