WEBVTT 00:01.040 --> 00:04.600 Okay, let's talk about JSON prompting in vo3. 00:05.040 --> 00:10.960 So first of all, if you don't know what JSON is, it's a type of structured data. 00:11.400 --> 00:13.000 So it's used in programming. 00:13.120 --> 00:15.040 You don't have to worry about it too much. 00:15.200 --> 00:17.480 And it's not a requirement for using Voe. 00:17.560 --> 00:19.680 You can just type what you want. 00:19.880 --> 00:28.800 Like you could say like an astronaut sleeps peacefully on a mystical in world. 00:29.040 --> 00:34.160 Whatever you want to type in, you could describe the camera angles, the environment, the character, 00:34.200 --> 00:42.640 etc. now the benefit of using something like JSON and I'm just going to paste some in here, is that 00:42.640 --> 00:44.280 it's a little bit more precise. 00:44.920 --> 00:49.640 And again, there is no actual decided upon structure for this. 00:49.640 --> 00:54.400 But what JSON does is it gives you this key value structure. 00:54.680 --> 00:57.040 So here the key is prompt. 00:57.360 --> 01:03.120 And then the value is this specific prompt which is what we just typed in. 01:03.560 --> 01:07.440 But it gives you the flexibility to control different parameters. 01:07.560 --> 01:09.920 Because Vo3 understands JSON. 01:09.920 --> 01:13.970 It understands instance programming, it can understand that it needs to be much more precise to follow 01:13.970 --> 01:14.850 these values. 01:15.010 --> 01:17.530 For example, here we have this config key. 01:17.690 --> 01:22.290 It there's nothing special about the word config other than it understands what config means. 01:22.330 --> 01:23.330 Configuration. 01:23.330 --> 01:25.370 And then within config we have camera. 01:25.850 --> 01:28.370 And again it just understands what a camera is right. 01:28.410 --> 01:30.650 There's no special magic keywords here. 01:30.650 --> 01:35.370 You can put whatever you want, but like that allows you to really describe more precisely what you 01:35.370 --> 01:35.730 want. 01:35.770 --> 01:41.730 So in this case we're communicating to it that we're going to use with for the camera configuration, 01:41.770 --> 01:43.570 we're going to use a wide shot angle. 01:44.090 --> 01:46.330 And then we want to slowly pan across the landscape. 01:46.330 --> 01:53.650 This is where having a little bit of video editing or video production or video shooting, I guess videography 01:53.690 --> 01:59.010 experience will be really useful because you understand the different types of shots you could use or 01:59.050 --> 02:06.810 how panning works, lighting, etc. but even if you don't know, you can actually ask ChatGPT or Gemini 02:06.810 --> 02:13.770 or whatever LLM you use to come up with some ideas for how to do the camera shots, you can describe 02:13.770 --> 02:14.370 the environment. 02:14.370 --> 02:19.330 So in this case, we have the astronaut sleeping peacefully on the mystical, mysterious, mystical 02:19.330 --> 02:20.250 alien world. 02:20.500 --> 02:24.900 but here we're describing a setting in more detail, saying it's a floating island. 02:25.100 --> 02:27.180 Neon blue and purple glowing flowers. 02:27.180 --> 02:27.900 Lights. 02:27.940 --> 02:28.460 Magical. 02:28.460 --> 02:29.100 Serene. 02:29.100 --> 02:30.780 And then we have the character. 02:30.820 --> 02:35.260 You can see you could just describe all of this in a prompt, but it might get a little bit confused 02:35.260 --> 02:37.620 as to whether you're talking about the environment or the character. 02:37.620 --> 02:38.940 It could be some bleed over. 02:38.980 --> 02:40.700 This kind of helps structure everything. 02:40.980 --> 02:47.820 I also find this really good for testing different ideas, because maybe you definitely want an astronaut 02:47.860 --> 02:52.660 sleeping peaceful in a mystical alien world, but you maybe want to try and describe like a different 02:52.660 --> 02:52.820 world. 02:52.860 --> 02:57.580 Maybe you make more of a chaotic one, or maybe you do the same astronaut, but with different camera 02:57.580 --> 02:58.140 angle. 02:58.540 --> 03:00.500 Or maybe they're not in a futuristic suit. 03:00.540 --> 03:04.660 They're in an old school suit, so you can start to think more systematically about your prompt this 03:04.660 --> 03:04.940 way. 03:05.100 --> 03:10.460 I'm just going to submit this, and it's going to take a couple minutes to run. 03:11.060 --> 03:17.820 The other thing we can do is I'm just going to take that same prompt, and then I'm going to change 03:17.820 --> 03:24.340 some element of it, maybe I would say anywhere with volcanic islands. 03:24.380 --> 03:31.830 And I'm going to say raw red and black Back pouring lava. 03:32.870 --> 03:39.550 Zero smoke and fog and just going to make the vibes chaotic and angry. 03:41.270 --> 03:41.790 Cool. 03:41.830 --> 03:43.830 So that's going to be really different. 03:43.870 --> 03:47.510 You're sleeping curled up on hard ash. 03:47.670 --> 03:47.910 Yeah. 03:47.950 --> 03:49.430 On volcanic ash, let's say. 03:50.390 --> 03:53.150 And gentle, glowing creatures hovering nearby. 03:53.430 --> 03:54.790 Just call them menacing. 03:55.510 --> 04:02.430 Glowing creatures hovering nearby are rocking their wings nearby. 04:03.350 --> 04:03.750 Cool. 04:03.910 --> 04:08.230 And the cool thing about via, by the way, is you can just generate new ones while you're waiting for 04:08.230 --> 04:08.990 the other one to generate. 04:08.990 --> 04:10.350 But it looks like this just finished. 04:11.150 --> 04:12.830 So let's see how this looks. 04:13.830 --> 04:16.070 So let's follow the prompts really well. 04:16.110 --> 04:17.630 Like we've got the floating islands. 04:17.630 --> 04:20.350 We've got everything here which is really helpful. 04:21.590 --> 04:24.270 And now we've got a completely different video. 04:24.270 --> 04:25.430 So let's play this one. 04:26.070 --> 04:27.510 Yeah very different vibe. 04:27.550 --> 04:28.510 And it's the same prompt. 04:28.510 --> 04:29.270 Exactly. 04:29.910 --> 04:31.390 It's got a lot of the same elements. 04:31.390 --> 04:38.110 But those elements that we changed, like the lava and the winged beasts is really made a difference. 04:38.110 --> 04:40.830 So it just gives you a little bit more control over your prompt.