I’ve been noticing a few videos in this style getting a lot of attention lately, and it’s easy to see why. One of them hit 4.9M views on Instagram in just the first two days. The concept is really interesting. These videos work because they trigger strong emotions in the viewer — a mix of curiosity, fascination, and even a little discomfort. When the final moment reveals the huge swarm of rats, the shock factor is big, which naturally makes people want to comment or share it with friends. Another reason it performs well is that it plays on our curiosity about the hidden ecosystems that exist inside cities. We rarely see what’s happening beneath the streets. The POV “rat-cam” perspective makes the video feel extremely immersive and different. It almost looks like raw found footage, which immediately grabs attention. The dark tunnel sequence, moving quickly through narrow spaces, builds tension and keeps people watching until the end to discover where it leads. That helps a lot with retention. What do you guys think? Would you like me to make a step-by-step tutorial showing how to create a video like this? 👀

One mistake I see a lot of people making is trying to create prompts completely from their imagination. You spend time describing the scene you want… but the result often looks very different from what you pictured. A much better approach is simple: use AI to help you build the prompt. Here’s what I usually do. First, find a reference image that represents the style or scene you want. It can be a screenshot, a photo, or any image that captures the look you're aiming for. Then upload that image to an AI like ChatGPT, Claude, or Gemini and ask something like: “Based on this image, create a detailed prompt I can use to generate a similar image in [your AI tool].” After that, take the prompt it generates and paste it into your image generator. You’ll notice that the results become much closer to the vision you had in mind. The reason this works so well is because AI understands visual references better than abstract descriptions. When you only describe something with words, there’s often a gap between what you're imagining and how the AI interprets it. But when you give it an image, it can analyze the lighting, style, composition, and mood much more accurately. It’s similar to showing someone a picture instead of trying to describe it. Using reference images and letting AI help build your prompts can save you a lot of time and frustration when creating visuals. — Marcos

Hello everyone.is there any tutorial or any training?

Hello everyone 👋 I see a lot of you creating great AI videos… but many of them lose viewers before the interesting part even starts. The reason is simple: the hook is too slow. The first 3 seconds of your video are everything. That’s the moment where people decide if they’re going to keep watching or scroll away. And the mistake most people make is starting with an intro like:“Hey guys, today we’re going to…” That kills the momentum instantly. Instead, begin your video with the most interesting, dramatic, or curiosity-driven moment you have. Drop the viewer straight into the action so they feel like they need to see what happens next. A simple structure that works well is starting with: • a question • a bold claim • or a confusing visual that makes people curious. Examples: “This $2 tool replaced my entire kitchen.” “You’ve been washing your face wrong your whole life.” “Watch what happens when I press this button.” That’s the hook. No intro. No long setup. Just straight into the moment that grabs attention. Next time you create a video, try moving your best scene to the very first frame. You’ll notice your retention improve. Don’t give people a reason to scroll. Give them a reason to stay 🔥 ~ Marcos

When is class starting?