Create Cinematic AI UGC Ads Easily With Seedance 2.5 (Full Guide)

With Cance 2.5, I'm going to show you how you can go from this to a full 30-second AIGC ad like this using just a single prompt. I'm also going to give you my exact prompting framework with a prompt master sheet that's going to allow you to create multiple different ads while you're following along. Now, this is the same exact workflow that I use for my e-commerce stores to get more consistent and usable AI UGC ads. The process is super simple. It's repeatable and really anyone can follow along. Okay, so to get started, I'm going to assume that you already have a product in mind. I already know what product I'm going to be drop shipping, but if you need some help, what I was doing is I was going through the handpicked product section, but if you do need some help, just go ahead and scroll through here, check out your different categories, and you'll be able to find a few different options. Now, I had this one already in mind, and the good thing is is that here you also already have some winning ad creatives. So, what we're going to do is we're going to take the first one that we have here and we're simply going to recreate it. Now, if you do want to see what it's like, you can just go ahead and click on play. >> Does anyone else's bought that snow and keep him awake all that? >> He talks a little bit about the product itself. He shows it. He puts it on. So, we're going to go ahead and recreate that. Now, the way that we're going to do this is by using Seedance 2.5. And we're going to gain access to Cedance 2.5 by using Higsfield. Now, Hicksfield is fantastic for this because we have options on there to use pretty much every single video generator that's out and they always have the newest releases, which for this we're going to be using Cance 2.5. So, let's go ahead and click on here. We're going to click on video and there's going to be two different options. So, what I showed you at the beginning with me transitioning through the different types of video styles and the different products. That is the Cance 2.5 edit. We're going to check that one out a little bit later. For now, we're going to create a completely original video based off of somebody else's video, which is the one that we just saw. So, what we're going to do here is by default, since it is the newer version, the newer model, it's already selected on here. We have Cance 2.5. And what we're going to need to do is we're going to need to add a few references. Now, when it comes to different references, we can add up to 50 different references, which gives us a lot of customization options. This is awesome because we can literally give it every single scene essentially in the way that we want it. And Cance 2.5 is going to put everything together, the way that it should go. But of course, this is all heavily going to depend on the prompt that we give it. So for that, I do have the prompt master sheet for you. If you want access to that, just go ahead and drop a comment down below letting me know what you're going to be making with it. What type of ads are you going to be making or what kinds of UGC are you going to be making along with the # AI UGC? And then once I see that you went ahead and did that, I'll reply back with a link to the sheet. So for now, what we're going to do is we're going to go ahead and first we're going to run over to chat GBT and then we're going to take one of the prompts that I have in the master sheet and we're going to paste it in here. Now the other thing that we're going to do is we're going to go ahead and start to prepare our different images that we're going to feed to Seedance. So for this, what we want to do is we want to have some clear product images and a few images of the product actually in use. And to do this is super easy. All we do is just go back to AutoDS and take screenshots of our image. So, we're going to go ahead and take this one. Try to get as many different detailed images as you can. Go ahead and take that one. I like this one cuz it actually shows with somebody using it. Go ahead and screenshot that one as well. And then if you want to just have the different color variations, you can go ahead and take screenshots of these as well. But for now, the three that we have should be enough because really with this one, we already have the front, we have the side, the back is pretty self-explanatory, and we have the inside of it. Now, the other thing that we're going to have to do is download this video. Now, the good thing is that for this, it's actually really easy to do. We don't have to have any fancy software or anything like that. All we have to do is click on see the original ad, and then we're going to take the link up here. We're going to cut it, and then we're going to look up snaptic.app. So, it's going to be this one over here, snaptic.app. We're going to go ahead and click on that and then we're going to paste in our video link. We're going to go ahead and click on download. Then once we have it downloaded, we're going to run back to chat GBT. We're going to upload directly to the chat that same video that we downloaded along with the prompt from our master sheet. And we're also going to include our three screenshots or however many screenshots that we took. Now, from here, we just go ahead and send it. And Chat GPT is going to analyze everything. It's going to analyze the video. It's going to check it out frame by frame. and it's going to put together a few different creatives for us. Now, after about 10 11 minutes, it's going to give us a video analysis. So, it's going to tell us what the video is about, but then it's going to give us three different creatives. So, if we scroll through, we're going to see exactly what the video is about. This is a minute and 9 seconds, which is a lot longer. It's actually double the size or the length of the videos that Cedense 2.5 can generate. CDN 2.5 can go up to 30 seconds, but that's fine because the prompt sheet minimizes the time to those 30 seconds or less. Now, if you want to read this, this gives you a total breakdown of the video itself. But if we keep going, we're then going to have the different creatives that I talked about. So, here we have creative one, which is a side sleeper swap, creative 2, the bedroom compromise, and creative three, the three-step windown. Now, from here, it's going to give you a pitch, the audience on who to sell to, the selling angle, the first 3 seconds, so the hook for the first three seconds, the presenter, the styling of the entire video, the performance, and the setting. And then it's going to give you timestamps along with the visuals and what the presenter is going