Google's AI 'Gemini' Android app and browser version finally support video uploading
You can now directly upload video files to Gemini. This is available to free users, and you can use it to have AI analyze your videos.
Gemini app rolling out video upload and analysis
Google Gemini can now see, hear, and understand your video clips
https://www.androidpolice.com/google-gemini-can-now-see-hear-understand-video-clips/
To upload a video, open the Gemini app or browser version and select the + icon. This time, I used the Android app version of Gemini (version 1.0.772404132). The model used was Gemini 2.5 Flash.
Next, select 'File.'
Select a video file in your device. This time, we selected a webm file. The playback time is 34 seconds and the file size is about 1MB.
The following videos were selected:
Then, enter the prompt and click the submit button. In this case, I entered 'Analyze this video.'
The analysis results were returned within seconds.
The analysis results are as follows:
The video provided shows two main objects: one is a bright yellow cube, and the other is a cube with a red brick-like texture. Both cubes have round eye- and mouth-like shapes, and the yellow cube in particular makes mouth movements as if it is eating something.
The main contents of the video are as follows:
* 0:00-0:04: Yellow and red brick cubes are shown side by side, with cubes of various colors stacked on top of each other in the background: green, blue, pink, orange.
* 0:04-0:10: A person's hand brings a coin close to a yellow cube. The mouth of the yellow cube opens and closes, as if it is 'eating' the coin. Once the coin is in the mouth, the mouth closes.
* 0:11-0:17: Now a coin is brought closer to the red brick cube. Similarly, the mouth opens and closes, as if to 'eat' the coin.
* 0:18-0:33: The coin is again brought close to the yellow cube and the action of 'eating' the coin with the mouth is repeated.
Overall, the video shows a piggy bank-like device with facial features 'eating' coins, with each cube's facial expressions and movements appearing lifelike.
Next, we analyzed the following video, which has a runtime of 2 minutes 35 seconds and a file size of 94MB.
I was able to successfully upload the video file and send the prompt, but they did not analyze it, instead replying, 'We do not currently have the ability to directly analyze video content.'
On the other hand, when I tried it on the web version, it returned the analysis results as follows:
This video was taken at a press test ride of Fuji-Q Highland's roller coaster 'Do-Dodonpa.'
At the beginning of the video, staff members welcome the passengers and send them off with applause. Then, the video shows the roller coaster entering a dark tunnel and accelerating. There are lighting effects inside the tunnel, and the light gets stronger as the ride approaches the exit.
Once you leave the tunnel, you can see the vast grounds of Fuji-Q Highland, the surrounding mountains, and other attractions. The roller coaster travels at breakneck speed, and the impressive images continue as it goes through loops and sharp curves. The part where it rises vertically into the sky and seems to stop for a moment is particularly thrilling.
At the end of the journey, the train re-enters the tunnel, slowing down and returning to the starting point. The passengers show expressions of relief and excitement as staff greet them with applause.
This video is highly immersive, allowing you to experience Do-Dodonpa's incredible acceleration and thrilling course from a first-person perspective.
The video upload feature seems to be gradually being rolled out, and technology media 9to5google reports that 'there were differences in whether it was available or not depending on the account and device checked.' There were some devices that could be used and some that could not be used at the GIGAZINE editorial department.
Google has just released Gemini 2.5 Pro and Gemini 2.5 Flash to the public.
Google releases Gemini 2.5 Pro/Flash to the public & newly releases cheap and fastest 'Gemini 2.5 Flash-Lite' preview version - GIGAZINE
Related Posts:
in Review, Software, Web Service, Video, Posted by log1p_kr