Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
12-04-2018, 04:43 PM   #1
Junior Member




Join Date: Dec 2018
Location: Ghent, East-Flanders
Posts: 26
Possible bug in firmware K1 with manual white balance and pixel shift?

Hi everyone, I'm new here and I am a very happy owner of a K1 (mk1).
Let me first thank everybody here on pentaxforums who spend (a lot of) time to help other people, that's awesome an I really appreciate this so much, I've learned a lot this way.

After many weeks of experimenting with all the possibilities of my K1, tonight I stumbled on a problem that I cannot fix or explain!
When I was trying to use the manual white balance in our living room with very warm colored LED light indoors, the results after taking a shot to measure with a white piece of paper and store the new white balance, were simply outstanding and as neutral as could be. So the manual way to measure the color temperature works incredibly well.

However, when I switched on pixel shift (with and without mc) to,test if Incould,invrease details and lower the noise, with the exact same settings, the results became a very yellowish image. And I mean really yellow, not a warm white balance. The histogram confirmed this color because the blue channel was almost completely pushed to the left.
When I swithed pixel shift of again, colors were perfect again, repeatable every time.
Even developing that RAW file in-camera caused a very strange behaviour. While developing the RAW pixel shifted image with the original settings and watching it in the background behind the settings, the colors looked again perfect, but the resulting jpg after I saved the image also became as yellow as could be, exactly like the pixel shifted RAW looked. The histogram was way of for the blue channel again, just as this was also the case with the pixel shifted RAW as it was recorded. When I removed the pixel shift setting in the RAW settings of the developers screen and saved the jpg, the colors of the image became perfectly normal.
So to me it lookes like the (or my?) K1 does something wrong while calculating a pixel shifted image, only while using a self created manual white balance.

Can anyone do the same test and confirm if this is a serious bug, or that it is only my camera that shows this strange yellow color.
I was only able to see this behaviour with manual white balance, not any other setting.
I like to use pixel shift but also the manual white balance and since it seems now I am not able to use that combination.

Thank you all very much in advance, have a good night (00.25pm here)
Greetings, Sven, Belgium

Gear, LX, K1, dfa 28-105mm, A*85mm f1.4, A50mm f1.2, A24mm f2.8, dfa 50mm f2.8 macro, 15mm f3.5

12-04-2018, 04:50 PM   #2
Administrator
Site Webmaster
Adam's Avatar

Join Date: Sep 2006
Location: Arizona
Photos: Gallery | Albums
Posts: 47,787
I would look at the 4 individual pixel shift frames. If the lighting appears different in any of them, this could explain the weird results when they are merged.

An easy way to test this would be to redo everything outdoors in natural light.

Adam
PentaxForums.com Webmaster (Site Usage Guide | Site Help | My Photography)



PentaxForums.com server and development costs are user-supported. You can help cover these costs by donating. Or, buy your photo gear from our affiliates, Adorama, B&H Photo, or Topaz Labs, and get FREE Marketplace access - click here to see how! Trusted Pentax retailers:

12-04-2018, 05:07 PM   #3
Otis Memorial Pentaxian
Loyal Site Supporter
stevebrot's Avatar

Join Date: Mar 2007
Location: Vancouver (USA)
Photos: Gallery | Albums
Posts: 31,344
QuoteOriginally posted by Svenneke Quote
When I was trying to use the manual white balance in our living room with very warm colored LED light indoors
Welcome to the Pentax Forums!

This sort of problem is reported on this site every so often in regards to household LED illumination and LED stage lights. Many LED lamps vary in spectral output on a set frequency such that a camera may "see" light that is biased differently for an actual exposure than for a white balance exposure or even other photos in a sequence. Whether there is a problem depends on the lamp. As Adam suggested, trying the same thing in natural light or with photo lamps should work better.


Steve
12-04-2018, 05:08 PM   #4
Pentaxian
photoptimist's Avatar

Join Date: Jul 2016
Photos: Albums
Posts: 2,726
What you describe may be an artifact of LED lighting which typically emits a wildly shifting color spectrum in time with AC power.

