WEBVTT 00:00:00.000 --> 00:00:02.419 have you ever been scrolling you know maybe late 00:00:02.419 --> 00:00:05.820 at night and you just stop you see this ai video 00:00:05.820 --> 00:00:07.940 it's kind of astonishing uh maybe it's glass 00:00:07.940 --> 00:00:10.560 breaking with these impossible sounds or like 00:00:10.800 --> 00:00:13.640 epic cinematic shot millions of views and you 00:00:13.640 --> 00:00:16.300 just think how like what's the secret there if 00:00:16.300 --> 00:00:18.480 you've felt that that sort of invisible wall 00:00:18.480 --> 00:00:20.739 between your ideas and getting that high quality 00:00:20.739 --> 00:00:23.420 viral AI stuff out there well you're not alone 00:00:23.420 --> 00:00:26.480 today that wall starts coming down welcome to 00:00:26.480 --> 00:00:29.059 the deep dive our mission today is to give you 00:00:29.059 --> 00:00:32.340 a detailed map really a clear compass for mastering 00:00:32.340 --> 00:00:35.399 AI video generation we're going to dig into Google 00:00:35.399 --> 00:00:38.060 VEO3 it's a breakthrough tool and it's really 00:00:38.060 --> 00:00:40.380 reshaping how we think about creating video Yeah, 00:00:40.439 --> 00:00:42.380 absolutely. And we're jumping right in. We'll 00:00:42.380 --> 00:00:45.719 unpack what VEO3 actually is, how it works, fundamentally. 00:00:46.039 --> 00:00:48.240 And we'll shift to thinking like an AI director 00:00:48.240 --> 00:00:50.299 when you prompt. Plus, we're going to lay out 00:00:50.299 --> 00:00:52.460 some proven viral video formats you can use, 00:00:52.859 --> 00:00:54.920 like right away we'll share advanced techniques, 00:00:55.159 --> 00:00:57.560 talk post -production polish, and touch on the 00:00:57.560 --> 00:00:59.020 ethics and the current limits. It's the whole 00:00:59.020 --> 00:01:01.859 journey, basically. Idea to a viral potential. 00:01:02.140 --> 00:01:04.680 It really is about moving past just that, wow, 00:01:04.739 --> 00:01:07.579 how did they do that? Moment to actually understanding 00:01:07.579 --> 00:01:10.099 the craft behind these things exactly and to 00:01:10.099 --> 00:01:12.819 really get VEOs power you gotta understand It's 00:01:12.819 --> 00:01:15.859 way more than just text -to -video VEOs this 00:01:15.859 --> 00:01:19.840 advanced AI It has like deep contextual understanding. 00:01:19.840 --> 00:01:21.739 It doesn't just read words It tries to figure 00:01:21.739 --> 00:01:24.120 out your intent the emotion the whole aesthetic 00:01:24.120 --> 00:01:27.319 you're going for what makes VEO 3 feel like a 00:01:27.319 --> 00:01:30.439 revolution from what we're seeing is Couple of 00:01:30.439 --> 00:01:32.599 key things, first, that native synchronized audio, 00:01:32.659 --> 00:01:34.900 that's a game changer, seriously. Other tools, 00:01:34.920 --> 00:01:36.540 you're kind of scrambling for sound afterwards. 00:01:36.819 --> 00:01:39.019 VEO can generate the whole soundscape right from 00:01:39.019 --> 00:01:41.579 the prompt. Dialogue with pretty decent lip sync, 00:01:41.859 --> 00:01:43.939 sound effects, footsteps, glass shattering, ambient 00:01:43.939 --> 00:01:47.500 noise, wind, traffic, all synced up. It's a huge 00:01:47.500 --> 00:01:49.260 leap forward. And it's not just sound, is it? 00:01:49.280 --> 00:01:51.840 It almost feels like it understands physics and 00:01:51.840 --> 00:01:54.780 film language. It does. It's got this intuitive 00:01:54.780 --> 00:01:57.700 physics simulation, plus what we're calling cinematic 00:01:57.700 --> 00:02:00.819 literacy. VEO. gets how the world works. Water 00:02:00.819 --> 00:02:03.760 flows, fabric billows, things have weight. But 00:02:03.760 --> 00:02:06.680 what's maybe more fascinating, it speaks filmmaker. 