# Youtube Shorts + Claude Automation = 21,857$/month Channel: Money Guy Video: https://www.youtube.com/watch?v=IKcAbOwV2SM Duration: 11 min Language: English Words: 2088 Transcript page: https://viewrankai.com/tools/youtube-transcript/IKcAbOwV2SM --- [0:00] There is a short niche making 2D animated content for sports that is pulling tens of millions of views on his videos. And from experience with my own channel, I can say that one video with that many views can make you from one to even $2,000. But this channel is making one mistake, and because of that mistake, he's leaving billions of views and money on the table. Yes, I'm not exaggerating because this guy basically took Zach D. Films format and made it 2D applied it to sports. But he missed one essential thing, which is uploading consistently. He uploads just once every 5 days, but still gets 22 million views [0:43] a month. If he or you could upload these types of videos every single day, you would blow up even more. So, I gave myself a challenge to find a way to create these types of videos in less than 20 minutes. Sounds impossible? That's what I thought in the beginning as well, but just look what I managed to [1:00] create with AI eventually. At 15, Cristiano Ronaldo was 5 ft 9 in. His son just hit 6 ft 3 in. Cristiano Ronaldo Jr. has grown a full 26 cm over the last 3 years and reached 6 ft 3 in at the age of 15. And because males height growth can typically continue until the ages of 18 to 21, [music] Ronaldo Jr. could end up being the same height as LeBron James by the time he turns [1:25] 21? You stupid. This footage was created with AI, and surprisingly, I used just one prompt to create it. I did it with Claude's newest fable 5 model and Higgsfield. Everything was generated and downloaded straight to my computer. But the craziest part is that when errors were popping up during the generation, the AI spotted them himself and fixed them. I literally just [1:48] watched and pressed buttons to continue. And that was honestly fun to watch. I think it will be extremely valuable for you as well because you'll be able to apply it to any niche and create content way faster and efficient. So, stay watching because I'm going to show you the full process step-by-step. So, before the magic happens, there's a setup part. And don't skip it because if you make this wrong, nothing after will work. First, you'll need to install Claude on your computer. So, just open Claude website, download the Claude, and that's it. Then, you'll need to connect Higgsfield to your Claude. So, just go on settings, connectors, add custom [2:26] connector, name it Higgsfield, and here you'll paste MCP link, which you can find on Higgsfield's website. Go on Higgsfield MCP, copy the link, paste it right here, and press add. I already added it, so I won't add it again. Then, you'll just have to connect your Higgsfield account, and that's pretty much it from the Claude side. However, this is not finished yet. Now, we need to add Higgsfield MCP to Claude code, and I already showed this step in my previous video. So, if you haven't watched it or watched it but don't remember it, I will play now that part of the video, so so you could follow the [3:00] step-by-step process because it's not that easy to do. So, pay attention and do everything step-by-step. In order to actually automate the image generation, we need to install the Higgsfield skill to Claude code. I know it might sound complicated. It was the same for me, and it is the same for me now. But since I've already gone through it, I'll show it to you. So, go back to your browser and search for Higgsfield. Once you log in, find the section for MCP and CLI, and click on CLI. Here, you can see three commands that we'll need to paste into your computer's terminal, which sounds easy to do, but here is where I [3:38] faced with a lot of issues. So, I'll show you how to fix those in case you come up with them as well. So, open your terminal, then paste the first command and hit enter. That should install the CLI, but for me, it showed an error that most of you probably going to face as well. After some research, I figured out that I needed to download an additional program called Node.js, which lets your computer run these JavaScripts outside the web browser. After installing it, I pasted the command and got another error. And to fix that error, all I had to do was add a simple .cmd text to my command. I don't know exactly [4:20] how that helps, but [music] at least it does. So, then go back, copy the second command, and paste it. If the same error happens, just add that .cmd text again, and it will work. At least that's what worked for me. After hitting enter, it'll ask you to authorize access with your account. And then the third and final command actually installs the Higgs Field skill that generates the images. I can paste the command, and if the same error happens, include the same .cmd. But, after this, your installation might still fail, cuz that that's what happened for me. So, again, I did some research and found that you'll need to [4:55] install one more program. So, you'll need to install Git, but you don't need to actually go to any website. You can just type this command, and it installs automatically. So, after installation, I retried the third command, and it finally worked. Now, after completing this step, you'll need your voiceover to be transcribed with timestamps. This is what the AI uses to know exactly what [5:16] needs to happen on screen every second. So, go to the tool like Turbo Scribe or any tool that transcribes your voiceover with timestamps. I did it with my voiceover right here, and as you can see, I got this short voiceover with timestamps. And now, the most important piece, the prompt. I uploaded my big and very detailed prompt together with realistic reference images of the main characters I need in the video. [music] And this prompt is super long, and it [5:43] basically commands the whole operation. First, it has to generate characters and character sheets based on the script and the reference I gave. And it can only use