WEBVTT 00:00:00.000 --> 00:00:03.640 What if everything you're about to see, even 00:00:03.640 --> 00:00:06.339 someone like me speaking directly to you, what 00:00:06.339 --> 00:00:08.960 if it could all be generated by AI? Imagine creating 00:00:08.960 --> 00:00:12.199 these amazing animated stories or even music 00:00:12.199 --> 00:00:14.699 videos that look totally pro. Without a crew, 00:00:14.839 --> 00:00:17.500 without animators, none of that. Just a single 00:00:17.500 --> 00:00:21.320 prompt. Welcome to our deep dive. Today, we're 00:00:21.320 --> 00:00:24.329 really getting into... Google VO3. Think of it 00:00:24.329 --> 00:00:27.429 like your own personal Hollywood studio right 00:00:27.429 --> 00:00:29.690 in your computer. Right. Our mission to unpack 00:00:29.690 --> 00:00:32.990 exactly how this thing works, this AI video tool 00:00:32.990 --> 00:00:35.149 and, you know, what it really means for creators 00:00:35.149 --> 00:00:36.979 like you. Absolutely. Yeah, we're going to cover 00:00:36.979 --> 00:00:38.979 the big breakthroughs, the stuff that makes VO3, 00:00:39.039 --> 00:00:41.179 well, kind of revolutionary. And then we'll walk 00:00:41.179 --> 00:00:43.219 you through how to actually use it, getting started, 00:00:43.399 --> 00:00:45.520 directing the AI actors, getting the camera moves 00:00:45.520 --> 00:00:47.219 right. We'll even get into different styles, 00:00:47.299 --> 00:00:50.359 how to write those perfect prompts and the practical 00:00:50.359 --> 00:00:52.299 side to how to make it work for real projects. 00:00:52.460 --> 00:00:54.380 The idea is you'll leave feeling like, OK, I 00:00:54.380 --> 00:00:57.700 get this. I can use this. OK, let's dig in, because 00:00:57.700 --> 00:01:02.399 for a while, AI video, it felt a bit, I don't 00:01:02.399 --> 00:01:05.150 know. Clunky. A bit frustrating sometimes. You'd 00:01:05.150 --> 00:01:07.189 get these silent clips, maybe a bit jittery. 00:01:07.269 --> 00:01:09.629 Interesting, yeah, but not really telling a story. 00:01:09.950 --> 00:01:13.709 Exactly. And this is where VO3 feels like a genuine 00:01:13.709 --> 00:01:17.409 shift, a real step change. The biggest thing. 00:01:17.530 --> 00:01:20.430 It's the synchronized audio generation. And that 00:01:20.430 --> 00:01:22.510 doesn't just mean the lip slap in time. It means 00:01:22.510 --> 00:01:24.930 the AI generates a voice that actually matches 00:01:24.930 --> 00:01:28.069 the lip movements. Sure. But also the facial 00:01:28.069 --> 00:01:30.430 expressions, the emotion in the dialogue. Ah, 00:01:30.469 --> 00:01:32.670 OK. So it's not just visuals plus some random 00:01:32.670 --> 00:01:34.629 sound. It's actual performance coming through. 00:01:34.730 --> 00:01:36.650 Right. It's like going from silent film to the 00:01:36.650 --> 00:01:39.010 talkies. You know, that's the leap. That feels. 00:01:39.239 --> 00:01:41.180 significant. It lets the AI characters genuinely 00:01:41.180 --> 00:01:43.620 perform. That's it. That's the key that unlocks 00:01:43.620 --> 00:01:46.519 real storytelling. It takes AI video from just 00:01:46.519 --> 00:01:49.000 being like neat visual tricks to something that 00:01:49.000 --> 00:01:51.480 can deliver a character's performance. So what's 00:01:51.480 --> 00:01:53.599 the single most significant breakthrough here, 00:01:53.680 --> 00:01:55.500 if you had to pick one thing? Synchronized audio, 00:01:55.700 --> 00:01:58.299 definitely. It lets AI characters truly perform. 