ColorChecker Digital SG - Which patch to use to create a White Balance?

[Dinarius]

I can't find the answer to this anywhere.

Which patch do I use to create a white balance with this chart?

Many thanks.

D.

[daicehawk]

I can't find the answer to this anywhere.

Which patch do I use to create a white balance with this chart?

Many thanks.

D.

It depends on how you expose and what camera you use. If you ETTR to the clipping and your camera is not linear in the top range then any mid-grey patch will go since spectrally all neutral patches of SG must be flat -the white, grey and black. Note however that the illuminantion across the target must be very even both spectrally and luminancewise.

[Bart_van_der_Wolf]

I can't find the answer to this anywhere.

Which patch do I use to create a white balance with this chart?

F5 should work, but there are relatively neutral patches all along the edge of the grid as well.

Cheers,
Bart.

[scyth]

I can't find the answer to this anywhere.

Which patch do I use to create a white balance with this chart?

Many thanks.

D.

http://xritephoto.com/ph_product_overview.aspx?ID=938&Action=Support&SupportID=5884

patch L* a* b* for new (>= 11/2014) SG

H6   60.04   0.09   0.05
K6   70.6   -0.24   0.07
G5   65.22   -0.27   0.16
I6   75.36   0.35   0.26
F5   79.65   -0.08   0.62

if you believe X-Rite data of course (nobody knows how exactly they measured and then it was, I bet, just one specimen)

[Alexey.Danilchenko]

I can't find the answer to this anywhere.

Which patch do I use to create a white balance with this chart?

RawDigger authors recommend G5 patch for white balancing

[scyth]

RawDigger authors recommend G5 patch for white balancing

G5 was probably for pre Nov 2014 target (with all the fine print about the measurements), based on its brightness (reflectance) vs deviation from neutrality.

E10   49.7   -0.19   0.01
A3   49.7   -0.18   0.03
B1   49.72   -0.19   0.02
K10   49.72   -0.19   0.02
N2   49.69   -0.18   0.03
A6   49.66   -0.2   0.01
B10   49.69   -0.2   0.01
E1   49.74   -0.19   0.03
H1   49.75   -0.21   0.01
H10   49.7   -0.2   0.02
N5   49.72   -0.18   0.05
H5   49.71   -0.2   0.03
K1   49.79   -0.2   0.03
M1   49.77   -0.19   0.04
M10   49.68   -0.19   0.05
A9   49.72   -0.2   0.04
N8   49.7   -0.19   0.07
K7   45.59   -0.05   0.23
E6   6.66   0   -0.3
A8   6.49   -0.02   -0.28
G5   65.05   0   -0.32
I5   35.26   -0.09   -0.24
N3   6.39   -0.04   -0.33
N9   6.47   0   -0.38
K8   20.3   0.07   -0.32
K6   70.76   0.07   -0.35
N6   6.49   -0.03   -0.41
I6   75.16   0.25   -0.2

PS: counting angels on the proverbial pinhead

[Bart_van_der_Wolf]

G5 was probably for pre Nov 2014 target (with all the fine print about the measurements), based on its brightness (reflectance) vs deviation from neutrality.

And there is the issue of brightness as well. White balancing is preferably done with a relatively bright patch, which also automatically improves the Signal to Noise Ratio of the measurement.

Cheers,
Bart

[Alexey.Danilchenko]

And there is the issue of brightness as well. White balancing is preferably done with a relatively bright patch, which also automatically improves the Signal to Noise Ratio of the measurement.

White balancing has less to do with brightness and more with neutrality. Perhaps if the raw converter use one click correction (you click the patch, it adjusts WB and exposure) then the brightness matter

[scyth]

And there is the issue of brightness as well. White balancing is preferably done with a relatively bright patch, which also automatically improves the Signal to Noise Ratio of the measurement.

Cheers,
Bart

one might also add the line about "WB tool" used - how many pixels it is averaging which depends on the software used (1x1, 3x3, 5x5, etc)

[Alexey.Danilchenko]

G5 was probably for pre Nov 2014 target (with all the fine print about the measurements), based on its brightness (reflectance) vs deviation from neutrality.