to be saying. Now this is very important to understand because the way that Cance 2.5 works is completely different than any other video generator out there. Cance 2.5 actually generates the audio at the same time that it's generating the video. Whereas other generators generate the video first and then they generate the audio. Now why is this important? Because this right here determines how realistic our video is going to be and how realistic the speech is. And that is insanely important. Me personally, if I see a video and the speech is off, maybe the lips aren't working or they're not talking or matching the words that are coming out, I get turned off from it and I just keep scrolling and we don't want that. So, what we're going to do from here is we're going to go ahead and read through everything and choose the creative that we want. Now, really quick, if you're enjoying this so far and you're actually taking something away from it and you're learning, make sure you hit that subscribe button, make sure you like this video, and after you like it, if you see a little hype button down there, go ahead and click that as well. that lets us know that you're interested in videos like this, which helps us start making a few more for you. All right, let's go ahead and continue. Now, I decided to go with the first creative. And I also decided, you know what? I want to put myself in the video. You don't have to do this. Chat GBT will automatically generate a random person, but I wanted to see how it would come out with myself. So, I went with let's go with number one. And let's make the person presenting this one. So, use these to generate the assets for the presenter and make it so that I'm wearing a nicer shirt because my shirt was a little bit messed up. or one more appropriate for sleep time. Now, after about 13 minutes, what ended up happening is that ChachiBT actually analyzed all of my images. It analyzed the original video and it came up with an entire prompt for us along with different assets. So, it gave us this. Here we have creative one is locked down. It's going to have the generated reference images. So, this is the way that it's going to map everything out. It's going to give us the final script and a bit more information along with the prompt that we're going to give seed dance which I'll show you exactly how to use that in just a minute. But then if we keep scrolling, we're going to have the different images that we're also generated. So these are the ones that we're also going to be feeding into Cance. So that way you can see what the video is already going to look like to begin with. So what we're going to do is we're going to go ahead and click on one image. Let's click on this one. Now we're going to go ahead and download all six images in the series. Then once they're downloaded, let's run back to our prompt. Let's go ahead and copy it. Then we're going to go to Higsfield and we're going to put everything in here. So let's go ahead and type in our prompt. Go ahead and paste it in here. Then under references, we're going to upload all of our different references. Then after they're uploaded, we're going to go ahead and select each one of them. You might have to verify them just to make sure that they're usable for commercial use. They're going to be usable, don't worry about that. But Higfield does require a quick verification. And if when you do have to do it, it only takes a few seconds. Now once that's done, let's go ahead and verify over here that we have all of our images, we have our prompt, and now we want to do is under model, we do want to verify that it's C down 2.5 and not 2.5 edit. So once we're good with that, we're going to take the duration and we're actually going to max it out to 30 seconds. Then we're going to switch this over to 9x6, which is portrait mode like a phone. And you can generate up to 1080p, but I'm going to leave it at 720p just so we can get our test video, see how everything looks, and really so we can use less credits as well because honestly, it does use a lot of credits. Right now, Cance 2.5 is currently using 195 credits through Higsfield and that in itself does take up a decent chunk of credits. It's maybe two to three times, maybe even four times more than some of the other ones. But then again, we are making a 30se secondond long video and it's actually very detailed and very well done, very well put together, so it's totally worth it. So now we're just simply going to go ahead and click on generate. It's going to take us to this page over here and it's currently processing and generating. So this is going to take anywhere between 15 to 20 minutes. Now again, this is being completely generated from scratch. It's using all of our different reference images, and it's putting together a 30-se secondond video, so it is going to take a little bit longer than most of the other models. So, we'll be right back. All right, so it did take about 15 minutes to generate this one video, but just check out how good this came out. >> Side sleepers still fighting your earbuds? Try this instead. It's a Bluetooth sleep mask so I can play rain sounds without anything stuck in my ears. The convert eye area gives my eyelids room, and the adjustable strap helps me get the fit right. It combines darkness and audio in one bedtime step. Tap below to check it out. >> So, overall, I think this was a very well done video for something that was completely generated with AI in less than 20, 30 minutes. It is absolutely mind-blowing. Now, the cool thing about this is that you can actually change the voice directly on here if you want to. I'm fine with the way that the voice came out. But, one more thing that I would suggest to do on this is probably speed it up just a little bit. maybe by like 1 to 1.10 times as fast because the way that everything was going through it was a little bit slower paced. So that is one thing that I would definitely do. I would speed it up but again not by much. 1.10x should be enough. All right. So for this next one, what I'm going to do is I'm going to show you how I created the video at the very beginning. So for this we're going to switch over to 2.5Eit and we're going to use the same exact video that we just generated, but I'm going to delete all these reference images. I'm also going to take a quick screenshot of this just so chatbt can give me a couple of different ideas on what we could do. Switch this over to 2.5 edit and delete the prompt. Now let's run back over to chatbt and well chatbt already has the different images so I don't need the screenshot but if you're using any other video something that you didn't feed chatbt earlier then just go ahead and take a screenshot