[00:09] this video, and this is the power that Sans 2.0 has.  Today I'm going to teach you everything I learned while testing this tool.  If this is your first time passing through the canal, I am Lorwin Enrique, captain of the artificial ark.  Let's begin.
[00:22] captain of the artificial ark.  Let's begin. Arcan released by Bike Dance, the owners of TikTok.  This tool has caused such a stir that there's an issue with Hollywood where they ordered restrictions on
[00:38] many of the things that could be done. But be warned, this has happened with Sora, in fact it happened with BO3 in its early days, so we have to wait a bit for the tide to go down, for time to pass to see how this tool fares.  I take this
[00:53] opportunity to thank Mo, who is part of the Youart staff, for allowing me to use and test the tool on their platform.  To begin with, if you know the channel, I'm always looking for the best formula when creating a
[01:06] promo to give it the best structure and get the best results.  And this time formula I found to get the best results in this test.  The first thing you have to do is put the identity, where you are going to talk about your
[01:20] character, if it is one or two, if it is a product, you are going to talk about everything here. Second, the specific scene, where the character is, where the product is, you will place everything without ambiguity. Three, the main action.  What's the
[01:34] idea?  Is he going to run, is he going to shoot?  Is the product going to drop in price?  Will the person float?  You're going to put all of that here. Progression.  You're going to give it a beginning, an evolution, and a seamless ending.  With camera changes, yes, but always
[01:48] continuous.  So, what exactly does the guided camera do ?  If there will be tracking, if it will follow the target, if it will give it a twist.  You're going to put all of this in the cameras. So, what are realistic physics like?   Do
[02:01] they interact with their environment?  You will put everything in this section.  And finally, the restrictions.  If you don't want distortion or blur, and you want to maintain an untouchable consistency, then you place it in this section.  If
[02:14] you want to add dialogue, use it this way.  Natural character A, okay?   You're going to put it in a short sentence or a medium-sized sentence, not too long.  And then the second character, if they are going to answer, make it a short sentence, one that is understandable
[02:28] with its commas and periods, so that the guide can interpret it in the best way.  The best thing about this is that he does it in Spanish and he does it very well.  In the end, the complete prom structure would be this.
[02:45] the community, both on Telegram and on YouTube, and they all gave me very similar answers, so I'm going to start with this test.  One of the things they mentioned most was character consistency, that he do it with my own
[02:58] personality and realistic cases.  So, the first thing I did was create an image for myself. I'm a soldier, I'm in a war, and I made two videos.  The first one was 15 seconds.  I used the option where Sidans Pro lets you use an initial frame and an
[03:14] final frame if you want.  In fact, you can see it here.  Here you can paste the prom, the start frame, and the end frame. However, in sedan omnir reference it's different, it only grabs references. But I'll explain this to you a little
[03:28] later.  The promo I used for this video was the following. run, the camera to give a 180-degree turn, and this was the result.
[03:58] the explosions are there, boom, although it's very close, the character in the back reacts, the turn is good, but in the end not much happens.  Then I wanted to try Sidans Pro too, what's the difference here?  So I tell him that I'm
[04:12] shooting, that the camera is following me, and he was very, very, very specific about the was very, very, very specific about the cameras, and this is the result.
[04:29] showed you at the beginning.  The cameras and the movement, the shooting, all of this makes it very beastly.  The sound is getting perfect, it's adapting perfectly.  Obviously there's an explosion very close to me, I don't fully react, but it's very well
[04:45] done.  Another one I did, where they're asking me for character consistency, was this one .  I was able to dry this one with OVN [clearing throat] reference because I put my [clearing throat] reference because I put my photo on it, I put an image of what the
[04:58] place where I was was like.  The promo I used for this was the following. Notice that here in the prom I'm putting image one is the
[05:11] character reference, image two is the reference environment, the reference place, okay?  As you connect them, if I, for example, release here, notice that it disconnects.  If I come and connect it, it takes it
[05:25] directly.  I put this there so that he could identify what each image was.  This was a video that I was going to take 15 seconds to make, but in the end I ended up taking it in 7 seconds and this is what I came up with.
[05:38] Today I will teach you how to master key elements of SD Dance 2.0 and it's superb.  To continue with the topic of consistency, you can see that by placing my photo it takes a video, it maintains consistency, my self, but it
[05:54] maintains the character during the seconds that the video lasts.  To show you another example of this, here is the following.  What I was going to try here was the product from the ark, which you already know if you've seen the channel.  with a photo of
[06:07] me.  You can see that I made it on a white background because that's another piece of information I'm giving you. as a reference, it's always best to have them with a white background so the model causes fewer problems when taking real people, and the same applies to the product.
