Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
03-30-2011, 10:13 PM   #436
Site Supporter
Site Supporter




Join Date: Jan 2007
Location: Fallon Nevada
Photos: Albums
Posts: 504
QuoteOriginally posted by falconeye Quote
Christine, your theory is a good one. I recall you (and some others) making the distinction.

I introduced the idea that both issues are actually the same, but the micro issue becomes a macro issue with dark backgrounds. The difference between micro and macro issue would be the EV value where the issue happens, not its magnitude.

The point is I never actually conducted systematic tests with a dark background and therefore, I cannot compare for this situation. My bad.
Hmmmm,

All of my recent tests were done with a LensAlign fixture against a black backdrop and either 5500k CFL bulbs and/or 2750K (nominal) 250W tungsten modeling bulbs. I did not test with white or grey backdrops, but I think I have both backdrops and could do so.

In my tests, 1.03 was clearly better than 1.02 in the same conditions (with a black back drop). The focal point shift with the FA 50 @ f1.4 after 1.03 was installed was clear.

However, my FA 31 ltd and FA 43 ltd lenses both BF significantly on the K5 in all lighting after 1.03.

In normal shooting, the K5 still misses sometimes for unknown reasons, but not by anywhere near as much as it did before 1.02 and it does not seem to be dependent on light levels.

Oddly, it seems to do so most often when the frame is mainly filled with a light colored face.

Ray

03-30-2011, 11:47 PM   #437
Veteran Member
Emacs's Avatar

Join Date: Oct 2010
Location: Moscow
Posts: 1,223
So, I've made my little and dirty research and it looks like the FF issue is inconsistent among all points but is consistent enough for the group of points. Let's see at the chart:
Code:
+ + +

| + + + |

+ + +
Both edge AF points seem to be useless and miss much of the time, so I excluded them from the «research».
Red group of cross AF points seem to have about the same FF much of time, even at daylight they have it.
Orange one has lesser but quite significant FF.
Yellow has the FF too, but it's less than of orange and red ones.
Green, the central AF point focused precisely most of the time.
I've got similar result with my FA 43Ltd at 1.9, 2.0,2.8 and FA* 85 at 1.4,2.0,2.8
The measurements were made in daylight.

Last edited by Emacs; 03-31-2011 at 11:55 PM.
03-31-2011, 02:54 PM   #438
Veteran Member
Smeggypants's Avatar

Join Date: Sep 2010
Posts: 1,536
As of today, I'm now the proud owner of TWO K-5s.

Thankfully the 2nd K-5 is alsos behaving itself and focusses fine in Tungsten light at EV2. Again confirming that my original first K-5 was a dud in the AF area.

Now I've tested THREE K-5s I can confirm that with V1.03 the AF is inconsistent. I would say that 3 times ouf of 10 it either back or front focusses. I've taken hundreds of real world test pics on all three bodies to confirm this. I never got the AF inconstancies with 1,01 and 1,02

This is at any light levels.
03-31-2011, 04:33 PM   #439
Veteran Member




Join Date: Dec 2007
Photos: Gallery
Posts: 8,237
Time for another comparison

QuoteOriginally posted by Smeggypants Quote
As of today, I'm now the proud owner of TWO K-5s.

Thankfully the 2nd K-5 is alsos behaving itself and focusses fine in Tungsten light at EV2. Again confirming that my original first K-5 was a dud in the AF area.

Now I've tested THREE K-5s I can confirm that with V1.03 the AF is inconsistent. I would say that 3 times ouf of 10 it either back or front focusses. I've taken hundreds of real world test pics on all three bodies to confirm this. I never got the AF inconstancies with 1,01 and 1,02

This is at any light levels.

Smegs, do you have any access to any other aps-c dslr body, Canon, Nikon, Sony? Any way you could borrow one for a day or so?

I think at this point - for everyone doing extensive k-5 testing, like you Ray, Falk, anyone else - a non-Pentax checkpoint sample should be introduced to see where the 1.03'd K-5 really stands.


.

03-31-2011, 04:56 PM   #440
Veteran Member




Join Date: Mar 2010
Photos: Gallery | Albums
Posts: 1,545
Can someone clue me in on how to tell what EV my lighting is at? I mean, I keep hearing all this stuff about EV2, etc. but am clueless as to how dark/light that is...
03-31-2011, 06:20 PM   #441
Pentaxian
Class A's Avatar

Join Date: Aug 2008
Location: Wellington, New Zealand
Posts: 11,250
QuoteOriginally posted by Ubuntu_user Quote
Can someone clue me in on how to tell what EV my lighting is at?
A light meter will tell you.

