The EOS 80D Replacement to be a Big Leap Forward [CR2]
- Thread starter Canon Rumors Guy
- Start date
Without getting into the weeds, Spot AF is amazingly accurate. I can point at a tiny bird nestled behind tons of foliage, much smaller than would useful for a photo and the camera will correctly autofocus on it.
What more could I ask for? I don't really care about the size of the box, really; I care about the camera focusing on what I perceive to be at the center of that box.
What more could I ask for? I don't really care about the size of the box, really; I care about the camera focusing on what I perceive to be at the center of that box.
Upvote
0
There’s some whiplash. An hour ago anything which displays the AF selection with a different size than the actual sensor/pixel group was sh*t. Now it’s 100% agreed that the size of the displayed box isn’t a care.
Orientation has never been mentioned but I’d offer that it’s more important than relative size.
In any case, quick turn progress, baby!
Orientation has never been mentioned but I’d offer that it’s more important than relative size.
In any case, quick turn progress, baby!
Upvote
0
fullstop said:
Typically when someone quotes something (cutting out part of the statement even) and says he agrees 100%, that’s what he means. I wasn’t picking a nit, but mea culpa for taking you at face value.
Ironically, talys’s statement essentially calls the size of an AF indicator a nit. Similarly, battery charge indicator scheme many people call a nit. Of course, they’re apologists.
Upvote
0
I do find it interesting that orientation of the AF sensor isn’t typically conveyed in the display. I’m not sure what would be a good method. Maybe different colors, but visual queues are typically best when colorblindness isn’t a contraindication. I expect knowing what the sensitivity associated with a given point is will improve success more than most anything you can show in the VF, beyond rough alignment to the center of the AF line. And that I’m not sure is even a given.
Upvote
0
3kramd5 said:
I'm getting confused. We seem to be talking about "mirrorslapper" 6DII phase detect AF more or less at the same time we are talking about the M's sensor based AF. Isn't the AF sensor size issue mainly an issue associated with sensor based AF?
Upvote
0
3kramd5 said:
The answer is no, there are not. For those that subscribe to the school of thought that Canon and other manufacturers are out to nerf their products and deliver sh!t to their customers, this is just par for the course and they can stop reading this post now.
Those who prefer factual information and would rather understand why an aspect of camera performance is the way it is, instead of just blaming the manufacturer, read on...
The reason there are no 'cameras whose display of the AF points are the “100% exact size” as the [AF points on the] AF sensors' is that the area of the image field which is sampled by an AF point is not constant, it's variable. When an a lens' maximum aperture is larger than the aperture baseline of an AF point, the image area sampled is also larger. So, for example, with a 'typical' f/5.6 AF point the AF area is larger with an f/1.4 lens than with an f/2.8 lens, which is larger than with an f/4 lens, etc. Similarly, with a variable aperture zoom lens (e.g., f/3.5-5.6) the AF point area gets progressively smaller as the lens is zoomed from wide to telephoto focal lengths. In theory, for a fully controllable display (i.e., the main/rear LCD or an EVF, but not an etched focus screen or the transmissive LCD in many newer DSLR OVFs), the camera could alter the relative display size of the AF point based on the lens' aperture.
But...aperture is not the only variable. There are 'masks' before the AF sensor that restrict the light falling on the sensor line pairs to light coming from the edges of the iris/aperture opening (that's the phase difference that underlies please-detect AF). Those masks are actually outside of the plane of focus, so the edges of the AF point area are, in effect, blurry. One consequence of that blurriness is that higher contrast features in that blurry zone can be detected by the AF system, whereas lower contrast features may not be detected. This variability is additive with the effect of lens aperture. So, for example, on a high-density AF array (5DIII, etc.) with an f/2.8 lens and an f/5.6 AF point, a high contrast feature (e.g., a point light source against a dark background or a black/white transition) can be locked onto when it's right in the middle of an AF point adjacent to the selected point (i.e., the actual area of a single AF point can, in some circumstances, cover 9 displayed AF points). However, with an f/5.6 lens and lower contrast features, the AF area will be basically restricted to the typical/current display size of AF points (i.e., the actual area of a single AF point can, in some circumstances, be pretty much identical to the single selected/displayed AF point). (Oh, and with an f/8 lens like a pin-taped 100-400 with a 1.4x TC and an f/5.6 AF point, the effective area of the AF point is actually smaller than the little box in the OVF.)