[06:22] As you can see here, it gave me an error. Because?  Because I think UFO Reference is the most used and well, times are going very, very crazy.  But to solve the
[06:34] problem of the vertical video that I wanted to try and that I wanted to test with product, I created this image where it's me or a virtual one dressed as a biker in a bar and I put the same promo that I'm using in Chinese that, look, the
[06:50] in Chinese that, look, the promo has dialogues in Spanish.  Bang, bang.  And here's another thing the crew wanted to see, which was how the model reacted to dialogues in Spanish, and the result is as follows.   Let me
[07:05] tell you something, not all of them taste the same.   Let's really like the lexing that's there, I like how it reacts.  It's really
[07:21] well done, you know?  the camera change there.  There was a mistake here when he said blah blah blah, but it's not understandable, but if I cut it out and such, there's no problem or if I could generate it again, there's no problem.  Oh, I
[07:34] forgot to tell you, the promo I used for this video was the following. Another test I did with the product was this one .
[07:52] Spanish prom and it worked perfectly.  The promo I used is the following. With this one, I encountered one of the most common problems I had during the entire test, which was the copy, like, look, you're using either a character or some brand and
[08:09] original product, it's not even this one.   It was a Nike, okay?  Product N. I had to remove the logo here.  Then I turned off Zoom X to see what would happen.
[08:21] However, the problem didn't come from the product, but from the person.  What can you do to fix this?  Next, I went to fix this?  Next, I went to Gemini and created a 3.3 grid to create
[08:35] different views of the product, and with the boy I also created a 3x3 grid where I got several views of the boy. This makes me interpret it as, "Oh, this isn't a real person, it's a model, it's just several photos."
[08:50] Then help him not to interpret it. It's like skipping the filter a bit when creating.  This information is super important, at least in Arto.  Here
[09:02] I also used the promino to see if I could bypass the entire filter.  If you're wondering how I switched prom to Chinese, just come to Deep L, okay?
[09:14] Here you paste your prom hello and it puts it in Chinese and here it tells you in Chinese.  It's that easy.  This mini commercial lasted 7 seconds with camera changes and this is seconds with camera changes and this is the result.
[09:34] prom and I was testing it.  Another piece of information I'll give you here is that I'm testing you here is that I'm testing Cance Omnireference here, but in this case, Cance Omnireference here, but in this case, I used something called Star DCE 2.0.   Let me
[09:47] explain, if they give version 2.0 it comes directly from the supplier in China and Star Dance changed it, okay?  Because it's an API, they're still the four S's , but if you look at it, one costs much more than the other, but we'll
[10:03] see about the prices.  Now, another one that was requested and that I had pending because I had already done it with Clean 3.0 when it came out, is one of a car transforming.  It's a super viral thing that's been done and I did it too.  In
[10:18] fact, I tested it here using the same frame I used in Clean.  If you saw the Clean video, you'll know which video I'm referring to.  I'll show you the comparison right now.  The promo I used for this video was the following.
[10:32] I'm using Sidans Pro 2.0 here.  And something I want to clarify here as well is that at first I was using the PROM directly from the video creation node
[10:44] , but then when I started placing the text promo and the image separately and it started connecting, notice that here it simply shows me the promo is attached to this node, so they came out much faster.  What's happening?
[11:00] Sidan Pro and Sidan 2.0 have too much traffic and take a very long time, but when I did it this way, it didn't take me long at all.  And the result of this video is the all.  And the result of this video is the following.
[11:23] When I did it in clean, the result was the following. other comparisons that I also want you to convince me of, to see which one you like
[11:39] best.  Another example I made to simulate what an avatar is, you know? Look, I'm promoting a product, although I didn't do it entirely that way, and I apologize because the same thing happened to me.  It was one of the first videos, and
[11:52] when it changed, boom, the nodes came loose and it produced a video without the reference.  But that's how it is .  The promo I used for this video was the following.
[12:04] The idea is that he would use this girl as a reference.  As I can say, it pointed out was the following. looked good, effortless, that felt
[12:17] light, but that made a difference.  And that's exactly it. The dialogue is very well adapted, the lexing is very well done, the dialogue is consistent with what I wrote, and the character is very realistic.  It's
[12:29] really, really well done, and I liked that.  It's a shame that I only realized afterwards that it had come loose and I couldn't try it.  I could have put a product here or the initial photo with a product like the one I made with the
[12:43] beer and it takes it out.  Notice that the beer maintained its consistency, just as it maintained the consistency of the beer or the can itself.  For this video I made a comparison with BO3 because it was one of the things that were also
[12:56] requested and this is the result. This is the lipstick I'm using right now .  I love how BO3's lips look in this aspect.  He did it
[13:08] perfectly, although there's the issue of "I did this," and then there are some weird pauses that did n't happen with Sidans.  However, the BO3 video is very well done.  What do because there are more of these comparison videos.  Another video that was requested
[13:24] was about dancing and something with anime. So, I mixed a little bit of So, I mixed a little bit of realism here with an animated character, so to speak, in a real area, I mean, it could be real.  What did I do with this
[13:38] video?  that they came walking and ended up dancing.  This is the result big a dream, but I can't stop imagining myself there dancing.  It's not just a dream, it's who you are and we're going to achieve it together.