00:02:07.180 --> 00:02:09.659 You can ask for a time lapse, an aerial shot, 00:02:10.020 --> 00:02:12.680 dolly zoom, and it actually translates those 00:02:12.680 --> 00:02:15.180 into proper camera moves. It's not just making 00:02:15.180 --> 00:02:17.400 video, it's kind of directing it. So that really 00:02:17.400 --> 00:02:19.319 demands a whole different way of thinking then. 00:02:19.479 --> 00:02:21.860 This isn't just typing a few words and, you know, 00:02:22.060 --> 00:02:23.819 crossing your fingers. You have to come at it 00:02:23.819 --> 00:02:27.129 with this AI director mindset. Every prompt isn't 00:02:27.129 --> 00:02:30.490 just a request. It's a clear directive to a virtual 00:02:30.490 --> 00:02:32.530 crew. You're the director, screenwriter, sound 00:02:32.530 --> 00:02:35.050 designer, all rolled into one. It's like VEO 00:02:35.050 --> 00:02:38.030 is just waiting for you to command it properly. 00:02:38.610 --> 00:02:41.490 So that deep contextual understanding, what does 00:02:41.490 --> 00:02:43.310 that actually look like when we're trying to 00:02:43.310 --> 00:02:46.110 create? Does it really grasp what we want beyond 00:02:46.110 --> 00:02:48.250 just the key words? It interprets the intent, 00:02:48.349 --> 00:02:50.689 the emotion, and the aesthetic you want to convey. 00:02:50.830 --> 00:02:52.810 That deeper understanding. Yeah, that makes the 00:02:52.810 --> 00:02:54.789 whole prompting philosophy just incredibly important. 00:02:55.069 --> 00:02:57.349 Oh, absolutely. In generative AI, the prompt 00:02:57.349 --> 00:02:59.689 is, well, it's everything. A good prompt is like 00:02:59.689 --> 00:03:01.909 the bridge from your imagination to what the 00:03:01.909 --> 00:03:04.069 AI can actually make. Think of it like a really 00:03:04.069 --> 00:03:06.669 detailed blueprint, a precise directive for that 00:03:06.669 --> 00:03:08.830 virtual crew we talked about. Got to be clear, 00:03:09.389 --> 00:03:11.990 layered, no room for guesswork. And one really 00:03:11.990 --> 00:03:13.770 effective way to build those tromps is what we're 00:03:13.770 --> 00:03:16.120 calling the layering technique. It's kind of 00:03:16.120 --> 00:03:18.840 like an artist sketching first, then adding the 00:03:18.840 --> 00:03:21.520 detail and color. So layer one, the foundation, 00:03:21.800 --> 00:03:24.360 that's your shot type, your aesthetic. The framework, 00:03:24.580 --> 00:03:27.520 perspective, style, the mood. So maybe you start 00:03:27.520 --> 00:03:30.759 with cinematic wide shot, epic fantasy aesthetic. 00:03:31.419 --> 00:03:34.020 Then layer two, the setting, build the world. 00:03:34.360 --> 00:03:36.599 Where's it happening? More detail, more believable 00:03:36.599 --> 00:03:40.830 video. So adding on. In a sprawling ancient forest 00:03:40.830 --> 00:03:43.689 at twilight, mist clings to the moss -covered 00:03:43.689 --> 00:03:46.050 trees. Okay, then layer three. Subject in action. 00:03:46.409 --> 00:03:48.270 Put your character or object in there. Give them 00:03:48.270 --> 00:03:51.069 something to do, a purpose. Continuing that example. 