Make Your First AI Video in 3 Hours: 5 Steps with Codex (No Experience Needed)
How to Make an AI Video in 3 Hours (Zero Experience Required) – Full Walkthrough
This guide breaks down the exact process used to create the first AI video in just 3 hours, without writing a single line of code or editing by hand. Whether you're a complete beginner or a curious creator, these five steps will save you time and frustration.
Step 1: Install the Right Tools
- Use Codex or Cloud Code (cloud version helps if you hit usage limits).
- Install skills – these are instruction manuals that tell Codex what to do for each job.
- Don't install entire large packages (e.g., the social media package has 106 skills). Instead, tell Codex what you're making and let it choose only the skills you need.
- For this project, only 7 skills were required (list shown in video). For more on AI creation basics, check out Unlocking Creation: How to Make Videos with Nvidia AI for Beginners.
Step 2: Write Your Script – The Foundation
Key insight: Most AI video tutorials jump straight to tools. Wait until you have something to say, then build.
- The script went through 16 versions before it felt right.
- Use different skills to review each other's drafts, then read everything yourself.
- Common pitfalls to avoid:
- Too corporate
- Point gets buried
- Doesn't sound like you
- Keep cutting until it feels authentic.
Step 3: Clone Your Voice (Tricky but Worth It)
- Fish Audio created the voice from just 2 minutes of personal recording.
- The process wasn't instant – version 2 almost failed, version 3 finally worked.
- How it worked: Codex used the Fish Audio API to generate each sentence as a separate MP3 file, named by sentence ID. This allowed:
- Perfect control over pauses
- Easy editing (change one sentence, only that file rebuilds)
Step 4: Create a Storyboard (Even If You've Never Done One)
- Ask Codex to generate a sketch for each scene first, using Hyperframes for general video.
- Tips for working with AI visually:
- Say what's wrong
- Say how you want it changed
- If you don't know, ask for options
- The final color palette matched the logo – achieved by asking for multiple options and choosing one.
Step 5: Build the Video – One Segment at a Time
- Critical mistake avoided: Codex once built in vertical (portrait) instead of landscape. Build one segment, review it, give feedback, then move to the next.
- Build sequence:
- Build segment 1 → play it → give concrete, small feedback → repeat
- Add captions
- Export
- Watch the entire video from start to finish
- Final checks before calling it done:
- Captions appearing too early
- Scenes not fully cleared
- Narration endings clipped off
For a more cost-effective approach to AI video, you might also consider How to Create Unlimited Free AI Videos and Midjourney-Quality Images with Meta AI.
Key Takeaways for AI Video Beginners
- Work incrementally: One segment at a time, with immediate feedback.
- Be specific with AI: Say what's wrong, how to fix it, or ask for options.
- Check orientation before building everything.
- Watch the full export – small problems are invisible during editing.
If you're looking for a broader learning path, see A Step-by-Step Roadmap to Mastering AI: From Beginner to Confident User.
Final Result
This 3-hour process produced the first finished AI video using zero code and zero manual editing. If you'd like a deeper dive into any single step, or want to see another AI project, drop a comment below.
Like and subscribe to Jellyhop Labs for more experiments in AI-powered creation. Let's learn together.
Without any video editing background, I made my first AI video in only 3 hours with codecs. And so can you. And through
the whole thing, I didn't write a line of code or do any editing by hand. Hi, welcome to Jelly Pop Labs. You found
my very first video. Come along and I hope my mistakes save you a few of your own.
In this video, I'll walk you through five steps to make a video like this one with codecs or anything equivalent.
Let's get started. Step one, install. You need codecs or cloud code. I used both only because hit
its usage limit halfway through. Then install the skills. A skill is basically an instruction manual for one
job. a text file that tells Codeex when to use it and what to do. You don't need to open it yourself. Codeex picks the
right skill based on what you ask it to do. Skills come in packages and some
packages are huge. The social media one, for example, has 106 skills. I only installed seven. So, don't just tell
Codeex, "Install this package." Tell it what you're trying to make and have it pick the skills you actually need.
Everything I used for this video is on screen. That's the whole list. Step two, the script.
Most AI video tutorials go straight to the tools. Hey Genen, fish audio, hyperframes, all these shiny things to
play with. But halfway through my first version, it hit me. What am I even trying to say?
So honestly, don't use a skill just because it looks cool. Wait until you have something to say, then build.
I had never written a video script before, so this one went through 16 versions.
I had a few different skills review each other's drafts, then read everything myself.
Some drafts were too corporate, some buried the point, some just didn't sound like me. I cut all of that, and slowly
it became what you're hearing now. Step three, the voice. The voice you're hearing right now is a clone made by
Fish Audio from about two minutes of my own recording and it almost didn't work. By version two, I almost gave up. Then
version three came back and it sounded a lot more like me. And that thing I said half out of frustration generated
sentence by sentence turned out to be the method. Codeex used a skill called fish audio API. Every sentence became
its own MP3 named after its sentence ID. So we could control the pauses. And if I changed one sentence, only that file got
rebuilt. Nothing else changed. Step four, the storyboard. I had never made a storyboard before, so
I asked codeex with this prompt. It used general video from hyperframes and gave me a sketch for each scene
first. I didn't like the colors much, but I had no idea what I wanted either. So, I
asked Codeex to give me a few options, and I chose the palette that matches my logo.
