How to Build a Brand With Dreamina – Step 5: Creating a 30-Second AI Commercial with Seedance 2.5

So if you think about it, pretty much every AI video that you've seen over the past few years has been a short clip. Five seconds, 10 seconds, maybe 15 seconds, and that, in a lot of cases, is why AI video looks like AI video. You're stitching four clips together, the lighting's different, the characters look slightly different. It's pretty obvious it was made with AI and it just kind of looks off. Now today, for part five of our six-part AI brand building series, I'm gonna make a 30-second seamless commercial, or at least that's the goal, for our wireless earbud brand, Aero, in a single generation. We're going to see if we can pull it off, make it actually look and feel seamless, like a true kind of flagship TV spot. It's gonna be a really cool one and build on a lot of the assets we have generated over the past four parts. So welcome to the world of Aero, if you're new, and let's get into the video. So to recap, if you're new here, like I said, I'm building a wireless earbud brand called Aero, completely using AI. So far, we've built a brand identity, a set of photos, video ads, an AI spokesperson or rep for our business, and then now, finally, we're putting all of that together into a holistic flagship TV 30-second commercial. Now this video was made possible by my partner, Dreamina, who is sponsoring the entire series. Also, ByteDance recently launched CDANCE 2.5 on Dreamina, and that is the video model we're using for a lot of this, one of the best video models out there. I have actually been meaning to do a full dive into the technical side of the updates, but that's for another video. And now, if you comment Dreamina under this video, we'll send you a link to go and build along yourself, or you can go to the link in the description and just take your idea, whatever's in your head, and learn how to do these things with AI by following along with this video. And that in mind, let's jump into my computer and get into the build. So friends, here we are back inside the Dreamina dashboard for the fifth time in the series now. And like I said, all of what we're about to do is only possible because of the new Dreamina CDANCE 2.5 model launch just came out very recently. So we can see Dreamina CDANCE 2.5 versus the previous versions here. Now I wanna start by just showing you the capability of the model, specifically the Omni reference in the 32nd generation. Then we're going to use more of our established references that we've created to get to this or build to this final commercial hero spot. Here are some of the assets in our Octo canvas that we have created prior. So we have our existing product shots over here, our models. Now, what I'll start with is just two reference images of, you'll see why I point that out in a minute, but for now let's jump back over to here. Let's upload these two images as references. And now I'm going to give it this prompt, create a 32nd cinematic commercial for Aero wireless earbuds. I'm describing the shots I want. And then actually we will need one logo shot. So let's grab our brand kit image there from the first video in the series. We'll pull that in and then we can go ahead and for example, where we say logo, we can reference the logo image. So this is the most basic level of references you could upload if this was your business, literally just an iPhone image or two of your product. We're going to keep this vertical for use on social media or ad platforms. I'm going to go all the way to 30 seconds right here. And let's take a first look at what Dreamin SD-DANCE 2.5 can do. Now I do want to double click really quickly on those editing features. They are very new and exclusive in large part to Dreamin and I think important to understand. So what I've uploaded here is a reference video of some people where you can see a sky in the background. They're on a bus. That is a completely real human generated video. And then a reference of a sunset photo here. And what I can do by clicking on smart edit under the people's omni reference tab here is say something like replace the sky behind the people in the bus with the sky in the sunset video. And then I can reference what that video is. So we use this already. Again, I just want to really clearly detail that process because there's a lot you can do there to kind of point at it or make very specific changes by swapping or adding or removing things to your video. So let's let that run. And then while that's generating, I wanted to show you the edit segment as well. Let's click pause there. Then we can click edit right here. And this lets us actually make very specific edits by scrolling through to the four second clip we want, clicking next and then editing just those four seconds. So instead of having to regenerate a 30 second video in its totality, I can just say, change the background to blue or green, for example, and it will do that just in the four second clip and then give me back the whole 30 second video with that update. So you can see a lot of these editing features are making AI video much more of the traditional editing suites and platform more of your Premier Pro style feel versus just tell the AI what you want. And then once it gives it back in one take, it generates an entire video from scratch and so on. So much more control that you get now as an editor. So now we're gonna go ahead and put together a lot more of the recent assets we've created, specifically the spokesperson that you can see here. So I'm going to go to references and click upload. And then you can see four of our generations previously. So we have our chosen model right here. We have her here as well, a gym, and then in a studio. I'm gonna upload all four of those as references with omni-reference, it caps out at 50, so plenty of space. And with that said, the prompt I'm giving it is create a statement 30 second cinematic commercial. We have our earbuds, we have our