00:03:51.689 --> 00:03:53.849 A lone elven warrior with glowing silver armor 00:03:53.849 --> 00:03:56.090 kneels by a stream, dipping her hands into the 00:03:56.090 --> 00:03:59.430 water. Right. And finally, layer four. The fine 00:03:59.430 --> 00:04:01.490 details. These are the final touches. Give it 00:04:01.490 --> 00:04:04.469 soul. Lighting, textures, and importantly, the 00:04:04.469 --> 00:04:07.650 specific sounds. Finishing it off. The water 00:04:07.650 --> 00:04:10.169 ripples around her fingers. The only sounds of 00:04:10.169 --> 00:04:11.969 the gentle gurgle of the stream and the chirping 00:04:11.969 --> 00:04:15.270 of unseen insects. Build it like that layer by 00:04:15.270 --> 00:04:17.850 layer and you've got a really comprehensive directive 00:04:17.850 --> 00:04:20.569 for VEO. And it's not just the structure, right? 00:04:20.829 --> 00:04:24.189 It's the words themselves, the power of words. 00:04:24.730 --> 00:04:27.310 Your word choice makes a huge difference. The 00:04:27.310 --> 00:04:31.250 AI picks up on nuance, adjectives, verbs surprisingly 00:04:31.250 --> 00:04:33.490 well. Think about like, a man walks down the 00:04:33.490 --> 00:04:36.509 street, okay, vague, versus a weary old man shuffles 00:04:36.509 --> 00:04:39.189 down a rain -slick, neon -lit cobblestone alley, 00:04:39.410 --> 00:04:41.790 that second one. It tells a story, gives you 00:04:41.790 --> 00:04:43.910 a mood, a setting, a character. Use strong verbs, 00:04:44.069 --> 00:04:45.769 sensory adjectives. Yeah, definitely. And it 00:04:45.769 --> 00:04:47.870 really helps to practice thinking in frames. 00:04:48.079 --> 00:04:50.000 Before you even type, like seriously, close your 00:04:50.000 --> 00:04:51.959 eyes, visualize the shot, ask those director 00:04:51.959 --> 00:04:54.000 questions, where's the camera, moving, static, 00:04:54.160 --> 00:04:56.579 what lens, light source. Once you have that clear 00:04:56.579 --> 00:04:58.660 image translating into detailed language for 00:04:58.660 --> 00:05:01.730 the AI, much easier, much more effective. That's 00:05:01.730 --> 00:05:03.589 the real difference, isn't it? Between just getting 00:05:03.589 --> 00:05:06.110 something random back and intentionally creating 00:05:06.110 --> 00:05:08.829 an artistic vision. So to really nail that vision, 00:05:09.209 --> 00:05:12.490 why is visualizing the shot first so vital? It 00:05:12.490 --> 00:05:15.490 helps translate your clear image into effective 00:05:15.490 --> 00:05:19.110 detailed language for the AI. OK, now this is 00:05:19.110 --> 00:05:20.810 the core of today's deep dive. We're going to 00:05:20.810 --> 00:05:22.970 break down some video formats that, well, they've 00:05:22.970 --> 00:05:24.930 proven to have a serious viral potential online. 00:05:25.370 --> 00:05:27.970 For each one, we'll look at the what, the how, 00:05:28.250 --> 00:05:31.170 and crucially, the why. the psychology, and how 00:05:31.170 --> 00:05:33.310 VEO can nail them. All right, let's jump in. 00:05:33.350 --> 00:05:36.209 First one. Unrealistic object ASMR videos. They 00:05:36.209 --> 00:05:38.649 seem absolutely everywhere. They do. And they 00:05:38.649 --> 00:05:41.310 hit two things our brains just love. ASMR, that 00:05:41.310 --> 00:05:43.930 tingly feeling from certain sounds, and our fascination 00:05:43.930 --> 00:05:46.029 with the absurd. See, our brains know physics. 00:05:46.689 --> 00:05:49.250 So when VEO shows like a croissant, but made 00:05:49.250 --> 00:05:51.610 of obsidian, and then shatters it, it's this 00:05:51.610 --> 00:05:54.689 weird, safe violation of reality. You can't look 00:05:54.689 --> 00:05:57.089 away. VEO's physics engine lets us play that 00:05:57.089 --> 00:06:00.220 trick for the prompt. Camera angle, object material, 00:06:00.699 --> 00:06:02.360 the main destructive action, how it interacts 00:06:02.360 --> 00:06:04.439 with the environment, and that ASMR soundscape. 