Create Multi‑Shot Cinematic Videos with SeaDance 2.0 (Full Transcript)

[00:00:00] Speaker 1: SeaDance 2.0 can create incredible cinematic videos with multiple different scenes within one single generation. So let me show you how to use SeaDance 2.0 and create cinematic multi-shot sequences inside of Eleven Creative. Let's begin. So inside of Eleven Creative we're going to go ahead and click on Image & Video and down here to create multiple shot sequences with SeaDance 2.0 we're simply going to switch to video and then we're going to change the model and select SeaDance 2.0 and right here is where we want to describe the video that we want to generate and input our prompt. And so when it comes to generating multi-shot sequences with SeaDance 2.0 it's all in the prompt but there's also a few details that you need to tweak as well. And so I could put in the prompt a cinematic trailer for a British period drama and there's a few things here we could do to make it a multi-sequence shot. First of all SeaDance 2.0 tends to create generations with multiple different scenes the longer it is. So if we went and selected 50 seconds which is the maximum duration at which you can generate with SeaDance 2.0 it's more likely to have multiple scenes. If we were to switch it to let's say 4 seconds it's less likely to be the case. The next thing is that SeaDance 2.0 uses the context of the prompt to decide whether there'll be multiple scenes. So if I change this back to 15 seconds when we're saying a cinematic trailer in trailers you usually have multiple scenes so it's more likely to generate multiple shots within your generation. We could also do something like add cinematic camera angles on the end of our prompt. Again here we're not even asking for multiple camera angles but because we're telling it that there should be cinematic camera angles plural SeaDance 2.0 will see that and will create multiple different scenes within the same generation. And so if we generate it we get a little something like this. And so as you can see by default because it's a long generation SeaDance 2.0 is generating multiple shots within the same sequence. If we wanted a single continuous shot we would add continuous single shot in the actual prompt. And you can actually use short prompts to your advantage during the ideation stage because sometimes SeaDance 2.0 might actually create and suggest and generate things that you wouldn't have thought of otherwise. But it is a bit hit or miss so when you want to have full control over the entire sequence you need to be more specific with the prompt. The second type of prompt that we have is a descriptive prompt and we covered this during the felt animation tutorial but this type of prompt essentially breaks down each individual element within your generation. So the aesthetic, the story, the characters, the environment, the action sequence, production brief and negative prompt. But for the action sequence obviously we're describing the different scenes that we want to see within the generation. So we're describing the sequence of shots that we want to see in that single video. As you can see here we've got the herd plods through swaying grass wind whips their ears silver rain threads fall the sun breaks through the silk ribbons and the calf splashes happily in the blue felt puddle. And the two parts to pay attention to when it comes to generating multiple shots within your sequence are the action sequence and the production brief. And these two parts of the prompt will give SeaDance 2.0 an idea of how many different scenes and shots that you want within your generation. And so again the more specific you are here the more likely you are to see each scene. But what if we want even more control over our generation? And this can look very similar except within the action sequence we break it down and we're as granular as possible. So for example here in the prompt you can see that we've got shot one shot two shot three shot four shot five and shot six. And for each of those shots we give a specific description as to what happens in that shot. And so it's almost like we're putting prompts within the prompt. And for each of these shots we are also giving it specific timestamps so it knows the exact duration and what it's supposed to generate from one specific second to another. And so here as you can see we've got from zero to three seconds top down close up he's asleep eyes closed face still. And here we don't have to be super specific because thanks to the rest of the prompt it knows and has full context as to what we're talking about. So when we say he's asleep eyes closed thanks to this character section in the prompt it knows who we are talking about. And using this type of prompt you can get some really cool sequences just like this. And as you can see we've got a generation that matches our prompts really really well. And the cool thing is that we could then actually go and place people or unique characters in these generations. So for example if we generated a specific character sheet like you can do with GPT image 2 using this prompt we could use the same descriptive prompt that we've used for the scene just before except this time drag in my character reference and so here I'm using me as the reference but we've created a reference sheet so my entire outfit and face is consistent throughout all of the different scenes generated and this is great when we're doing multiple generations to have consistent characters but we also want to make sure that we change this in the character. So for example here instead of saying one man I could say man from and then tag the specific reference so here I'm doing man from at image one and now we can go ahead and generate the same thing. And of course we want to adjust the settings so here I want to generate a 1080p we want to make this 15 seconds long because we want to match the timestamps that we're giving it if we make it four seconds but we're giving it 15 seconds of timestamps SceneArts 2.0 will get very confused but once we've adjusted all of our settings and we can include audio we can actually generate two generations at a time and then we simply click generate and we get something just like this. And as you can see I put myself in the generation that has the exact same sequence of shots than the first one and that is thanks to the fact that we are so detailed within the prompt and if we put them side by side you can actually see that the tempo and description of each scene within the sequence matches except this time it's me within the generation. And a cool thing to highlight within this prompt in shot five from second 10 to second 12 we asked for three rapid match cuts and so that's a different way that you can pack different cuts within the same type of scene that way it felt a little bit more continuous and they didn't feel too different or have too much contrast between them. And we've actually found after playing around with SceneArts 2.0 that the sweet spot is between five to seven scenes per generation that lasts 15 seconds. You can do more for example for this parkour one there's up to 15 different shots but sometimes it can feel too quick too dense and some of these shots don't make sense within the generation but the more descriptive you are about each specific shot within your prompt the better the results. And so if you want to start creating multi-shot sequences you can click the first link in the description down below and use SceneArts 2.0 inside of Eleven Creative. If you have any questions let us know in the comment section down below and if you enjoyed this video hit that like button and don't forget to subscribe. Thanks for watching.