00:01:58.579 --> 00:02:01.939 Right. So once you see that potential, the next 00:02:01.939 --> 00:02:03.620 thing you're thinking is, okay, how do I try 00:02:03.620 --> 00:02:05.920 this? Sounds like a big deal, this personal Hollywood 00:02:05.920 --> 00:02:09.310 studio. Is it hard to get started? Surprisingly, 00:02:09.310 --> 00:02:11.949 no. It's actually pretty simple. You just search 00:02:11.949 --> 00:02:15.490 Google VO3 online, click the official link. Then 00:02:15.490 --> 00:02:17.430 you're looking for buttons like try and flow 00:02:17.430 --> 00:02:20.590 and then create with flow. That's basically your 00:02:20.590 --> 00:02:23.509 way in. OK. And what about access? Is it expensive 00:02:23.509 --> 00:02:26.069 right off the bat? Well, they have a pretty generous 00:02:26.069 --> 00:02:28.849 trial. You get a full month free. You might need 00:02:28.849 --> 00:02:31.110 to put in payment details, but they don't charge 00:02:31.110 --> 00:02:32.949 you up front for that first month. A whole month 00:02:32.949 --> 00:02:35.569 free. That's decent. What about, say, students? 00:02:36.250 --> 00:02:38.270 Any breaks for them? Oh, yeah, that's a great 00:02:38.270 --> 00:02:40.370 point. There's actually an amazing program for 00:02:40.370 --> 00:02:43.129 university students. If you've got a .edu email 00:02:43.129 --> 00:02:46.189 address, you could get up to 15 months, one to 00:02:46.189 --> 00:02:49.409 five. Fifteen months of free access. It's huge 00:02:49.409 --> 00:02:51.870 for learning and experimenting. Fifteen months. 00:02:52.330 --> 00:02:55.310 Wow. OK, so once you're in, what are the first 00:02:55.310 --> 00:02:57.009 things you need to set up? Any crucial settings? 00:02:57.389 --> 00:02:59.969 Yeah, a couple of key things. First, set your 00:02:59.969 --> 00:03:03.340 output to one per prompt. That focuses the AI 00:03:03.340 --> 00:03:05.360 for better quality rather than giving you lots 00:03:05.360 --> 00:03:08.000 of variations. And make sure you choose text 00:03:08.000 --> 00:03:10.319 to video as the main way you're generating. Got 00:03:10.319 --> 00:03:13.159 it. Output one, text to video. Anything else? 00:03:13.240 --> 00:03:17.280 And the big one. Always, always select VEO3 quality. 00:03:17.520 --> 00:03:19.439 That's what gets you the good stuff, the synchronized 00:03:19.439 --> 00:03:22.539 audio, the cinematic look. Don't skip that. Right. 00:03:22.659 --> 00:03:25.240 VEO3 quality. Makes sense. Okay. Now the fun 00:03:25.240 --> 00:03:28.500 part. Making characters actually talk. How does 00:03:28.500 --> 00:03:31.120 that work? It's pretty intuitive, actually. You 00:03:31.120 --> 00:03:32.919 just include the dialogue right in your prompt. 00:03:33.039 --> 00:03:36.020 You use little cues like she says or maybe he 00:03:36.020 --> 00:03:39.620 whispers. The key things to include are a good 00:03:39.620 --> 00:03:41.939 physical description of the character, what they're 00:03:41.939 --> 00:03:43.900 doing, the action the speech indicator says, 00:03:44.300 --> 00:03:46.800 and then some emotional context. Can you give 00:03:46.800 --> 00:03:50.159 an example? Sure. Like, imagine a man in tattered 00:03:50.159 --> 00:03:52.860 clothes sits at a metal table across from a masked 00:03:52.860 --> 00:03:56.159 stranger. Okay, that's the scene. Then the action 00:03:56.159 --> 00:03:59.110 and speech. Voice trembling, he says. You know 00:03:59.110 --> 00:04:02.110 what I want. Bring my sister back. See? You add 00:04:02.110 --> 00:04:05.270 that voice trembling part. VO3 will try to generate 00:04:05.270 --> 00:04:08.090 audio that matches that nervous trembling quality. 