Jump to content
UltravioletPhotography

Strange white balance results with my GH-1?


Andrea B.

Recommended Posts

Hi everyone -

 

When I was using my nifty Panasonic Lumix GH-1 to shoot a long filter series for you (Filter Series: Scarecrow, Insect Hotel and Foliage - GH1 - WB & Raw Comp, I observed that the GH-1 in-camera white balance measurement against a Spectralon rectangle was not what I expected. I'll repeat here two photos from that long page to show you what I'm trying to figure out.

 

This first example, in Infrared, is made using the longpass Red + Infrared filter glass Schott RG695. We use this filter to shoot IR with some nice bits of false colour after channel swapping or some white balance technique. The RG695 raw recording (upper left) has a pale orange appearance indicating that the red channel predominates, followed closely by the green then the blue channel. When the GH-1 measures WB against the Spectralon, I get the expected blue false colours shown in the upper right. When I convert the raw file in Photo Ninja and use its white dropper, I get the expected blue false colours shown in the lower finished photograph. So, both the GH-1 and the Converter produce the same white balance. This IR result was also true for the other Infrared-pass filters I used in the test.

 

Both the GH-1 (upper right) and the Converter (bottom) produce the same IR white balance false colours.

rg695.jpg

.

.

.

Now, the second example, in Ultraviolet, is made using the BaaderU. The raw recording (upper left) has the typical false colours of a substrate peaking around 360nm or so indicating that the red and green channels predominate followed by the blue. But, when the GH-1 measures WB against the Spectralon, I get this dark cyan false colour. When I convert the raw file in Photo Ninja and drag the white dropper across a UV-neutral portion of the photo or, alternately, when I use a white balance profile made from Spectralon standards, I get the expected gray/blue/yellow false colours. So, the GH-1 and the Converter produce different white balances. WHY? This UV result was also true for the other Ultraviolet-pass filters I used in the test.

 

The GH-1 (upper right) and the Converter (bottom) produce different UV white balance false colours. WHY?

baaderU.jpg

Link to comment

Might be a problem related to computer&mathematics?

 

You calculate (simplified) e.g. the new white balanced blue Bwb from the original 3 colors Ror Bor & Gor by multiplying them with factors fR, Fb and Fg: Bwb = Ror x fR + Bor x Fb + Gor x Fg

 

Now when one of your input value is very small (from the blue channel) and has to meet up with the other two colors (to get white) your multiplying factor needs to be large. Then depending on your calculating processor & input data length (8 bit, 14bit 16bit, 32 bit, 64bit) you can get quite different results with respect to even smaller differences in your input data.

 

I see this effect, when I try to do a white balance on a (not yet white balanced) picture coming with 8bit or with 16bit. With a 16bit, the results are better.

Link to comment
  • 2 months later...

I can't know the source of the white balance error in the GH1, but I can say for certain that the GH1 will not correctly measure a white balance under some of my filters. I am posting this note because for a long time I thought it was something *I* was doing wrong.

 

This remark is about the Lumix GH1 only. That was an early model. I do not know how later converted Lumix cameras might handle white balance under UV-pass or other filters.

 

I'll also note that my converted Pentax K1 also does not correctly measure white balance with some filters.

 

I suppose the take-away here is that you must verify your in-camera white balance measurement in a good converter which has a white balance tool capable of handling large shifts in colour. Not all converters can do this. To ensure both camera and converter are providing the best white balance, you must measure and convert against a diffuse white balance standard which is stable under UV, Vis or IR illumination. As Werner notes in the preceding post, staying with 16-bit conversions is also useful, perhaps mandatory, for the best white balance outcomes.

 

[Kindly do remember that making a white balance adjustment, per se, is not mandatory in UV/IR photography. We perform it here on UVP for our documentary photos in order to illustrate uniform results across different camera/lens platforms.]

Link to comment

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
×
×
  • Create New...