LumoLabs: Pentax K-5 low light focus study is published

Ray Pulley · 03-06-2011, 10:15 AM

Originally posted by dlacouture

AWB is out of the equation, as it is computed afterwards from the image taken...
A manually selected WB could be used, but I doubt it...

Here's a blurb on Nikon's color metering:

Nikon Imaging | Nikon D Technology THE POWER TO CHANGE YOUR PHOTOGRAPHY

The color metering system clearly is designed to set exposure, WB, and according to Nikon, improve AF, which makes perfect sense.

Before all the boo-birds start, yes I am aware of the AF iconsistencies with the new D7000.

The Samsung wavelength grating device might have advantages over simply using a color metering chip, but I am not sure what those would be. I suspect that it is simply a work-around for patents related to color metering that sets AWB and helps with AF tuning.

It also seems to me that just setting an average offset based upon manually chosen WB would be good enough in many cases. After all, fast action situations that move from tungsten to daylight are probably a tiny percentage of the shooting scenarios most will encounter.

Even an offset based upon a WB reading taken off of a WB reference like the WhiBal card would be fine. It could be stored as a user setting and called up quickly.

Methinks the designers (users too) are often just too enamored of the "do everything for me" camera to include simple manual fixes that could well work for almost everyone.

Ray

falconeye · 03-06-2011, 10:34 AM

Originally posted by Ray Pulley

Samsung, applied for 2009, issued 2010:

WAVELENGTH DETECTING APPARATUS AND FOCUS DETECTING APPARATUS HAVING THE SAME

I suspect that this is a bit closer to what is in the K5 than the older Hoya patent.

Ray,

thanks for the link but I beg to disagree.

The layout seems to be the standard for metering stuff. Pentax, in their patent, describes a way around an earlier 2003 Canon patent (US 7595475 B2). IMHO, Pentax does it the Canon way. The Canon patent even describes a 7x9 metering sensor with 21 embedded colorimetric sensors while, I believe, Pentax uses 7x11 metering sensors with 11 embedded colorimetric sensors. They patented another way to do it (separating both sensors) in order to avoid Canon claims but I think they didn't use it in the SAFOX+ system. Heck, they may just use their own patent to lower any license fee to Canon

In their patent, Pentax honors the necessity to measure color at every AF point. The Samsung device may or may not be able to do this: It could create a rainbow image of the scene which may be good enough to determine AF point color. Therefore the Samsung patent may be a way around the Canon patent which remains the most obvious way to do it. But the K-5 lacks the diffractor plate which is at the heart of the Samsung patent (I only see an IR cut filter which is normally used to replicate the AF module IR cut filter's behaviour). So no, I think Pentax does it the Canon way rather than the Samsung way (and does it matter, really?).

As a side note. I notice the Canon patent uses much simpler graphics. The work spent into preparing those patent documents probably is overblown.

Ray Pulley · 03-06-2011, 01:44 PM

Originally posted by falconeye

Ray,

thanks for the link but I beg to disagree.

The layout seems to be the standard for metering stuff. Pentax, in their patent, describes a way around an earlier 2003 Canon patent (US 7595475 B2). IMHO, Pentax does it the Canon way. The Canon patent even describes a 7x9 metering sensor with 21 embedded colorimetric sensors while, I believe, Pentax uses 7x11 metering sensors with 11 embedded colorimetric sensors. They patented another way to do it (separating both sensors) in order to avoid Canon claims but I think they didn't use it in the SAFOX+ system. Heck, they may just use their own patent to lower any license fee to Canon

In their patent, Pentax honors the necessity to measure color at every AF point. The Samsung device may or may not be able to do this: It could create a rainbow image of the scene which may be good enough to determine AF point color. Therefore the Samsung patent may be a way around the Canon patent which remains the most obvious way to do it. But the K-5 lacks the diffractor plate which is at the heart of the Samsung patent (I only see an IR cut filter which is normally used to replicate the AF module IR cut filter's behaviour). So no, I think Pentax does it the Canon way rather than the Samsung way (and does it matter, really?).

