Index  •   • New posts  •   • RTAT  •   • 'Best of'  •   • Gallery  •   • Gear  •   • Reviews
New posts  •   • RTAT  •   • 'Best of'  •   • Gallery  •   • Gear  •   • Reviews
Register to forums    Log in

 
FORUMS Canon Cameras, Lenses & Accessories Canon EOS Digital Cameras
Thread started 11 Apr 2008 (Friday) 21:20
Prev/next
sponsored links
(this ad will go away when you log in as registered member)

An Easier UniWB for EOS Digital Cameras (at least the XT)

 
E-K
Senior Member
981 posts
Joined Sep 2006

Canada
MORE INFO

There are a few threads here and elsewhere on the internet discussing UniWB. Briefly, UniWB provides a mechanism that results in no (or very little) adjustment to the individual colour channels when the camera creates the embedded JPEG on which the camera's histogram is based.

This in theory provides a more accurate representation of what is actually happening with the raw data -- useful when trying to expose to the right.

Guillermo Luijk provides an excellent analysis at http://www.guillermolu​ijk.com/article/uniwb/​index.htmexternal link.

Now if you run PhotoBola's Raw Image Analysis tool on the sample UniWB images in the above link you will notice that for the areas of interest the raw values for each of the channels are essentially equal.

This is effectively what we are striving for. When you set a custom WB in the camera, you are asking the camera to figure out how it needs to adjust the raw data for the provided custom WB image to make these channels equal (i.e. white/gray). If the channels equal each other in the CWB image, than the camera thinks it doesn't need to adjust anything as for the given "lighting condition" the image is already gray.

G. Luijk's method does work, but it requires a little effort to execute. I'm lazy, so I looked for an easier solution ;).

Now I can think of two other "naturally" occurring scenarios where channels will have equal values (or approximately equal): a dark frame and a completely saturated frame.

Using the PB Raw Image Analysis tool on a dark frame and saturated frame returned average RGGB values of (255, 255, 255, 255) and (4095, 4095, 4095, 4095) respectively. The dark frame had more variance in values whereas the saturated frame was 4095 all the way through.

So far so good. Now the only thing to see was if the camera would take these images for use as a custom WB. On the Rebel XT, the answer is yes in both cases.

Using the dark frame resulted in WB coefficients for an image of (1.006, 1.000, 1.006). The saturated frame resulted in WB coefficients for an image of (1.008, 1.000, 1.004).

The obvious benefit with the dark frame is that you can do it at any point in case you lost your UniWB image/setting -- just shoot a frame with the lens cap on. The saturation method is a little more difficult as sometimes it can be difficult to achieve full saturation in all channels for some lighting conditions. For mine, I used a flash.

I would be interested to know if any other camera models can apply the same method.

e-k

Apr 11, 2008 21:20



LOG IN TO REPLY
sponsored links
(this ad will go away when you log in as registered member)
tdodd
Goldmember
tdodd's Avatar
3,731 posts

Joined Jun 2006

Essex, UK
MORE INFO

I tried your approach described above and found that the dark image gave results close to the uniWB balance - certainly the green cast is there - but the white image seems to have been recorded using a default WB that is equivalent to the daylight balance.

Here are comparative shots in LR and DPP of the white and black calibration images and a "real world" scene photographed with my uniWB, the light-uniWB and the dark-uniWB. Unfortunately the light conditions altered just a little from shot to shot - we have mixed sun and cloud here at the moment so light levels are fluctuating. I'll have to perform a more controlled test to really check the results out properly.

HOSTED PHOTO
please log in to view hosted photos in full size.

HOSTED PHOTO
please log in to view hosted photos in full size.
Apr 12, 2008 07:27



LOG IN TO REPLY
tzalman
Fatal attraction.
tzalman's Avatar
12,927 posts

Joined Apr 2005

Gesher Haziv, Israel
MORE INFO

Very clever, E-K. On the basis of the 40D histogram I can't see much difference from panopeeper's CWB, using the blackframe method.

Apr 12, 2008 08:42

Elie / אלי

LOG IN TO REPLY
RebelTasha
Junior Member
21 posts
Joined Apr 2008
MORE INFO

This is waaay over my head but it sounds interesting...

Apr 12, 2008 08:47

Rebel XT Flickrexternal link
Rebel 35mm Portfolioexternal link

LOG IN TO REPLY
E-K
THREAD ­ STARTER
Senior Member
981 posts
Joined Sep 2006

Canada
MORE INFO

tdodd wrote in post #5312923external link
I tried your approach described above and found that the dark image gave results close to the uniWB balance - certainly the green cast is there - but the white image seems to have been recorded using a default WB that is equivalent to the daylight balance.

