[clackers]

How can I agree with the results from the test you posted

Same answer for you, Dan: Show us newer objective tests.

They always beat supposition, anecdote and subjective tests, which the internet seems to thrive on, and your post, I'm afraid to say, is exactly the same.

Here is just one set of *objective* results with a reproducible method from a German lab. Incredible, isn't it?

Attached Images

 

---

Last edited by clackers; 03-12-2020 at 05:22 PM.

[mikesbike]

The burst and auto-select multi AF point mode is not going to be reliable unless the technique is there to support it. With that system, the camera will select exactly which subject to focus on, which can turn out to be the wrong one unless guided in some way. The most precise way to take advantage of a good tracking system for action is to have the camera on a panning tripod to follow action. Since most multiple-point AF systems are basically geared for central-priority then if nothing can be had, defaulting to nearest, it would be important to nail focus at a central point where subject is moving across, then the tracking can follow as subject moves off-center.

Based on so many reports, my impression is that Nikon has a better tracking system, if used with good technique, compared with a Pentax system. But as clackers says, it is still gonna be tricky, yet will improve as one gains experience. And yes AF does need an edge to latch onto, or some contrasting detail, whether a wrinkle, player number, belt, chin, or whatever. I am also not especially a fan of having a very fine tracking outfit and burst rate of say 10 FPS, then hoping for best results, I would not look forward to sitting to sift through hundreds if not thousands of shots to determine which are the best ones. If highly paid by a sports mag, that's a different story.

One way or another, there are going to be takeouts.

@ThorSanchez: The shots you post are fine. Here are some observations- you are working with plenty of available light, which makes things much easier. A good slow lens is all you need. The f/5.6 or even f/6.3 will give plenty of shutter speed, and you still get some important DOF even though wide open, and all this still at a modest increase in ISO.

But you could have even more range, and even better effects. Although a shutter speed of 1/1250 sec is fine to stop fast action, I think it is more than needed for kids running with a ball. When I shoot college-level or adult club floor roller-hockey, the indoor lighting is not near the outdoor daylight advantage you have, and the action, believe me, is much faster. I use a fast f/2.8 zoom lens wide open or nearly so, yet I need to boost ISO to around 2,000 or 2,500, or even 3,200 or more just to get 1/500 sec to 1/1,000 sec shutter speed. Yet as fast as those players go, I am able to freeze body action quite well even at 1/500 sec. I am panning while shooting hand-held with spot AF, and working my half-press AF finger rapidly. For me, the front finger half-press works quicker than the rear AF button, as I find for me using one finger for all is faster than co-odinating my thumb and front finger. I can be so fast, I can get good shots even if I choose AF-S instead of AF-C.

I can slightly alter shutter speed (I don't have a whole lot of laterality here, though much more now with the KP) so most often I can freeze the puck in mid air. Keep in mind it is hard to tell just when a player will decide to hit the puck-and then it is sooo quick! Or, I can freeze the player's body, and the puck in-air is recognizable, but shows blur- which is fine! Then still photography is not too still! Better yet, the player's stick shows some blur also, maybe his arm too, but his face at least is sharp. Great! It is a true action shot!

Just some things to think about in developing technique for control to vary your outcome for different effects.

[Dan Rentea]

Same answer for you, Dan: Show us newer objective tests.

If you want me to show you lab tests with wrong Af settings like the german lab did, I can. Even then it will be hard to beat those results with any camera they used if you give me time to learn the Af from each. But what's the point in showing bad results due to wrong Af settings? Above I showed you real life examples and below it's the very next burst from the series. I have around 400 images from this shooting with around 280-300 images taken in bursts of 5-12 shots. I usually don't take bursts with more than 5-7 shots, except when it comes to very fast subjects. Out of these 280-300 images, I have around 9 images out of focus and this is mostly because I never shoot with focus priority on the first shot. Knowing very well the cameras I work with, I know what the Af system is capable of and I know that I rather lose the first shot knowing that the next ones will be in focus rather than miss the entire burst by prioritizing the first shot by selecting focus priority for Af. That's why when I test the Af system I always take singular shots also to see how fast and accurate the camera I'm using will acquire focus in challenging situations. I see lots of people using focus priority for the first shot and with fast subjects you may miss the opportunity in lots of occasions. This is another thing that the german lab didn't mentioned in their "respectable" test regarding Af settings.