00:04:08.490 --> 00:04:10.090 Okay, so you can guide the emotion. What about 00:04:10.090 --> 00:04:12.750 the actual sound of the voice? Like pitch? Yeah, 00:04:12.849 --> 00:04:15.229 you can influence that too with simple words. 00:04:15.310 --> 00:04:17.790 High -pitched desperate voice or deep booming 00:04:17.790 --> 00:04:20.430 voice. Things like that. Just a heads up though, 00:04:20.490 --> 00:04:22.850 sometimes the AI might... Well, it might ignore 00:04:22.850 --> 00:04:24.689 you if your description clashes too much with 00:04:24.689 --> 00:04:26.930 the character's look or the scene. Ah, okay. 00:04:27.029 --> 00:04:28.750 So it tries to make sense of the whole picture. 00:04:28.970 --> 00:04:31.949 Exactly. And honestly, I still wrestle with prompt 00:04:31.949 --> 00:04:34.370 drift myself sometimes, you know, where you ask 00:04:34.370 --> 00:04:36.110 for one thing and the AI gives you something 00:04:36.110 --> 00:04:39.449 slightly different, especially with nuanced emotions 00:04:39.449 --> 00:04:43.430 and voices. It's a bit of an art. So beyond those 00:04:43.430 --> 00:04:46.230 basic commands and the prompt, how can you really 00:04:46.230 --> 00:04:48.829 nail the emotional delivery, get that fine control? 00:04:49.319 --> 00:04:51.480 Okay, for really pro -level control, there's 00:04:51.480 --> 00:04:54.180 a kind of two -step process people use. It's 00:04:54.180 --> 00:04:56.759 more advanced, but very effective. Like a workflow. 00:04:56.959 --> 00:05:00.579 Yeah. First, you generate the video in VO3. Focus 00:05:00.579 --> 00:05:02.519 on getting the visuals right, the lip movements 00:05:02.519 --> 00:05:04.579 perfect. Then you download that video. Okay. 00:05:04.639 --> 00:05:07.000 Then you take that video file and upload it to 00:05:07.000 --> 00:05:10.519 a specialized AI voice tool. Eleven Labs has 00:05:10.519 --> 00:05:13.339 a great voice changer feature for this. Ah, so 00:05:13.339 --> 00:05:15.939 you're replacing the voice. Exactly. You replace 00:05:15.939 --> 00:05:18.980 the original VO3 voice with a new one you generate 00:05:18.980 --> 00:05:21.980 in Eleven Labs. You can pick a voice, tweak the 00:05:21.980 --> 00:05:25.000 pitch, the stability, even how exaggerated the 00:05:25.000 --> 00:05:27.060 emotion is. And it stays in sync with the lips. 00:05:27.439 --> 00:05:30.019 Perfectly. That's the magic. So you get VO's 00:05:30.019 --> 00:05:33.279 visuals and lip sync combined with super detailed 00:05:33.279 --> 00:05:36.120 voice control from 11 Labs. That sounds powerful. 00:05:36.439 --> 00:05:39.100 Combining VO3 visuals with dedicated tools like 00:05:39.100 --> 00:05:41.699 11 Labs for that real granular control. That's 00:05:41.699 --> 00:05:44.199 how you get Hollywood -level results. Okay, let's 00:05:44.199 --> 00:05:46.259 talk about consistency. If you're making something 00:05:46.259 --> 00:05:48.899 longer than one clip, maybe a short film or a 00:05:48.899 --> 00:05:51.660 series, how do you make sure your character looks 00:05:51.660 --> 00:05:54.860 the same from scene to scene? Ah, yes. crucial 00:05:54.860 --> 00:05:57.100 point there's basically one golden rule here 00:05:57.100 --> 00:06:00.199 the physical description you use for a character 00:06:00.199 --> 00:06:02.839 it needs to be absolutely identical every single 00:06:02.839 --> 00:06:05.079 time you prompt for that character identical 00:06:05.079 --> 00:06:09.740 like word for word word for word the best way 00:06:09.740 --> 00:06:12.300 to manage this is to create a separate document 00:06:12.300 --> 00:06:15.360 like a character sheet like whiters use exactly 00:06:15.360 --> 00:06:18.180 you list out everything core description hair 00:06:18.180 --> 00:06:21.279 color and style face shape eye color any accessories 00:06:21.279 --> 00:06:24.000 maybe their typical clothes Be super detailed. 