_MacTwin_
Offline
•
Upload & Sell: Off
|
 p.30 #14 · Sony 400-800 f/6.3-8 G It's here!! | |
octo wrote:
Seems like you don't understand at all how it works so I have to give more explanations.
Phase IS NOT based on contrast, it's the light phase, basically if light is a waveform, it tells you how many waves it has.
The PDAF is splitting the image into 2 images: left and right.
By comparing left phase vs right phase for an equivalent pixel, you can tell the distance of that pixel (or how much the lens needs to move to get that pixel into focus).
The only reason that some form of contrast is needed is because the camera needs to be able to map all the pixels in the left image with all the pixels in the right image, and it does that by finding edges and mapping them, this creates "pixel pairs".
However, even if some pixels cannot be mapped (ex: no edge was found) then an intelligent system like the one in the A1 can still find the missing pixel pairs, because it can use surrounding pixel pairs to guess the missing ones.
As I wrote above, even a very small difference is enough for the camera to be able to do the mapping.
Also when the sensor as 759 PDAF points, in fact those are logical points, physical points are much greater in number, like 1% of the pixels would be PDAF points meaning 500k, meaning 660 physical points for every logical point, this allows to deal with noise and have even more reliable measurements, and also accelerates the camera processing by having fewer points.
The "blur" is not being analyzed at all, what are you talking about? This is not "Depth from Defocus", which is a form of contrast detection autofocus (CDAF).
If you need more explanations, I can give, but I don't want to turn this topic into a "PDAF" topic. Maybe PM is possible then.
Edit: I want to specify that the PDAF gives DIRECTLY the distance. There is no "iteration" process like what you're alluding to. Phase doesn't just give the direction but also the exact amount for the lens. The iterative process is used by CDAF. So it looks like you think PDAF is just CDAF but with a direction. It's not, it gives the final answer directly....Show more →
First, the edge is possibile if there is contrast, if you don’t have contrast you cant see any edge.
The waweform that is read by the pdfaf is base on difference between 2 adjacent areas, based on contrast, you probably miss the simple definition of contrast, the receptors read “lines” (in fact they read horizontal or vertical lines…or cross-shaped), if the surface has no contrast points the PDAF system (that works like on receptors used in the submirror reflex af) does not know where to go at all. The operation is similar to that of a leica rangefinder, if you do not have contrast areas the rangefinder cannot give the vision of the phase shift, it is not a laser rangefinder, contrast is the basis of what you continue to “try” to explain, trivially without contrast there would be no waveforms, the reading would be a flat line, and the two lines used by the phase receptor would receive the same information.
Receptors can determine position, not distance (that is calculated by the camera…,not from the sensor, only after finding the focus, and with information provided by the optics used), and without information provided by the optics used….only a laser or ultrasoud rangefinder can determine distance…
Anyway I don't feel like arguing, I'm sorry there's some misinformation. Good luck.
|
I hope to snap some shot of him soon, before he sheds his antlers. All in all... a nice morning out. 