As a side note. I notice the Canon patent uses much simpler graphics. The work spent into preparing those patent documents probably is overblown.

Thanks for the comments.

Camera and autofocus apparatus

It looks like Oly pre-dates the Canon and Pentax work with a color sensor feeding back color temp infor to the AF system to correct the wavelength focus error issue.

Maybe they are all working around (and catching up to) Oly in this area? Either way, the issue is the same and the solution of applying a correction basd upon sensing color temp seems also to be consistent, so there really should be no reason this design concept cannot be made to work properly by Pentax.

One wonders why the long delay on a K5 fix given that?

I look forward to comparing a K-7 that is on its way to me to my K20 and the K5.

Edited to add: None of this matters, including all of the testing. The design is what it is, and Pentax will agree that there is a problem and fix it or not no matter what theories we dream up or discuss. The only thing that matters is sales revenue numbers. If Pentax perceives that the K5 revenue is being hurt enough by this, then they might address it. I am fast running out of faith on that point, however.

Ray

falconeye · 03-06-2011, 04:29 PM

Hi friends,

I upgraded the paper (as linked in the OP) to version 1.1. I did so to take all the new insight into account, esp. with respect of the inner workings of the colorimetric sensor.

You'll hopefully find some new and interesting stuff in the upgrade.

Enjoy,
Falk

Here are the links (from th OP) again for easier reference:

->Falk Lumo: LumoLabs: Pentax K-5 low light focus
www.falklumo.com/lumolabs/articles/k5focus (HTML version)
www.falklumo.com/lumolabs/articles/k5focus/K5Focus.pdf
(PDF version for download and offline reading)

ursamajor · 03-07-2011, 06:04 AM

Your study has already a celebrity, its own.

You'll find it in <link to RH removed>.
Enjoy Pentax's "qualities".

areidjr · 03-07-2011, 06:56 AM

Thanks for version 1.1. I gave it a read and there is lots of good info there.

I'm beginning to wonder if Pentax is going to fix this issue or just let it go as is. That also makes me wonder if the follow up camera will be any better. I never before had any doubts about my decision to invest in Pentax cameras and lenses, but I'm now having serious doubts as to the wisdom of my actions.

ogl · 03-08-2011, 08:56 AM

K-5 FF Issue - Good News: Pentax SLR Talk Forum: Digital Photography Review

ursamajor · 03-08-2011, 10:13 AM

I also had that 2 days ago call and the person promised to send me a solution to be received next day (so yesterday) as a priviliged PENTAX user.
One day later than promised I received a package from PENTAX with a brand new MAGLITE with a hot shoe connection ! Soo happy with it .

Lol, they began to bribe the customers.

simico · 03-09-2011, 07:57 AM

Interesting study, but I still don't understand how can my first and second K-5 be significantly different: first produced significant front focus that was obvious even on downsized pictures, under the same conditions with same subjects and same lenses the second K-5 doesn't produce noticeable front focus (at least I can't see it even at 100% pixel-peeping).
I'm not sure I want to install the new firmware when it comes out, I'm afraid it may screw my cam

Winder · 03-09-2011, 10:45 AM

Originally posted by simico

Interesting study, but I still don't understand how can my first and second K-5 be significantly different: first produced significant front focus that was obvious even on downsized pictures, under the same conditions with same subjects and same lenses the second K-5 doesn't produce noticeable front focus (at least I can't see it even at 100% pixel-peeping).
I'm not sure I want to install the new firmware when it comes out, I'm afraid it may screw my cam

The fact that the problem is not consistent across all K-5 cameras, and some are worse than others makes me wonder if it is a firmware issue. It sounds like some cameras might have a bad color meter.

Can the color meters used for AF be calibrated? Maybe the firmware will simply recalibrate the color meter for improved accuracy.