You can also use the camera settings and calculate EV as follows:
EV = log2(f-ratio^2)-log2(shutter_speed)-log2(ISO/100)
Your calculator might not have a "log2" function in which case you can use
EV = (log(f-ratio^2)-log(shutter_speed)-log(ISO/100))/log(2)
Note that here "shutter speed" has the unit "seconds", i.e., you need to put in "1/200" for a "1/200" exposure, not "200".

However, and this is a problem of most of the EV values reported, this EV number only gives you the "camera settings" exposure value. What you really need instead is the photometric exposure value of the part of the scene that the AF module focuses on. Only if the latter part meters to an 18% gray, the two EV values will be identical. If someone uses a mostly white background, their reported EV values will be about 2EV higher than the correct figure.

Read Falk's report to see how he properly calibrated his camera setting EV to a photometric EV by taking image brightness into account. Unless the latter is done, EV values taken from the camera settings are just rough ballpark figures.

BTW, using the image brightness also takes care of any exposure compensation you might have used. If you don't use an image brightness calibration make sure you either shoot with exposure compensation set to 0 or subtract the exposure compensation value from the camera settings EV value.

Last edited by Class A; 03-31-2011 at 06:26 PM.
03-31-2011, 06:39 PM   #442
Veteran Member
Smeggypants's Avatar

Join Date: Sep 2010
Posts: 1,536
QuoteOriginally posted by jsherman999 Quote
Smegs, do you have any access to any other aps-c dslr body, Canon, Nikon, Sony? Any way you could borrow one for a day or so?

I think at this point - for everyone doing extensive k-5 testing, like you Ray, Falk, anyone else - a non-Pentax checkpoint sample should be introduced to see where the 1.03'd K-5 really stands.


.
The only non Pentax body I can get hold of is a FF Canon 5D MKI, my shooting buddy's machine. He's jealous of the K-5's performance

Frankly though I don't care too much about how Canikonylymusasonic performs. I just want Pentax machines to perform to spec.

03-31-2011, 06:48 PM   #443
Veteran Member
Smeggypants's Avatar

Join Date: Sep 2010
Posts: 1,536
QuoteOriginally posted by Ubuntu_user Quote
Can someone clue me in on how to tell what EV my lighting is at? I mean, I keep hearing all this stuff about EV2, etc. but am clueless as to how dark/light that is...
I keep meaning to enter the formula into Excel. But I don't ahve excel installed on my computer these days.


I use this ....

Ultimate Exposure Computer


but Class A is right, that the metered value of the AF area is more exact. My EV2 test shots have been done where the light level is fairly consistent across the frame.


btw - I just threw a RAW into PhotoMe and it gives a graphic of the metering segments and the LV value for each segment, pretty cool!!

Search the EXIF for 'AE Metering Segments' and click on <graphic>
03-31-2011, 09:03 PM   #444
Pentaxian
Class A's Avatar

Join Date: Aug 2008
Location: Wellington, New Zealand
Posts: 11,250
QuoteOriginally posted by Smeggypants Quote
btw - I just threw a RAW into PhotoMe and it gives a graphic of the metering segments and the LV value for each segment, pretty cool!!
That would be a convenient way of getting a reading, if it is reliable.

Falk, could you check a couple of your test shots to see whether the LV values recorded by the camera coincide with the values you calculated?

Maybe you'll even find a correlation between the LV values recorded and the misfocus produced?

Last edited by Class A; 03-31-2011 at 09:19 PM.
04-06-2011, 04:34 AM   #445
Veteran Member
falconeye's Avatar

Join Date: Jan 2008
Location: Munich, Alps, Germany
Photos: Gallery
Posts: 6,871
Original Poster
QuoteOriginally posted by Class A Quote
That would be a convenient way of getting a reading, if it is reliable.

Falk, could you check a couple of your test shots to see whether the LV values recorded by the camera coincide with the values you calculated?

Maybe you'll even find a correlation between the LV values recorded and the misfocus produced?
The problem of course is that it is this light meter exactly which starts to fail when we see the problems emerge

Moreover, I rather keep control over color to intensity calibration.

