機械翻訳 / Machine-translated
@aifriends
AI Friends（https://aifriends.jp）のクロスポスト公式アカウント。AIツールの紹介・使い方・できることを、中学生でもわかるやさしい日本語で届けます。
What you'll learn from this article
I started using Veo in April 2026, right around the time it became free through Google Vids. Up until then, Sora had been my go-to AI video generator, but when it was announced in March 2026 that Sora would be discontinued, I found myself looking for the next alternative.
The deciding factor was the ability to generate audio alongside the video. With Sora, I had to create the footage first and then separately hunt for music, but with Veo, a single prompt — the instruction text you give to the AI — automatically generates background music, sound effects, and even conversation lip-sync (synchronizing mouth movements with speech). As someone who frequently makes vertical videos for social media, this level of efficiency was a real selling point.
My honest impression after the first week was: "This is easier to operate than I expected." All you do is enter a prompt in the Google AI Studio interface, and a few minutes later you have a finished video with audio. That said, I was surprised to find that the initial video generated was only about 8 seconds long — my reaction was, "That's it?"
However, I soon learned that the scene extension feature (which lets you lengthen a video incrementally) can stretch clips up to a maximum of 148 seconds, and that made sense of the format. Because it works by chaining together multiple short clips, it felt a little tedious at first, but once I got used to it, I actually found it convenient for making fine adjustments like "I only want to redo this one part."
I was also pleasantly surprised that even the free version lets you choose 720p or 1080p resolution and even use 4K upscaling (a feature that enhances image quality).
#1: The ease of generating audio together with the video
Dialogue, background music, ambient sounds like footsteps or rain — just write them in your prompt and they're generated automatically. I make short videos for Instagram fairly often, and not having to separately search for audio was a genuine relief. The lip-sync is also quite natural, making it easy to create videos where characters appear to be speaking.
#2: Support for vertical video (9:16)
Being able to create portrait-oriented videos for social media posts directly is a quietly useful feature. It saves time by eliminating the need to shoot in landscape (16:9) and then crop. It's especially great for anyone who wants to pump out content for TikTok or YouTube Shorts.
#3: Free to use starting April 2026
Anyone can access the Veo 3.1 model for free through Google Vids, a Google service. There are also paid tiers — Veo 3.1 Fast and Veo 3.1 Lite — but being able to try it out for free first lowers the barrier considerably, which I think is a big plus for beginners.
#1: Each generation only produces 8 seconds of footage
Because the workflow involves repeatedly extending short clips and stitching them together, making even a 30-second video requires at least four separate generations. Each one takes several minutes, so when I was in a hurry it felt stressful. Sora could produce 60 seconds in a single generation, so Veo clearly falls short in this regard.
#2: Writing effective prompts takes practice
Simply typing "a cat running" won't produce the video you have in mind. You need to be specific — something like "a calico cat running happily in a park at dusk, camera tracking from the side, upbeat piano background music." Until I got the hang of it, I found myself regenerating videos over and over, which ate up a lot of time.
#3: Character consistency is somewhat lacking
Even though this was improved in the January 2026 update, as you keep extending scenes, the same character's face or clothing can shift slightly. This became noticeable when I tried to create videos with a continuous storyline.
Compared to OpenAI Sora, which I used previously, my impression was that Veo feels geared toward professional video production, while Sora felt more suited to personal hobby use. Sora was simple — you could get a 60-second video in one go — but audio had to be added separately. Veo, on the other hand, supports comprehensive production including audio, but requires more detailed configuration and extension work.
Veo also comes in three versions — Veo 3.1, Veo 3.1 Fast, and Veo 3.1 Lite — which you can choose based on your needs. It breaks down roughly like this: 3.1 if you want the highest quality, 3.1 Fast if you want a balance of speed and quality, and 3.1 Lite if you want to generate large volumes quickly. I found this flexibility to be a strength that Sora didn't have.
Recommended for:
Not recommended for:
Personally, I felt that the ability to generate audio-inclusive videos for free alone makes it worth using. That said, it's not the best fit for those who want to produce longer videos in one shot. I'd suggest starting with the free version to see whether it matches your particular needs.
This article is a cross-post from AI Friends.
