How to Edit Videos With Claude Code (ChatCut Tutorial)

So I typed this one line into Claude Code and it became a video editor where I can say things like remove the bad takes and the pauses or add captions and motion graphics and it can do that on an actual timeline. And in this video I'm going to show you exactly how to do that. So how to install it, how to let it edit your videos, add captions, highlights, resize footage, and a whole lot more. And if you are not currently using AI to edit your videos at all, you are going to want to watch this. I promise it will save you time. It's going to be a fun one. Let's go. So the way we're doing this is with an app called Chatcut that has an MCP or basically lets us integrate the app and the timeline into Claude Code. So let's start by doing the installation and then we'll jump into demo. So that said, let's get into it. Now, before we jump into the AI side of it, this right here is just the CapCut website, so your AI video editor. Now, we are going to use it in the agent in just a second, but you can see right here, all of this is functionally what we're going to do just by not having to touch this website, just directly through the tools you already use, and that does include AI motion graphics, which is pretty epic. So that is Chatcut, and now let's get that installed in Claude. Okay, so here we are at the Chatcut website at forward slash Claude, and I want to start with the installation process to use it within your existing AI tools. Keep in mind, you can use it with ChatGPT as well and TrayWork. If you've never heard of TrayWork, I know you guys all know GPT and Claude. We did make a video previously on that. Now, to install Chatcut for Claude, we're going to open Terminal, and then we're going to copy this top line. We're going to paste it. And what this is doing, basically, is just installing the code from GitHub. That's done, let's paste our second line and this is now installing the plugin of Chatcut. Now we need to authorize our Chatcut account. So we can enter this line, you can see that right here. Go enter. Then once you authorize your account, this page is gonna open up. You can go ahead and log in with Chatcut. And there we go, we are authenticated. And then to confirm, let's open up a new terminal tab. We'll paste this line right here, and we can go ahead and see Chatcut at chatcut.inc. So now I've opened a new terminal tab. Let's enter Claude Code. Let's go to the full screen renderer, and now we're inside of Claude. And let's just say, can you access Chatcut, and what can you do with ChatCut? So we can see that the MCP server is successfully loaded. We can do everything ranging from creating new projects, to the editing, the reorganizing, and then our speech, captions, motion graphics, library, export, the whole nine yards of video editing. And now let's get into some demos of what we can do once it's installed just by talking to Claude. So essentially just going through this list of use cases. Now what I have here is a sample video that's kind of a base track for a video like this. So it's just talking head footage, It's me standing here and there are a lot of, you can see pauses where I'm looking down. And so what we want to do is what normally would take an editor or myself half an hour to do. This is just a two minute video and a longer video, a lot more time than that of removing your bad takes, your pauses, all of that kind of very easy, but just manual and time consuming editing work before you even get to things like motion graphics. So I have this video file right here. Now I've dragged and dropped that into Claude Code, and then I'm going to give it this prompt. Watch this footage, cut the repeated takes and filler words, keep the best take of each line, put the clips in an order that makes sense, and then add captions and music. So again, this is a lot of work, right? This takes a human editor a lot of time to do well, and we're going to see if we can do it just by talking to Claude. This is demo one. Let's go. So we can see it by default loaded the ChatCut plugin. So that's not something I had to prompt it to do. It'll just do that by default. Now we can see it's figuring out a plan. Now we're going to get to the motion graphics and everything in a minute. I think the base application of AI for video editing is just getting rid of work like this. That 80% of menial work, cutting bad takes and fillers. It has to be done for every video. I don't need to tell you this if you live in this world. You know how exactly how annoying it is. And you also know exactly how big of a lift it is for AI to be able to do this now. Okay, we can see it applied the edits, so the failed retake and the duplicates. It's 48% shorter. So that's the volume of work done just on a two and a half minute clip. And now it's doing the music. Looks like it should be ready in a second. Okay, so it removed two verbal. Fillers and 18 long pauses. Let's go ahead and open this clip. And here we go. We can see the resulting work and all the clips that were done. This is just a visual demonstration of the work that the AI already did. And let's give it a watch. Okay, it's done. There's no UI you might have to pass. But it's very well done. Yep. So that right there was a perfect cut. We removed probably four or five seconds of silence. And I've muted this track so you can hear the music for a second a little bit louder. So that's also exactly the type of backtrack you would actually add as a human editor. So overall, and then finally, you can see the captions right here. So this text, let's go ahead and mute this. And that is your captions track as well. And if you want to edit the style or the look of the