00:06:24.259 --> 00:06:26.240 And then you just copy and paste that whole block 00:06:26.240 --> 00:06:28.560 of text into every new prompt featuring that 00:06:28.560 --> 00:06:31.060 character. Precisely. If your first prompt says, 00:06:31.139 --> 00:06:33.360 a weary detective with a crumpled trench coat, 00:06:33.540 --> 00:06:35.959 salt and pepper hair, and tired blue eyes. Hmm. 00:06:36.259 --> 00:06:38.800 Your second prompt, even if he's now in a different 00:06:38.800 --> 00:06:41.519 location, needs that exact same description. 00:06:41.899 --> 00:06:43.939 That makes sense. It forces the AI to remember. 00:06:44.199 --> 00:06:46.699 It dramatically increases the chances the AI 00:06:46.699 --> 00:06:48.699 will render the same looking person. It's not 00:06:48.699 --> 00:06:50.939 foolproof, but it's the best method we have right 00:06:50.939 --> 00:06:53.399 now. So what's the key to making an AI character 00:06:53.399 --> 00:06:56.180 feel like the same person across different scenes? 00:06:56.740 --> 00:07:00.139 Exact, detailed character descriptions used consistently 00:07:00.139 --> 00:07:02.800 every single time. Copy and paste is your friend 00:07:02.800 --> 00:07:07.579 here. Got it. Now, beyond the characters, what 00:07:07.579 --> 00:07:09.660 about the camera? Making it look cinematic isn't 00:07:09.660 --> 00:07:11.500 just about the actors, right? Oh, absolutely. 00:07:11.639 --> 00:07:14.339 How the camera moves tells a huge part of the 00:07:14.339 --> 00:07:18.399 story. It sets the mood, directs attention. It's 00:07:18.399 --> 00:07:20.759 fundamental. And VO3 gives you quite a bit of 00:07:20.759 --> 00:07:23.579 control over this virtual camera. So how do you 00:07:23.579 --> 00:07:25.939 direct the AI camera? Do you just type, make 00:07:25.939 --> 00:07:29.079 it look cool? Hey, if only. No, you use standard 00:07:29.079 --> 00:07:31.379 filmmaking terms. It really helps to know a few 00:07:31.379 --> 00:07:35.060 basic ones like dolly. That's moving the camera 00:07:35.060 --> 00:07:37.959 closer to or further away from the subject. Pan 00:07:37.959 --> 00:07:40.579 is swiveling left or right. Tilt is swiveling 00:07:40.579 --> 00:07:42.980 up or down. Pretty simple. What about more dynamic 00:07:42.980 --> 00:07:45.360 shots? Yeah, you can do things like orbit where 00:07:45.360 --> 00:07:47.120 the camera circles around the subject. Really 00:07:47.120 --> 00:07:49.680 dramatic. Or a trapping shot, which follows a 00:07:49.680 --> 00:07:51.759 character as they move. Or even a crane shot. 00:07:52.430 --> 00:07:54.490 the camera physically up or down like on a big 00:07:54.490 --> 00:07:56.610 crane gives you those sweeping views you just 00:07:56.610 --> 00:07:58.670 add these words into your main prompt that's 00:07:58.670 --> 00:08:00.889 the easiest way yeah directing via the script 00:08:00.889 --> 00:08:02.990 basically just add a simple instruction like 00:08:02.990 --> 00:08:06.129 a slow dolly in on her face simple enough is 00:08:06.129 --> 00:08:09.170 