Is CDAF really derived from in-cam JPG adjustments?

I was enjoying reading this thread (Including this Posters Reply) until below little gem turned up.

micksh6 said:

The two folks missed such simple test. You know, why?

If a person can't explain the issue in simple terms that most people would understand, that means the person doesn't see the issue as a whole, from different angles. Such person is limited by his little domain.

An individual that can only use scientific terms that are relevant to his/her domain wouldn't expand horizon to higher level, just as he would never think to expand his vocabulary to a level understandable by more people.

Click to expand...

Yes, it's easy for me to ignore, but must be a bit hard to swallow for the OP and other Poster trying to get to grips with this complex issue raised by the OP.

Several other Threads going on why perfectly genuine theads deteriorate with unnecessary "Amateur Psycholgy" and/or veiled insults being inserted.

Other than that, interesting thread.

André

Andr2 said:

I was enjoying reading this thread (Including this Posters Reply) until below little gem turned up.

Click to expand...

Andr2 said:

micksh6 said:

The two folks missed such simple test. You know, why?

If a person can't explain the issue in simple terms that most people would understand, that means the person doesn't see the issue as a whole, from different angles. Such person is limited by his little domain.

An individual that can only use scientific terms that are relevant to his/her domain wouldn't expand horizon to higher level, just as he would never think to expand his vocabulary to a level understandable by more people.

Click to expand...

Yes, it's easy for me to ignore, but must be a bit hard to swallow for the OP and other Poster trying to get to grips with this complex issue raised by the OP.

Several other Threads going on why perfectly genuine theads deteriorate with unnecessary "Amateur Psycholgy" and/or veiled insults being inserted.

Other than that, interesting thread.

Click to expand...

Yes. It has been interesting. I especially appreciate the informative input from Timur Born , and the insightful input from Chris Noble . Have definitely learned some things from this thread's existence - which is exactly why I initiate such threads. And we already know "clown psychology" as well as (I, myself) have any desire to. People who reveal that they are unaware that engineers call it "damping" (as opposed to "dampening") reveal themselves as being "all wet" very early on ...

He that knows least commonly presumes most .
- Thomas Fuller

Best Regards,

DM ...

Andr2 said:

I was enjoying reading this thread (Including this Posters Reply) until below little gem turned up.

micksh6 said:

The two folks missed such simple test. You know, why?

If a person can't explain the issue in simple terms that most people would understand, that means the person doesn't see the issue as a whole, from different angles. Such person is limited by his little domain.

An individual that can only use scientific terms that are relevant to his/her domain wouldn't expand horizon to higher level, just as he would never think to expand his vocabulary to a level understandable by more people.

Click to expand...

Yes, it's easy for me to ignore, but must be a bit hard to swallow for the OP and other Poster trying to get to grips with this complex issue raised by the OP.

Several other Threads going on why perfectly genuine theads deteriorate with unnecessary "Amateur Psycholgy" and/or veiled insults being inserted.

Other than that, interesting thread.

André

Click to expand...

Yeah, little domain, I stand by my words. The terminology is so specific that even engineers from adjacent domains may have difficulties understanding it without more extensive explanation. Wider audience without technical backround has even less chance understanding it.

I'm glad you enjoyed this thread. But I don't think I'm an ultimate evil that ruined perfectly going discussion. I'm not that powerful to destroy people's minds. The discussion was going to fade anyway.

There isn't much more to be discovered from technical speculations. For all practical reasons we are at the point when it's better to test our own cameras in our own style and shooting conditions.

You can prove me wrong by opening another thread, hoping that camera designer will pop up and shed the light on the issue, but I don't think it'll happen.

Or, you better prove me wrong by shooting action photos better than I posted. This is a photography forum, right? It's all about photos. Even if my photos aren't top-notch, try something similar. The thing is that capturing running people and acquiring decent focus with CDAF requires some understanding of CDAF focusing mechanism. This thread won't give you a better understanding. Your own practice will serve better.