Sorry...I know that was a long and technically detailed explanation. If anyone who routinely bashes Canon ignored my above advice to stop reading and has gotten to this point, feel free to just ignore the technical facts and continue to blame Canon for displayed AF points that don't 100% accurately reflect the image area being sampled as just a manufacturer nerfing their cameras to screw over their customers. Those who are inclined to accept facts and reality are welcome to blame physics for this, among others, of life's trials and tribulations.
Upvote
0
ok. But question:
What about mirrorless cameras / DSLRs in LiveView mode?
there is no separate Phase AF unit in play. Will "physical AF fields" not be constant in size then?.
Drilling down.
* non-DP AF sensor, hybrid AF i-plane Phase AF + Contrast AF [like EOS M]
* DPAF sensor
If variable size even in those cases, would it not be feasible then to either
* set AF field boxes to maximum size the respective physical AF field can take? [Box larger than physical AF field is less of a problem than box smaller than actual AF field.]
* or - more complex - use in-camera software to re-size overlay marker box/es dynamically to match size of physical AF-field for current settings
?
What about mirrorless cameras / DSLRs in LiveView mode?
there is no separate Phase AF unit in play. Will "physical AF fields" not be constant in size then?.
Drilling down.
* non-DP AF sensor, hybrid AF i-plane Phase AF + Contrast AF [like EOS M]
* DPAF sensor
If variable size even in those cases, would it not be feasible then to either
* set AF field boxes to maximum size the respective physical AF field can take? [Box larger than physical AF field is less of a problem than box smaller than actual AF field.]
* or - more complex - use in-camera software to re-size overlay marker box/es dynamically to match size of physical AF-field for current settings
?
Upvote
0
fullstop said:
Good questions.
The on-sensor 'hybrid' PDAF systems with dedicated/masked AF pixels have the same limitations/challenges as the dedicated secondary AF sensor PDAF systems in terms of lens aperture. Canon's DPAF could theoretically overcome that because with every pixel available for PDAF (well, 80% of them since the frame edges aren't included) the baseline aperture of the AF point can be dynamically adjusted based on the max aperture of the attached lens.
But...the variability resulting from high-contrast features being detectable further from the center of the AF point than lower contrast features affects all PDAF systems. Since that variability is dependent on the scene being imaged as opposed to the camera/lens hardware, there's really no feasible way to compensate for that (well, I suppose the camera could analyze the scene and display AF points as differently sized boxes depending on the content of the scene, growing or shrinking each box independently as the scene changes and/or the camera is moved...but I really don't see that as feasible).
For all practical purposes, the only way to have an AF point display that exactly matches the area being sampled is by restricting AF to contrast-detect only (with the slow focus and racking back-and-forth that comes with it).
Sometimes physics sucks.
Upvote
0
3kramd5 said:
Fair point. Let me rephrase....sometimes the consequences of physics suck, particularly when those consequences conflict with our desires.
Upvote
0
Talys said:Neuro, very informative post. I did not know, well, any of that
Live and learn. Thanks!
His post over at TDP is great. I reference it often.
https://www.the-digital-picture.com/Photography-Tips/canon-eos-dslr-autofocus-explained.aspx
Upvote
0
3kramd5 said:Talys said:Neuro, very informative post. I did not know, well, any of that
Live and learn. Thanks!
His post over at TDP is great. I reference it often.
https://www.the-digital-picture.com/Photography-Tips/canon-eos-dslr-autofocus-explained.aspx
Awesome. I'm at TDP all the time and I never saw this. Very informative passage on why PDAF setups can't cover more of the frame, Neuro.
(spell check on 'vingetting', btw
)- A
Upvote
0
3kramd5 said:Talys said:Neuro, very informative post. I did not know, well, any of that
Live and learn. Thanks!
His post over at TDP is great. I reference it often.
https://www.the-digital-picture.com/Photography-Tips/canon-eos-dslr-autofocus-explained.aspx
This is a great article (I hadn't read it previously, obviously). That's the first I've seen a picture of a 1DXII PDAF sensor, I think. Excellent explanation as to the reasons why we get the coverages that we do with dedicated PDAF sensors, and the differences between various camera bodies and their AF sensors, and why the 1DXII has so much better performing AF with less light.
I wonder if (when?) we'll ever have DPAF or other on-sensor solutions that are as sensitive with little light as the 1DXII, giving us that sort of AF performance.
Neuro, excellent article, thanks again
Upvote
0
Similar threads