Again, straight out of camera burst of 9 images. As long as the german lab wasn't using as a subject the mighty Usain Bolt, getting that bad results when shooting faces is doable only if you work hard to fail or if you don't take time to learn a complex Af system with 6 Af cases, multiple Af point selections and various tweaks for each of the 6 Af cases.

They always beat supposition, anecdote and subjective tests, which the internet seems to thrive on, and your post, I'm afraid to say, is exactly the same.

The above images are images taken for fun and given the colors of the horses and given the backgrounds, in a few situations it's hard for the Af system to lock focus (white horse with white background, white horse with people with contrasty clothes in background, etc.). And it did focused for 2 reasons:
1. I know how to set the Af system to get the best out of it
2. I know how to track because it's something I worked on very hard during the years, mostly with 400mm + lenses and by shooting wide open with them rather than fool myself by shooting at f8-f11 to have enough DOF to keep an entire elephant in focus

Here is just one set of *objective* results with a reproducible method from a German lab. Incredible, isn't it?

It's incredible that it's so easy to cancel their tests just by looking at the Af settings they used. I showed the print screen of yours to some of my friends and the Nikon shooters were surprised also by the setting used. This means to me that the german lab just used a combination of wrong af settings, like:

1. Af case not suited for the subject
2. Af point selection not suited for the subject
3. Af configuration for each case they used was untouched or again, used wrong given the subject
4. Af priority vs. Release priority used wrong
5. Lots of wrong combinations from the above 4 points, which is what they did in my opinion.


But, given that we talk about tracking, Auto Af it's not recommended for faces or for birds, with a few exceptions when comes to birds. Next, when shooting sports with a slow lens, it doesn't matter if the focus is on the player's shirt because at f6.3 there is enough DOF to get the face in focus also. This is why the photographer should concentrate in keeping the subject in the frame by using either single point af or 1+4 Af combination. If the camera doesn't have the settings that allow tweaks like ignoring obstacles when tracking af is active, then you rather use single point af to maximize the results. This technique needs practice and also needs a photographer who knows the sport because anticipation is another great "tool" available. When comes to birds, there are lots of variables that can make the Af system to fail but there are also (in some cameras) lots of features that when used right the results are very good.

As I said, the repeatable results of the german lab can be successfully ignored in my opinion given the wrong settings they used from the print screens you posted. Using wrong Af settings will give repeatable wrong results over and over again. The useful thing to keep in mind from all the comments you made in this topic is regarding the D5 comment of yours where experienced photographers know very well both the subject they are shooting and the camera they use and based on these 2 things they know when the action is so intense that anticipation and pre-focusing can get you the winning shot instead of relying on Af system always. This is a very useful comment for both experienced and inexperienced photographers and you should have bolded it to draw attention to it. The german lab tests are just for clicks and views.

[Dan Rentea]

Here is also a very challenging situation due to lack of contrast on the horse and due to the fact that the sun decided to show up from the clouds and makes my life harder. The horse was coming fast towards me (you can see that by looking at the legs) and the focus worked because it was set properly, among other things. I have tons of examples far more challenging than what the german lab did in their comparison, but as long as some people are willing to believe lab tests over real world examples, then no wonder why so many people complains all over the internet about Af of their cameras, being Canon, Nikon, Pentax, Sony, Olympus, Fuji.

Image straight out of camera




The same image with 100% shadows recovery




100% crop from the above image with shadow recovery. Focus is where it should be.




Next image from the burst, straight out of camera




The above image with 100% shadow recovery




The third image from the burst, straight out of camera




The above image with 100% shadows recovery



And there are 2 more images in the burst, all of them in focus. As I said, repeatable results I can give you tons, with af settings that works given the circumstances and the subjects photographed. I can't give you as bad results as the german lab if you want me to be objective given what I shoot.

---

Last edited by Dan Rentea; 03-13-2020 at 03:08 AM.

[ThorSanchez]

Well, Thor, show us newer objective tests.

They always beat supposition, anecdote and subjective tests, which the internet seems to thrive on.

Good data always beats supposition and anecdotal evidence. But bad or biased data can be misleading or worse than none at all. And a single test by a single individual or group that contradicts consensus at least requires further study.

I've been an avid student of analysis in baseball since I was a teenager. There are countless cases where the best available data was just wrong, and the anecdotal data was better. Traditional fielding metrics, for example. Or even the first generation of more advanced fielding metrics from the 1980s and 1990s. Of course there are many of the opposite case, too, where conventional wisdom was wrong and only through careful analysis did the truth become apparent.