And I don't mind when someone calls me a clown. I've been called worse on this and other forums. No reason to be a drama queen. I don't hate anybody either, even DM I'm trying to be open. It's internet, be more relaxed.

I'm not sure, but did I mention yet that zooming into the Magnify box (aka looking at full magnification on screen) seems to affect focus speed? Happened to me once reproducible (zoomed in would focus slower), but did not check again since then.

Timur Born said:

Timur Born said:

Zooming into the Magnify box (aka looking at full magnification on screen) seems to affect focus speed? Zoomed in would focus slower.

Click to expand...

The focus loop feedback signal is luminance variation, and there is less of that in a smaller area.

Yes, but it only happens when you actually magnify into the area to make it full screen, not just by decreasing the box size (increasing zoom leven).

Timur Born said:

Yes, but it only happens when you actually magnify into the area to make it full screen, not just by decreasing the box size (increasing zoom leven).

Click to expand...

Can you explain the two different focusing scenarios in more detail?

Timur Born said:

Timur Born said:

Timur Born said:

Timur Born said:

Timur Born said:

Zooming into the Magnify box (aka looking at full magnification on screen) seems to affect focus speed? Zoomed in would focus slower.

Click to expand...

Click to expand...

Click to expand...

Click to expand...

Timur Born said:

Timur Born said:

Timur Born said:

Chris Noble wrote:

The focus loop feedback signal is luminance variation, and there is less of that in a smaller area.

Click to expand...

Click to expand...

Click to expand...

Timur Born said:

Timur Born said:

Timur Born wrote:

Yes, but it only happens when you actually magnify into the area to make it full screen, not just by decreasing the box size (increasing zoom leven).

Click to expand...

Click to expand...

Chris Noble said:

Can you explain the two different focusing scenarios in more detail?

Click to expand...

Based upon an assumption that array of photo-sites sampled for luminance info remains the same:

Unless the subject-matter has fractal-like properties (of the spatial-frequencies remaining constant at any scale), it seems to me that most (but perhaps not all) subject-matter may well have less luminance variations within their structure when the zoom lens projects a more magnified view ?

This effect would be somewhat reduced by the fact that when optical magnification takes place, additional fine-details existing within the subject-matter are brought into a range no longer limited by the spatial-frequency sampling resolution limitations imposed by the size of the photo-sites used.

So far I come to the following interpretation from the behavior and image displayed on screen/EVF during focusing. My battery keeps running out on me, so the analysis is slowed down. Maybe I should post an article and corresponding new thread on this?!

I analyzed this further today and found the differences between magnified view and normal (focus/zoom box) view as well as a new look on the two different focusing methods of the E-M5.

General

There are at least two focusing methods and four different modes/combinations of using these two methods.

(1) Method one changes exposure, contrast/gradations and gain/time during focusing. It moves the focus motor faster than the other method, but surprisingly may fails on some targets where the other method succeeds.

(2) Method two keeps all JPG settings active and does not seem to change any settings (i.e. exposure, contrast) while focusing. I have to check that again once my battery allows me to. This method may succeed with acquiring focus where the first method fails!

S-AF only uses method (1), C-AF first uses method (1) on first acquire (including in between Burst L shots) and then switches to method (2) with a special twist (see below), Magnified only uses method (2) and "Full time AF" only uses method (2) at a slower activation rate than C-AF (also used for video recording as it seems).

• My speculation as to why (2) succeeds where (1) fails is that this may be due to longer exposure of single live view/focus frames when (2) is used. I assume this on seeing increased noise in shadows with (1) even when they are pulled down (darker) during focusing.

• Picture mode and Saturation are kept active (including Monochrome and Art Filters) during focusing with both focusing methods. I could not reproduce any impact on focusing for both focusing methods yet, unless the preview caused screen/processor lag itself (Art modes preview fully turned on). In "Vivid" the increase of red vibrance may or may not have a positive impact on some targets, same goes for the pushing of shadows and corresponding lowering of contrast.