Do you mean the recorded raw values for the saturated image are not (4095, 4095, 4095) or that it did but when you used it for CWB it produced a result similar to daylight balance?

It's a pain to get the channels fully saturated. I had to use ISO 1600 and a flash on a white wall. ISO 100 was problematic as was trying to use a CF bulb (blue channel wouldn't saturate).

Thanks for giving it a try.

e-k

Apr 12, 2008 14:21



LOG IN TO REPLY
_GUI_
Senior Member
_GUI_'s Avatar
342 posts

Joined Aug 2007

Madrid (Spain)
MORE INFO

E-K, I am Guillermo Luijk. Congratulations, you have achieved in about 10secs what took me a whole night to generate, and in addition to this I achieved a more precise UniWB for my Canon 350D with a saturated shot:

multipliers 1.006856 1.000000 1.005877 1.000000

0.69% deviation for the R channel is the max error, the closest to a perfect neutral WB I can achieve (probably because of some hot pixels or whatever).

It works fine for my camera. Please tell me your name so I can credit you in the original article.

Apr 12, 2008 14:27

http://www.guillermolu​ijk.comexternal link to subscribe click hereexternal link

LOG IN TO REPLY
Perry ­ Ge
Batteries? We don't need no...
Perry Ge's Avatar
12,176 posts
GALLERY: 67 photos

Joined Jul 2007

Hong Kong
MORE INFO

Oh that is COOOOOL. Thanks a lot for sharing this - perhaps the biggest challenge in exposing to the right is getting the embedded JPEG histogram to accurately reflect the channels in RAW, and this is a cool solution.

Apr 12, 2008 14:33

Perry | www.perryge.comexternal link |flickrexternal link | 500pxexternal link | C&C always welcome | Market Feedback & Gear | Sharpening sticky | Perspective sticky

LOG IN TO REPLY
tdodd
Goldmember
tdodd's Avatar
3,731 posts

Joined Jun 2006

Essex, UK
MORE INFO

E-K wrote in post #5314703 (external link)
Do you mean the recorded raw values for the saturated image are not (4095, 4095, 4095) or that it did but when you used it for CWB it produced a result similar to daylight balance?

It's a pain to get the channels fully saturated. I had to use ISO 1600 and a flash on a white wall. ISO 100 was problematic as was trying to use a CF bulb (blue channel wouldn't saturate).

Thanks for giving it a try.

e-k

Here are screen prints from rawanalyze. As far as I can tell I have blown everything. I attempted to white balance within the rawanalyze software and got perfect multipliers of 1.0000 for all channels.

When I used ExifTool to check out the white balance data that had been recorded in the exif - the data which I believe is actually used by software to work out white balance adjustments - the "measured" white balance values were exactly equivalent to those for daylight.

HOSTED PHOTO
please log in to view hosted photos in full size.

HOSTED PHOTO
please log in to view hosted photos in full size.
Apr 12, 2008 14:35



LOG IN TO REPLY
tdodd
Goldmember
tdodd's Avatar
3,731 posts

Joined Jun 2006

Essex, UK
MORE INFO

Here is the RGB WB data in the exif information from the blown image (first attachment). You will note that the RGGB values are identical for "Auto", "Measured" and "Daylight".

The same data from my custom uniWB file (second attachment) clearly shows the nearly equal values for "Measured", which are very different from the "Daylight" values.

HOSTED PHOTO
please log in to view hosted photos in full size.

HOSTED PHOTO
please log in to view hosted photos in full size.
Apr 12, 2008 14:36



LOG IN TO REPLY
tdodd
Goldmember
tdodd's Avatar
3,731 posts

Joined Jun 2006

Essex, UK
MORE INFO

I've looked further with rawnalyze and, if I'm reading the stats correctly, it seems that every single one of my sub-pixels is at 4071. I don't know the significance of that information. I guess it means that every single sub-pixel was as saturated as could be, but for some reason the sub-pixels top out at 4071 instead of 4095. What consequences would that have for attempting an easy uniWB with oversaturation?

Here's a screen print....
.
.

HOSTED PHOTO
please log in to view hosted photos in full size.
Apr 12, 2008 15:18



LOG IN TO REPLY
_GUI_
Senior Member
_GUI_'s Avatar
342 posts

Joined Aug 2007

Madrid (Spain)
MORE INFO

tdodd wrote in post #5314963external link
I've looked further with rawnalyze and, if I'm reading the stats correctly, it seems that every single one of my sub-pixels is at 4071. I don't know the significance of that information. I guess it means that every single sub-pixel was as saturated as could be, but for some reason the sub-pixels top out at 4071 instead of 4095. What consequences would that have for attempting an easy uniWB with oversaturation?