What you're looking at now is the version after that. So, working with AI really comes down to three things. Say
what's wrong, say how you want it changed, and if you don't know, ask for options. And it's not just for video. It
works the same way for a website, an app, anything you build with AI. Step five, build. This is where the
video actually gets made. I hit one big problem in my practice version. Codeex never asked whether I
wanted landscape or portrait. It just built a 45se secondond vertical draft. I only found out when the whole thing was
done. Redo. So this time build one segment, play it, give feedback, then move on to the next
one. Keep the feedback small and concrete, then add captions, export, and watch the
whole thing from start to finish. That's where I caught the last problems. A caption showing up early, a scene that
hadn't fully cleared, one narration line with its ending clipped off. Only then is the video actually done.
So, that's the first finished video I made with these five steps. Thanks for watching all the way to the end. If
you'd like a more detailed walkthrough of any one step or there's another project you'd like me to try with AI,
tell me in the comments below. This is Jellyop Labs. If you want to see what I try next, like and subscribe.
Let's learn together. See you next time.
You’ll primarily use Codex (or the cloud version, Cloud Code) to orchestrate the process. Instead of installing large packages, let Codex choose only the necessary skills—in this project, just 7 skills were required. Avoid the social media package with 106 skills; tell Codex what you’re making, and it will install only what’s needed.
Start with your core message, then refine it through multiple drafts—here, the script went through 16 versions. Use AI skills to review each other’s drafts, but always read everything yourself to ensure authenticity. Avoid sounding too corporate, burying your point, or losing your personal voice; keep cutting until it feels genuine.
Use Fish Audio, which creates a voice clone from just 2 minutes of personal recording. Codex will automate the process via Fish Audio’s API, generating each sentence as a separate MP3 file named by sentence ID. This gives you perfect control over pauses and easy editing—if you change one sentence, only that file rebuilds.
Yes, even if you’ve never done one—ask Codex to generate a sketch for each scene using Hyperframes. For visual feedback, tell the AI exactly what’s wrong, how to change it, or ask for options if unsure. For example, to match your logo’s color palette, request multiple visual options and choose the best one.
Build one segment at a time, review it, give concrete feedback, then move to the next segment. A critical mistake to avoid is orientation—once Codex built in vertical instead of landscape. After all segments are done, add captions, export, and watch the entire video from start to finish to catch issues like early captions, uncleared scenes, or clipped narration endings.
Be specific and incremental: first say what’s wrong, then suggest how to fix it, or ask for options if you’re unsure. Work segment by segment rather than trying to fix everything at once. This approach ensures small issues are caught early and prevents the AI from making broad assumptions that don’t match your vision.
Watch the full export carefully—small problems like captions appearing too early, scenes not fully clearing, or narration endings being clipped off are invisible during editing. Also verify orientation (landscape vs. portrait) and ensure the color palette is consistent. These final checks can save you from redoing large parts of the project.
Keep this summary
Save it to LunaNotes and it becomes a real note in your library — editable, searchable, and ready to turn into flashcards or a diagram. Free to start.
Save to LunaNotesOr summarise for another video.
This summary and transcript were automatically generated using AI with the Free YouTube Transcript Summary Tool by LunaNotes.
Related summaries
Unlocking Creation: How to Make Videos with Nvidia AI for Beginners
Learn how to easily create captivating videos using Nvidia AI technology, from scripting to voiceovers, in just minutes.
How to Create Unlimited Free AI Videos and Midjourney-Quality Images with Meta AI
Discover a step-by-step method to generate long, engaging AI horror story videos and Midjourney-quality images for free using Meta AI. This guide covers scriptwriting with ChatGPT, image and video generation on Meta AI, voiceover enhancement, and seamless video editing techniques.
A Step-by-Step Roadmap to Mastering AI: From Beginner to Confident User
This video provides a comprehensive roadmap for anyone looking to start their AI journey, emphasizing the importance of understanding core concepts before diving into tools. It offers practical tips on building an AI learning system, developing critical thinking skills, and strategically selecting AI tools to enhance productivity.
Free AI Method: Turn YouTube Lectures into Handwritten Notes Fast
Struggling to focus on long YouTube lectures? This tutorial reveals a completely free, step-by-step AI-powered method to convert any lengthy video into structured handwritten notes. Learn how to use transcript extraction tools, Claude AI, and Google Gemini to create beautiful, printable notes for faster revision.
Build AI Agents with Microsoft Copilot Studio – No Code Guide
Learn how to create custom AI agents using Microsoft Copilot Studio without writing any code. This step-by-step tutorial covers building a customer service bot, connecting it to company documents, automating email responses, and publishing it for your team.
Most viewed summaries
A Comprehensive Guide to Using Stable Diffusion Forge UI
Explore the Stable Diffusion Forge UI, customizable settings, models, and more to enhance your image generation experience.
Kolonyalismo at Imperyalismo: Ang Kasaysayan ng Pagsakop sa Pilipinas
Tuklasin ang kasaysayan ng kolonyalismo at imperyalismo sa Pilipinas sa pamamagitan ni Ferdinand Magellan.
Mastering Inpainting with Stable Diffusion: Fix Mistakes and Enhance Your Images
Learn to fix mistakes and enhance images with Stable Diffusion's inpainting features effectively.
Pamamaraan at Patakarang Kolonyal ng mga Espanyol sa Pilipinas
Tuklasin ang mga pamamaraan at patakaran ng mga Espanyol sa Pilipinas, at ang epekto nito sa mga Pilipino.
How to Install and Configure Forge: A New Stable Diffusion Web UI
Learn to install and configure the new Forge web UI for Stable Diffusion, with tips on models and settings.
Found this summary useful?
Take it with you. One click puts it in your own LunaNotes library.
Save to LunaNotes