AI Video Visuals Without Slop — Step-by-Step Guide & Transcript

[00:01] content in videos? Visualizing what you are currently saying or explaining. Everything you say can be forgotten within seconds. Unless you visualize it. In this video,

[00:15] I'll show you how, with the help of Cloud and the Hixfield MCP, you can not only automatically add suitable visualizations to your raw video, which can really

[00:35] and then go through it step by step. With Hickfield AI, you now have the option to use an MCP, a Model Context Protocol. This allows us

[00:47] Context Protocol. This allows us and then utilize their functions via our AI. So, to put it

[01:02] cloud as an example. We have the option here to simply copy this MCP URL . I'll just use the Cloud Desktop App for that. And theoretically, you could also do it via Cloud

[01:16] Code. You can do it via the cloud in your browser. It is important to browser. It is important to MCP, once using your own account. We can easily do this here, for example in the game at Cowork, by customizing it

[01:30] Connectors tab at the bottom. Here I have the option to add various connections. Among other things, you can now easily add the Hixfield MCP.

[01:42] Then we can call this whole thing Hfield, for example, and then here's the URL we just copied. Then click on " Add" and now you have to connect. Then a window will open asking you again if

[01:57] you allow access. Confirm everything with "allow" and we will then receive the success message connecting to Hixfield. And from that moment started. That means if I simply write down what I

[02:10] want here and have also activated the Hixfield collector, I can have the first animations created . And a legitimate question might already arise for you: why not Remotion or

[02:25] Hyper Frames? I've already introduced Remotion before, for example, to create game animation in my videos. The important thing is to recognize where I use Remotion, for example, and where I use Hickfield. There is actually a relatively

[02:41] simple way to demonstrate this. Remotion is pretty good for graphs if you want to blend them, while HF does much more complex things. But . I've structured it so

[02:55] , expand on it further, and in the end you'll see a way to create really good animations for your entire video. So, my very simple starting point is, if I

[03:10] select Fable 5 here, for example, which is included in my Max Planment, then I can describe relatively simply what I want. Let's do that look like. Create a suitable clip for me using Hixfield. A streamer sits

[03:25] Twitch. We look at him from behind and see a very clean setup equipped with a lot of equipment, and so on. I'm not going to read the yourself. What actually happens when we fire it is that

[03:38] Claud Fable 5 uses it to build the correct instructions and sends the whole thing via the MCP to Hixfield, where our clip is then created. Let's just do it . You can see that the thought process starts immediately after sending the message.

[03:53] Then loaded tools Hixfield generate video. And now the question is being asked here: Is this allowed to be done? Yes, and the communication continues. Here, they are requesting that the data be returned by Hixfield . Hey, I suggest you

[04:06] . Hey, I suggest you use the Ins Dark Prompt. Yes, let's just do it .

[04:24] minutes. What Claud and Hickfiel are actually doing here is that a lot of and Hickfiel are actually doing here is that a lot of time [laughter][gasps] But after a short time, or as I said,

[04:38] a few minutes, our finished clip comes out. Just a heads-up, I did finished clip comes out. Just a heads-up, I did n't even know what kind of preset that was. Leave out the preset. This was intended more for

[04:52] aware of that. I edited the whole thing again without the preset, and this is what the clip looks like. That's pretty ingenious, I'd say. You can really create some awesome clips that

[05:05] you can then incorporate into the video. And if, for example, I explain a great streaming setup and I talk about using three screens , bam, I have my clip without having to spend

[05:18] half an hour setting it up or recording myself, but I can generate it. And that, I think, is actually the crucial point. I'm not doing any AI slop nonsense; I have good

[05:32] content and simply want to enrich it with suitable clips, and this fits extremely well. But let's move on to a completely different example, and that would be example 2 for advanced learners. We don't just want to

[05:44] We don't just want to create clips, we want to add a little something extra to our videos. If you edit your videos in Adobe Premiere Pro or other tools, you might find that there's

[05:57] a part you want to change. Let's assume I want to change this part now . Then I could go in and have it printed out as a clip . If I do that,

[06:10] course, I can rename it now. That's not what this is about at all. [music] But I can use it now. I can use it now. I can then send the clip to the Hickfield MCP via Cloud and have it

[06:23] add cool effects. Let 's add the file here first. Then that it's a clip of mine from the video and that I want to edit the clip with Hixfield . A brief instruction: the ceiling is to collapse, a rope like that used in

[06:36] drop down, and I grab the rope and am pulled upwards out of the picture . Okay, now I'm exaggerating a bit. Yes, then let's get started . And once the whole process starts here, you'll see

[06:48] what the result is. Pretty cool. You can also see what instructions Cloud Fable 5, for example, then sends to Hickfield. If you open this [music] section, you'll see that my information has been

[07:02] expanded considerably. This will then run through for a short spoiled it, but if you watch the clip , [laughter] I can't help but laugh. It's so awesome.

[07:22] wanted, to add another small effect to the video. Of options once I've edited my clip . So, as you can see, it's then zooms out a bit upwards and looks incredibly realistic.

[07:36] scale the whole thing down here. That means I can say, yes, please downscale to 4K. And it will be sent back to Hexfield and they will create another really awesome, juicy 4K

[07:49] version of this clip for me. And of place it exactly where I wanted my animation to be. That means I'll just put it here . Zack.

