Transcription
AI storytelling reels are exploding right now, pulling millions of views across Instagram, Tik Tok, and YouTube shorts. So, in this video, I will show you how to create two completely different styles using one universal workflow that works for any kind of storytelling reel.
Hello, good people. My name is Mhoki and this is where creativity meets technology, one topic at a time. If you enjoy creative experiments with AI, feel free to subscribe. I share new workflows and discoveries as I explore this space.
Okay, let's face it. I know that a lot of creators feel a little bit overwhelmed with AI tools, but once you understand this workflow, everything becomes so so so simple. And trust me, just stick to the video and you're going to be fine.
Before we create anything, we need references. Storytelling reels, especially the viral ones, aren't random. They follow patterns in pacing, tone, writing, visuals, and sound design. This is actually one of the most important steps and most of the creators skip it and that's exactly why their reels feel random.
Okay, so the two niches I want to try today will be a true crime story about Jeffrey Dahmer and the second one will be an explainer documentary style reel about human senses.
Okay, I start by searching for accounts that post storytelling reels. I'm typing tags like documentary explainer AI storytelling AI reels AI crime story horror and similar keywords. When I see something that catches my eye, I click through and check the account for best performing reels.
For Jeff Dahmer reel, I found this account and I instantly loved the dark 2D animation style. The colors, the atmosphere, it's exactly the vibe I had in mind. So, we've got really good contrast here because we've got 2D animation with true crime story of serial killer. So, yeah, we've got a huge potential here. I'm looking at their top performing reels and this one has over 900k views and more like 41k likes, which is totally insane. So, I downloaded using any free Instagram reel downloader. And I don't stop at one. Let's download a few more so we have enough data to understand the pacing, the tone, the style, and later to train our GPT.
Next, I'm doing exactly the same thing with the documentary explainer reel about human senses. I found this account that's really popular right now. So, let's check their best performing reels. And as you can see, this one about the human eye has almost 16 million views and over 2 million likes. I guess that's what we call viral. Oh my god, I don't know what to say. It's crazy. Okay, let's download this reel and do more from this account. The more examples we collect, the easier it is to understand the pacing, the tone, and the visual language of this niche.
So now we can take some screenshots from our downloaded reels. These screenshots became extremely useful later when we match style, lighting, and color. And remember, we are not copying the style. We're just borrowing the structure and refine into our own vision. It's simply a good starting point for finding our own style by mixing and learning from the best out there.
So once we have enough reference collected, it's time to turn that inspiration into a script. After downloading the reels, I run them through a transcription tool. You can use Riverside, Whisper, or any AI transcriber you like and extract the full script. I'm using Riverside because you can use their free trial. So basically, you don't need to pay for anything for 7 days.
Then I move everything into chat GPT and start a new project. Next, I add a prompt that explains the full context and what I want the AI to do. Right below that prompt, I paste the transcriptions we prepared earlier. This gives ChatGPT real examples of pacing, tone, and structure so it understands exactly how I want the new scripts to be written. All the prompts I use in that step can be found in the description below.
Once the context is set, I generate the first draft of my script. From there, I refine the script a couple of times until it feels tight and cinematic. And remember, you don't need to rely fully on AI. It's always better to trust your gut. Move things around, cut lines, and shape the story until it feels right. And if you want to speed up this whole process, you can also try my custom GPT that's already trained for writing storytelling reels. Just type your idea and it builds the structure for you. The link is below the video.
Once the script is ready, the next step is the voice over. And this part is more important than people think. You can have the best visuals in the world, but if the voice doesn't carry the story, people will scroll away instantly. In storytelling reels, the voice sets the entire atmosphere. It controls the pacing, the emotion, and how the viewer feels while watching.
For this, I use 11 labs. I paste my script inside, choose the voice I want, and adjust the tone depending on the style of the reel. For Jeff Dahmer's story, I've used a darker, dramatic voice, slower and more cinematic.
"Every building has that one apartment, the one neighbors whisper about. In Milwaukee, it was 213."
And for the science reel, I use a calmer, clearer voice with a clean documentary style tone.
"What if I told you your senses are lying to you every second of your life? You think you're seeing reality."