00:06:04.800 --> 00:06:06.759 Think of Siddy and Croissant crumbling sharp 00:06:06.759 --> 00:06:09.339 cracks, or a ruby pomegranate spilling cladowing 00:06:09.339 --> 00:06:11.699 gems. Pitfall, though. Keep the action simple 00:06:11.699 --> 00:06:13.879 in an eight -second shot. One decisive thing 00:06:13.879 --> 00:06:16.759 works best. Too much gets messy. Okay, next. 00:06:17.000 --> 00:06:18.740 Selfie -style vlogger videos. There's something 00:06:18.740 --> 00:06:21.079 really engaging about that format. Totally. Humans 00:06:21.079 --> 00:06:24.279 connect, right? This format taps into that parasocial 00:06:24.279 --> 00:06:26.199 thing. Makes it feel authentic, like a direct 00:06:26.199 --> 00:06:28.420 chat. And the humor comes from the contrast. 00:06:28.959 --> 00:06:31.100 Put an amazing character in a mundane spot. Imagine 00:06:31.100 --> 00:06:33.079 an astronaut in a supermarket, totally baffled 00:06:33.079 --> 00:06:36.699 by avocados. Or a Viking king versus a vending 00:06:36.699 --> 00:06:39.399 machine, you know, yelling for his crisps. The 00:06:39.399 --> 00:06:41.160 EO's good at those subtle character reactions, 00:06:41.420 --> 00:06:44.009 prompt -wise. Declare selfie style, describe 00:06:44.009 --> 00:06:46.069 the character in detail, add self -film touches, 00:06:46.290 --> 00:06:48.610 modern setting, action dialogue showing conflict, 00:06:48.889 --> 00:06:51.170 ambient audio. Just keep dialogue short, like 00:06:51.170 --> 00:06:54.629 under 10, 15 words. Long speeches tend to confuse 00:06:54.629 --> 00:06:57.170 the AI. And then satirical street interviews. 00:06:57.449 --> 00:06:59.829 What's the magic there? Subverting expectations. 00:07:00.089 --> 00:07:02.649 We know the news interview format. So when the 00:07:02.649 --> 00:07:05.220 content is suddenly absurd, Hilarious! It lets 00:07:05.220 --> 00:07:07.660 you do satire or just pure nonsense. Viewers 00:07:07.660 --> 00:07:10.240 like feeling they're in on the joke. VEO can 00:07:10.240 --> 00:07:12.660 create those realistic backdrops and expressions. 00:07:13.100 --> 00:07:15.519 Think the invisible pet show, people grooming 00:07:15.519 --> 00:07:18.699 thin air, or a 1920s interview about the shocking 00:07:18.699 --> 00:07:21.420 new trend of sliced bread. Hello. Your prompt 00:07:21.420 --> 00:07:24.100 needs the format, shot setup, absurd backstory, 00:07:24.699 --> 00:07:26.600 interviewee details, and a short Q &A script. 00:07:26.879 --> 00:07:28.879 Key thing. Keep the premise simple. The humor's 00:07:28.879 --> 00:07:31.279 in the basic absurdity. So what about other formats 00:07:31.279 --> 00:07:34.079 that use similar ideas, like incongruity or just 00:07:34.079 --> 00:07:35.860 strong visuals? I'm thinking news reports with 00:07:35.860 --> 00:07:38.199 a twist or even those really cinematic videos. 00:07:38.660 --> 00:07:40.300 Well, yeah, absolutely. They tap into similar 00:07:40.300 --> 00:07:42.579 core psychology. Take the news report with a 00:07:42.579 --> 00:07:45.540 twist. That's all about comedic timing and incongruity. 00:07:45.939 --> 00:07:48.579 Our brain focuses on the person talking. Right? 00:07:48.980 --> 00:07:51.120 So when something crazy happens in the background, 00:07:51.579 --> 00:07:54.139 unnoticed by them, it shatters expectations. 