there a more advanced way There is. For more 00:08:09.170 --> 00:08:11.069 precise control, you can use something called 00:08:11.069 --> 00:08:13.870 the motion control rig. This works with the frames 00:08:13.870 --> 00:08:16.810 to video option. How does that work? You upload 00:08:16.810 --> 00:08:19.470 a starting image, then you select a predefined 00:08:19.470 --> 00:08:22.569 camera movement from a menu like dolly in out 00:08:22.569 --> 00:08:25.970 or pan left. Then you add your text prompt as 00:08:25.970 --> 00:08:29.050 usual. The AI will apply that specific motion 00:08:29.050 --> 00:08:31.790 you chose to the image based on your text. It's 00:08:31.790 --> 00:08:33.769 more technical, but gives you really fine control. 00:08:34.740 --> 00:08:37.299 Okay, let's shift gears a bit. Talk about style. 00:08:37.580 --> 00:08:40.399 How versatile is VO3? Can it only do realistic 00:08:40.399 --> 00:08:42.919 stuff? Not at all. It's actually really good 00:08:42.919 --> 00:08:45.320 at different artistic styles. The key, again, 00:08:45.440 --> 00:08:47.320 is just being super clear in your prompt about 00:08:47.320 --> 00:08:49.759 what you want. Like music videos. Got to handle 00:08:49.759 --> 00:08:51.799 that with the singing and everything. Definitely. 00:08:52.269 --> 00:08:55.250 You can prompt for, say, an energetic Afrobeats 00:08:55.250 --> 00:08:57.970 dance scene in a bustling urban street market 00:08:57.970 --> 00:09:01.110 at sunset. Add a camera move like a fluid tracking 00:09:01.110 --> 00:09:03.470 shot. And because of that synchronized audio, 00:09:03.750 --> 00:09:06.110 you can get really believable lip syncing and 00:09:06.110 --> 00:09:08.429 vocal performances. Makes it feel very authentic. 00:09:08.730 --> 00:09:12.070 What about animation? Can it do cartoons? Yep. 00:09:12.090 --> 00:09:14.730 The animation wing, as we could call it, is fully 00:09:14.730 --> 00:09:17.269 operational. If you want that Pixar look, you 00:09:17.269 --> 00:09:20.049 literally just start your prompt with a Pixar 00:09:20.049 --> 00:09:22.889 -style 3D character. So like a Pixar -style 3D 00:09:22.889 --> 00:09:25.789 character of a clever young inventor and describe 00:09:25.789 --> 00:09:28.009 the scene. Exactly. And the quality is usually 00:09:28.009 --> 00:09:30.210 pretty high, less of that weird morphing you 00:09:30.210 --> 00:09:33.450 sometimes see in AI video. Cool. What about classic 00:09:33.450 --> 00:09:36.190 cartoons like Looney Tunes style? You could do 00:09:36.190 --> 00:09:38.409 that, too. Just specify vintage Looney Tunes 00:09:38.409 --> 00:09:41.129 cartoons and crucially ask for the sound effects. 00:09:41.309 --> 00:09:44.250 Include rapid footsteps, comical slips and playful 00:09:44.250 --> 00:09:46.789 xylophone chase music. Get us at the whole vibe. 00:09:46.929 --> 00:09:49.669 Huh. OK. And even like comic book style. For 00:09:49.669 --> 00:09:52.980 sure. But again, be specific. Mention vivid colors, 00:09:53.179 --> 00:09:56.200 thick black outlines, energetic motion lines. 