In those numbers yes H6 looks better but I prefer to measure my targets with Spectrolino or i1 Pro

[scyth]

with Spectrolino

and how do you keep reheating the lamp there (in the said spectrolino) to deal with tungsten deposits while measuring the target like SG ?

[Doug Gray]

And there is the issue of brightness as well. White balancing is preferably done with a relatively bright patch, which also automatically improves the Signal to Noise Ratio of the measurement.

Cheers,
Bart

Yes, that's why I use the brightest gray patch. Another issue with the SG is specular reflection of a tinted illuminant that can be from a colored, bright surface in the environment that happens to line up with reflective SG surface. That effect is particularly important to avoid when using the SG to create a camera profile as it can desaturate and shift colors on the darker patches a large amount.

[Alexey.Danilchenko]

and how do you keep reheating the lamp there (in the said spectrolino) to deal with tungsten deposits while measuring the target like SG ?

I don't (I average several target measurements) - does it have the same problem like i1 Pro v2? Any measurements/data to support that?

[scyth]

I don't (I average several target measurements) - does it have the same problem like i1 Pro v2? Any measurements/data to support that?

yes, it has a noticeable drift, you saw that topic = http://forum.luminous-landscape.com/index.php?topic=108238

for i1Pro* there are API calls (or workarounds) to instruct the spectrometer to heat its lamp from software on demand

[Alexey.Danilchenko]

yes, it has a noticeable drift, you saw that topic = http://forum.luminous-landscape.com/index.php?topic=108238

What I saw there was this statement from Graeme

I've seen similar effects on all my i1Pro's, and also Spectrolino's,
although none as dramatic as the Rev E was.

And my experiments do not show a large discrepancies benween the measurements on my Spectrolino and i1 Pro rev D that I have.

[scyth]

What I saw there was this statement from Graeme

And my experiments do not show a large discrepancies benween the measurements on my Spectrolino and i1 Pro rev D that I have.

that topic has Spectrolino test with graph... and iPro2 test with graph... " b* " drift for Spectrolio is big.

[Alexey.Danilchenko]

that topic has Spectrolino test with graph... and iPro2 test with graph... " b* " drift for Spectrolio is big.

Have you read the post with follow up Spectrolino diagrams after that one? Where calibrations were performed?

With Spectrolino automated measurements it does force recalibrations fairly frequently - possibly to account for the drift.

[scyth]

Have you read the post with follow up Spectrolino diagrams after that one? Where calibrations were performed?

which post ? I am referring to the graphs in the following post http://forum.luminous-landscape.com/index.php?topic=108238.msg892565#msg892565

With Spectrolino automated measurements it does force recalibrations fairly frequently - possibly to account for the drift.

with spectroscan table you most probably keep the lamp heated good enough by measuring printed target patch by patch in a quick succession - you can't do the same manually with SG...

whether standard calibration helps or not like the heating request - I think we can do the test... say execute calibration every 10th (or may be even every other) measurement... I need to dig out my Spectrolino... what is clear though that you can't calibrate once before measuring the whole big target

PS: what is also interesting to test is to see how different is drift say between repeated measurements of dark patch (allegedly the lamp is on for a longer duration) and bright patch (allegedly the lamp is on for a shorter duration)

[Alexey.Danilchenko]

which post ? I am referring to the graphs in the following post http://forum.luminous-landscape.com/index.php?topic=108238.msg892565#msg892565

That one - and it shows b* drift of 0.15.

with spectroscan table you most probably keep the lamp heated good enough by measuring printed target patch by patch in a quick succession - you can't do the same manually with SG...

whether standard calibration helps or not like the heating request - I think we can do the test... say execute calibration every 10th (or may be even every other) measurement... I need to dig out my Spectrolino... what is clear though that you can't calibrate once before measuring the whole big target

Why cannot I do the same manually with SG? Not as fast but not far enough

[scyth]

Why cannot I do the same manually with SG? Not as fast but not far enough

does it work (as "fast" as strip reading by sliding spectrometer manually over printed target via ruler) with non flat targets (SG patches are sunken) ? I do not have the table, so may be I am wrong...