00:07:54.660 --> 00:07:57.120 Serious foreground, chaotic background, classic 00:07:57.120 --> 00:08:00.319 visual comedy. Like a construction safety expert 00:08:00.319 --> 00:08:02.819 talking confidently while someone perfectly falls 00:08:02.819 --> 00:08:05.360 into wet cement behind him. Or an environmental 00:08:05.360 --> 00:08:07.720 activist giving a speech as the wind blows all 00:08:07.720 --> 00:08:10.819 his fliers away in a clean park. VEO handles 00:08:10.819 --> 00:08:13.220 those complex scenes pretty well. Then totally 00:08:13.220 --> 00:08:15.500 different vibe. Cinematic style videos. That's 00:08:15.500 --> 00:08:18.589 about emotion. beauty, grandeur, visual storytelling. 00:08:19.009 --> 00:08:21.829 We don't just see it, we feel it. The high production 00:08:21.829 --> 00:08:24.829 value VEO gives feels professional. Picture a 00:08:24.829 --> 00:08:27.370 sci -fi scene, Blade Runner vibe, maybe an android 00:08:27.370 --> 00:08:30.029 with one tear, slow crane reveals a future city, 00:08:30.170 --> 00:08:32.230 or the lone samurai, Kurosawa style, drawing 00:08:32.230 --> 00:08:34.850 his sword on a cliff. VEO gets those complex 00:08:34.850 --> 00:08:37.289 camera moves and lighting. We also see huge engagement 00:08:37.289 --> 00:08:39.669 with stuff like the animal Olympics. That's pure 00:08:39.669 --> 00:08:41.929 anthropomorphism, giving animals human traits, 00:08:42.049 --> 00:08:43.789 especially in a serious context like the Olympics. 00:08:44.169 --> 00:08:46.620 Instant comedy. We laugh at a grizzly bear weightlifting 00:08:46.620 --> 00:08:49.440 or Siamese cat gymnastics, but we kind of root 00:08:49.440 --> 00:08:52.580 for them too. And for controlled chaos fans, 00:08:53.419 --> 00:08:55.899 satisfying object destruction. Our brains like 00:08:55.899 --> 00:08:58.120 patterns, but also get satisfaction from controlled 00:08:58.120 --> 00:09:00.980 destruction. Slow motion, which VEO does okay, 00:09:01.159 --> 00:09:03.559 shows off the physics details. Watching something 00:09:03.559 --> 00:09:06.159 go from order to chaos predictably, it's almost 00:09:06.159 --> 00:09:09.539 meditative. Think ink. Blooming in milk, creating 00:09:09.539 --> 00:09:12.240 those cool fractal patterns. Or a time lapse 00:09:12.240 --> 00:09:15.019 of a melting ice car. And finally, maybe a personal 00:09:15.019 --> 00:09:17.379 favorite. Historical figures in modern situations. 00:09:17.879 --> 00:09:20.399 Classic fish out of water. Anachronism. Humor 00:09:20.399 --> 00:09:22.620 from the time clash. Works best when their reaction 00:09:22.620 --> 00:09:24.659 fits their personality or achievements. Think 00:09:24.659 --> 00:09:27.000 Nikola Tesla at the Apple Store, annoyed by cat 00:09:27.000 --> 00:09:29.480 pics instead of wireless power. Or Marie Curie 00:09:29.480 --> 00:09:31.159 tries an energy drink and discovers some new 00:09:31.159 --> 00:09:33.460 energy. OK, so if you pull all those different 00:09:33.460 --> 00:09:36.879 formats together. What's the common thread? What's 00:09:36.879 --> 00:09:39.600 the core idea uniting their success? They all 00:09:39.600 --> 00:09:42.740 tap into universal human psychology for vital 00:09:42.740 --> 00:09:44.679 appeal and engagement. Alright, we talked about 00:09:44.679 --> 00:09:47.039 making great -looking videos, understanding the 00:09:47.039 --> 00:09:50.279 psychology. Now let's talk about control. Controlling 00:09:50.279 --> 00:09:53.320 consistently, efficiently, moving beyond just 00:09:53.320 --> 00:09:56.600 looks good to making it predictable, repeatable. 00:09:56.700 --> 00:09:58.100 Yeah, this is super important if you're doing 00:09:58.100 --> 00:10:00.360 a series or telling a story over multiple shots. 