• Sharpening is visible on screen if you look for it, especially in 14x Magnified view. But even then it makes only small fraction of information and I could not reproduce any impact on focusing yet. I am not sure if Sharpening is turned off during focusing with method (1), but I doubt that it makes any difference anyway.

S-AF

• Only uses focus method (1), moves focus motor the faster than the other modes, but may fail on some targets.

• Always shoots in focus when "Shutter Release Priority" is turned off (no shooting allowed without focus)! But both the green light and focus box light up too late, sometimes not at all when you fully depress the shutter right away. Focus audio beep works correctly and thus beeps way before the green lights turn up.

• "High" Frame Rate can have a positive impact on focus motor speed, but a negative impact on reliability in low light.

• AEL (exposure lock) has no impact on focus, because method (1) uses its own exposure. Exposure compensation may have an impact in that it seems beneficial to set very different exposure to what method (1) uses itself (which is useless in practice, because you don't want to mess with image exposure just for that).

Magnified view

• This seems to be the most reliable way of focusing! It only uses focus mode (2), is thus slower performing than S-AF, but sometimes faster than C-AF. Not only does it successfully acquire focus in situation where S-AF (1) fails, but when it shows the green light or allows to release shutter with Release Priority off you can be pretty sure to be in focus.

• Ignores AEL (exposure lock) and exposure compensation for live view and always runs with automatic screen exposure.

• Ignores C-AF mode and always only allows to focus once per shutter half-press and then stops focusing (allows recomposing even with C-AF active).

• Limits "Rattlesnaking" to only the very, veeery brightest spots and tries to keep the aperture as open as possible compared to all other modes. Even pointing at my flickering display set to full brightness keeps the aperture wide open in "Rattlesnaking" mode, which is quite astounding because said flickering display is one source for making the E-M5 turn to "Rattlesnaking" to begin with.

• Seems to always use "Normal" Frame Rate and thus ignore the "High" Frame Rate setting.

C-AF

• Uses focus method (1) on first focus acquire (shutter half-press) and in between shots when shooting "Burst L" series. If method (1) fails to acquire focus or the shutter if held half-pressed it switches to focus method (2), but with a special and maybe buggy twist.

• Seems to track the last "in focus" area and keeps signaling "in focus" (constant green light) until it detects a change of that area. If no change is detected the focus motor will settle down after a few seconds (stops hunting) and the green light stays on constantly. If a change is detected the focus motor will increase hunting action and the green light starts blinking. Furthermore the very first detected change is signaled by a second audio beep.

• Is the most unreliable method of shooting even when Release Priority is turned off, because it signals "in focus" (green light) even while focus is still hunting back and forth. Eventually it does settle down and stops hunting, but there is no way to know/discern unless you keep watching live view (which is useless for bursts and fast action). And since it allows to release shutter even while focus keeps hunting you easily get out-of-focus images. I consider this a bug!

• Can be as fast as S-AF if shutter is fully depressed and focus method (1) succeeds. Can be slower than both S-AF and Magnified when method (1) fails, because it always only tries method (2) after method (1) failed.

Full Time AF

• Only uses focus method (2), but usually uses a slower hunting and focusing method than C-AF. This also seems to be used for video once you start recording.

Side notes

Exposure metering makes Live View on the E-M5 freeze for a brief moment, just like focusing does on the Fujifilm X10. This seems to take longer in Magnified view where it may also show a brief flash at the end of metering. This may result in different focus behavior/performance when exposure metering is turned off for the shutter button and thus could use some testing.

"High" Frame Rate seems to keep the E-M5 from switching to "Rattlesnaking" mode even when pointed directly towards flickering light sources. Once the camera has switched to "Rattlesnaking" it doesn't help to switch Frame Rate, though, you still need to turn off the camera to get rid of "Rattlesnaking".

One more thing. Magnified mode in M needs to switch screen exposure to auto exposure (which is used permanently in PAS) during focusing if the Info button was used to simulate image exposure (time + aperture).