captions, you can also just talk to Claude to get that done. So overall, that's a massive chunk of your video editing workflow completely. Done with AI now, but we're not stopping there. And for the next demo, I want to double click on that motion graphics side of things. And to really illustrate that, I want to create an ad that is just motion graphics. It is purely CGI, just generated by Claude, no base footage to pull from. All I'm going to give it are some reference images or videos to pull from. So we're going to jump back into our terminal session here. Now, for those who are familiar with the channel, we have this fictional coffee brand that we use. make a lot of different assets for what that looks like is this it's the Alto Cold Brew so i have some different images right there of what that looks like including the logo files i'm going to drop this into the chat and then we're going to separate this process into two different prompts the first is going to be this so pull out the strongest three selling points for the cold brew brand using those reference assets we uploaded and then write a 30 seconds ad script So all we're doing, you can do the same thing for your business, upload a photo of the product or the service, and then we're starting out by just saying, hey, write an ad script for me. What we're going to do after it is ask the AI through ChatGPT to just generate an entire 30 second ad with motion graphics for us, soup to nuts, and I'll let you judge the quality of that work. Okay, there we go. We have our three selling points. So speed, taste, and value. And our 71 word ad script. And now we're going to say this. Build that ad inside of ChatGPT using these product shots. I'm going to re-add the references right here. And let's let it run. Okay, so we can see the finished version here. We have our motion graphics of things like $7 versus $2.50 per cup on subscription. That looks good. And then let's give the voiceover in Musicbed a test. About 2.58 o'clock on Twitter. Cool, so voiceover sounds good, music sounds good. I would want this formatted in more our 9 by 16 kind of horizontal format. For that we can just go back and prompt the agent to make the change. You can also do it, if you want to, inside of the Chatcut agent. And then if we want to edit any of these graphics as well, we can drag and drop here or talk to Claude and say make Whatever changes it is that you want now also everything we're doing in this video demo wise is pretty much leaving the AI just completely open ended if you have references or ideas that you want to describe with text for things like motion graphics or editing you can also communicate that to the agent and it can specifically bring that idea or reference into existence on your video or there are some of the most essential kind of lower third style. motion graphics but i want to show you the more advanced motion graphics that you can make as well and the way in which i want to do that is through this clip and what this is 50 seconds of me talking and then what i've done is i've gone to the chatcut app and i've asked chatcut to add seven different styles of motion graphics to this clip so we're going to review what some of those different styles are so the first one you can see right here this is the editor And if we scroll along here, we can see a plan being built. Right, so this motion graphic overlay, you can see that here made in one prompt on chat cut, excuse me, and you can do this through talking to Claude as well for any of these motion graphics. And so what we're seeing here, not just a lower third, but this clip is about a plan that we're reviewing in the clip of the video where we're looking at the new X subscription. And you can see that plan clearly laid out here with those Overlays, again, that's one prompt, and if we want to change that or improve that, we just keep talking to the AI. That's number one. Number two, right here, you can see the AI decided to add two different overlays. One is right here, so more of a title card. It's great. And then that same plan that we've seen, and the AI chose to just do it in plan number one, plan number two. Here's number three and here the AI didn't think the plan was the best because this is more of a title motion graphic that wouldn't work with that much text so it added three of those titles instead. So really impressive work there. Here's number four, we can see another plan again a different visual style right there and this cool effect here and you can see right this isn't a template this is AI intelligently applying what it is that it thinks is best for the video. So that I think summarizes the new era we're in of video editing better than anything. So here we can see a similar style to the first one, but an alternative. So a little expansive. Option number six right here. This is one of my favorites. I really like the big red text right here. We have this timeline along with the visual. I'll go ahead and let that play a few seconds. And then here, finally. X subscription lower third and research as well. And the way these are being accessed is motion graphic styles right here. So these are just some of the ones that are out there. Whatever you want, as complex as you want, talk to Claude or like we've gone through prior, just talk to Claude and all of this. So everything we looked at is part of the production process for a video, right? And that is a majority of the work you do. But there is one final portion of that process that's a little more into the distribution side of things versus production. But I want to show you how to do with CapCut. So let's jump back into Quad. Now, what I'm going to do is upload as a reference or just