If everyone says that Adrian Beltre is the best fielding third baseman in the league but the data says he's awful the data better be pretty comprehensive and rock-solid. Just as is the case if you give us a single study that suggests cameras that have been reviewed by multiple sources and praised for their autofocus are actually no better or worse than cameras that have been reviewed and graded as having subjectively poor autofocus.

[Dan Rentea]

Good data always beats supposition and anecdotal evidence. But bad or biased data can be misleading or worse than none at all.

If only the data from the german tests were accurate... If I post those print screens with the results of the german lab to my Facebook page saying that those are my results, all the photographers who knows me would think that I've either gone mad or that I'm joking. There is so much wrong data there that even someone who just read the Af manual from Canon 7D Mark II can give them some useful suggestions. They even managed to get 50-60% less sharp images with the Canon 70-200mm f2.8 shooting a "bird" at 70mm vs. the cheap (150$) and slow 50mm f1.4 lens shooting the same subject. Wait, the 150$ 50mm f1.4 lens beats the Canon 70-200mm f2.8 lens at 70mm, 95mm and 200mm in terms of keeper rate. That's a performance that even my 11 years nephew would have hard time to match, even doing the tests repeatedly. And my nephew read the manual of my 5D Mark IV, twice. Yet, these tests are like Youtube influencers and we have to live with them, unfortunately.

But since we talked about repeatability, in real world for me this means repeatability when shooting the same subjects. The only image which is not in focus (it's acceptable) from the set of 16 images taken in a burst is the one I highlighted in red. The af lost the horse for a split of a second because the horse basically stopped for a split of a second before starting to run again like crazy. It's a z-direction burst if the diagonal ones are not good enough.



It was a training and an event that day that's why you see one person making the horse running. Me and the woman you see in the below image were lucky to be able to shoot that day the event. She is a horse trainer and I have some beautiful shots with her at the end of the event.



So you see Clackers, I have another 250 images like that from that day that makes me say that the german lab test is for clicks and views, not for real photographers interested in Af performance. And I have a lot more images with different subjects if I want to really make a point about it. Heck, if I post the bursts of images taken with the old 6D when I was out shooting my nephews running fast towards me on their roller skates I will "destroy" all the so called repeatability of the german lab tests because I have more keepers than them with and old 11 af points camera. You know that I have images to back up my comments when necessary. You just have to tell me which ones do you like:
- running in z-direction like above
- running straight at me like above
- running side way
- running to me in diagonal
- running from me in diagonal like the below one

I have all kinds of repeatable burst images.

[clackers]

And it did focused for 2 reasons:
1. I know how to set the Af system to get the best out of it
2. I know how to track because it's something I worked on very hard during the years

I think, shame on you, @Dan Rentea because that's what I said back in post 25, and you should be agreeing with me and preaching all that to Thor Sanchez, not me, because that's already my position - your strategy is baffling!

I think the Cameraville guy says it very well, "received wisdom" doesn't count when it's by poor photographers or photographers who don't understand their own camera's settings:

Youtu.be

Youtu.be

If you want me to show you lab tests with wrong Af settings like the german lab did, I can.

Dan, again, in your enthusiasm, you have made a mistake. It's the problem when you make assumptions.

The first set of results I showed you were not the 2017 German lab tests ... they're here, there's no other study you can show us that beats it for taking the photographer out of the test and showing what the hardware is capable of:

https://www.image-engineering.de/content/library/conference_papers/2017_03/a...erformance.pdf

If you try and bring up the DPR bicycle test, you should probably volunteer to leave this forum.

---

Last edited by clackers; 03-18-2020 at 10:03 PM.

[clackers]

Good data always beats supposition and anecdotal evidence. But bad or biased data can be misleading or worse than none at all. And a single test by a single individual or group that contradicts consensus at least requires further study.

I think you should see both the links I've just given Dan above, Thor, and if afterwards you still think that someone can buy their way to taking better pictures instead of improving their skills ... I don't know what else to say, to be honest! I've stood beside a Canon user at an air show who was shattered by the problems in the pictures he was reviewing in his viewfinder.

[ThorSanchez]

I think you should see both the links I've just given Dan above, Thor, and if afterwards you still think that someone can buy their way to taking better pictures instead of improving their skills ... I don't know what else to say, to be honest! I've stood beside a Canon user at an air show who was shattered by the problems in the pictures he was reviewing in his viewfinder.

