Original video in Chinese.
Key Takeaway
- The three most important elements in video creation are: lighting, microphone, and only then the camera and lens.
- Good lighting and setup can significantly improve video quality; even a phone can produce great results.
- Sound quality is crucial to the viewer experience and should not be ignored.
- I share my experience upgrading from the Sony A6400 to the FX30, emphasizing the FX30’s advantages in image quality, stabilization, and heat management.
- The video creation workflow includes topic selection and planning, script writing, recording, editing, subtitles, cover design, and uploading.
- AI can help in content creation, such as Claude providing inspiration for topic planning and script writing, and DaVinci being used for automatic subtitle generation and masking.
- I believe text-to-video currently has limited value, and AI should be integrated into existing production workflows in an auxiliary role.
- In the AI era, narrative ability and coding ability have become important levers for ordinary people to change their lives.
Full Content
I’d like to share the equipment and workflow for my video content creation, as well as my views on AI in content creation.
I started making videos in 2018. Back then I was still a Bilibili gaming-category creator, and my content was mainly centered on Nintendo games. If you’re interested in digging through the archives, you can search for this account. Looking back now, I can barely bear to look at it—it feels so awkward.
If you compare videos from a few years ago, you’ll notice that from the production quality to how I speak in front of the camera, there’s a huge difference—this is a process of continuous accumulation and improvement.
My gear was purchased in several separate rounds; I didn’t buy everything at once. My shooting and editing have also been learning and iterating continuously. I watched a huge number of YouTube videos, and only this year did I finally optimize a workflow that feels really comfortable to me.
So, the most important thing in video creation is: start doing it! All problems are solved through practice. I hope this episode can give everyone some encouragement and inspiration.
Let’s start with the equipment.
I believe 99% of beginners think that when filming videos, the camera and lens are the most important. But I want to tell you that for video content, the three most important things are: first, lighting; second, microphone; and only third, camera and lens.
Take my current setup as an example. I’m using two lights.
The first light is the Aputure Amaran F21x, which I bought for a little over 1,900 RMB. This is a key light; it’s compact, foldable, and especially space-saving, so it’s very suitable for a setup like mine where the desk is backed up against the wall. Its output is sufficient—I only turn it on a little and it’s already very bright.
I place it at about a 30 to 45 degree angle, shining down from above and to the side, so the other side of my face has some shadow and the overall look is more three-dimensional, not so flat.
The second light is the Elgato Key Light Air, which I bought for about 1,000 RMB. I bought it a few years ago for game streaming, and now I place it overhead as a rim light.
Let me show you a comparison. If I turn it off, it looks like this. Turn it back on, and you can see that my hair and the whole outline are lit up. That way, the person and the background are separated.
Besides these two lights, I actually have another one: the Aputure 120d. I bought it in 2021 for over 5,000 RMB. When I need to film from the angle with my back to the desk, I use it.
The 120d is extremely powerful, and the results are very good. But once you add the softbox, it also becomes very large. To make deployment easier, I mounted it on three wheels. When I’m not using it, I can just push it to the side. I don’t take the softbox apart, otherwise filming every time would be too much trouble.
With good lighting and proper setup, even with an iPhone you can get a very decent result. So lighting absolutely has to come first, and everyone should spend their budget there first.
The second priority is the microphone. Many beginners overlook the value of sound. If a 10-minute video has mediocre sound quality, it’s really hard to keep viewers listening until the end.
I have three audio recording setups. The one I’ve used the most over the years is the RØDE VideoMic NTG. I bought it for a little over 1,700 RMB. As a shotgun microphone, I’m very satisfied with its sound quality. Most of the videos on my channel were recorded with it.
When I’m operating the computer while recording at the same time, I use a PreSonus dynamic USB microphone. I bought it years ago specifically for streaming games. Compared with a shotgun mic, its sound is fuller and warmer, so it’s very suitable for real-time operating scenarios, whether it’s live-streaming games or recording a tutorial video.
To unlock more filming scenarios, I also bought the RØDE Wireless Pro, which I just received. It’s currently the best wireless microphone. I bought it for two reasons: 32-bit float and onboard recording. With it, the NTG shotgun mic in front becomes an assistant, to be used with the second camera.
OK, that’s lighting and microphones done. Finally, the camera and lens.
I bought my first camera in 2019, the Sony A6400, for 7,000 RMB. This series is a very classic one; it’s fine for both photography and video. Over the past few years, I probably recorded more than a hundred videos with it. It wasn’t until this year that I bought my second camera, also Sony APS-C, the FX30, which is the one currently recording this video.
The reason I chose APS-C rather than full-frame was, first, size and weight: full-frame bodies and lenses are simply too large and too heavy; second, lenses: APS-C lenses are very abundant; third, price: the difference is huge. So after weighing everything, I still chose APS-C. And in this regard, the FX30 is top-tier.
First, in terms of image quality, the FX30 supports 10-bit color recording. Compared with the A6400’s 8-bit, it has more delicate color transitions and a wider color range. The improvement in video quality is very significant.
