NEX 6 fails video test from Sonyrumors

Started Nov 14, 2012 | Discussions thread
BasiliskPhoto Senior Member • Posts: 1,008
Re: What "test" has been done?

John Koch wrote:

My guess is that the ideal solution, from the vantage of video, would be to use sensors with exactly 1920x1080 pixels, which would be "perfect" resolution for video, and avoid the need for messy ditigal downsampling, but too low for stills.

Actually downsampling is good. Look at unprocessed footage from the Canon C300 which has a 4K sensor, downsampled to HD, and each frame looks like a photographic still. Compare with a frame from the Sony EX1 - which was considered crisp in its day, and at the pixel level you will see the difference. It just is not in the same league of sharpness. Your best chance of reducing moire and aliasing, while retaining sharpness, is oversampling, and using a high quality resizing algorithm (like Photoshop bicubic for example). Unfortunately the best resizing algorithms are processor intensive, and if you have to do them 30 times a second, even more so. That is presumably why the C300 only works at 720p at 60 fps. If they could get the same quality at 1080p, surely they would.

So the resizing algorithm used in the NEX 5N looks like it is a bit cheap and cheerful - fine for soft images, a bit of a giveaway on parallel diagonal stripes, and to be honest, like most DSLR footage, closer to 720p footage in real definition. Sony are not going dedicate a superfast chip to resizing video footage in a camera at this size and price. But this camera smaller than my hand can also do pretty good slow motion, and costs about 1/20th of the price of the C300!

Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow