How I Made a Short Film by Talking to an AI | Dreamina Octo Vibe Creation

So I didn't write a single prompt to make that. I built it by talking the idea itself through with an AI on one canvas from a rough thought to a finished short. And most AI tools nowadays work the same way where you give it a prompt, you get something back, you improve the prompt, you try again, you do that a few times and eventually you try and get to the thing that's in your head. So in this video we're covering a different approach from Octo that is Dreamina's brand new creative agent. And the idea is that you work with it on a single canvas where it understands the context of what's already been discussed or worked on on that canvas and then it will answer back with questions and suggestions over time. And so as opposed to filling in a form it's much more like you're talking through the breadth of the idea with a partner so to speak that understands all the nuances of it and anything prior that is important to recall when creating what's involved now. Dreamina calls this vibe create which I like. And so creatively what I want to show you in this video is that process of vibe creating. And then we'll get into a little bit about what this new Octo update entails. If you have no idea what Dreamina is we've covered it in some prior update videos. Either way if you work in any creative space I think you'll find this a useful one and let's get into it. So in terms of what the Octo creative agent is specifically it lives on the Dreamina canvas. You can talk or type to it. And then under the hood it runs three frontier models at once. So you have Dreamina Seedance 2.0 running for the video. GPT image 2 for the stills. And then Gemini for reasoning. So the agent itself depending on the task at hand under the hood will delineate tasks to these different models. And so now let's jump into my computer. I want to show you a full demo on what this entails and how to use it. And thanks to Dreamina for the early access to try out this feature and for sponsoring this video. If you want to follow along as well so try out the agent while you're watching this video you can go to this link on the screen or in the description. And now let's get into the first demo. Okay so here we are inside the Dreamina dashboard. Now we've made tons of videos at this point created all kinds of cool stuff on the platform. So if you're not familiar you can go ahead and watch those. For now I just want to jump straight into the Okto canvas and for that we can press right here on the beta link. And you can already get a feeling for kind of the vibe of this. It's definitely a different vibe for interacting with AI and much more conversational attuned to a conversation. So if we hit our forward slash right here we get a little hi there from the agent. And we'll just start talking. And what I want to start this flow or this conversation off with with the AI. And we'll jump more into the features in a second but I want you to understand the high level means of engagement the modality of Okto. And so I've just said hey and then what I want to say is I have a little idea. Now my idea is a lone figure walking through a rain soaked neon city at night. That's all it is. Help me develop this. So you don't have to have some big vision in mind. All you need is an idea to start. Let's go ahead and expand this little portal. And so the agent has responded and said hey are we seeing this more like a hyper realistic cyberpunk scene or more dreamlike. So you can see it's taking actions on behalf of me. It's going and experimenting with some visuals. We can see he just generated one right here. That actually looks really good. I'm a fan of that. What I'm looks great. So we can go on its own initiative create images videos text basically a whole range of design assets. And that's what I recommend for you. If you have a kernel of an idea for your brand for your business for client work just sit down and just talk for 20 minutes and develop the scope of it just like we're doing here. And you can start to see some of the actual features or the generations. For example if we click on a generation like this we can go ahead and edit with a prompt. We can generate a different type of media from it or we can go ahead and have some more specific editing applied. Okay so now it's asking me a question. It's saying do we want to figure out a character profile. Let's check out a short scene list to see what they're doing in the city. So let's say what are your ideas for the story of the person. What could they be up to that makes for a good story. And now we're going to do something pretty cool. It's also pretty unique to any kind of AI design platform. Okay and I like this one idea of a digital scavenger. So it's a scavenger in a city. There's a very material vibe that's being felt out here. Now it's asking me do we want something specific or is the search itself the focus. I think the search itself should be the focus. And then here's where we get into one of the really cool features and that is locking characters for sake of consistency. What I'm going to say is let's build out a character profile a visual style in the environment. And this is where we really get into the bulk design and generation abilities. Whereas if you're in something like GPT and you want an image you'll get one image at a time. Something like this it can give you 20 30 40 assets at a time if you really want it to. And we can see that being created here. So it has a cool Kaelin if that's how you pronounce that name for the character. And then this is a very material kind of character training set for the AI that is creating. And then it will recall and use again and again and again with this kind of 3D model look. And also the personality and arc and appearance. So we can edit all of this if we want to. But then this is going to be reused across every shot. And what's cool is we can go ahead and reference it. So when I type this app right here I can reference the character and I can say for example let's make an image with and then reference the character doing such and such. So that's pretty cool. And I like this idea of a seamless. So let's kind of expand this story a little bit into the domain of more visuals first off and imagery and then ultimately a video a short drama piece as well. So let's say yeah turn it into a seamless or think think through that story. Okay a sequence of discoveries. I'll tell it that sounds good it can go ahead. Okay and it's developed before we go there it pushed back a little bit which is really cool. It's kind of the design partner feel on hey let's get the story figured out before we more storyboard visually. So we have three layers the sub-level trawl the near miss and then the fragment. Let's say that sounds good. Let's get it onto the canvas. Okay it's creating the seamless now. Okay and we can see that three scene breakdown here. We have some reference images and now we want to turn this into more of that storyboard. Now one cool thing while it's working on that