Second, it has in-body stabilization. What I disliked most about the A6400 was that it had no stabilization. So if I wanted to shoot handheld, I could only choose a stabilized lens, such as Tamron’s 17-70. Now with the FX30, I can turn on enhanced stabilization, and pair it with a wide-angle lens. Even with the crop, it’s completely sufficient for filming self-shot vlogs.
Finally, heat management. The FX30 has a built-in fan, so I don’t have to worry about overheating and shutting down during long outdoor recordings or live streams. Cameras released in the past two years, like the ZV-E1 and the 6700, don’t have built-in fans, so for long recordings you still have to attach an external fan, which is really ridiculous.
As for lenses, over the years I’ve been using two Sigma lenses: the 16mm F1.4 and the 30mm F1.4. The one currently recording is the 16mm.
These two lenses are recognized worldwide as excellent lenses. If you go on YouTube and watch recommendations from major creators, they’ll definitely mention them. The extremely large aperture creates background blur, which is very suitable for my talking-head content. The framing distance of 16mm and 30mm is also just right.
If I were to buy another lens, I’d choose between the Sigma 10-18 F2.8 and the Sony 11mm F1.8. I still haven’t decided whether I want a zoom lens. I guess I’ll probably still choose a prime lens in the end, after all, the aperture is larger, and paired with the FX30, the stabilization effect is better.
OK, that’s the main gear I use for recording videos. The scene you’re seeing right now is the one I use most often for filming. I bought a camera stand and mounted both the FX30 and the teleprompter on it. The FX30 is powered by a dummy battery, which completely solves the battery life problem. The teleprompter is from Elgato, connected to the PC via USB, so I can operate it directly on the PC—it’s super convenient. For monitoring the image, I use an iPad with Monitor+, which can connect wirelessly to the FX30, so I don’t need to buy a monitor, saving me more than 1,000 RMB. And the two lights mentioned earlier are fixed to the wall with Smallrig mounts.
The value of this whole setup is that it minimizes the preparation work before recording.
When I want to record, I just sit down, do a few simple operations, and I can start. By contrast, my other setup—the one at the beginning of the video, where I’m facing away from the desk—still requires setting up the lights and camera, which is a bit more troublesome, so I’ve gradually become too lazy to use it.
Of course, recording is only one part of the whole process. Looking forward, I first need to plan topics and write scripts; looking backward, I need to edit, add subtitles, make cover images, and upload the video to various platforms. In this whole workflow, AI can indeed help, but not that much.
First, in the topic-planning and script-writing stage, I use Claude. Give it an initial idea, and it can inspire me, as well as help me think of supporting examples.
I should have recommended this product in the community many times already. Claude is currently the only Chatbot product that can truly help me in actual production. Its Artifacts feature is very, very good. By comparison, ChatGPT falls quite a bit short—it only says nice-sounding empty things, and I’ve already unsubscribed. If GPT-5 doesn’t come out soon, OpenAI is going to be in a bit of danger.
Second, in the later stages, I now do everything from editing to color grading in DaVinci. I strongly recommend that whether you’re using Final Cut Pro or Adobe Premiere, you switch to DaVinci as soon as possible. For individual creators like us, DaVinci is the ultimate choice.
There are two AI features I use. One is automatic subtitle generation. Before, I would export an audio file and send it to Jianying or some online platform to generate a subtitle file, and then import it back into DaVinci. Now it generates directly, which is much more convenient. The second is automatic masking, namely Magic Mask. This feature is really powerful. You just roughly outline it, and the software will cut out the person, after which you can adjust the color separately or add those overlay effects.
I think this is the more practical approach right now: letting AI join existing video production processes in this way. As for text-to-video, the idea is beautiful, but if it can’t pass for the real thing, its value is very limited. I really can’t think of any use for the bizarre AI-generated videos we’re seeing now.
If you’re a giant like ByteDance, you can use platform-level ambition to plan and develop a disruptive new mode for the AI video track; if you’re just a small nobody, you should find a niche and get rooted there first. For example, intelligent video editing, A Roll and B Roll in different styles—that’s a real need. Then if you can integrate with major software like DaVinci and survive through a subscription model, that has a lot of promise.
Finally, one more thing: why do I care so much about video creation ability?
Over the past twenty years, there have been two levers that everyone is familiar with, but only a few people have mastered: one is labor leverage, and the other is capital leverage. Labor and capital propelled the great development of the last era.
By today, especially in the rise of AI, things have changed somewhat—narrative and code have become the most powerful levers. Look around: among the big names still active on the front lines today, which one doesn’t have narrative and coding ability, and hold these two levers in their hands?
Conveniently, ordinary people can also master these two levers through their own efforts alone. Not to make the claim too grand, but at least they can change the course of life. So I’m willing to keep honing these two abilities year after year. After this, I’ll make a community-exclusive video to talk about it in more detail.
OK, that’s it for this episode. See you next time!