And: C-AF allows to depress the shutter right after waking from sleep/turning on and hopefully get a sharp shot. Using Sequential L can also help to get that shot right after wake/on.

With AF-S there seems to be a bug that keeps the E-M5 from focusing when you depress the shutter right after wake/on. It won't even work with Sequential L until you first released the shutter, everything will be OOF.

Timur ,

You, my friend, are downright amazing ! I have never seen anybody even come close to so thoroughly analyzing any camera model, ever, anytime, anywhere !

This is all good, and quite commendable. DPReview really shoud contract you (as in paid services) in order to play with cameras. Your sincere and dedicated involvement with your E-M5 has led to a veritable wealth of (seeming useful, as well as interesting) detailed information as a result. Excellent

Regarding all this information, if I was E-M5 owner, I would likely study it very thoroughly. In that I am not, it is all a bit overwhelming. That is not any sort of criticism, though. Excellent information !

Best Regards,

DM ...

Likely less than a handful of people will likely read this or even care for it. If you ever buy a Fujifilm EXR type camera you should read my X10 summary article.

http://www.dpreview.com/articles/3785306838/summary-of-fujifilm-x10s-functions-and-issues

Richard agreed to help me get in contact with someone at Olympus who is able to understand and care for the list of bugs and quirks that I collected.

Among the most unnerving for me personally are:

• Bad face detection (especially compared to my Fujifilm X10). Aka detecting faces where none are present and losing lock on real faces far too easy.

• Signaling "in focus" and allowing shutter release in C-AF when in reality the focus is still hunting and out of focus.

As another side-note: Both Face Detection and Tracking entirely depend on Live View exposure and thus exposure metering. If exposure (preview) changes to being considerably darker or considerably brighter/washed out you will lose detection/tracking. Using a combination of AEL (keeps screen from changing exposure preview) and Live View Boost (keeps screen from changing exposure compensation preview) helps to keep the focus on target.

Timur Born said:

Likely less than a handful of people will likely read this or even care for it. If you ever buy a Fujifilm EXR type camera you should read my X10 summary article.

http://www.dpreview.com/articles/3785306838/summary-of-fujifilm-x10s-functions-and-issues

Richard agreed to help me get in contact with someone at Olympus who is able to understand and care for the list of bugs and quirks that I collected.

Among the most unnerving for me personally are:

• Bad face detection (especially compared to my Fujifilm X10). Aka detecting faces where none are present and losing lock on real faces far too easy.

• Signaling "in focus" and allowing shutter release in C-AF when in reality the focus is still hunting and out of focus.

As another side-note: Both Face Detection and Tracking entirely depend on Live View exposure and thus exposure metering. If exposure (preview) changes to being considerably darker or considerably brighter/washed out you will lose detection/tracking. Using a combination of AEL (keeps screen from changing exposure preview) and Live View Boost (keeps screen from changing exposure compensation preview) helps to keep the focus on target.

Click to expand...

You are, of course aware that "the truth has nothing to do with" these faux "gear-head" forums, right ? And that those who dare to speak the truth risk eternal stalking from raving, drooling fanpeople whose very self of self worth is based on their techno-tribal belonging to a family-unit of like-minded neanderthals, right? With all that firmly in hand, my brave son, godspeed and good luck

Sincerely yours,

DM ...

Detail Man said:

DPReview really shoud contract you (as in paid services) in order to play with cameras. Your sincere and dedicated involvement with your E-M5 has led to a veritable wealth of (seeming useful, as well as interesting) detailed information as a result. Excellent

Click to expand...

I'm not good at writing entertaining articles that focus on the essentials. In my collaboration with Richard I threw heaps of information at him and he filtered it out.

Instead companies should cont(r)act me to play with their equipment and get detailed quality assurance feedback on issues and quirks. Most companies don't want to invest in this kind of QA, but some do pay me for exactly that. So feel free to recommend me to any tech related company, I've got a second baby on its way and can use the extra dough. Hehe.

