"What is hidden behind this picture?
Nowadays, making Short videos and using AI to generate videos from pictures is becoming more and more popular. You can take a picture and let AI turn it into a moving video. This technique is now called "image-driven video generation."
You may ask, how is this different from previous AI videos? Previous technologies had to be described in words or given a pile of material. Now using pictures directly, AI can automatically analyze the content of the picture and generate dynamic effects by itself. For example, for a landscape photo, AI can add the effect of floating clouds and swaying leaves.
How did this technology come from? The earliest is 2023, some major manufacturers will start research. The results were still very rough at the time. But by 2024, the model has been upgraded. Now, as long as the picture is clear, AI can identify the subject, background, and light, and then add reasonable dynamics.
How to operate it specifically? For example, you see the "stop-motion animation" effect on TikTok. In fact, many of them are generated using AI. When you upload a photo of a person, AI will automatically ask the person to do some small actions. Such as blinking or turning your head. These details used to have to be done manually, but now AI does it itself.
What should I pay attention to? First, picture quality is important. Fuzzy graph AI is difficult to deal with. Second, complex backgrounds can generate failure. For example, when a person is standing in a crowd, the AI will not know which part to move. Third, the effect is highly random. If the same picture is generated twice, the results may be completely different.
What tools are currently on the market? For example, Runway's Gen-2 and Pika Labs. They all support uploading pictures to generate videos. However, the free amount is limited. If you want to make a lot of videos, you have to consider a payment plan.
In terms of cost, using AI to generate a 15-second video is now about a few yuan. It's much cheaper than hiring someone to edit it. But whether the effect is good or not depends on the picture itself. Complex scenarios may need to be retried several times.
Many people are worried that this technology will put editors out of work. Actually not. AI is just an auxiliary tool. For example, if you use AI to generate the first version and then manually adjust the details, the efficiency will be the highest. Relying entirely on AI, videos can easily become the same.
Future trends? It is estimated that AI will become smarter and smarter. Maybe you can upload a picture in the future to generate a 3D effect. Now some tools can add a simple sense of depth. For example, let the foreground move faster than the background to simulate a real shot.
In short, if you want to use AI to make a video from pictures, first find a good picture. Then try those popular tools. Try a few more times and you will find the pattern., AI is not perfect yet, but it can help you save a lot of time. "
