And I took the opportunity to experiment with video processing. This is the result of almost the best quality I can currently produce for social media platforms. Here's the method:
Recent Pixel phones have a feature called Video Boost. If you enable it and record in 4K the video processing is performed for you in Google's cloud, using Google's AI.
Your phone stores a 1080p preview while you're waiting for the AI version to process. Apparently the processing time varies but I had to wait TEN HOURS for Google to perform the processing and make this one minute-long clip available. What struck me though is the file sizes:
163 MB - 1080p preview
2.0 GB - Video Boost file as recorded (4K 10-bit HDR)
357 MB - Google AI product in 4K
209.5MB - This video (Google AI version downsampled to 1080p)
The advantage of this approach is -at least nominally- greater clarity of the final output from an improved source.
There is one step higher I could go beyond this, by using the AI to upsample the 4K to 8K for processing. I don't think it's worth it for a what seems to me marginal quality improvement and the time consumption of processing delays.
Simon's verdict: great to be improving the technology but at this point in the development cycle it's more a gimmick than a useful utility.
Share this post