Until it is actually fixed........ who knows. This is why I am never a first adapter. I use to say wait until the 2nd or 3rd firmware release to decide on buying a camera, but now I am leaning towards the 4th or 5th.

jstevewhite · 03-09-2011, 11:09 AM

Just received my K-5 yesterday, and it exhibits the exact behavior described in this report. Even the 255um variance agrees with my experience, as the shorter the focal length the greater the focus error; my FA 50mm 1.4 focuses well in my office, while my 18-50 Sigma 2.8 does well at 50, but is off by several feet at 18mm. My 10-20 Sigma is off by a repeatable, RIDICULOUS amount; @ 10mm it FF by ~10 feet - focussed on the window, the markings on the barrel say it's focussed at between 1 and three feet (and images support that assertion). It's consistent and repeatable. In addition, it doesn't seem to affect Live View (as others have also stated).

I've already ordered a Katzeye screen for it; I dont' often use autofocus in low light levels with my K20D, either. *shrug*. It would be nice if the AF would work at those light levels, but I've never had a camera that did, so I'm supposing I won't miss it.

I would be really angry if the camera wasn't so stunning in every other way. Instead, I'm a bit put off and hoping they'll have a software fix for it.

areidjr · 03-09-2011, 11:13 AM

Originally posted by Winder

This is why I am never a first adapter. I use to say wait until the 2nd or 3rd firmware release to decide on buying a camera, but now I am leaning towards the 4th or 5th.

I bought the DL when it first hit the market. No problems and 1 FW update to accommodate SDHC. I bought the K100D Super the first week it was available. Zero FW updates and zero problems. I bought the K-5 in December and exchanged twice and finally giving up in February and returning for a refund.

I guess that makes me 2 for 3 as an early adopter.

jstevewhite · 03-09-2011, 11:28 AM

Originally posted by Winder

The fact that the problem is not consistent across all K-5 cameras, and some are worse than others makes me wonder if it is a firmware issue. It sounds like some cameras might have a bad color meter.

You just inspired me to ask for a replacement. If it doesn't affect all K-5s, then by definition, mine is defective. If it's *all* K-5s, then I'd still buy it - I own several manual focus lenses that I use frequently anyway, so I'm no stranger to doing it, but if some work properly, I want one

falconeye · 03-09-2011, 04:27 PM

Originally posted by simico

Interesting study, but I still don't understand how can my first and second K-5 be significantly different: first produced significant front focus that was obvious even on downsized pictures, under the same conditions with same subjects and same lenses the second K-5 doesn't produce noticeable front focus (at least I can't see it even at 100% pixel-peeping).
I'm not sure I want to install the new firmware when it comes out, I'm afraid it may screw my cam

The message about the forthcoming firmware fix relates back to Pentax Europe which the Greece distributor called (if info at DPR is correct). Let's see...

On sample variation: Within (my) standard theory (as laid out in v1.1 of the study), there is a simple mechanism which can cause the variation: How fast the colorimetric sensor signal decreases with fading luminosity. I then would expect the threshold to vary, not the focus shift as such. Also note that it isn't easy to repeat the experiment: Same EV level is only a small part in the overall equation.

mrjamesabels · 03-09-2011, 06:28 PM

The new K5 1.03 Firmware is available in pentax japan. Still waiting for the K-r firmware update.

03-06-2011, 10:15 AM	#301
Ray Pulley Veteran Member Join Date: Jan 2007 Location: Fallon Nevada Photos: Albums Posts: 504	Nikon Originally posted by dlacouture AWB is out of the equation, as it is computed afterwards from the image taken... A manually selected WB could be used, but I doubt it... Here's a blurb on Nikon's color metering: Nikon Imaging \| Nikon D Technology THE POWER TO CHANGE YOUR PHOTOGRAPHY The color metering system clearly is designed to set exposure, WB, and according to Nikon, improve AF, which makes perfect sense. Before all the boo-birds start, yes I am aware of the AF iconsistencies with the new D7000. The Samsung wavelength grating device might have advantages over simply using a color metering chip, but I am not sure what those would be. I suspect that it is simply a work-around for patents related to color metering that sets AWB and helps with AF tuning. It also seems to me that just setting an average offset based upon manually chosen WB would be good enough in many cases. After all, fast action situations that move from tungsten to daylight are probably a tiny percentage of the shooting scenarios most will encounter. Even an offset based upon a WB reading taken off of a WB reference like the WhiBal card would be fine. It could be stored as a user setting and called up quickly. Methinks the designers (users too) are often just too enamored of the "do everything for me" camera to include simple manual fixes that could well work for almost everyone. Ray

