On 2022-11-25 07:46, Dainiushas wrote:
galiu tik pasakyt, kad 10x kamerą naudoju labai dažnai.
Kažkas įdomaus pas kaimynus vyksta? :] Jei rimtai, tai čia matyt nuo asmeninio fotografavimo stiliaus priklauso. Aš pvz. nei ant DSLR dažnai maudavau televyką, nei su mobilniku telephoto naudoju. Gal kokį 1% nuotraukų padarau su 4x kamera. O, va, wide ir ultrawide naudoju pastoviai.
Bet, kaip sakoma, geriau telephoto turėt dėl viso pikto, nei neturėt iš viso, tai todėl ir PRO versija kišeniuje ;)
p.s. guglo gcam AI sugeba panaudot duomenis iš visų kamerų:
https://www.knowyourmobile.com/news/google-pixel-7-pro-camera-specs/
The Pixel 7 Pro has three cameras in its main module, including the new 48MP telephoto lens. With respect to its overall zooming capabilities, the Pixel 7 Pro recruits all of its lens to create a near-continuous zoom range from 0.5x to 30x and the image is then “treated” by Google’s new Super Res Zoom algorithm.
What is this Super Res Zoom algorithm? It’s essentially yet more super-smart camera software that works by capturing multiple frames and, through the magic of machine learning, composites them to create a fuller, more detailed image free from noise and/or other artefacts that are usually present when zoomed in on a subject.
Going deeper in the Pixel 7 Pro’s zooming capabilities, with 2x zoomed shots, the Pixel 7 Pro crops into 50MP primary sensor to create a 12.5MP image which is then enhanced by Google’s HDR+ algorithm as well as re-mosiaced. Beyond 2x, the Pixel 7 Pro uses ML fusion which combines data from its main sensor and telephoto lens for optimal photo quality.
Up to 5x, the Pixel 7 Pro uses optical zoom and pixel binning to improve both its performance in low light conditions as well as the image’s overall quality and dynamic range. Beyond 5x, the Pixel 7 Pro switches back to software-lead zoom which is assisted by Google’s machine learning algorithms, and this process remains the same right up to 10x zoom.
When shooting between 20x and 30x zoom, the Pixel 7 Pro leverages Google’s Super Res Zoom algorithm and combines this with the Tensor G2 CPU’s new ML upscaling component create and build supremely detailed images free of noise and the usual mess you experience when zoomed in at the 20-30x range.