Try the test again (including doing the white balance with a white piece of paper) outdoors at sunset (reddish lighting) or sky-lit shade (bluish lighting) to see if it still does weird things.

12-04-2018, 05:25 PM   #5
Junior Member




Join Date: Dec 2018
Location: Ghent, East-Flanders
Posts: 26
Original Poster
Goodnight Adam, for me at least, and thank you very much for your quick answer.

I had about the same idea in mind. Tomorrow after work, if I find the time, I hope to do the same test during daylight. I will also will try to look in Rawtherapee what the results look like with pixel shift settings there.

The lighting indeed, as you suggest, does look very strange when I look at the 3 color histograms on the display of my K1, only with the pixel shift enabled.
The blue channel is almost completely pushed together on the left side of the histogram when I compare it to the non pixel shifted version. The green one on the other hand is strechted in width and is a little wider in range, compared to the non pixel shifted one, and the red channel seems almost identical to the non pixel shifted.

When I have tested this further with daylight, I will come back to you for feedback.
Thank you so far and I wish you a good night, or day depending on your location, I must go to bed now.
Sven
12-04-2018, 07:11 PM - 1 Like   #6
Loyal Site Supporter
Loyal Site Supporter
pschlute's Avatar

Join Date: Mar 2007
Location: Surrey, UK
Photos: Gallery
Posts: 2,406
Try using a shutter speed slower than 1/60 and see if that makes a difference. Certainly with fluorescent light it does. It is about capturing all the light output during the electricity cycle.
12-05-2018, 01:24 AM   #7
Junior Member




Join Date: Dec 2018
Location: Ghent, East-Flanders
Posts: 26
Original Poster
Good morning everybody, thank you for all tour help and suggestions.

The lights are IKEA dimmable LED lights with a color from cool to warm. I know that at short shutter speeds these lights causes weard striped patterns in my images and I understand the reason why that happens.
But the shutter speed in my tests from yesterday evening was 3 seconds, with and without pixelshift. Without pixelshift, there are no stripes or bands visible with that shutter speed, but with pixel shift, my blue channel becomes so narrow and skewed to the left side of the histogram that it becomes unusuable.
I try to upload some pictures of my display so you can see how it looks.

The first one (zoomed out) clearly shows the different color rendering with pixel shift. (The first of the five images is with automatic white balance to show how hard it was to get the colors right). Image two to five are all taken with the same manual white balance setting, no ps, with ps, no ps, with ps.
The second image looks normal on my display and the white balance is very neutral because of the manual setting after a measurement.
The third one does not look what I expected to be. The only difference is that I enabled pixel shift.
I hope this helps more to explain what is happening.

Now I must hurry to work, thanks everybody for your time. Tonight I try to do a test with a natural light source outside as asked by a few of you.
This site is great and you all are great people, realy!

Greetings Sven
Attached Images
View Picture EXIF
TG-5  Photo 
View Picture EXIF
TG-5  Photo 
View Picture EXIF
TG-5  Photo 
12-05-2018, 01:58 AM   #8
pid
Senior Member




Join Date: Jan 2010
Photos: Gallery
Posts: 268
if you wish stable Color with LED Light try the Rotolights

12-05-2018, 02:06 AM   #9
Junior Member




Join Date: Dec 2018
Location: Ghent, East-Flanders
Posts: 26
Original Poster
Hi all, some additional info.
While taking the pixel shifted shots, my display always shows four perfectly normal color pictures, not thise strange yellow ones.
Only after the camera has calculated the four images together, the strange yellow color become visible.
This only happens when in manual white balance mode, not with any other mode I've tested. Always with a long shutter speed of minimum 3 seconds.
have a beautiful day!
Greetings, Sven
12-05-2018, 04:49 AM   #10
Junior Member




Join Date: Dec 2018
Location: Ghent, East-Flanders
Posts: 26
Original Poster
QuoteOriginally posted by pid Quote
if you wish stable Color with LED Light try the Rotolights
Thank you PID,
But the color is stable, even with these cheap IKEA LED lights, but only of I don't use pixel shift in combination with manual white balance.
See my attached pictures and the weird histogram change when activating pixel shift.
All other settings were exactly the same.