them after my approval. So, if it shows me a generated character and it doesn't look how I want, I can freely ask to change it before anything else gets made. Then, it analyzes and generates the backgrounds the video will need. And this part is also very crucial because this is what keeps characters and backgrounds consistent through the whole video instead of your character or background changing every shot. So, everything is uploaded, reference are in, let's see what actually happens when [6:21] you click enter. And this is where you either save credits or waste them. So, watch closely. The first thing it gives me is a shot list. It analyzes the script, picks the best images and video generation models for the job, and tells me approximately how much it will cost. Right here, you can see it's 500 credits. But, don't worry, it will cost you way lower. For some reason, the shot list wasn't showing for me, so I just confirmed the Nano Banana Pro and models, uploaded my reference image [6:49] again, and just look how clear this is. Three main backgrounds. For every single shot, you can see which characters it will use, what action they'll do, what graphics will be shown, even the framing. Everything laid out before a single credit gets spent. So, if something looks off, you can change it now instead of paying for a bad generation. For me, everything seems [7:09] fine, so I write approved and continue. And now the magic begins. It starts generating the character sheets, and as you can see, it immediately runs into a problem. It can't generate. Maybe it's copyright, maybe it's because it's a famous celebrity. I don't know exactly. But, what I do know is that AI will find a way and keep going until it works. And you'll even see it fixing major mistakes [7:32] in the photo himself. Just look at this. LeBron is wearing number 44 on his shirt, and the text says veterans, which is complete nonsense, right? After the generation, the AI explains everything itself. The safety filter blocked all photo reference conversions. So, all three sheets were built from the appearance locked text description instead. But, this is the part that surprised me the most. I already caught and fixed two problems myself. Full text frame and fake team name on LeBron's first version. It caught its own [8:03] mistakes. No, like nobody told it to. So, now I look at the images and they're exactly the way I wanted. Yeah, maybe Ronaldo Jr's hair is a bit too big. But, honestly, for me, that's even better because from my personal channel and experience on going viral multiple times, I can say that it gives that wow factor and people will comment about his hair. And comments boost engagement and views. So, I will leave it like this. If you wanted to change it, you just write [8:31] in the chat and ask to regenerate. That's it. It also shows how many credits it used so far. For me, everything is okay, so I just press approve. And with that one word, it starts generating the backgrounds. And after that, the shots themselves based on the script. After generating, it even flags which shot might be bad and asks if I want to regenerate. For example, here, the second shot. But, for me, it looks fine, so I'll just keep it. I kept approving and after some time, I had all nine shots. There was one shot I wanted to regenerate. LeBron James was generated standing in front of [9:04] basketball rim and covering it, so I asked for a new one and got a good clip. And remember that 500 estimate? The whole thing cost just 103 credits because it switched the model to cling. I don't know why, maybe AI just analyzed and thought it will be better. But, these 103 credits is basically $5 for one video. And some people might complain that it costs money, but just look from the other perspective. If you will have to hire an animator for this job, you would spend 50, maybe even $100 for one video. Now, you just spent $5. I don't know if you can get it any [9:41] cheaper. So, the AI part is done. I asked it to download every video to my folder, and here they are. Now, the last 15 minutes is putting everything together. I put everything together in my editing software the way I want. I added my voice over, simple arrows, sound effects, some memes for engagement, and background music that I used from Epidemic Sounds. I zoomed in some funny faces, and I got this final result that you already saw in the beginning of the video. That took me an extra 15 minutes, but I'll be honest with you, that's because I already know how to edit fast, and I have all my [10:16] keybinds and shortcuts set up. And the 20 minutes of real work doesn't count the waiting time while things generate. I'm not going to pretend that the video makes itself while you sleep. You still have to confirm things and write extra one to two words, you know? But, 20 minutes of actual hands-on work for an animated video that used to take days for animators, that's just unreal. So, that's the full system. Apply to your niche and create your own unique style to stand out, find gap in the market, and go viral. Create your unique prompt, or if you want mine, I'll leave my exact prompt in my school community along with [10:55] everything else about growing with shorts. But, I know some people can't afford spending money and still would like to look at the prompt fully. [music] So, here's the deal. If we reach 2,000 likes on this video, I will give the exact full prompt I used to my newsletter subscribers, which which is free to subscribe. So, like the video and subscribe to the newsletter and this channel if you actually need it. I'll [11:18] see you in the next one. Peace. --- About this transcript Read from YouTube's own caption track and laid out by ViewRank AI (https://viewrankai.com). ViewRank AI finds the videos already beating a creator's own average on Instagram, TikTok and YouTube Shorts, transcribes them from the audio itself in more than 60 languages, and turns what worked into new ideas and scripts. Free transcript tools, no account needed: https://viewrankai.com/tools How to read any video this way: https://viewrankai.com/llms.txt