00:09:56.659 --> 00:10:00.159 Then describe your scene like a masked vigilante 00:10:00.159 --> 00:10:03.059 leaping from a rainy rooftop. Wow, that's a lot 00:10:03.059 --> 00:10:06.940 of range. But with all these options, how do 00:10:06.940 --> 00:10:09.039 you make sure the AI actually understands your 00:10:09.039 --> 00:10:11.220 creative vision? How do you get what's in your 00:10:11.220 --> 00:10:13.429 head onto the screen? That really boils down 00:10:13.429 --> 00:10:15.990 to prompt engineering or maybe just good screenwriting, 00:10:16.049 --> 00:10:18.250 honestly. It's that old computer science idea. 00:10:18.409 --> 00:10:20.750 Garbage in, garbage out. The quality of your 00:10:20.750 --> 00:10:23.509 output is directly tied to the quality of your 00:10:23.509 --> 00:10:26.190 input. Makes sense. So is there a formula, a 00:10:26.190 --> 00:10:28.490 structure for writing good prompts? There is, 00:10:28.590 --> 00:10:31.289 yeah. Think of it like a recipe with key ingredients. 00:10:31.429 --> 00:10:34.450 You need style specification first, like a realistic 00:10:34.450 --> 00:10:37.750 cinematic video or a Pixar -style 3D animation. 00:10:38.009 --> 00:10:39.850 Okay. Then your detailed character description 00:10:39.850 --> 00:10:41.450 pulled from that character sheet we talked about. 00:10:41.899 --> 00:10:43.960 Then the scene description environment, lighting, 00:10:44.139 --> 00:10:46.820 mood. Then the action description, what's actually 00:10:46.820 --> 00:10:49.179 happening. If they're speaking, add the dialogue 00:10:49.179 --> 00:10:51.960 in quotes. Then any camera movement instruction. 00:10:52.600 --> 00:10:55.340 And finally, other audio elements, sound effects, 00:10:55.580 --> 00:10:58.539 background music style. So you're basically writing 00:10:58.539 --> 00:11:00.820 a mini screenplay for every shot. Pretty much. 00:11:00.980 --> 00:11:03.899 A complete package. Leaving less ambiguity for 00:11:03.899 --> 00:11:06.379 the AI. Detailed structured prompts are key. 00:11:06.620 --> 00:11:08.700 Thinking like a director really helps you get 00:11:08.700 --> 00:11:11.299 there. And here's a really neat trick, meta -prompting. 00:11:11.629 --> 00:11:14.370 You can actually use another AI, like ChatGPT, 00:11:14.490 --> 00:11:17.970 to help you write better VO3 prompts. Use an 00:11:17.970 --> 00:11:21.350 AI to prompt an AI? Yeah. You give ChatGPT a 00:11:21.350 --> 00:11:23.809 template, tell it to act like an expert VO3 prompt 00:11:23.809 --> 00:11:26.129 writer. Then you give it their basic idea. Like, 00:11:26.149 --> 00:11:29.669 you tell ChatGPT, my idea is, in a dynamic comic 00:11:29.669 --> 00:11:31.889 book style, Spider -Man swings through the city. 00:11:32.669 --> 00:11:35.149 ChatGPT then takes that simple idea and fleshes 00:11:35.149 --> 00:11:36.929 it out into a detailed structured prompt using 00:11:36.929 --> 00:11:39.990 all those ingredients we just listed. Whoa. Okay, 00:11:40.049 --> 00:11:42.230 that's pretty meta. Imagine scaling that. You 00:11:42.230 --> 00:11:44.450 could, like, outline an entire animated series 00:11:44.450 --> 00:11:47.450 concept, feed it to ChatGPT block by block to 00:11:47.450 --> 00:11:49.509 generate the VO prompts. It's like having a full 00:11:49.509 --> 00:11:51.850 pre -production team and studio powered by AI. 00:11:52.190 --> 00:11:54.370 Mind -blowing. Right, it really does feel like 00:11:54.370 --> 00:11:57.610 a full studio at your fingertips. Okay, so we've 00:11:57.610 --> 00:12:00.289 covered the creative side, but using this effectively 00:12:00.289 --> 00:12:02.929 also means thinking like a producer. There are 00:12:02.929 --> 00:12:05.070 practical things to consider. Right, like rules 00:12:05.070 --> 00:12:08.330 and costs. Exactly. First up, content policies. 00:12:09.100 --> 00:12:12.080 Google, like most AI platforms, has filters. 