00:10:00.460 --> 00:10:02.580 If your character suddenly looks different, it 00:10:02.580 --> 00:10:05.139 just breaks everything. The fix is what we call 00:10:05.139 --> 00:10:07.100 the character sheet, not just a description. 00:10:07.500 --> 00:10:10.080 It's a standardized block of text, your character's 00:10:10.080 --> 00:10:13.899 DNA. Needs extreme detail. Age, build, hair, 00:10:13.940 --> 00:10:16.700 eyes, skin, signature outfit gets specific on 00:10:16.700 --> 00:10:18.960 fabric, color, style. And defining features. 00:10:19.519 --> 00:10:22.360 Scar, tattoo, mold, maybe a specific walk, an 00:10:22.360 --> 00:10:24.480 accessory, whatever makes them unique. I still 00:10:24.480 --> 00:10:26.700 wrestle with punch drift myself when I try to 00:10:26.700 --> 00:10:28.700 maintain character consistency across scenes. 00:10:28.960 --> 00:10:30.960 It's tougher than it looks. It's amazing how 00:10:30.960 --> 00:10:33.179 just changing black leather jacket to black jacket 00:10:33.179 --> 00:10:35.759 middle leather can mess things up. So this character 00:10:35.759 --> 00:10:38.690 sheet... It sounds essential, but how do we make 00:10:38.690 --> 00:10:41.750 it really work every time? Precision. Total precision. 00:10:42.169 --> 00:10:44.149 You write the sheet once, maybe in a text editor. 00:10:44.730 --> 00:10:47.470 Then, for every single prompt with that character, 00:10:47.909 --> 00:10:50.950 you copy and paste the exact block of text. 100 00:10:50.950 --> 00:10:53.700 % accurate. Beginning of the prompt. Even a comma 00:10:53.700 --> 00:10:56.240 changes things sometimes. Beyond characters, 00:10:56.379 --> 00:10:58.580 there's mastering the camera. For really impressive 00:10:58.580 --> 00:11:01.059 shots, speak cinema language. Think Dolly add 00:11:01.059 --> 00:11:03.860 camera moves closer or further. Crane shot, smooth 00:11:03.860 --> 00:11:06.059 vertical moves, tracking shot, moves alongside 00:11:06.059 --> 00:11:08.779 the subject. Rack focus, shifts focus within 00:11:08.779 --> 00:11:11.100 the shot. Dutch angle, tilted camera for unease. 00:11:11.580 --> 00:11:13.220 He is describing the camera's journey and not 00:11:13.220 --> 00:11:15.840 just one word. Don't just say, Dolly in, try. 00:11:16.120 --> 00:11:17.899 Camera starts close on hands, assembling a device, 00:11:18.279 --> 00:11:20.320 tilt up smoothly to their face. Then Dolly's 00:11:20.320 --> 00:11:23.080 out to show them alone in a vast workshop. More 00:11:23.080 --> 00:11:26.379 nuanced. And sound. Sounds huge. VEO lets you 00:11:26.379 --> 00:11:28.000 direct that, too. It's important to know the 00:11:28.000 --> 00:11:29.879 difference between diegetic and non -diegetic, 00:11:29.899 --> 00:11:32.259 right? Exactly. Diegetic is what characters hear 00:11:32.259 --> 00:11:34.740 dialogue, footsteps, rain, part of their world. 00:11:35.240 --> 00:11:37.519 Non -diegetic is just, for the audience, musical 00:11:37.519 --> 00:11:40.799 score, narrator, enhances emotion. Okay, finally, 00:11:41.019 --> 00:11:43.419 credit -saving strategies. AI credits aren't 00:11:43.419 --> 00:11:46.269 cheap. Got to work smart. Use fast mode if the 00:11:46.269 --> 00:11:49.250 EO has it for testing ideas, drafts, quick lower 00:11:49.250 --> 00:11:51.950 res versions, good for iterating before the big 00:11:51.950 --> 00:11:54.730 render. And here's a big pro tip. Before you 00:11:54.730 --> 00:11:56.950 burn credits on an eight second video, take your 00:11:56.950 --> 00:11:59.409 core visual description, pop