Eventually, the LV reading still only is correct if the target was 18% gray and IF it is exposed 18% gray. At the same time, a white part of the target must expose 100%. Therefore, I rather use the white parts directly and calibrate for the amount of deviation from 100% in one or multiple color channels. In doing the calibration based on image feature brightness, one must absolutely avoid all non-linear development settings (brightness, contrast, levels, blacks etc.). The LR default setting is NOT good to do the calibration! And finally, one must convert from sRGB to a linear color space. Only then we can know the colorimetric properties of the focus feature.
04-06-2011, 05:17 AM   #446
Veteran Member
falconeye's Avatar

Join Date: Jan 2008
Location: Munich, Alps, Germany
Photos: Gallery
Posts: 6,871
Original Poster
Hi Ray,

thanks for the feedback. I didn't notice the BF issue with 1.03 and my FA 31. It's actually in the testing data: with sufficient light, the FA 31 performed well, both under 1.02 and 1.03.

It's getting stranger and stranger.

Interesting, you observed a significant improvement with 1.03 and a black backdrop. What was the distance? Do you have an before/after 1.03 shot for the FA50 or FA31 with identical lighting? Raw preferred. I'd like to take a closer look.

A LR JPG export of the center region, levels, contrast, brightness, blacks all to zero; but sharpening set to 100%/0.5px radius would suffice though.

QuoteOriginally posted by Ray Pulley Quote
All of my recent tests were done with a LensAlign fixture against a black backdrop and either 5500k CFL bulbs and/or 2750K (nominal) 250W tungsten modeling bulbs. I did not test with white or grey backdrops, but I think I have both backdrops and could do so.

In my tests, 1.03 was clearly better than 1.02 in the same conditions (with a black back drop). The focal point shift with the FA 50 @ f1.4 after 1.03 was installed was clear.

However, my FA 31 ltd and FA 43 ltd lenses both BF significantly on the K5 in all lighting after 1.03.

In normal shooting, the K5 still misses sometimes for unknown reasons, but not by anywhere near as much as it did before 1.02 and it does not seem to be dependent on light levels.

Oddly, it seems to do so most often when the frame is mainly filled with a light colored face.

Ray
04-06-2011, 06:56 AM   #447
Pentaxian
Class A's Avatar

Join Date: Aug 2008
Location: Wellington, New Zealand
Posts: 11,250
QuoteOriginally posted by falconeye Quote
The problem of course is that it is this light meter exactly which starts to fail when we see the problems emerge
Sure, but I was wondering what values it reports before it becomes "blind".

Also, I didn't mean to suggest that your thorough EV calculation approach should be replaced by inspecting LV data with PhotoME. I was just interested in the utility of the LV data for people who are not going to make precision measurements as you did.

As it stands, if someone (not using your precision approach) says "starts failing at EV 2", this could mean a lot of things... So perhaps, the recorded LV values could be useful to achieve a more standardised way of reporting the brightness of the focus target.

QuoteOriginally posted by falconeye Quote
Eventually, the LV reading still only is correct if the target was 18% gray and IF it is exposed 18% gray.
Are you sure? Couldn't the values indicate absolute brightness levels independently of camera exposure settings? They could be a poor man's (reflected light) light meter?

P.S.: I hope you had a chance of doing some non-measurement shooting with the K-5. Do you have a handle on whether or not it focuses better or worse in low light than your K-7?
04-06-2011, 11:43 AM   #448
Veteran Member
Eruditass's Avatar

Join Date: Jul 2009
Photos: Gallery
Posts: 2,207
QuoteOriginally posted by Class A Quote

However, and this is a problem of most of the EV values reported, this EV number only gives you the "camera settings" exposure value. What you really need instead is the photometric exposure value of the part of the scene that the AF module focuses on. Only if the latter part meters to an 18% gray, the two EV values will be identical. If someone uses a mostly white background, their reported EV values will be about 2EV higher than the correct figure.
What about spot meter mode?
04-06-2011, 09:17 PM   #449
Veteran Member




Join Date: Mar 2009
Location: Midwest
Photos: Gallery
Posts: 1,407
QuoteOriginally posted by Class A Quote
Are you sure? Couldn't the values indicate absolute brightness levels independently of camera exposure settings? They could be a poor man's (reflected light) light meter?
I've been wondering this myself. One might presume that, for a correct exposure, ISO+f-stop+shutter speed (not additively, but from those bits of data) one might derive the EV of the scene within some fairly reasonable error margin. The opposite, say, of the EV charts we're all familiar with. For instance, EV1 appears to be the light level at which one obtains a proper exposure at ISO 3200 and 1/15 sec @f2 . This alone is enough to tell me when I'm testing in EV-1(ish) areas.