Don't be surprised for this, some cameras do not saturate at the max value of their expected integer range (2^n-1 for a n-bits camera). Some saturation points:

- Canon 350D: 4095
- Canon 30D: 3398
- Canon 5D: 3919
- Canon 40D: 13823

The important thing is that R=G=B in your RAW file so the UniWB should be properly calculated; try it. Then shoot anything and check the resulting multipliers (this can be done with DCRAW doing: dcraw -v -w foto.cr2)

Apr 12, 2008 15:22

http://www.guillermolu​ijk.comexternal link to subscribe click hereexternal link

LOG IN TO REPLY
E-K
THREAD ­ STARTER
Senior Member
981 posts
Joined Sep 2006

Canada
MORE INFO

_GUI_ wrote in post #5314737external link
It works fine for my camera. Please tell me your name so I can credit you in the original article.

Glad it works for you. I'm not worried about the credit but if you want to put something then just say e-k on the POTN forum or some such.

e-k

Apr 16, 2008 07:46



LOG IN TO REPLY
E-K
THREAD ­ STARTER
Senior Member
981 posts
Joined Sep 2006

Canada
MORE INFO

tdodd wrote in post #5314776external link
Here is the RGB WB data in the exif information from the blown image (first attachment). You will note that the RGGB values are identical for "Auto", "Measured" and "Daylight".

The same data from my custom uniWB file (second attachment) clearly shows the nearly equal values for "Measured", which are very different from the "Daylight" values.

Sorry to dredge this up, but when you used the saturated images as the CWB image. For images you shot using it as the CWB, what was the "As Shot RGGB Levels" set to?

e-k

May 07, 2008 08:15



LOG IN TO REPLY
tdodd
Goldmember
tdodd's Avatar
3,731 posts

Joined Jun 2006

Essex, UK
MORE INFO

I just shot a fresh blown image with my 40D (the previous efforts above, which seemed not to work properly with the blown image, were shot with my 30D) - 30" exposure at f/5.6 and 200 ISO, waving the lens in the general direction of the sun on a clear day. I then used this image to set a custom WB and the camera warned me that the image might not be suitable for WB but I went ahead anyway and then fired a sample "real world" shot. Here are the results....

Blown image
Raw Image Analyse gives RGB coefficients of 1.0000, 1.0008, 1.0015
Exif Tool give "As Shot" values of 1045, 1039, 1039, 1052

Real world image with blown custom WB
Exif Tool give "As Shot" values of 1024, 1037, 1037, 1024

Maybe these figures suggest that the blown uni-WB does work after all, at least for the 40D.

Here is the image and histogram for the blown image and the "real world" image. I did not take care to maximise exposure with the real world shot - I just pointed the camera and centred the exposure meter manually.

HOSTED PHOTO
please log in to view hosted photos in full size.

HOSTED PHOTO
please log in to view hosted photos in full size.
May 07, 2008 08:57



LOG IN TO REPLY
tdodd
Goldmember
tdodd's Avatar
3,731 posts

Joined Jun 2006

Essex, UK
MORE INFO

Here's the real world image after setting the WB to the daylight preset in Lightroom. No other edits/adjustments.

HOSTED PHOTO
please log in to view hosted photos in full size.
May 07, 2008 09:00



LOG IN TO REPLY
sponsored links
(this ad will go away when you log in as registered member)

39,604 views & 0 likes for this thread
An Easier UniWB for EOS Digital Cameras (at least the XT)
FORUMS Canon Cameras, Lenses & Accessories Canon EOS Digital Cameras

NOT A MEMBER YET? CLICK HERE TO REGISTER TO THE FORUMS
Registered members get all the features: search, more forums, private messaging, thread follow, notifications, own gallery, own reviews...




Send feedback to staff    •   Jump to forum...    •   Rules    •   Index    •   New posts    •   RTAT    •   'Best of'    •   Gallery    •   Gear    •   Reviews    •   Polls

COOKIES DISCLAIMER: By using this site you agree that some cookies will be stored on your browser. For unlogged users we store one session id cookie. For registered members we store (in addition to login session cookie) only cookies that are essential for required functionality. We do not store any personal tracking data in cookies or other browsers' data storage methods.

POWERED BY AMASS 1.33version 1.33
made in Finland
by Pekka Saarinen
for photography-on-the.net
Spent 0.0017 for 4 database queries.
PAGE COMPLETED IN 0.03s
Latest registered member is robpopper
771 guests, 644 members online
Simultaneous users record so far is 5175, that happened on Jun 16, 2015