it into an image 00:11:59.409 --> 00:12:01.669 generator like in Gemini for your cheap, lets 00:12:01.669 --> 00:12:03.789 you quickly check aesthetic, composition, character 00:12:03.789 --> 00:12:07.029 design without spending video credits, and always, 00:12:07.029 --> 00:12:09.490 always finalize prompts in the external text 00:12:09.490 --> 00:12:12.370 editor, Google Docs, whatever. Avoid accidental 00:12:12.370 --> 00:12:15.470 generations from typos. great tips there. Which 00:12:15.470 --> 00:12:17.850 one would you say offers the biggest efficiency 00:12:17.850 --> 00:12:21.009 gain, saves the most time and resources? Testing 00:12:21.009 --> 00:12:23.210 core visual descriptions with an image generator 00:12:23.210 --> 00:12:26.389 before costly video generation. So once you've 00:12:26.389 --> 00:12:29.269 got that VEO output, remember, it's raw material. 00:12:29.490 --> 00:12:31.269 Post -production is where you polish it, turn 00:12:31.269 --> 00:12:32.809 it into something professional, ready to go. 00:12:33.149 --> 00:12:35.190 So the workflow is kind of like, VEO generates 00:12:35.190 --> 00:12:37.250 the clip, then an upscaler, then an editor, then 00:12:37.250 --> 00:12:40.450 publish? Pretty much. VEO usually outputs 720p, 00:12:40.509 --> 00:12:43.409 maybe 1080p. For top quality, you want to upscale 00:12:43.409 --> 00:12:47.559 to 4K. AI upscalers like Topaz Video AI are great. 00:12:47.820 --> 00:12:49.679 They don't just stretch pixels, they intelligently 00:12:49.679 --> 00:12:53.120 paint in details, denoise, sharpen, looks natural. 00:12:53.679 --> 00:12:56.240 Then, editing and color grading. DaVinci Resolve 00:12:56.240 --> 00:12:58.779 is amazing, powerful free version, great for 00:12:58.779 --> 00:13:00.960 color. CapCut's good too, beginner -friendly 00:13:00.960 --> 00:13:03.539 mobile desktop. Color grading is vital. Adjust 00:13:03.539 --> 00:13:05.919 colors for mood, cool blues for horror, warm 00:13:05.919 --> 00:13:08.460 oranges for sunset. And even with VEO's audio, 00:13:08.720 --> 00:13:10.720 adding extra layers in post makes it more immersive. 00:13:10.919 --> 00:13:13.179 background music, extra sound effects, a whoosh 00:13:13.179 --> 00:13:15.320 for movement, stuff like that. And then reformatting. 00:13:15.399 --> 00:13:17.679 That's a big one for social media. Video is horizontal, 00:13:17.879 --> 00:13:21.799 16 .9. TikTok reels need vertical, 9 .16. Cropping 00:13:21.799 --> 00:13:24.120 just butchers it usually. Yeah, cropping loses 00:13:24.120 --> 00:13:27.440 too much. The solution. AI video outpainting 00:13:27.440 --> 00:13:29.860 tools. Runway, Luma Labs have these. They use 00:13:29.860 --> 00:13:32.399 AI to paint the missing space above and below 00:13:32.399 --> 00:13:35.179 your horizontal video, creates a seamless vertical 00:13:35.179 --> 00:13:38.279 frame, keeps your original content intact. It's 00:13:38.279 --> 00:13:41.059 like magic for social media. That is a powerful 00:13:41.059 --> 00:13:44.139 trick. So thinking about post -production for 00:13:44.139 --> 00:13:46.679 social, what's the biggest mistake people tend 00:13:46.679 --> 00:13:49.460 to make? They often just crop horizontal videos 00:13:49.460 --> 00:13:51.879 losing most of the actual frame. Okay, so with 00:13:51.879 --> 00:13:54.440 all this incredible creative power, there's also... 00:13:54.360 --> 00:13:57.440 responsibility, especially now in the AI era. 00:13:57.700 --> 00:13:59.779 Oh, definitely. We have to talk about misinformation, 00:14:00.139 --> 00:14:02.460 deep fakes. It's a real risk. This tech can be 00:14:02.460 --> 00:14:05.200 misused, fake news, deceptive videos, harmful 00:14:05.200 --> 00:14:07.379 stuff. As creators, we kind of have to commit 00:14:07.379 --> 00:14:09.720 to using it for art, entertainment, positive 00:14:09.720 --> 00:14:12.950 storytelling, never to deceive or harm. And transparency 00:14:12.950 --> 00:14:16.110 is key. Always tell your audience it's AI generated. 