03-06-2011, 10:34 AM	#302
falconeye Veteran Member Join Date: Jan 2008 Location: Munich, Alps, Germany Photos: Gallery Posts: 6,871 Original Poster	Originally posted by Ray Pulley Samsung, applied for 2009, issued 2010: WAVELENGTH DETECTING APPARATUS AND FOCUS DETECTING APPARATUS HAVING THE SAME I suspect that this is a bit closer to what is in the K5 than the older Hoya patent. Ray, thanks for the link but I beg to disagree. The layout seems to be the standard for metering stuff. Pentax, in their patent, describes a way around an earlier 2003 Canon patent (US 7595475 B2). IMHO, Pentax does it the Canon way. The Canon patent even describes a 7x9 metering sensor with 21 embedded colorimetric sensors while, I believe, Pentax uses 7x11 metering sensors with 11 embedded colorimetric sensors. They patented another way to do it (separating both sensors) in order to avoid Canon claims but I think they didn't use it in the SAFOX+ system. Heck, they may just use their own patent to lower any license fee to Canon In their patent, Pentax honors the necessity to measure color at every AF point. The Samsung device may or may not be able to do this: It could create a rainbow image of the scene which may be good enough to determine AF point color. Therefore the Samsung patent may be a way around the Canon patent which remains the most obvious way to do it. But the K-5 lacks the diffractor plate which is at the heart of the Samsung patent (I only see an IR cut filter which is normally used to replicate the AF module IR cut filter's behaviour). So no, I think Pentax does it the Canon way rather than the Samsung way (and does it matter, really?). As a side note. I notice the Canon patent uses much simpler graphics. The work spent into preparing those patent documents probably is overblown. Last edited by falconeye; 03-06-2011 at 10:52 AM.

03-06-2011, 01:44 PM	#303
Ray Pulley Veteran Member Join Date: Jan 2007 Location: Fallon Nevada Photos: Albums Posts: 504	Oly Appears to Have Been First Originally posted by falconeye Ray, thanks for the link but I beg to disagree. The layout seems to be the standard for metering stuff. Pentax, in their patent, describes a way around an earlier 2003 Canon patent (US 7595475 B2). IMHO, Pentax does it the Canon way. The Canon patent even describes a 7x9 metering sensor with 21 embedded colorimetric sensors while, I believe, Pentax uses 7x11 metering sensors with 11 embedded colorimetric sensors. They patented another way to do it (separating both sensors) in order to avoid Canon claims but I think they didn't use it in the SAFOX+ system. Heck, they may just use their own patent to lower any license fee to Canon In their patent, Pentax honors the necessity to measure color at every AF point. The Samsung device may or may not be able to do this: It could create a rainbow image of the scene which may be good enough to determine AF point color. Therefore the Samsung patent may be a way around the Canon patent which remains the most obvious way to do it. But the K-5 lacks the diffractor plate which is at the heart of the Samsung patent (I only see an IR cut filter which is normally used to replicate the AF module IR cut filter's behaviour). So no, I think Pentax does it the Canon way rather than the Samsung way (and does it matter, really?). As a side note. I notice the Canon patent uses much simpler graphics. The work spent into preparing those patent documents probably is overblown. Thanks for the comments. Camera and autofocus apparatus It looks like Oly pre-dates the Canon and Pentax work with a color sensor feeding back color temp infor to the AF system to correct the wavelength focus error issue. Maybe they are all working around (and catching up to) Oly in this area? Either way, the issue is the same and the solution of applying a correction basd upon sensing color temp seems also to be consistent, so there really should be no reason this design concept cannot be made to work properly by Pentax. One wonders why the long delay on a K5 fix given that? I look forward to comparing a K-7 that is on its way to me to my K20 and the K5. Edited to add: None of this matters, including all of the testing. The design is what it is, and Pentax will agree that there is a problem and fix it or not no matter what theories we dream up or discuss. The only thing that matters is sales revenue numbers. If Pentax perceives that the K5 revenue is being hurt enough by this, then they might address it. I am fast running out of faith on that point, however. Ray Last edited by Ray Pulley; 03-06-2011 at 01:55 PM.