It would also seem that we can adjust exposure in post and obtain a reasonable (retroactive) view of the EV at the time, as Lightroom shows adjustment in EV.

Not a precision scientific instrument, mind you; just a guide. Just a thought.

That said, I'm pretty sure my K-5 is slightly slower than advertised ISO; using the Sunny 16 rule, I consistently see 1/4 to 1/2 stop underexposure, regardless of metering mode.
04-06-2011, 10:12 PM   #450
Site Supporter
Site Supporter




Join Date: Jan 2007
Location: Fallon Nevada
Photos: Albums
Posts: 504
QuoteOriginally posted by falconeye Quote
Hi Ray,

thanks for the feedback. I didn't notice the BF issue with 1.03 and my FA 31. It's actually in the testing data: with sufficient light, the FA 31 performed well, both under 1.02 and 1.03.

It's getting stranger and stranger.

Interesting, you observed a significant improvement with 1.03 and a black backdrop. What was the distance? Do you have an before/after 1.03 shot for the FA50 or FA31 with identical lighting? Raw preferred. I'd like to take a closer look.

A LR JPG export of the center region, levels, contrast, brightness, blacks all to zero; but sharpening set to 100%/0.5px radius would suffice though.
Falk,

Test distance was 25x FL (FA 50 f1.4 mounted) measured from the target test face to the white mark on the top of the camera body near the mode dial.

I was shooting jpeg out of the camera at the default settings on the body (setup was as received in other words) but right now I can only find the 1.01 EV2 jpeg (before) shot and not the (after) 1.03 EV2 jpeg.

I had been testing 14 lenses on 3 bodies and also had to shoot a company function or two since then so it may be that I no longer have the after shot other than the cropped/embossed version.

The light level was measured at the target face with a Sekonic L-358 and the color temp was measured at the same point using a Gossen color meter. The light level at EV2 on the Sekonic meter was just at or slightly above the area where the camera meter starts to blink, so I would estimate that the camera was seeing very close to EV0.

I did my best to not disturb anything between the two tests and I did manually focus the lens to infinity between each shot. I actually shot EV9, EV 4.5 and EV2 sequences 1.01 and 1.03 and all showed the same shift to BF after the update under equal test conditions.

If I find the after shot I will PM you and send you both.

After shooting the K5 a bit in real life situations since I did these tests, I have no doubt that the 1.03 update made a significant improvement in AF accuracy versus the 3 other K5 bodies I tested. That is not to say that it is perfect as it still inexplicably misses now and again, but when it does so now, it is "just" out versus the FF step I saw quite often before the upgrade.

I am also pretty sure that the 31 and 43 will need to go to Pentax with the K5 in order to correct the BF. My Sigma 18-50 f2.8 also BF significantly on the K5. All 3 could not be fully corrected with all AF adjust dialed in after the 1.03 update.

Ray
Reply

Bookmarks
  • Submit Thread to Facebook Facebook
  • Submit Thread to Twitter Twitter
  • Submit Thread to Digg Digg
Tags - Make this thread easier to find by adding keywords to it!
copy, ev, focus, front, issue, k-5, light, pentax, pentax news, pentax rumors, plane, study
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
Any tips for low light focus with K-5 designinme_1976 Pentax K-5 & K-5 II 5 11-21-2010 08:38 PM
focus hunting in low light sorin Pentax DSLR Discussion 12 07-17-2010 02:20 PM
Low Light auto focus JohnKSA Troubleshooting and Beginner Help 6 03-10-2010 04:19 AM
Pentax Low Light Focus indy1984 Pentax DSLR Discussion 15 02-17-2010 09:42 AM
EV low light focus question tarsus Photographic Technique 3 06-26-2008 08:22 AM



All times are GMT -7. The time now is 04:49 PM. | See also: NikonForums.com, CanonForums.com part of our network of photo forums!
  • Red (Default)
  • Green
  • Gray
  • Dark
  • Dark Yellow
  • Dark Blue
  • Old Red
  • Old Green
  • Old Gray
  • Dial-Up Style
Hello! It's great to see you back on the forum! Have you considered joining the community?
register
Creating a FREE ACCOUNT takes under a minute, removes ads, and lets you post! [Dismiss]
Top