spokesperson, we have our brain design. We're building on what we've created up to this point in this series all with AI and now generate this final 30 seconds piece that adheres to, again, the logo, the product, and the spokesperson. And we wanna make sure in what we're looking for is that all these assets are continuous and that they make sense and look the same across the shots. That's been a big downside of AI for years now on the video and image generation side of things. And so now we're gonna see, can we overcome that? And if so, what does that unlock as far as the quality of result goes? We're gonna keep all these settings the same and let's let it run. Okay, it's done. Let's go ahead and give it a walk. These are Aero. And they cost less than the big names. I have not taken mine off in a week. Aero, elevate your sound. Cool. I love the fade in at the end, super fun. So there you go. You can see the human looks realistic, no errors. The sounds, the music, the sound effects, the narration, all of that layered, it's leveled, sounds good. So we can edit anything here we want to, but honestly, that's all we need. But let's jump into the even beefed up version of what you just saw. Here we go. These are Aero. 40 hours of battery, noise canceling that actually works, and they cost less than the big names. Aero. So a few things. I mean, I think the fade in on the logo, again, super awesome. The big difference, or one of the big differences in this version versus the previous is that text that you can see on screen actually renders correctly, which is good. And overall, again, I mean, just the realism across the board is awesome. Now, what we might want to do here is play around with the script, but again, keep in mind, this is the script that I fed it. So you can play around with that as much as you want. I think I may do that in a little bit before we go ahead and actually launch the brain. So now what's pretty incredible about AI is that we can now take this one 30-second commercial that we have generated, and then just translate it or edit it as much as we want. And there's two features there that really matter. One is just, again, the ability to translate for that. We just upload our reference, and we just quite literally say regenerate this, but have the person speak, and then whatever language it is that you want, I'll go ahead and say Spanish. So you can quite literally say that. I'll do that now, and we can take a look at the results. We want to make sure, number one, the lip sync matches. Number two, everything is the same from the original, just with the different language. Okay, and here's that Spanish version. The other feature I want to dive into briefly is called Smart Edit. That lets us make specific changes to these commercials or these videos if we want to. So let's take a look at that now as well. So let's go ahead and click on Edit Segment. Let's click Next, and let's say change the background color to green. Let's let that run. And so any super granular changes you want can be achieved with Gremina C-DANCE 2.5. That's one of the big updates of the model where you don't need to regenerate a whole 30-second video. You can just edit that one, for example, little four-second clip that you want there, and we'll let that finish processing, and then we can take a look at the update. These are Aero. And so there we go. We're back on the canvas where we started. That's it. We have generated upward of 100, if not more than that, assets at this point. All that's left is to take everything we've created and get it out to the world. The next video will be very exciting, and that wraps our 30-second commercial generations with Gremina C-DANCE 2.5. So there you go. That's the generation. There are two things I wanna quickly mention that we didn't demo, but you should know exist. One is long video mode that goes up to three minutes. That's if you're making kind of a brand short film instead of a spot, so something with more of a story arc. Then each clip there gets two extension attempts in post if 30 seconds runs out. Also, this isn't relevant to earbuds, but it might be for you. It will take 3D white models straight out of Blender or Maya through a plugin. So if you work in game CG or something like architectural visualization, then you aren't gonna wanna take a look at that. Okay, so overall take on limitations here, I think number one, timestamps are directional. They're not accurate on a per frame basis. So if you need to cut on a literal specific frame for syncing something or whatnot, then you might need to trim in post. For everything else, it is pretty much good enough. Number two, typically what I still like to do is generate a few variations and pick my favorites because 30 seconds is still a very long generation time for these newer models. They can do it now, which is awesome and they couldn't before. But again, I like to just give it a few shots on goal and then pick my favorite from there. And then number three, smart edit is great, but still again, it takes a few tries to get right. So it is still much faster than regenerating the whole asset, but it's not just one click and you're done, which I do hope is where this feature is going. But there we go. Aero is coming more and more to life. We now have our 30 second commercial spot. And that again, the way I view it is our flagship asset that we've generated for the brand up to this point. So we're getting there, right? The next episode, part six of this series is launch day. It's taking everything we've created and putting it out there into the real world. Now, again, if you want to build yourself, comment Dreamina under this video or go to the link on the screen. And coming soon, I'm excited for this one. Again, we're gonna take Aero and put her out there in the real world and put these assets to the test. That in mind, I hope you're learning something from this series and enjoying yourself. I appreciate you being here and I will see you in the next one. Cheers.