00:12:12.500 --> 00:12:15.139 It's generally best to keep your prompts relatively 00:12:15.139 --> 00:12:19.220 family -friendly. Avoid overly violent or sensitive 00:12:19.220 --> 00:12:22.440 topics. If a prompt gets blocked or fails, don't 00:12:22.440 --> 00:12:25.059 panic. Just try rephrasing it. Maybe use less 00:12:25.059 --> 00:12:27.440 aggressive language. Instead of a brutal fight, 00:12:27.620 --> 00:12:29.919 try an intense confrontation or something similar. 00:12:30.159 --> 00:12:33.110 Good tip. What about the cost? You mentioned 00:12:33.110 --> 00:12:35.830 VEO3 quality is best, but is it always necessary? 00:12:36.169 --> 00:12:38.169 That's the budget question. You usually have 00:12:38.169 --> 00:12:40.809 a choice. VEO3 quality gives the absolute best 00:12:40.809 --> 00:12:42.730 results, especially with that synchronized audio, 00:12:42.889 --> 00:12:44.909 but it's the most expensive option per generation. 00:12:45.659 --> 00:12:48.720 Then there are usually older VEO2 options available. 00:12:48.899 --> 00:12:50.860 These are cheaper. They're great for just testing 00:12:50.860 --> 00:12:53.120 ideas, maybe generating some visual -only background 00:12:53.120 --> 00:12:54.799 shots, or if you need to create a lot of clips 00:12:54.799 --> 00:12:58.379 quickly for experiments. So VEO2 for rough drafts 00:12:58.379 --> 00:13:01.580 or visuals, VEO3 for the final cut, especially 00:13:01.580 --> 00:13:03.620 with dialogue. This is a smart way to think about 00:13:03.620 --> 00:13:06.909 it, yeah. Use VEO2 for concept testing, B -roll, 00:13:07.029 --> 00:13:10.090 high volume stuff. Switch to VEO3 for your final 00:13:10.090 --> 00:13:12.409 production videos where quality and audio sync 00:13:12.409 --> 00:13:15.090 are paramount. Strategic choices save money. 00:13:15.309 --> 00:13:17.549 And beyond just making cool videos for yourself, 00:13:17.750 --> 00:13:19.909 are there real business opportunities here? Oh, 00:13:19.909 --> 00:13:22.110 absolutely. This opens up a ton of possibilities. 00:13:22.450 --> 00:13:25.090 Think about faceless YouTube channels. Where 00:13:25.090 --> 00:13:27.149 you don't show yourself on camera. Right. You 00:13:27.149 --> 00:13:30.110 could use VEO3 to animate folktales, explain 00:13:30.110 --> 00:13:32.570 historical events, create fictional narratives. 00:13:33.710 --> 00:13:35.570 without needing to film yourself. Interesting. 00:13:35.789 --> 00:13:38.769 What else? Commercial uses. Yeah, big time. Companies 00:13:38.769 --> 00:13:42.950 could create custom AI brand mascots or generate 00:13:42.950 --> 00:13:45.149 corporate training videos using consistent AI 00:13:45.149 --> 00:13:48.210 instructors or even just quickly mock up video 00:13:48.210 --> 00:13:50.429 concepts for clients to approve before investing 00:13:50.429 --> 00:13:52.730 in a full live action shoot. It's incredibly 00:13:52.730 --> 00:13:54.690 versatile for business. Okay, so for getting 00:13:54.690 --> 00:13:56.889 the absolute best final product, you mentioned 00:13:56.889 --> 00:13:59.409 combining tools earlier. What does that workflow 00:13:59.409 --> 00:14:01.899 look like? Right, the quality optimization workflow. 00:14:02.240 --> 00:14:04.740 You'd use VO3 to generate the core video with 00:14:04.740 --> 00:14:07.500 good visuals and lip sync. Then maybe use 11 00:14:07.500 --> 00:14:09.700 Labs for that top -tier voiceover we talked about. 