[13:58] They come walking, he speaks, you have lip sync.  It's very nice. I think the only mistake is when he says, "We're going to achieve it" instead of " achieving it is very good."  And when they dance, there is synchronization between the two and
[14:11] synchronization between the two and they make the same movement.  And I liked this a lot.  For this I also made a comparison with Clean and with BO3 and I'll show them to you . Sometimes I feel like it's too
[14:26] big a dream, but I can't stop imagining myself there dancing.   I'm going to do it.   Of course .
[14:40] Which one do you think had the best result when it came to dancing?  I think all three.  What's my problem or what's my drawback?  I see three did what he wanted in the sense that I gave him the dialogue, I never told him it had to be a
[14:54] song, but that they should dance to a song and he took the dialogue as a song and that's what he did.  In Clint's case, I had to reduce the text to what they saw so that he could understand the Spanish much better.  However, when it came to
[15:08] dancing, Can was perfect, in fact, BO3 was perfect and Sidans, well, perfect.  I can give the highest score to Sidans because he did everything, and in fact the dialogue was much longer and he took it on. Obviously in Sidans it was 15 seconds, BO3
[15:25] only lets you do eight, and in Clean I did it in 10 seconds.  To wrap things up, one thing I was asked about was camera movement, so to speak, abrupt camera movements, and that's why I made this video.  This is a video I
[15:38] made from a prom, meaning I didn't look for an initial image or anything, I just kept using Sidans Pro because I was going with a direct PR and this is the result.
[15:59] passing through a modern city.  I think the camera movements are amazing, and it's something we 're already used to seeing with Sidans 2.0 because, well, thousands of people who have made videos have seen how
[16:14] well the camera movements are done.  I wanted to try it and I know you also wanted to know what it was like, if it really happened or was just people talking nonsense, and the truth is that there are some very good camera movements.  My
[16:31] verdict after this test I did in Art, where I can tell you that I spent 5,000 credits that they gave me for this test, is that Sidans has a this test, is that Sidans has a gigantic power.  This nerf, so to
[16:46] speak, that they made it so that real characters can't appear and such, is a strong limitation.  However , there are ways to achieve it.  As you can see, I took some pictures of myself, and this didn't happen
[17:01] at the beginning in BO3, if you remember.  Later, as they started coming out and all that, months passed, and that's when they started allowing people to create videos with real people.  On that side, the power is brutal.  The
[17:15] camera movement is brutal.  How it respects camera movements.  It's very well done.  The dialogues are well done.  I'd like to try making a video of 1 minute, 2 minutes, or 5 minutes testing the consistency of
[17:31] dialogues and characters, but a complete story, not just clips to see how it works.  Also, if you're watching this video and enjoying it, please like it. This will help me gain more exposure and get
[17:45] more collaborations that give me more credit, allowing me to make more videos like this and share my workflow and how I made it so you can benefit from it too. Something I had mentioned was seeing them
[17:59] in the credits.  If you make 15-second videos, they're going to take away 300 credits, at least here on Are.  And if you use the model, which is the API, it's much
[18:11] more expensive.  In fact, I made about two videos of these and that's what cost me the most. two videos of these and that's what cost me the most. It was 630 for each video.  If you use 5 It was 630 for each video.  If you use 5 seconds it's 168 or 80. 10 seconds is
[18:23] about 200 credits or so.  So, to give you an idea of ​​roughly how much you can spend.  The plans aren't expensive at all.  It costs $10 and they aren't expensive at all.  It costs $10 and they give you 1,000 credits and €30, a good
[18:38] investment for 3,300 credits.  I think you can do something really good with 3,300 credits, especially if you want to try it out.  At the time of recording this video, platforms like Dreaminia have already expanded the countries where they can be
[18:52] used.  It wasn't available yesterday, but I checked this morning and it is now available, so at least in Europe it's ready to use.  And another platform that also has it is Martini, so you have several
[19:05] options.  I have no problem with that and I really appreciate it.  If you can try it, I'll leave you Artar's description.
[19:18] subscribe and hit the notification bell, share this video, comment on it, like it, and spread the word so we can reach more people.  And as I told you, this way you'll help me get
[19:31] better collaborations and do something much bigger.  On the other hand, I Telegram community.  It's a completely free community.  The link will be in the description .  And finally, if you have a project that you don't know how to handle
[19:44] look, I don't know how to do the workflow, I have one-on-one consultations where you and I will sit down for 75 minutes to work on your project.  I'll also leave that in the description.  And that's how this video ends.  I hope you
[19:59] liked it.  If you want to continue learning about artificial intelligence focused on images or videos, you have this video here and here you'll find my face so you can subscribe directly to the channel.  Remember to share and we'll see you in
[20:14] channel.  Remember to share and we'll see you in the next video.