If something doesn't sound right, I just generate a few variations until the voice matches the feel of the story. Once the voice over is ready, I export it as a MP3 file, and that becomes my backbone of the entire reel.
Now, it's time to turn the script into visuals. And this is where the real creative work starts. This is the step that turns your idea into something visually real. And it's also the step where AI can easily ruin your vibe if the prompt isn't clear.
I take my script and I ask ChatGPT to turn each sentence into a detailed cinematic image prompt using a very specific structure. You can find the full template in the description below. So now ChatGPT acts as a prompt optimizer for AI image generation. Each output must describe a single still frame, no movement, and it must follow the format, scene, visuals, composition, and details.
Once ChatGPT gives me the prompts, I drop it into Midjourney. That's the tool I use for generating consistent, high-quality still images. One of the reasons I use Midjourney is because it has Omni reference and style reference. Omni reference keeps the character consistency across the whole reel and style reference lets me match the exact aesthetic using a screenshot or uploaded image. This is honestly one of the easiest way to make your images look intentional instead of random.
Then I generate the image, adjust if needed, and move on to the next line. One line equals one image. And every time I generate an image I'm happy with, I upscale it to the highest possible quality. And remember, the better your input is, the better your image to video results will be. I repeat this process until I have the full set of frames for my reel.
And once the all images are ready, we can finally turn them into the motion. I'm using the Kling video model inside the InVideo platform. It's simple, fast, and gives you clean cinematic motion without needing advanced settings. I upload my image, and click. It understands the whole scene very well on its own. So, I keep the prompt really simple. And if I want subtle camera movement, I add commands like camera dolly in, dolly out, pan left, or tilt up. Because small controlled camera movements always feel the most cinematic. And when I have two shots that naturally connect, I use the first frame and the last frame option. Kling blends the two images into smooth, natural-looking transition. This instantly makes the whole reel look more intentional and polished.
Once all the motion clips are ready, this is where everything starts coming together. Editing is the moment where the story finally takes the shape. I bring my footage into the timeline and start by dropping in the voice over. The voice over sets the pacing, so everything else has to match it. Then I place the animated clips in order, following the script line by line. I keep the cuts really simple. Remember, clarity always wins.
Next, I add music. Something atmospheric, subtle, or cinematic, depending on the tone. Sound design also plays a huge role. Even one or two well-placed effects like a heartbeat or soft whoosh, a distant echo can make the visual feel much more immersive. Then I move on to captions. I keep them really clean and minimal. Short lines, simple font, a bit of glow if the style needed, of course. And once everything feels cohesive, the reel is ready to export.
"What if I told you your senses are lying to you? You think you're seeing reality, but your brain is actually remixing it, turning waves, vibrations, and chemicals into a world that only feels real. Light doesn't have color. Sound doesn't have pitch. Touch doesn't have texture. Those are illusions, electrical signals painted inside your mind. When you hear your favorite song, it isn't your ears that love it. It's your brain interpreting tiny air vibrations. When you smell rain, you're detecting bacteria waking up after the storm. Some people taste words. Others see sounds. Some humans literally hear color. To a dog, your home is a library of smells. To a bat, it's an echo map. To a snake, it glows with heat. Every species lives inside its own filtered version of reality, shaped by senses, not the truth. So maybe the world isn't what you see. Maybe it's what your brain lets you see. The real universe still waits, untouched, unseen, and probably nothing like you imagine."
"He worked at a chocolate factory, quiet, polite. But when the world slept, his apartment came alive. Every building has that one apartment, the one neighbors whisper about. In Milwaukee, it was 213. Camera flashes, faint music, and sometimes voices that stopped too soon. When police broke in, the air was heavy, sweet, sour, still. Red bulbs lit the room. A blue barrel stood sealed tight. hoping that... and on the couch sat a man calm as still. They asked him why. He looked up and whispered, 'I just wanted them to stay.'"
And that's the full workflow, guys. Two completely different styles created with the exact same process. And I really want to emphasize this. It's the workflow, not the tools that makes these reels work. Once you understand the system, script, voice, images, motion, and edit, you can create any storytelling reel you want.
I hope this breakdown helps you create your first storytelling reel. If you enjoyed this video, feel free to subscribe and let me know in the comments what style I should try next. Thanks for watching and I'll see you in the next.