drop into the chat the video for demo number one that was cleaned up. So the version exported from ChatCut. And then I'm going to give it this prompt. So take this finished video and cut three vertical shots from it that have the strongest moments with captions, size for TikTok. And Instagram Reels, and then write a LinkedIn post, right? So again, this is part of that distribution process of our video editing is done, but we want to get it out in a lot of different formats across the internet. And we can do this too, because it is that same fundamental process of being able to resize, to edit, to add captions, and so on. So let's let it do that. Okay, we got a little update of where it's at. It is. Just about done and just working on the timelines. Now, while it's finishing up, I do want to say the model you choose to do the editing does matter, right? There's this Pareto frontier of cost versus performance. If you're using Quad, I think Opus roughly is the sweet spot. You can go Fable for more advanced motion graphics to really work through things. You can go Starnet if you want something a little more cost-effective for easier edits. But I think roughly Opus is on par with the... Difficulty of most of the edits that people like yourself and me are dealing with and then finally while this is wrapping up We've been talking about quad this whole time but if you want to do this same exact thing in ChatGPT you can all you do is you go to chatcut.io forward slash ChatGPT And then you copy this link you paste it and then you install it from this repo So you should just use the system you already do all of your AI work in and then again just I think use a model that's cheaper but also performative somewhere in the middle unless you have a really difficult motion graphics challenge in which case again use Fable or the Astra equivalent for GPT. And now it's on captions for the final short number three. Okay, it is done. We have our three shorts, Built in Public, Talk to Claude, No API Keys. And then we have our LinkedIn post written as well. Let's open the project timeline. And here we can see our timeline for short number one, short number two, and short number three. And for each of them, we have our captions here in a different style. We have the audio track, and then we have the cleaned video as well. You can get an idea from this for the potential of AI video editing as it currently stands. This is one of the frontier tools in that regard. And my goal here was to give you a little taste of each part of the editing process, right? Your basic editing, just splicing, cutting, fixing errors of the base track on one side of things. Second, your motion graphics, so adding. Layers on top and then finally the distribution and just that basic menial work that needs to get done that's time intensive of resizing and cropping and pulling out clips for shorts. Now I have three notes on how you should use a system like this. Number one, output quality will track the quality of the prompt. Basically, if you want to get the best use out of a system like this, you should have a vision for what you want it to be, and then just specifically prompt the AI to do so. I don't think these systems yet have kind of the taste element of video editing, but they can do the work, right? They can just brute force these things if you tell them exactly what you want. So there is going to be a quality spread of results based on how you prompt. And then number two, I've touched on this already, but the model setting does matter. So if you go Stable, extra effort, or same with GPT-Astra, you are going to get a better result than if you use a much cheaper model So just keep that in mind. Obviously, those platforms are more expensive, or those models, and I would just say, depending on the difficulty of the edit, that should determine what kind of model you use to complete the task Now as far as pricing goes for this, the plugin is free to install and a lot of what it does stays completely free. For example, to do everything that you just saw on a demo account, that was all completely free to do. Now if you wanted to generate footage, so it's using external tokens, or you want to pass motion graphics to another editor, that's when you're going to want to use the ProPlan. And overall, just if you do a lot of video editing, obviously this is going to save you a lot of money and time and effort for your business. I would say just subscribe to the ProPlan. If you just want to give it a feel, test out AI editing, then just stick with the free plan and you're good to go. I want to give one other shout out to the Desktop Shotcut platform. So the Desktop platform lets you run all the different AI models inside of it locally, and then you can locally process files on your computer so you don't have to upload download time. So if you have really big projects, you can proceed with that. Personally, I find it much better to just roll it into the background of my existing Claude code, codex workflows And so there you go, that's the video. Now, if you do want to install this right now, this is the link to go to. You go there, you copy the line, run it in Claude, just like we did, and you're done. And overall, I think video editing is such an obvious use case for AI for me, but I haven't seen a ton of innovation in this space. I'm personally just really excited to see where this continues to go and how these tools continue to develop over time. If you work in the space, I'm sure you have a similar experience. And so kudos to Chatcut for making this happen, and I will keep y'all updated on tools like this in the future. That said, I appreciate y'all being here, and I'll see you in the next one.