and include it in here. So let's go ahead and think about this really fast. What can we do? Let's switch it over to sunglasses with speakers in them. I think that's good. So, I'm just going to be giving ChBT a little bit of context. So, let's rework the prompt to take the existing video that was generated as a reference video and make it so that I'm on a balcony in a high-rise at night. Keep all of the spoken language the same and add any details that you think would be relevant for that type of scene. We'll be using Cance 2.5 edit, which uses a video as a reference to recreate. Let's go ahead and send that. All right, so it gave us back our prompt. So, let's go ahead and copy this. Let's run back over to Higsfield. And we're just going to go ahead and post it in here. Now, what we do need to do also is we need to upload the video that we just generated. Well, first, let me download it. Now, once it's downloaded, just go ahead and do add a video to edit. Then, we're going to upload the one that we just did. Go ahead and click on upload. Or actually, you can go ahead and click on video generations, and you'll be able to find it here as well. My bad. So, let's click on that. It'll show up over here. Kind of look like a model there, huh? And then under prompt, we're going to go ahead and put our prompt in there. Perfect. So, we have our prompt, Cense 2.5 edit, 720p. And the bit rate, we're going to keep that at high. Don't touch that. And we're good to go. So, let's go ahead and generate for this one. And now, it's going to be processing. And let's just give it about 15 minutes. And we'll be right back. Now, while those are generating, let's go ahead and run this one. Now, this is one that I had done previously. Pretty much the same as a sleep mask, except for this one, I didn't use myself. I actually just went with whatever generations chatbt came up with and this was the avatar that it came up with. So, let's go ahead and check this out. >> Wait, I'm only doing half my face so you can actually see the difference. This is Hydramas foundation and this side is staying completely bare. It blends in fast and makes everything look smoother without covering up my skin. Okay, now look straight down the middle. Hydramos is on this side. This side has nothing. And here it is in natural window light. See the difference? Tap to check it out. >> So, in my opinion, this came out absolutely fantastic. There was one little thing at the end where she turned one way instead of turning the other way. Now, this is an easy fix. All you really have to do is just with a little bit of editing, just go ahead and cut that last part out and flip it, and you're good to go from there. But, as you saw, this was very realistic looking. It looked really well. The way that she actually put the foundation on and started spreading it around, it looked real. There was really not much on here that was telling me that she was made completely with AI. Okay, so the first one just finished generating. And I did another one just to give you guys a little bit of a bonus. So, let's see what this one came out looking like. Side sleepers still fighting your earbuds? Try this instead. It's a Bluetooth sleep mask so I can play rain sounds without anything stuck in my ears. The conferred eye area gives my eyelids room and the adjustable strap helps me get the fit right. It combines darkness and audio in one bedtime step. It combines darkness and audio in one bedtime step. Tap below to check it out. >> Okay, so this one worked out pretty well. It wasn't as good as the first one and it had a couple of repeated steps. So, what I would do from here is I would simply either just edit the prompt or I would tell ChatGBT what's wrong with the prompt or even easier just go ahead and download this and upload it directly to ChatGBT and tell it to fix the prompt based on what it's analyzing from the video. That is going to be your best friend. Always remember, my biggest tip for this is if there's something wrong with your video, instead of you trying to describe what's wrong with it, if it's getting hard for you to be able to actually describe that, just upload the video to JBT. JGBT is going to analyze the entire video. It's going to take maybe about 10 minutes to do so, but it's going to give you hands down the absolute best results. >> Side sleepers still fighting your earbuds? Try this instead. It's a Bluetooth sleep mask so I can play rain sounds without anything stuck in my ears. The conferred eye area gives my eyelids room and the adjustable strap helps me get the fit right. It combines darkness and audio in one bedtime step. It combines darkness and audio in one bedtime step. Tap below to check it out. >> Okay, this one actually came out really good. I still find it kind of strange that I'm sleeping on a balcony, but that was my fault because I simply told Chachi BT, "Put me on a balcony instead of giving it a bit more detail and telling it, you know, put me on a balcony just as if I was sitting on a chair rather than I mean, I guess it kind of works. I just don't see myself going to sleep on a pillow on a balcony like this one." But I think this one did pretty good. Now, from all of the different ones that we just saw, I think the best ones that I would go with would be this one over here, the original one, but also the Hydro Mass one. Now, that one I think came out really good. Again, they both would use just a little bit of touching up, a little bit of editing to really have them nailed down, but at the end of the day with AI, there's always going to be a little bit of human intervention that we need to have. It's never going to come out perfect the first prompt, and it's never going to be a fully finished product. Although, it will be super close. We can upload these already as they are. But personally, I would like to just gloss them up just a little bit and maybe have them look something like this. Side sleepers still fighting your earbuds? Try this instead. It's a Bluetooth sleep mask so I can play rain sounds without anything stuck in my ears. The conferred eye area gives my eyelids room and the adjustable strap helps me get the fit right. It combines darkness and audio in one bedtime step. Tap below to check it out. >> So remember, if you do want to check out the prompt that I used for ChachiBT to start all of this, go ahead and drop a comment down below with the # AIUGC and let me know what type of UGC ads you would be creating. Now, if you want to learn how you can actually make this a lot faster and automate it even more by being able to connect either chatbt or claude directly to Higsfield and do everything through the chat, go ahead and check out this video right Yeah.