Have a nice day.
Greetings,
Sven

---------- Post added 12-05-18 at 04:52 AM ----------

QuoteOriginally posted by pschlute Quote
Try using a shutter speed slower than 1/60 and see if that makes a difference. Certainly with fluorescent light it does. It is about capturing all the light output during the electricity cycle.
Thank you pschlute, as I wrote above, the shutter time was 3 seconds and the non pixel shifted version looked perfect.
Have a nice day.

Greetings,
Sven

---------- Post added 12-05-18 at 05:04 AM ----------

QuoteOriginally posted by photoptimist Quote
What you describe may be an artifact of LED lighting which typically emits a wildly shifting color spectrum in time with AC power.

Try the test again (including doing the white balance with a white piece of paper) outdoors at sunset (reddish lighting) or sky-lit shade (bluish lighting) to see if it still does weird things.
Hello photoptimist,
Also a big thank you for your time and answer.
I made a test this morning with the tungsten continuous light of our bearded dragon and as expected, the pixel shifted version was all right this time.

I wonder why this only happens after merging the four images together and only with manual white balance, when using LED lights?
Could the software in-camera screw things up when calculating the four perfect looking shots?
When I look at my diplay of the K1 while taking these pixel shift shots, they all have a perfect and neutral, accurate color, it is only after doing the calculations the combination becomes yellow as you can see in the pictures I posted above.

Enjoy your day.
Greetings,
Sven
12-05-2018, 05:58 AM   #11
Junior Member




Join Date: Dec 2018
Location: Ghent, East-Flanders
Posts: 26
Original Poster
QuoteOriginally posted by stevebrot Quote
Welcome to the Pentax Forums!

This sort of problem is reported on this site every so often in regards to household LED illumination and LED stage lights. Many LED lamps vary in spectral output on a set frequency such that a camera may "see" light that is biased differently for an actual exposure than for a white balance exposure or even other photos in a sequence. Whether there is a problem depends on the lamp. As Adam suggested, trying the same thing in natural light or with photo lamps should work better.


Steve
Hello Steve, thank very much you for your time.

I understand what you mean. But I wonder, if I first manually measure the color temperature with the camera and store that value, shouldn't that setting be a constant value when calculating the same, but pixel shifted version?
Look at the very obvious shift in the histogram when I activate pixel shift.
As I wrote above to photoptimist, using a continuous light source such as that tungsten llght I described don't causes any problem at all.

I know tha LEDlights can cause strange patterns (or colors) when using short shutter speeds due to the frequency of emitting light, but why does it not show up in the four recordings during exposure on my K1 display, but only after the math is done?

Thank for all your help everybody, have a nice day.

Greetings,
Sven
12-05-2018, 08:54 AM   #12
Junior Member




Join Date: Dec 2018
Location: Ghent, East-Flanders
Posts: 26
Original Poster
Hello everybody,

As some members asked and promised, I did a test outside with daylight and it works as it should, with and without pixel shift in combination with manual white balance.

After that I went back inside and took the same picture with a combination of the IKEA LED lights and daylight that still was entering the living room. No problems at all.

As a third test I closed the curtains completely and tested it with only the LED's and those yellow color was back after calculating the pixel shift, but it was not visible during the four exposures on my screen. These four exposures looked perfect again, until the calculation was finished and then the same images combined were as yellow as could be.

As long as the manual white balance was mixed with enough daylight, the results stayed normal, but as soon as I only had the LED's as light source, the histogram for the blue channel was completely skewed to the left allost leaving no data at all, resulting in a yellow image.

I will not investigate this further, but I want to warn everybody that this can happen under certain circumstances. If this happens, simply use any other white balance setting that fits the needs, or disable pixel shift. These results are reproducable every time with my K1. Perhaps someone can do the test and see if it is an exception with my camera or if it happens in all camera's. (Updated tot firmware 1.43)

Happy shooting everybody and thank you all for the help.
Greetings,
Sven
12-05-2018, 08:55 AM   #13
Otis Memorial Pentaxian
Loyal Site Supporter
stevebrot's Avatar

Join Date: Mar 2007
Location: Vancouver (USA)
Photos: Gallery | Albums
Posts: 31,344
QuoteOriginally posted by Svenneke Quote
Hello Steve, thank very much you for your time.