I don't think I ever said anything of the sort. I've worked on my skills doing sports with Pentax for the better part of a decade and I often get very good results. I take hundreds of pictures of my kids' U13 travel soccer team at the same time as another parent who has a small photography business where she advertises sports as one of her specialties. She shoots Canon. Frankly, my results are every bit as good or better than hers.

But none of that means that I shouldn't be curious as to how other brands' more highly-regarded autofocus systems might squeeze more keepers out if the photographer is skilled, too.

[Ian Stuart Forsyth]

If you want me to show you lab tests with wrong Af settings like the german lab did,

One of the thing I have notice with their data is with the Nikon cameras is they list in there table that they use for mode they Dyn 9 and AF setting as 3d tracking. I find this kind of hard to do as Dynamic 9 is a AF setting just as 3d tracking is a setting you cannot set dynamic 9 and use the function of 3d tracking in Nikon camera.

Also in 3d tracking there is no auto mode and I have a feeling that they don't know what 3d tracking is really doing and how to use it as there is not auto. You must lock onto the subject you want to track with the selected AF sensor that you have selected before you start 3d tracking.

[ThorSanchez]

What needs to happen is for us to have our own Pentax Forums autofocus shootout with the K-3ii, KP, K-1, Nikon D5 and D500, Canon EOS 1D X Mark II, Fuji X-T3/4, Olympus OM-D E-M1 II, Sony A7 III, Sony A9, Panasonic Lumix G9, etc. Make it totally comprehensive. Various subjects, people playing soccer/football, birds, rare arctic tigers, zigzagging bicyclists approaching us, meteors, the F-35 Joint Strike Fighter, hypersonic cruise missiles, your morbidly obese Uncle Wayne collapsing into the couch with a can of Miller Lite. All kinds of users from 30-year pros to random bums hanging out in abandoned railway yards.


Then we'll have reams of data to apply confirmation bias and conclude that our preconceived notions were right all along.

[Dan Rentea]

What needs to happen is for us to have our own Pentax Forums autofocus shootout with the K-3ii, KP, K-1, Nikon D5 and D500, Canon EOS 1D X Mark II, Fuji X-T3/4, Olympus OM-D E-M1 II, Sony A7 III, Sony A9, Panasonic Lumix G9, etc. Make it totally comprehensive. Various subjects, people playing soccer/football, birds, rare arctic tigers, zigzagging bicyclists approaching us, meteors, the F-35 Joint Strike Fighter, hypersonic cruise missiles, your morbidly obese Uncle Wayne collapsing into the couch with a can of Miller Lite. All kinds of users from 30-year pros to random bums hanging out in abandoned railway yards.


Then we'll have reams of data to apply confirmation bias and conclude that our preconceived notions were right all along.

That will be nice, but hard to implement. I would love such a topic because I extract data regarding af by looking at both random images and burst images. On random images (or singular images) I look to understand and learn things like how difficult was the subject photographed based on:
1. settings used
2. background (clean, busy, contrasting, etc.)
2. size of the subject and how fast and erratic moves
3. etc.

Same for bursts when accuracy of the af system is also visible.

It would be nice to start a topic similar with the "300mm plus lens club" where everyone can post action images with settings used and if someone really wants to say something or if the one who post ask for feedback, a private message would be preferable so that the topic can stay clean.

[normhead]

What would be nice would be if a few of the folks who claimed the tests posted above were inaccurate would do some comparisons, with other settings showing if different settings would make any difference to the outcome of the test.

So far we know...

The test didn't come out favourably to some people's cameras.
Using the settings they did, their cameras did not fair as well as they believed they should.
Some people have claimed their brand of camera would have done better with better settings.
Yet there is no information where any one of them has done a comparison with different settings with information on whether better settings would have increased or diminished the performance achieved in those tests.

What do we know? If you don't read the whole manual and take courses etc. your Canon ,Sony Nikon whatever isn't any better than anyone else's camera? That much has clearly been established.

Based on your response it's been suggested that some of the camera setting were wrong.
But that is a very weak criticism, without any corroborating evidence to say there are circumstances where they could have done better.
The whole conspiracy theory that they did it just to try and discredit these cameras is also a complete fabrication with no justification.