Detail Man said:

You are, of course aware that "the truth has nothing to do with" these faux "gear-head" forums, right ? And that those who dare to speak the truth risk eternal stalking from raving, drooling fanpeople whose very self of self worth is based on their techno-tribal belonging to a family-unit of like-minded neanderthals, right?

Click to expand...

I came over from the Fujifilm Talk forum, m43 Talk is happy-go-lucky in comparison. The alpha males over there seem to h_te me to the bone, but I only care about sharing and getting information and helping some people to make up their own minds along the way.

Timur Born said:

Detail Man said:

You are, of course aware that "the truth has nothing to do with" these faux "gear-head" forums, right ? And that those who dare to speak the truth risk eternal stalking from raving, drooling fanpeople whose very self of self worth is based on their techno-tribal belonging to a family-unit of like-minded neanderthals, right?

Click to expand...

I came over from the Fujifilm Talk forum, m43 Talk is happy-go-lucky in comparison. The alpha males over there seem to h_te me to the bone, but I only care about sharing and getting information and helping some people to make up their own minds along the way.

Click to expand...

Does DPReview block the word "hate" when included in posts ?

While all fanboys know something of a serious disconnect from reality, every once in a while we see one who is (nevertheless) a very nice person. My fine friend Digby Dart endlessly defended the noisy FZ100 on the Panasonic Talk Forum (resorting to stacking huge numbers of images recorded at base ISO=100 and the like, just to make the camera seem more viable). I see that his re-incarnation as Max Metz on the Fuji forum (and his alleged endless attempts to deny the existence of the dreaded "orbs") actually appeared to have gotten him banned (at one point recently, and for an unknown time period) for (dig this) "brand preening" (so to speak) - as opposed to being banned for the alleged "sin" of "brand bashing" ... This should be a cautionary note to any and all fanpersons ... excessive stroking of one's "unit" may result in a DPR ban ! ...

Nah, I usually keep away from using strong words on public forums, especially since I am writing under a real name while being a self-employed tech specialist.

I know Max from the X10 forum, very benign guy, but surely guilty of being a fan not just occasionally.

Timur Born said:

Nah, I usually keep away from using strong words on public forums, especially since I am writing under a real name while being a self-employed tech specialist.

Click to expand...

Understood. Web-bots are ever-present, ever searching for a chink in one's armour. Nightmarish ...

I really do think that DPReview should contract you to give closer attention to new camera models than they have time for. Your deep scrutiny of the E-M5 clearly brought forth a great deal of user-relevant information. Their schedules are swift and time-limited due to the quantity of tasks that they take on. Sometimes I do wonder, though, whether camera manufacturers would be so eager to have their products (from best to worst, from strengths to weaknesses) found/published.

Timur Born said:

I know Max from the X10 forum, very benign guy, but surely guilty of being a fan not just occasionally.

Click to expand...

He is a fine fellow, indeed. Never seen someone gall so many angry people by being ever-pleasant. However, reality does not seem to be his primary forte. Love him just for who he is, nevertheless.

Especially if you consider that my forename + surname combination seems to be unique on the web. I just tried to google myself and found this very DPR thread here right on page 1 fourth entry.

Feel free to ask DPR paying me regular money, but in practice they needs folks to write finished articles and from my position they are sitting at the other end of the world. And Olympus Germany isn't even able yet to fulfill my preorder or tell me a date (instead they sent a mail telling me to regularly check their web-store now), so I doubt that companies want to send out two units to different places.

I did work in journalism for brief period, both for print (newspaper) and web (computer display review site), but find myself too lazy/uninterested/untalented to write entertaining and get to the point articles. What I do best is to provide other people - preferably developers - with lots of information and have them draw their conclusions.

Unsurprisingly I get lots of Beta invitations for all kinds of software, but of course none of these pay in other ways than providing me with free licenses. Fortunately there some companies who do pay for my service of ripping their products apart for good. And surely I'd like to get more customers of that sort.