[WONTFIX] Implementing autofocusing algorithm from Mark III in Mark II

[Harsher]

Hello
I discovered ML not so long ago, and it is great, and after the big 7D upgrade,
I began to wonder about something.
I know that I will be talking about something that might not be possible given the very deep difference
between the mark III and II at digic level,
but I think that there is no fundamental obstacle to improve autofocusing in markII
at software level by porting the algorithm used in the markIII.
Indeed, this is not so simple since we have a different disposition of autofocusing points
and things like that which could be important when different points are used for this instead
of one, but I think that we have a real chance to improve mark II there,
even if indeed, hardware is still a limiting factor, if we can improve quickness,
or precision (and not wasted shots) per even 20%, it would be great.
In the same line of idea, I am wondering about aperture sensitivity of autofocusing points,
is it a hardware or software matter?
Because one of the biggest problem of autofocusing precision in Mark II is
that when working at bigger aperture than f/2.8 (the best sensitivity achieved by central focusing point),
there is a lot of imprecision.

Can we also improve noise management (similar to mk III), maximum burst for RAW images
(similar to 7D update), File name customisation (already asked elsewhere but I group all ideas here),
and implement in camera chromatic aberration correction,
and differential microadjustment based on serial number of the attached lens (instead of the type of the lens?)

It is a lot of features, but it revolves around the same idea, taking from the state of the art
firmware of Mark III interesting features or up to dates algorithms and implementing these in mark III.

I know that not all should be possible, but most could.
And if somebody will tell me that if I want a mark III, I should buy a mark III,
indeed, I plan to do it, but still, a lot of mark II users could benefit from this,
and honnestly, even if it improves mark II, it won't be comparable to mark III,
so we will just polish the mark II and make it the way Canon should have done it initially,
because a professional camera that can waste 1/3 of shots while fixated on a tripod and with good settings (for experiment),
just because of thin depth of field and not so good autofocus hardware is something that should not be.
I love the mark II for everything else, but as I can't manual shoot because some eyes problems,
and line of work, I can only depend on the autofocus system. And it spoils my work.
So, I fight hard to make the best of it, but in some conditions I can't do better than allowed by rom and hardware.
But we can improve the rom.

So please, let me know all what you think about this.
Thanks by advance.
Damian

[a1ex]

but I think that there is no fundamental obstacle to improve autofocusing in markII

Prove it.

[1%]

Didn't they alter the sensor to get that new af? We haven't even dug through the firmware yet, AFAIK.  So this is all speculation.

[Harsher]

Indeed since an idea is that at first.
But I mean, no, if these algorithm doesn't require something very specific to the 5D mk 3,
like calculation operations that requires something new in the architecture of digic 5 instead of the 4 of mark II,
or pieces of hardware that can't be replaced (like some metering system not present in the mark II),
that should not be unmanageable.
I mean, they, Canon, don't do it because it would impair their sells of mark III to the benefit
of used pieces of a camera that is now deemed to be crap because of this old focusing system
(while it performs pretty well everywhere else).

But, they did improve the 7D (because for some reason, I don't think they plan to release a 7DII soon, or they did suffer
drawbacks in development but want their audience to keep being focused on their line up of camera).
So they doubled raw burst rate, just by software means.
For me, it does not seems crazy that we could achieve same effect on mark II.
And that we could use other algorithms for focusing than the original ones.
Since on Mark III you can reduce the number of points used, I think that is not just a mean of making the
number of points more handlable by the user, but that these points are deactivated for measuring.
So, the algorithm should take this into account.
Even if not, it must be possible (if not simple, I admit it), to downgrade the model of the algorithm used for using only
9 AF points instead of 61.

For me the sensor has not too much to do with this, pixel density augmented little, and otherwise,
other improvements improve image quality for us (fringing, banding, aliasing), but that does not mean
it contributes to a better mettering and focus.

I may be utterly wrong. But that surprises me that you guys who achieved so much from the original firmwares
don't expect such things to be possible while the concept is not so much more crazy than changing bitrates and resolutions
and so on... (because I do expect such things to be much more complicated than the "still part" of the camera.)

[Francis]

Autofocus ability comes from a separate specific sensor, not the image sensor. This why it works even when the mirror is down. It is not just a software difference but a hardware difference. Software won't make the older autofocus sensor of the 5d Mk II more sensitive to light.

http://www.dpreview.com/reviews/canon-eos-5d-mark-iii/3

[Harsher]

I fully understood that before even posting.
But while sensitivity of the sensor can't be improved on a hardware basis,
signal processing for what is detected could be.
It would not make sense that it could not be.