You and I have discussed these things before, you criticized pictures on aesthetic grounds that were posted as examples of technical ability, pictures that were perfect examples of tracking you dismissed as "cheating". You always have a reason why you're right but you defend those criticism giving the camera you use breaks you don't give competing cameras. You never acknowledge others points and observations. Why would I subject myself to that?
But I agree, the 300mm club is a great place for folks to analyze images. If they are on flickr, settings and exif will be somewhat availalbe. Lot's of great BiF images. The first thing you'll learn is, criticisms of Pentax AF are vastly overstated. People willing to read 157 pages of manuals can get it done.

If birding is your primary use ( that's a whole separate category of use with it's own demands irrelevant to most photography) you buy a birding camera just for that purpose.

That has absolutely no relevance to Pentax shooters. We bought field cameras with some tracking and birding capability. And that's what the focus of the forum is. None of us hear wants to buy a $10,000 system, just to take pictures of birds. None the less I take lots of pictures of birds, and when I've had rare opportunities to get images of birds in flight the camera has been up to the task, but I wouldn't buy a camera for that. I'd post them as examples but you've already insulted them all and explained why they are irrelevant (yours are always relevant somehow) so what would be the point?

[ThorSanchez]

So far we know...

The test didn't come out favourably to some people's cameras.
Some people have claimed their brand of camera would have done better with better settings.
Yet there is no information where any one of them has done a comparison with different settings with information on whether better settings would have increased or diminished the performance achieved in those tests.

It's reasonable to make a few generalized assumptions, such as:

- Correct autofocus settings should provide better performance than incorrect ones.
- Cameras that have subjectively been reviewed and judged to have superior autofocus should usually, generally, have better autofocus performance than ones judged to have poor autofocus performance
- There is at least some kernel of truth to the idea that there are meaningful performance differences between cameras

The baseline assumptions for this discussion shouldn't be that all autofocus systems are equal. When a single study from five years ago indicates counter-intuitive results such as a Leica being on par with the best sports cameras that tells me that the results should be further analyzed and much more data gathered instead of taking it as the ground truth.

To me it's unwise to dismiss the conclusions of the entire body of subjective camera reviews in favor of a single comparative study from five years ago. The German study discussed here is one data point out of thousands; it slightly adds to what we know.

[normhead]

It's reasonable to make a few generalized assumptions, such as:

What do you want, actual research or a few generalized assumptions.
Personally, I don't give in to anyone's generalized assumptions. That's not scientific. Science is there to test generalized assumptions, and it'a amazing how often they are wrong.

It is quite reasonable to assume there will be some difference. Whether or not it's a significant difference requires research, how much the proper settings would make is also unknown. So no, it's not reasonable to consider "generalized opinions" If we did that we'd all just buy A9s 1Dxs or D4's and our photography for 99% of us would suffer greatly.

"Let's post images taken with modern lenses on a K1, what do you think about that? I have some portraits taken with Pentax 70-200mm f2.8, some action images with a guy on rollers taken with Pentax 24-70mm f2.8 and some street images taken with Pentax 15-30mm f2.8. Let's also talk about the poor af of Canon and Nikon with your example that seems to match the conclusions of the test. Come on, let's see images because I have them and you can even say that I'm a cheater if you don't like them. I have images for the cameras I used and talked about here. Do you have anything to show us regarding my so called " you know little about modern Pentax gear" or about your knowledge about Canon and Nikon af system?"

Why? So you can weigh in with your incredibly biased make up the rules as you go commentary? No one wants that.
That ship has sailed Dan, I've given you many chances for reasonable discussion. You're just one of those people who defends his opinions at all costs and couldn't say something nice about a system competing with the one you own to save your life. If you wanted that kind of dialogues, it was offered many times, and you couldn't stop yourself from writing pages of partisan non-sense.

Now you want to appear as if you're some kind of champion of reasoned thought.

Volume of words does not equate to volume of knowledge.

Just your analysis of the above test is so off as to serious suggest you lack the intellectual rigour to participate in such an exercise. You confuse objections to methodology, with research. The only answer to flaws in research is better research. Not some guy going on and on about why it all isn't fair.

I don't understand why this is an issue. If I found a similar error in Pentax testing, I'd just go out, take a series of images with both settings, and document how one setting was better than the other.

I definitely wouldn't claim a system could do better with better settings without establishing that. That's science. What you're doing is trying to turn socially/ cultural narratives into science, when they are perhaps the least scientific of all data collected.

Once you have have hard data, what people think goes out the window.

So give us some hard data to support your case. Page 157 in the manual is not performance data.

Full stop until that happens. Enough weasel wording your way through justifying your unsupported opinions.