The color cast correction saga

[Hannes Lummes]

Advertisement (gone after registration)

I have lately been thinking about why exactly was there a color cast correction issue, and why did it take so long to correct....*partially*.

 

This is my logic now:

 

1. As anyone knows the correcting mask is pretty straightforward calculation. How is it possible to design it badly?

2. I am pretty sure it was done right. BUT:

3. The micro lens form has undergone some minor mods. Perhaps unintended.

4. I strongly suspect the answer lies in not-so-controlled factors, because:

5. After downloading the new firmware and testing with 28/2.8 asph, I can see generally quite OK result, but extreme right side seems to now get a bit *too much* magenta removal, resulting in slight cast to green along the border. Aperture makes no difference. This makes sense, there seems to be some disagreement among users about how good the new correction is.

6. Why, well,perhaps because every sensor is a bit different. The micro lens structure might be a little too sensitive to manufacturing tolerances, and Leica has had to find an average. (Makes no sense to blame the lenses, I'm sure they are standard enough.)

 

Totally another issue and a very troubling one is *why did Leica initially tell that the ray angle issues are gone* and *why did a certain early tester claim that they really are gone* What exactly has happened after the late 2012?

 

Yes, I am a bit disappointed, but mainly I am just a very curious mind, and would like to understand why Leica can miss such an easy signal processing thing. My answer is sensor variability and/or changes.

 

Wild card: Was the data calculated from design or actually "measured"? (Just like you create correction masks in various software)

[tobey bilek]

Corner fix.

 

Camera software is for one aperture only. You will need to correct the others manually.

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[Guest porco1962]

I have lately been thinking about why exactly was there a color cast correction issue, and why did it take so long to correct....*partially*.

 

This is my logic now:

 

1. As anyone knows the correcting mask is pretty straightforward calculation. How is it possible to design it badly?

2. I am pretty sure it was done right. BUT:

3. The micro lens form has undergone some minor mods. Perhaps unintended.

4. I strongly suspect the answer lies in not-so-controlled factors, because:

5. After downloading the new firmware and testing with 28/2.8 asph, I can see generally quite OK result, but extreme right side seems to now get a bit *too much* magenta removal, resulting in slight cast to green along the border. Aperture makes no difference. This makes sense, there seems to be some disagreement among users about how good the new correction is.

6. Why, well,perhaps because every sensor is a bit different. The micro lens structure might be a little too sensitive to manufacturing tolerances, and Leica has had to find an average. (Makes no sense to blame the lenses, I'm sure they are standard enough.)

 

Totally another issue and a very troubling one is *why did Leica initially tell that the ray angle issues are gone* and *why did a certain early tester claim that they really are gone* What exactly has happened after the late 2012?

 

Yes, I am a bit disappointed, but mainly I am just a very curious mind, and would like to understand why Leica can miss such an easy signal processing thing. My answer is sensor variability and/or changes.

 

Wild card: Was the data calculated from design or actually "measured"? (Just like you create correction masks in various software)

 

I think most of the variation of correnction are coming from sesor variation especially misalignment of microlens and sensor pixel.

 

But to say frankly, back focus length of M-mount is too short compared to that of SLR digital camera, so the ray angle coming into sensor surface becomes too sharrow especially in the corner are and it's difficult to change ray direction with microllens only, so artifacts arrises.

 

So if very small misalignment happens on the sensor, you can easily see the difference.

 

I think the best way to correct this is to use LCC function in Capture One PRO. It perfectly removes any color casting when you open raw file.