03-06-2011, 04:29 PM	#304
falconeye Veteran Member Join Date: Jan 2008 Location: Munich, Alps, Germany Photos: Gallery Posts: 6,871 Original Poster	Lumolabs study upgraded to version 1.1 Hi friends, I upgraded the paper (as linked in the OP) to version 1.1. I did so to take all the new insight into account, esp. with respect of the inner workings of the colorimetric sensor. You'll hopefully find some new and interesting stuff in the upgrade. Enjoy, Falk Here are the links (from th OP) again for easier reference: ->Falk Lumo: LumoLabs: Pentax K-5 low light focus www.falklumo.com/lumolabs/articles/k5focus (HTML version) www.falklumo.com/lumolabs/articles/k5focus/K5Focus.pdf (PDF version for download and offline reading)

03-07-2011, 06:04 AM	#305
ursamajor Banned Join Date: Jan 2011 Posts: 242	Your study has already a celebrity, its own. You'll find it in <link to RH removed>. Enjoy Pentax's "qualities". Last edited by Parallax; 03-07-2011 at 10:21 AM.

03-07-2011, 06:56 AM	#306
areidjr Veteran Member Join Date: Sep 2007 Location: Pittsburgh, Pa, USA Photos: Gallery Posts: 529	Thanks for version 1.1. I gave it a read and there is lots of good info there. I'm beginning to wonder if Pentax is going to fix this issue or just let it go as is. That also makes me wonder if the follow up camera will be any better. I never before had any doubts about my decision to invest in Pentax cameras and lenses, but I'm now having serious doubts as to the wisdom of my actions.

03-08-2011, 08:56 AM	#307
ogl Banned Join Date: Dec 2007 Location: Sankt Peterburg Photos: Gallery \| Albums Posts: 8,382	K-5 FF Issue - Good News: Pentax SLR Talk Forum: Digital Photography Review

03-08-2011, 10:13 AM	#308
ursamajor Banned Join Date: Jan 2011 Posts: 242	I also had that 2 days ago call and the person promised to send me a solution to be received next day (so yesterday) as a priviliged PENTAX user. One day later than promised I received a package from PENTAX with a brand new MAGLITE with a hot shoe connection ! Soo happy with it . Lol, they began to bribe the customers.

03-09-2011, 07:57 AM	#309
simico Veteran Member Join Date: Jan 2009 Location: Budapest Posts: 821	Interesting study, but I still don't understand how can my first and second K-5 be significantly different: first produced significant front focus that was obvious even on downsized pictures, under the same conditions with same subjects and same lenses the second K-5 doesn't produce noticeable front focus (at least I can't see it even at 100% pixel-peeping). I'm not sure I want to install the new firmware when it comes out, I'm afraid it may screw my cam

03-09-2011, 10:45 AM	#310
Winder Veteran Member Join Date: Apr 2010 Location: Tennessee Posts: 6,617	Originally posted by simico Interesting study, but I still don't understand how can my first and second K-5 be significantly different: first produced significant front focus that was obvious even on downsized pictures, under the same conditions with same subjects and same lenses the second K-5 doesn't produce noticeable front focus (at least I can't see it even at 100% pixel-peeping). I'm not sure I want to install the new firmware when it comes out, I'm afraid it may screw my cam The fact that the problem is not consistent across all K-5 cameras, and some are worse than others makes me wonder if it is a firmware issue. It sounds like some cameras might have a bad color meter. Can the color meters used for AF be calibrated? Maybe the firmware will simply recalibrate the color meter for improved accuracy. Until it is actually fixed........ who knows. This is why I am never a first adapter. I use to say wait until the 2nd or 3rd firmware release to decide on buying a camera, but now I am leaning towards the 4th or 5th.