[08:03] 'm talking here and then the ceiling breaks open and I'm pulled out of the picture . So [laughter] yes, I think it's clear what I mean . This is a great place to easily revise the video

[08:18] . Whether flames erupt behind me or simply the door opens and something flies in. I can use it to lighten things up can use it to lighten things up or focus on something in

[08:31] order to better explain to the viewer what I actually I'm trying to say. Before we go through an entire video to an entire video to

[08:44] , I'd like to make this point again . I already mentioned made a video about before , namely the generation of, let's say , animations with data, and the difference between when

[08:59] Hixfield makes sense and when Remotion makes sense . To communicate this clearly and honestly, I simply gave the same specifications for both and had graphs generated. You can see it here again

[09:13] . Everything is stated here, including the correct terminology, and this is what Hixfield makes of it. So it's actually not usable, intended use case for Hixfield, and Hixfield isn't meant for that anyway. If

[09:28] which we have here below, I'll open it now , then we get a really nice graph that we simply had generated with Remotion via the cloud and can use that for our video.

[09:41] If I want graphs and text, and I want to visualize them, then I don't use HFD, I use Remotion. I use Hfield when I'm doing exactly what Remotion can't do, namely building really complex animations

[09:54] where I'm also involved, where I might generate broll footage of a , which we've already done. Hickfield is incredibly strong in that area, and we can really benefit from it. But now, as promised, let's

[10:08] transition to the part where we're going to edit an entire video, and there are approaches to doing that. For example, I have now

[10:20] and have them transcribed by Cloud via Whisper. So what we need is a transcript of our video. However, many gaps in the video at some point, and these will be cut out, we can also generate the transcript here

[10:36] using, for example, text, then " generate static transcript", then of course select German and then the audio line, i.e., the track. Have the whole thing built. This process is relatively quick, and we can feed this information into

[10:50] Clord and then tell Cloud that we would like to know the places where it makes sense to show clips, and also which clips. Once this is transcribed, I'll simply take this text (Export, export

[11:06] to Text File) and insert it into my project. Then I go to Cloud and here I can open a new chat and simply select the file. That means, take the file again, my transcript, and then

[11:18] kept it relatively simple. Read the attached transcript. [music] Then analyze the places where it makes sense to help visualize my explanations. Exactly what I'm always preaching.

[11:33] , please list them for me in a table description of what you want to show in the clip . This is important so that you have another kind of review step. So, you can see again what would be

[11:46] generated, what would be shown for which text, with which explanation, and then you can edit this table before you release it. And that's exactly when we say, only when I approve the clips, will I

[11:58] start generating them. So let 's do this. I also received a list here after a relatively short time. I had them shortened to just 6 because I didn't want to

[12:10] go through a list with you all for ages. And you can see some really cool stuff there. For example, you can stream not only in landscape format but also in portrait format simultaneously; you can simply rotate the whole thing and then the clip stops. Hand holds smartphone with running

[12:23] simply switches to landscape format full screen and a sharp desk with purple LED ambience in the background, and you can see various things simply have built. Let's assume I've read through it

[12:37] and it's exactly how I want it. Then I simply say, "Get started." want it. Then I simply say, "Get started." And Clud immediately starts And Clud immediately starts

[12:53] few minutes, I have all my finished clips. And here's also shows that you have to be a little careful. For example, I have some pretty cool clips, like this one where a gamer is standing or sitting in front of his screen

[13:07] and then starts playing. This is really well done with camera movement. We also have a cool clip showing the difference between landscape and mobile phone. Here, the whole thing is simply rotated, and in the end,

[13:21] it visualizes exactly what is being explained, wonderfully so. But when we then move on to topics like graphs like these, where we don't have any text at dpi, which can be done wonderfully, and is also superbly constructed, but if we now have text involved,

[13:36] look here for example, then you'll see that the text is sometimes not use that, because the text isn't important here . But when you have things like this, you can

[13:51] . But when you have things like this, you can , you also have to pay attention to that a little bit. Cloud Fable 5 does an extremely good job of thinking about what needs to be in a prompt for

[14:09] Hickfield to generate something good. So the collaboration between Cloud Fable 5 and collaboration between Cloud Fable 5 and Hixfield via the MCP is working really, really well. But I would recommend having the cinematic

[14:21] clips professionally generated, because they aren't record at home. Here's another issue. Incredibly awesome. If you were to build that, it would take hours. So let that be

[14:35] said. HF itself, of course, offers many more things. That means if I take a look here, the whole thing is also available for explainers . That means if I make explainer videos and I want to

[14:48] have such explainer graphics, I have many other image tools, I have many other image tools, I have video tools and can combine all of them with Cloud Fable 5, and Cloud Fable 5 is of course

[15:02] incredibly ingenious to think about and to enter everything in a really meaningful way. So, the advantage is, why should I use the MCP ? If I simply enter my prompt, then I need to know exactly what I

[15:16] want to create. But if I use Cloud Fable 5 to explain what I want to achieve, and Cloud Fable 5 generates the prompt for me and generates the prompt for me and passes it to Hickfield via the MCP, nice, really, really

[15:30] cool. All information about Hixfield and the Hixfield MCP can of course be found below in the video description, and I would say, see you in the would say, see you in the next video. Ciao. Ciao.