I understand what you mean. But I wonder, if I first manually measure the color temperature with the camera and store that value, shouldn't that setting be a constant value when calculating the same, but pixel shifted version?
Look at the very obvious shift in the histogram when I activate pixel shift.
As I wrote above to photoptimist, using a continuous light source such as that tungsten llght I described don't causes any problem at all.

I know tha LEDlights can cause strange patterns (or colors) when using short shutter speeds due to the frequency of emitting light, but why does it not show up in the four recordings during exposure on my K1 display, but only after the math is done?

Thank for all your help everybody, have a nice day.

Greetings,
Sven
As you have determined, the results you get with a particular LED lamp may change from exposure to exposure with higher shutter speeds being more prone to showing unusual color casts and/or banding. Manually determining the white balance with the camera from a gray card or from a gray/white object in the frame and then fine tuning on the rear LCD will give you a set value, but remember that it was derived from and is being applied to other than a consistent light source. You may do better with using AWB where the adjustment is done based on analysis of the actual capture data.

As for why the processed pixel-shift has a color cast while the individual exposures appeared not to, that is a difficult question to answer without doing some fine analysis on the four exposures* and/or trying to replicate using a neutral subject such as a blank wall or 18% gray card and analyzing the four exposures for spectral composition. Even attempting to suggest possible reasons would take an abundance of words. As @photoptimist suggested above, the output from an LED lamp is very complex.


Steve

* The four exposures may be extracted from a PEF or DNG RAW file using the PDCU software that came with the camera or other tools such as dcraw.
12-05-2018, 09:38 AM   #14
Junior Member




Join Date: Dec 2018
Location: Ghent, East-Flanders
Posts: 26
Original Poster
QuoteOriginally posted by stevebrot Quote
As you have determined, the results you get with a particular LED lamp may change from exposure to exposure with higher shutter speeds being more prone to showing unusual color casts and/or banding. Manually determining the white balance with the camera from a gray card or from a gray/white object in the frame and then fine tuning on the rear LCD will give you a set value, but remember that it was derived from and is being applied to other than a consistent light source. You may do better with using AWB where the adjustment is done based on analysis of the actual capture data.

As for why the processed pixel-shift has a color cast while the individual exposures appeared not to, that is a difficult question to answer without doing some fine analysis on the four exposures* and/or trying to replicate using a neutral subject such as a blank wall or 18% gray card and analyzing the four exposures for spectral composition. Even attempting to suggest possible reasons would take an abundance of words. As @photoptimist suggested above, the output from an LED lamp is very complex.


Steve

* The four exposures may be extracted from a PEF or DNG RAW file using the PDCU software that came with the camera or other tools such as dcraw.
Hello Steve, thank you very much for your answer.

AWB was not possible because the LED's emmit a very warm color, so warm that even with AWB the colors were much too warm and far for the neutral white that manual white balance privided me with.

In the mean time I did the test to open the (corrupt looking?) yellow files in my software of choice, RawTherapee.

The miniatures that are visible straight from the camera show the same yellow colors and indeed again only with the pixel shifted versions.
However, when I open these files in the developers panel in RawTherapee they look perfect and are exactly like the non pixel shifted versions, leaning perfect white balance and absolutely not a hint of tha yellow color it shows in-camera.

I opened all four individual images of the ps version in RawTherapee and they all are perfectly exposed and colored.
So it seems that my camera is rendering or interpreting something wrong when combining these four, but the RAW data in the DNG files are fine and perfectly usable.

That is good to know because now I know that these faulty yellow looking images do contain the right data in the RAW file and are perfectly usable, only not straight out of the camera.
Perhaps PENTAX should look into that if they can reproduce these circumstances?

Thank you for your time and effort to help me.