00:14:16.409 --> 00:14:19.149 Hashtags like hashtag AI generated, hashtag made 00:14:19.149 --> 00:14:21.629 with VEO, or just a little text overlay builds 00:14:21.629 --> 00:14:23.950 trust. Helps keep the whole AI creative space 00:14:23.950 --> 00:14:26.370 healthy. And VO3, even as powerful as it is, 00:14:26.450 --> 00:14:28.409 it still has limits, right? It's not perfect 00:14:28.409 --> 00:14:31.940 yet. No tech is. VEO still struggles with hands 00:14:31.940 --> 00:14:34.440 sometimes. Extra fingers, missing fingers, it 00:14:34.440 --> 00:14:37.259 happens. Physics simulation is good, but complex, 00:14:37.379 --> 00:14:40.379 illogical stuff can still look weird. And crucially, 00:14:40.700 --> 00:14:42.759 text in video. It just can't do legible text. 00:14:42.879 --> 00:14:45.279 It comes out garbled. Workarounds for hands, 00:14:45.580 --> 00:14:47.659 use gloves, make fists, hide them, frame them 00:14:47.659 --> 00:14:50.299 out. For physics, keep actions simpler, plausible. 00:14:50.659 --> 00:14:53.240 For text, always add it in post using your editor. 00:14:53.440 --> 00:14:56.039 Don't even ask VEO to generate it. So connecting 00:14:56.039 --> 00:14:58.879 this back to the bigger picture. What would you 00:14:58.879 --> 00:15:01.879 say is the most significant ethical concern with 00:15:01.879 --> 00:15:04.419 this tech right now? The potential for misinformation 00:15:04.419 --> 00:15:07.240 and those deceptive deep fake videos. We've covered 00:15:07.240 --> 00:15:09.220 a lot of ground in this deep dive from really 00:15:09.220 --> 00:15:11.659 understanding VEOs power, the prompting philosophy, 00:15:12.220 --> 00:15:14.500 to viral formats, advanced techniques and putting 00:15:14.500 --> 00:15:16.720 together a whole strategy. Yeah. And remember, 00:15:16.919 --> 00:15:19.519 the text is just a tool. The real power is your 00:15:19.519 --> 00:15:22.360 storytelling ability, the detail in your directives. 00:15:22.720 --> 00:15:25.620 A well -crafted prompt makes art, not just random 00:15:25.620 --> 00:15:28.379 outputs. And things are moving so fast. Whoa! 00:15:28.899 --> 00:15:31.000 I mean, imagine the future. Perfect character 00:15:31.000 --> 00:15:33.919 consistency across dozens of scenes, real -time 00:15:33.919 --> 00:15:36.960 video generation. The creator's role is really 00:15:36.960 --> 00:15:39.360 shifting, isn't it? More like a true AI director, 00:15:39.559 --> 00:15:42.460 a narrative designer telling the AI exactly what 00:15:42.460 --> 00:15:44.840 complex idea to build. You've got the map now. 00:15:44.919 --> 00:15:46.759 You've got the toolkit. It's really your turn 00:15:46.759 --> 00:15:48.960 to explore, push the limits, become a creator 00:15:48.960 --> 00:15:51.200 shaping what's next. Pick a format we talked 00:15:51.200 --> 00:15:53.080 about. Make it yours. Make that first video. 00:15:53.379 --> 00:15:55.559 Don't worry about failing. Every generation teaches 00:15:55.559 --> 00:15:57.679 you something. The future of storytelling is 00:15:57.679 --> 00:16:00.700 here, and it's literally in your hands. Now go 00:16:00.700 --> 00:16:01.240 create.