[Emacs]

I think the AF system will need work, it will have to match what Canon had in the 1D MKIIN camera at the very least, though asking Pentax to match what Canon and Nikon have going at the moment might be a bit too much to ask.

It's pointless. The PDAF is still alive only because of its relative speed. The accuracy and functionality of this method is nowhere near good and thus it will be abandoned in few years, just after better software AF approaches which are already available now (these are not CDAF) will be developed further. And it will mean the death of SLR too.

[Kunzite]

"Software AF"? How would that work? No software can "fix" a fully out of focus image.

[Clavius]

It's pointless. The PDAF is still alive only because of its relative speed. The accuracy and functionality of this method is nowhere near good and thus it will be abandoned in few years, just after better software AF approaches which are already available now (these are not CDAF) will be developed further. And it will mean the death of SLR too.

Seriously: Why do you come here anyway?

[Kunzite]

To predict the death of the SLR, over and over again - or so it seems.

[Zygonyx]

It's pointless. The PDAF is still alive only because of its relative speed. The accuracy and functionality of this method is nowhere near good and thus it will be abandoned in few years, just after better software AF approaches which are already available now (these are not CDAF) will be developed further. And it will mean the death of SLR too.

Would you be named Nostradamus in the real life ?

Seriously, i dont under-estimate the limitations of PDAF, but i don't know either many "intelligent" technological devices that have disappeared from the photography world since more than 100 years.
The photography market is everything but a "standard products only" place.

[Clavius]

"Software AF"? How would that work? No software can "fix" a fully out of focus image.

It's like the software shake-correction some point-and-shoots have. (had?) And imho, just as useless.

[Kunzite]

Clavius, contrary to Pixel Track (which does have some kind of effect, and it can be improved) a "software AF" just cannot work. I know there are algorithms which can "correct" small amounts of defocus, but what can they do when the subject is completely out of focus, and just a coloured irregular spot?

[Emacs]

"Software AF"? How would that work? No software can "fix" a fully out of focus image.

I have given a link several times already. http://jburge.cps.utexas.edu/research/Home_files/BurgeGeisler_PNAS_2011.pdf
In short, they are trying to mimic human eye focusing system, which is superior to any PDAF (blazing fast, accurate in even very low light conditions, isn't bind strictly to the specific areas, superior tracking capabilities).
This article is very basic now, the method needs a lot of adjustments, but I already can see how good it will be. First, I wonder if it will be useful to move consideration from original image to its laplace pyramid decomposition. I believe this will help a lot. Anyway, I'll research this myself too. Don't know if I will succeed, but it will be a fun

PS I only want a death of SLRs. It's time to get rid of them and their limitations, such a huge flange distance, bulk, etc. The only SLR advantage today is the speed of PDAF. The MF is already easier with EVF. With better AF pros will move to EVILs and thus SLRs will die.

[Kunzite]

From what I see, that article is only talking about defocus estimation. So you'll still have your blurry image, but you'll know how blurry it is
The human eye's "AF" is both software and hardware; yes, it is actually focusing.

I'll ask you again: who the h*ll are you to decide what I should use? Find a Sony forum (since that's what you're using now) and go there, instead of nagging us.

[Clavius]

From what I see, that article is only talking about defocus estimation. So you'll still have your blurry image, but you'll know how blurry it is
The human eye's "AF" is both software and hardware; yes, it is actually focusing.

I'll ask you again: who the h*ll are you to decide what I should use? Find a Sony forum (since that's what you're using now) and go there, instead of nagging us.

Agreed, apart from that: The human eye is a very poor optical device. E.g. The brain fills in all the gaps in the images the eyes produce. It's only sharp in the very center and has extremely low boarder sharpness. Mimicing the human eye would be a very stupid thing to do.

About Emacs: I've never ever seen any positive note from him. Nor have I ever seen him actually post a photo in the gallery. Hé's just here to troll.

[Kunzite]

I'd say the camera is just like the human eye, but with much worse processing and better optics

[Emacs]

defocus estimation.

PDAF is performing defocus estimation too, that's why it's so fast
Unfortunatelly, it has too many shortcomings such as it needs precise calibration and doesn't work well in low light.

The brain fills in all the gaps in the images the eyes produce. It's only sharp in the very center and has extremely low boarder sharpness. Mimicing the human eye would be a very stupid thing to do.

It's the most obvious way now. CDAF proved to be … not to be a good method LOL
It's not even mimicing but rather approach based on learning. Mimicing because human infants can't focus as well at the very begining.

[Emacs]

About Emacs: I've never ever seen any positive note from him. Nor have I ever seen him actually post a photo in the gallery. Hé's just here to troll.

I didn't and won't post anything into gallery. Put some images in lens club topics and lens samples archive though (Shitty ones at Distagon 21, 15, 31, 43, 77, A* 85, shitty ones with FA* 85, A50/1.2).

[Kunzite]

So, you don't even realize that defocus estimation is just the first step, that you actually need to mechanically set the lens to the proper distance?

[Emacs]

So, you don't even realize that defocus estimation is just the first step, that you actually need to mechanically set the lens to the proper distance?

Estimation is the hardest task.