00:14:09.879 --> 00:14:12.360 Then find an AI music generator for a custom 00:14:12.360 --> 00:14:14.500 soundtrack. Finally, you bring all those pieces 00:14:14.500 --> 00:14:16.320 together in a standard video editor, something 00:14:16.320 --> 00:14:19.360 like CapCut or Adobe Premiere Pro to assemble 00:14:19.360 --> 00:14:21.740 the final product, add titles, transitions, etc. 00:14:22.120 --> 00:14:25.639 So VO is one powerful piece, but it works best 00:14:25.639 --> 00:14:27.850 as part of a larger toolkit. For professional 00:14:27.850 --> 00:14:30.950 results, definitely. And quickly, troubleshooting, 00:14:30.990 --> 00:14:33.809 if a generation just fails. Probably a content 00:14:33.809 --> 00:14:35.750 filter issue. Simplify your prompt language. 00:14:36.289 --> 00:14:38.450 Characters look inconsistent. Your description 00:14:38.450 --> 00:14:41.129 wasn't detailed or copied exactly. Audio sounds 00:14:41.129 --> 00:14:43.549 bad or out of sync. Double check you selected 00:14:43.549 --> 00:14:46.529 VEO3 quality and maybe add more emotional cues 00:14:46.529 --> 00:14:49.230 to the prompt. So what's one key piece of advice 00:14:49.230 --> 00:14:51.509 for turning all this creative power into actual 00:14:51.509 --> 00:14:54.730 practical value? Think like a producer. Understand 00:14:54.730 --> 00:14:56.909 the content rules, manage your costs by choosing 00:14:56.909 --> 00:14:59.409 the right model, and know when to combine VO3 00:14:59.409 --> 00:15:01.870 with other specialized tools for the best outcome. 00:15:02.029 --> 00:15:03.889 Okay, wrapping things up. It feels like Google 00:15:03.889 --> 00:15:06.350 VO3 isn't just, you know, another incremental 00:15:06.350 --> 00:15:10.450 update to AI tools. It feels... It really does. 00:15:10.590 --> 00:15:12.970 It's more like a complete paradigm shift, putting 00:15:12.970 --> 00:15:15.690 capabilities that used to require massive teams 00:15:15.690 --> 00:15:18.870 and budgets into, well, potentially anyone's 00:15:18.870 --> 00:15:21.450 hands. The real key, it seems, is learning to 00:15:21.450 --> 00:15:23.970 think differently, not just typing a sentence, 00:15:24.070 --> 00:15:26.289 but thinking like a director, a cinematographer, 00:15:26.409 --> 00:15:29.610 a storyteller, giving the AI the detailed instructions 00:15:29.610 --> 00:15:31.809 it needs. Exactly. That's how you elevate your 00:15:31.809 --> 00:15:34.389 creations from just experiments to actual compelling 00:15:34.389 --> 00:15:37.429 content. The future of content creation really 00:15:37.429 --> 00:15:40.190 feels like it's arriving now. And it's way more 00:15:40.190 --> 00:15:43.269 accessible, more powerful than I think many of 00:15:43.269 --> 00:15:45.509 us imagined even a short time ago. The potential 00:15:45.509 --> 00:15:48.669 is just enormous. So a final thought for everyone 00:15:48.669 --> 00:15:50.970 listening. Imagine all the stories out there 00:15:50.970 --> 00:15:53.230 that haven't been told simply because the tools 00:15:53.230 --> 00:15:55.730 were too complex or expensive. Now, anyone with 00:15:55.730 --> 00:15:58.870 an idea and the ability to write a detailed prompt 00:15:58.870 --> 00:16:02.450 can potentially become a filmmaker. What stories 00:16:02.450 --> 00:16:05.370 will you create when the main limit is just your 00:16:05.370 --> 00:16:07.759 own imagination? Thanks so much for joining us 00:16:07.759 --> 00:16:09.919 on this deep dive today. We really encourage 00:16:09.919 --> 00:16:13.100 you to check out Google VO3 if you can. Explore 00:16:13.100 --> 00:16:15.220 its capabilities and start your own creative 00:16:15.220 --> 00:16:15.940 journey with it.