03-09-2011, 11:09 AM	#311
jstevewhite Veteran Member Join Date: Mar 2009 Location: Midwest Photos: Gallery Posts: 1,407	Just received my K-5 yesterday, and it exhibits the exact behavior described in this report. Even the 255um variance agrees with my experience, as the shorter the focal length the greater the focus error; my FA 50mm 1.4 focuses well in my office, while my 18-50 Sigma 2.8 does well at 50, but is off by several feet at 18mm. My 10-20 Sigma is off by a repeatable, RIDICULOUS amount; @ 10mm it FF by ~10 feet - focussed on the window, the markings on the barrel say it's focussed at between 1 and three feet (and images support that assertion). It's consistent and repeatable. In addition, it doesn't seem to affect Live View (as others have also stated). I've already ordered a Katzeye screen for it; I dont' often use autofocus in low light levels with my K20D, either. shrug. It would be nice if the AF would work at those light levels, but I've never had a camera that did, so I'm supposing I won't miss it. I would be really angry if the camera wasn't so stunning in every other way. Instead, I'm a bit put off and hoping they'll have a software fix for it.

03-09-2011, 11:13 AM	#312
areidjr Veteran Member Join Date: Sep 2007 Location: Pittsburgh, Pa, USA Photos: Gallery Posts: 529	Originally posted by Winder This is why I am never a first adapter. I use to say wait until the 2nd or 3rd firmware release to decide on buying a camera, but now I am leaning towards the 4th or 5th. I bought the DL when it first hit the market. No problems and 1 FW update to accommodate SDHC. I bought the K100D Super the first week it was available. Zero FW updates and zero problems. I bought the K-5 in December and exchanged twice and finally giving up in February and returning for a refund. I guess that makes me 2 for 3 as an early adopter.

03-09-2011, 11:28 AM	#313
jstevewhite Veteran Member Join Date: Mar 2009 Location: Midwest Photos: Gallery Posts: 1,407	Originally posted by Winder The fact that the problem is not consistent across all K-5 cameras, and some are worse than others makes me wonder if it is a firmware issue. It sounds like some cameras might have a bad color meter. You just inspired me to ask for a replacement. If it doesn't affect all K-5s, then by definition, mine is defective. If it's all K-5s, then I'd still buy it - I own several manual focus lenses that I use frequently anyway, so I'm no stranger to doing it, but if some work properly, I want one

03-09-2011, 04:27 PM	#314
falconeye Veteran Member Join Date: Jan 2008 Location: Munich, Alps, Germany Photos: Gallery Posts: 6,871 Original Poster	Originally posted by simico Interesting study, but I still don't understand how can my first and second K-5 be significantly different: first produced significant front focus that was obvious even on downsized pictures, under the same conditions with same subjects and same lenses the second K-5 doesn't produce noticeable front focus (at least I can't see it even at 100% pixel-peeping). I'm not sure I want to install the new firmware when it comes out, I'm afraid it may screw my cam The message about the forthcoming firmware fix relates back to Pentax Europe which the Greece distributor called (if info at DPR is correct). Let's see... On sample variation: Within (my) standard theory (as laid out in v1.1 of the study), there is a simple mechanism which can cause the variation: How fast the colorimetric sensor signal decreases with fading luminosity. I then would expect the threshold to vary, not the focus shift as such. Also note that it isn't easy to repeat the experiment: Same EV level is only a small part in the overall equation.

03-09-2011, 06:28 PM	#315
mrjamesabels Senior Member Join Date: Jan 2010 Location: NEW JERSEY USA Photos: Gallery \| Albums Posts: 198	The new K5 1.03 Firmware is available in pentax japan. Still waiting for the K-r firmware update.

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