Greetings,
Sven
12-05-2018, 10:12 AM - 1 Like   #15
Junior Member




Join Date: Dec 2018
Location: Ghent, East-Flanders
Posts: 26
Original Poster
QuoteOriginally posted by Svenneke Quote
Hello Steve, thank you very much for your answer.

AWB was not possible because the LED's emmit a very warm color, so warm that even with AWB the colors were much too warm and far for the neutral white that manual white balance privided me with.

In the mean time I did the test to open the (corrupt looking?) yellow files in my software of choice, RawTherapee.

The miniatures that are visible straight from the camera show the same yellow colors and indeed again only with the pixel shifted versions.
However, when I open these files in the developers panel in RawTherapee they look perfect and are exactly like the non pixel shifted versions, leaning perfect white balance and absolutely not a hint of tha yellow color it shows in-camera.

I opened all four individual images of the ps version in RawTherapee and they all are perfectly exposed and colored.
So it seems that my camera is rendering or interpreting something wrong when combining these four, but the RAW data in the DNG files are fine and perfectly usable.

That is good to know because now I know that these faulty yellow looking images do contain the right data in the RAW file and are perfectly usable, only not straight out of the camera.
Perhaps PENTAX should look into that if they can reproduce these circumstances?

Thank you for your time and effort to help me.

Greetings,
Sven
Dear all,

I noticed another no so normal behaviour. When I opened the file browser in RawTherapee, I saw the original pixel shifted files as yellow miniatures, just like on the display of my K1. As soon as I opened and developed them, they seemed normal and afterwards after exporting a jpg, without making any changes on the RAW data, the miniatures looked normal. See my 2 attached print screens begore and after.
Maybe someone has any idea what can cause this behaviour and solve this problem.

Since these yellow colors are only visible in the camera and the minitaures, I am inclined to think it is a rendering problem. The RAW files are perfectly normal.

Have a good night.
Greetings
Sven

---------- Post added 12-05-18 at 10:15 AM ----------

Sorry, the pictures are reversed in order, first the yellow without opening, afterwards the neutral version after saving without any change.
IPad's are not that handy for posting things...
;-)

---------- Post added 12-05-18 at 10:59 AM ----------

Hi everybody,

Determined to solve the reason why this happens I found out that it solely depends on which color the IKEA LED's are set.
The IKEA TRÅDFRI dimmable led's can be switched between warm white, neutral and cool white.
The strong yellow color as seen in the images above, are only visible when the LED TRÅDFRI is set to warm white, even if I have made a manual white balance beforehand. As soon as I change it to neutral or cool white, the problem went completely away. Color temperature is alway set with a new corresponding manual white balance. The results with the last two settings are perfect and there is no color shift whatoever, the same long shutter speeds are used to prevent the influence of the 50Hz flicker.

For some strange reason, the in-camera rendering goes wrong when set to a manual measurement with the (very) warm white IKEA TRÅDFRI LED's. I could not reproduce this with any other lights available.

Thank you all for thinking with me and suppose ideas.
Have a good night.
Greetings from Belgium
Attached Images
   
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!
balance, behaviour, camera, color, colors, custom, data, day, dfa, file, files, greetings, image, jpg, light, miniatures, pentax help, photography, pixel, settings, shift, shutter, test, time, troubleshooting
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
Pixel shift with pixel artefacts Dericali Pentax K-1 4 07-18-2018 12:57 PM
Lloyd Chambers Prefers Pentax Pixel Shift To Sony's Pixel Shift Fenwoodian Pentax K-1 2 07-16-2018 02:15 PM
K1 II Dynamic Pixel Shift as stabilizer ? madbrain Pentax K-1 11 04-28-2018 01:04 PM
How do Pentax K1 pixel shift images compare with Fuji GFX images? Fenwoodian Pentax K-1 75 04-06-2017 06:31 AM
K1/Samyang 24mm Tilt/Shift and Pixel Shift DDoram Pentax K-1 3 06-25-2016 02:05 PM



All times are GMT -7. The time now is 06:19 AM. | 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