I’m a heavy AI creator who uses AI tools throughout the entire video production process. Before I knew it, I’ve been making AI videos for 1,898 hours. Technical skills develop from the time invested, so naturally, creating videos has become smoother, but there are still some daily routines.
A long time ago, back in the Stable Diffusion 1.5 era, I had a phobia of bad hands. Every time I generated an image, the first thing I’d do was count how many fingers there were. Usually, the entire hand would be distorted, and someone even made a horror film based on this. I’d put the bad hands into Photoshop AI for repair, but sometimes it would create even worse hands. Fortunately, my phobia of bad hands has been completely cured by Flux.
Then came the card drawing – when generating images, Midjourney requires you to “draw cards” first. However, with Midjourney, using the same prompt and drawing multiple times can exhaust its creativity. Therefore, if you continue drawing and remain unsatisfied, you will still need to modify the prompt. After that comes video card drawing, watching credits get consumed with each click. Some tools are becoming increasingly expensive – it still hurts the wallet.
Some tools, like Conch AI, offer unlimited draws. If you buy unlimited draws for a month, you’re in a tense state the entire month, always thinking about drawing as much as possible. Every time you click for more than two draws, you’re afraid of wasting even a little. After rushing through one project, you must immediately start the next one. Sometimes the pressure really gets overwhelming.
In the Suno 3 era, I would finish all 500 songs I paid for each month. It had many flaws back then and couldn’t compare to the current 4.5 version, but just to use them all up… Now, many AI tools clear credit after each month, which adds invisible pressure to produce as much content as possible.
When working, I usually sit in front of the computer for an entire day and then produce a video that lasts 30 seconds to 1 minute. Long working hours are indeed the norm for all workers in the art and design field. So sometimes I’ll open chat groups to talk a bit, to relieve the fatigue from such long working sessions.
While working, I’ll have more than 10 browser tabs open, because current AI tools don’t have a single tool that can handle the entire production process; it’s a combination of multiple tools. So you need the ability to integrate and understand, constantly sensing what one tool excels at and what another tool can do.
Generating images and videos across multiple tools results in videos with some style inconsistency. Here I want to emphasize again the importance of color grading. If you want a consistent style in your work, then color grade the images once and color grade the videos once. Image color grading is relatively easier.
After finishing a day’s work, I still have to turn on the TV and start watching movies or TV shows. Because the volume of content consumption is a long-term accumulation for creating good works, without a certain amount of viewing, there’s no basis for aesthetic judgment. Finally lying on the sofa, striking Ge You’s lying pose, watching the bad films that frequently appear in the current market…
Finally, it’s time to sleep, but before bed, I still need to browse Bilibili and YouTube to catch up on the latest technical developments. If I discover a completely new tool, I immediately bounce back up from bed, return to the computer, and start learning the new tool. I know some people think technology isn’t important for AI film and video – that having an aesthetic sense is enough. But for me, it’s not like that.
All my videos are created and edited by me alone. I don’t have others helping me discover new technologies every day, so I need to do so myself. Many new technologies, once they appear, completely revolutionize previous methods. For example, current multi-reference completely transforms the image generation process.
After learning new technology, I can finally lie down, and then the next morning, when I wake up, I have to write for my WeChat public account. Now, with platform traffic restrictions, only the WeChat ecosystem is relatively open. Therefore, I need to focus my efforts on public accounts and video accounts. I usually write a public account article either first thing in the morning or when I’m very tired from work. Many AI creators write public accounts, so this is also a necessary step.
After writing for the public account, I rest for 1-2 hours, then start writing novels. Writing long novels is very mentally exhausting because you not only have to immerse yourself in everything the characters experience, but also maintain overall logical coherence and consistency. By the way, yesterday I had AI analyze my novel, and the style tags it generated were: rational and restrained, as well as epic and weighty science fiction narrative style.
After finishing the novel, I started making AI films, and the daily cycle began again. Is this tiring? Yes. But how much time and how many opportunities does a person have to do the creative work they want to do? Some people despise me for not following market trends, but do they really understand the market? Catering to the market doesn’t guarantee the market will love you. So if there’s an opportunity, you must definitely do your own creative work – a unique style might very well be what the market wants.
Follow me, and let’s explore AI filmmaking together.