if we press right here we can actually use voice input. So when we say this is conversational you can actually just literally have a conversation and kind of talk back and forth using your voice instead of typing with the agent. It's also a pretty cool feature there. Okay we can see the first of those come through for kind of scene design. That is looking epic. So that's the flooded industrial pipe. I'll say that looks good. Let's do the other scenes for kind of storyboarding. I'll say that looks good. Let's build out the other scenes. We can see that being generated here. Okay also a super awesome design style there and if you want a different style again you know you just talk to the agent and also while this is generating you can see we can move these assets around and line them up. I think just the little like lineup features that's a really nice touch for me at least just visually UI UX wise. Here we can see the final storyboard coming in there. So we have all of our scenes now. So I'm going to say yeah let's develop this more fully. So add narration, a script, a full proper storyboard, whatever you think is best to fully flesh it out. Let's see what it says there. So the agent again had more of an idea here where it's saying hey I have the shots the character and the flow why don't we just jump straight to the video. So I'm going to say okay just generate the clips for the three shots directly. So let's give that a minute to process. Okay and we can see that those clips have been completed. So we can see actual videos here that we'll check out in a second as opposed to the images or the storyboarding before that. So you can see that throughout this conversation without kind of a plan in mind we've gone through from the idea to a few mood board style visualizers then into the full character that we're now locking certain elements right then the full-on storyboard. So we can see the scenes that are being generated and then now we've turned that into even more design aspects in the form of this video. So let's give a few of these scenes a watch. So very cool it's this kind of trippy surrealist style which I really like. If you ever want a different style again you just talk to the agent. I'm a fan of this and you can hear the layered audio in there as well. Let's watch the second one. Really good audio here really well done and that's a sick ending shot right there. Okay cool third one let's see what this does. I'm not a fan of this one as much so I'm clicking on this edit button right here and I'm going to it and then I think we'll be good to put them into one video and that can be the final deliverable here of kind of a mini that we've generated from this idea to potentially turn into something later or to publish as is. We can see this is in the video generation queue right now and I'm saying nice job there to the agent you got to give the AI credit and now it's asking hey do you want to combine them to the final sequence so let's just get this final clip made and then we'll be good to do that. Now also more formally what the agent is working with in terms of structured elements that can be returned to and locked to stay consistent that's your style reference character object environment and then custom elements like objects. So those are the five things where if you have say a short series on social media or you make brand videos where certain elements like a logo or a mascot has to look the same in every single piece of media then it's those elements that you lock and if you want something to be locked and to be reused you just talk to the agent. Okay the revised version is done let's give that a watch. So I like the pacing better more of a surrealist theme we're good to go with that one let's go ahead and delete this so that gives us our three scenes right here in order so he finds the device he's looking at something he goes up he's avoiding the drone things and then he comes across the device here to be continued perhaps let's say yeah I have the three clips lined up combine them into one final sequence here we go okay and it's also going to add transitions then it'll render that put it on the canvas and also you can see right here that we can hit if we want to add the timeline and then we can do that ourselves but this is the age of AI we don't want to do it ourself and so we'll let the OctoAgent let's click continue okay that export is fully complete we can see that full 21 second video with the audio here and then something else I asked the agent to do is a cool follow-up here just to illustrate the scale of what you can do with this stuff is basically just make a second version so it went through and made another environment a whole search a script all scenes here with the image storyboard and then it created these three shots right here so I actually want to video so you can also get an understanding of why it's so important to bulk generate right because you have these different styles that you can pull out of the agent through conversation so we can see the first clip right here let's give this one a listen very cool vibe my favorite is the third see that in a second uh human looks awesome in this this is a super realistic human so he's searching for the thing and then here because look at the mud in this shot I think this is extremely well done and so there we go that is a similar story finding the device from a different angle and this is also what I mean when you're talking about the canvas kind of bulk generation I think anyone working in design AI nowadays you shouldn't be using the chap offs to generate point images one image one video at a time for this reason because you can get your 5 10 15 20 plus assets created at the same time octo is a great solution for that there are others as well but regardless you want to be bulk generating nowadays it is a much better iterative method to be more productive but also just be more creative and that in mind that's octo and we're going to finish this up by just saying thank you octo because if anything is the theme of this it is that octo is a partner for you in design and so kudos to the dream and the team in getting this out there and I look forward to continuing to see it improve so I think what's important here is how this fits into your workflow and this is much closer to that kind of creative partner when creating with AI as opposed to that for give x get y try and iterate through that process and overall it's cool to see where dream and us taking these new tools they're making a lot of improvements for sure as well as many other players in the space and so I think now is a really awesome time to work in in any kind of creative space really I can say that personally and I'm sure you've experienced that as well and again if you think that would make you more productive if you want to try it out you can go to this link on the screen or in the description we do help to support the channel when you do that so go create something cool that in mind appreciate you being here and I will see you in the next one cheers