The GetDPI Photography Forum

Great to see you here. Join our insightful photographic forum today and start tapping into a huge wealth of photographic knowledge. Completing our simple registration process will allow you to gain access to exclusive content, add your own topics and posts, share your work and connect with other members through your own private inbox! And don’t forget to say hi!

Why are LCCs Without Movements Not Symmetrical?

dchew

Well-known member
Ok, this has been bugging me forever and I finally have the guts to ask: I always thought symmetry was kind of important in physics. What is it about a pixel that is not symmetric?

Dave

35xl / IQ3100 no movements:
 

Wayne Fox

Workshop Member
yeah, sort of strange.

Of course, if you upgrade to the IQ4 150, almost all of that color cast will disappear ;) :LOL:
 

k-hawinkler

Well-known member
Ok, this has been bugging me forever and I finally have the guts to ask: I always thought symmetry was kind of important in physics. What is it about a pixel that is not symmetric?

Dave

35xl / IQ3100 no movements:
Yuck. :thumbdown:
 

dchew

Well-known member
Yuck. :thumbdown:
I know it looks bad, but they usually clean up pretty well in C1. Here is an example that is a pretty good test:


I'm not complaining so much as just really curious. Bill Claff describes something on his 2D FT page that might be relevant to all this, I just have to find time to do more digging.

Dave
 

dchew

Well-known member
My guess is higher temperature of some components near the sensor.

M
Could be. It would have to be temperature effects that target photons at crazy Schneider obtuse angles, not straight-on Rodenstock photons. Hmm. I wonder how temperature affects micro lenses or sensor well walls.

Dave in the Weeds
 

JimKasson

Well-known member
Ok, this has been bugging me forever and I finally have the guts to ask: I always thought symmetry was kind of important in physics. What is it about a pixel that is not symmetric?

Dave

35xl / IQ3100 no movements:
Not sure how you set this up, but I'm guessing we're looking at CFA crosstalk. CFA's aren't symmetric.

Jim
 

Paul2660

Well-known member
The 35XL really should go down in history. An amazing lens and I am so glad I kept mine back. The 32 Rodie surpasses it in many areas but has strong retro focus distortion and has to be handled so delicately due to the mass in front of the shutter. And it’s heavy. I guess the new aperture only versions will be less fragile.

The 35XL results on the IQ4 are impressive. Many times you can shoot without the CF if you can keep the ISO down.

Paul C.
 

dchew

Well-known member
Not sure how you set this up, but I'm guessing we're looking at CFA crosstalk. CFA's aren't symmetric.

Jim
Thanks, Jim. Do you mean the array itself being rggb isn’t symetric, or the way it is interpreted isn’t symetric or both? Why is it more uniform in the long dimension vs the short dimension?

Dave
 
Is the LCC a lens cast correction, correcting anomalies in the lens, or a sensor cast correction, correcting anomalies in the sensor? It would seem to be the latter, since it is not needed (or needed less) with the IQ4.
 

JimKasson

Well-known member
Thanks, Jim. Do you mean the array itself being rggb isn’t symetric, or the way it is interpreted isn’t symetric or both? Why is it more uniform in the long dimension vs the short dimension?

Dave
I mean the array itself is not symmetric. The way the colors look on a flat field image is going to depend on which filters spill over on which pixels, and on the white balance of the illuminant. Maybe to some degree on the demosaicing algorithm. Certainly a lot on the lens design and aperture. A classic CFA crosstalk case is the "Italian Flag" and that sure isn't radially symmetric.

As has been said, going to a BSI sensor pretty much eliminates this effect, but that may not be convenient or cost-effective for you. You may be reduced to flat-fielding corrections in post.

Jim
 

Paul2660

Well-known member
You still need the LCC with movements for any tech wide. At least I prefer it. However the color cast is really non existent or 95 percent gone on all wides so now a LCC from one shoot can apply to all images from that particular lens. The biggest problem with color cast was how it varied between shots due to angle of light, amount of light, amount of pure blue or any other color.

The freedom of not having to capture an LCC after each shot is wonderful.

The P1 process works well on all cameras and I use it all drone photography as DJI has some really harsh color cast in their cameras. Correction varies since it’s impossible to capture a LCC after a shot in flight. But it’s worth the extra step.

Paul C
 

dchew

Well-known member
Is the LCC a lens cast correction, correcting anomalies in the lens, or a sensor cast correction, correcting anomalies in the sensor? It would seem to be the latter, since it is not needed (or needed less) with the IQ4.
Definitely not the lens that is asymmetric. Since I can rotate the lens every 90 degrees, I tried that...

 

dchew

Well-known member
I mean the array itself is not symmetric. The way the colors look on a flat field image is going to depend on which filters spill over on which pixels, and on the white balance of the illuminant. Maybe to some degree on the demosaicing algorithm. Certainly a lot on the lens design and aperture. A classic CFA crosstalk case is the "Italian Flag" and that sure isn't radially symmetric.

Jim
I get that the four squares of the bayer array are not symmetric. Certainly it is easier for red and blue to spill into green than it is for red to spill into blue because they only touch at one point (and vice versa). What I don't get is if I fly down from my mini-helicopter and land on a sensor, I cannot tell which way is North, South, East or West unless I know precisely which row and column I am on. No matter which direction I face, I can easily get into a position where there is a red filter behind and to my left, a blue filter in front to my right and two greens in the opposite corners. All I have to do is walk around one or two pixels. That's why it hurts my head to see the global difference in color cast across the short dimension.

I suppose the sensor does know precisely which row and column each pixel is. Maybe that is the point.
:banghead:

Dave
 
Last edited:

Boinger

Active member
What camera is this on?

I have a similar issue when I shot lcc's but I thought it was because of slight deviation in parallelism.
 

dchew

Well-known member
What camera is this on?

I have a similar issue when I shot lcc's but I thought it was because of slight deviation in parallelism.
Alpa STC. I’ve done some reasonably critical testing of each corner by both rotating the lens and by using Jim Kasson’s corner-recompose approach. I am quite sure the lens plane is parallel with the sensor. The 35xl doesn’t have room for the tilt adapter so that isn’t in the mix either.

Dave
 

onasj

Active member
Interesting, and of course noticeable even just in shifted photos taken with FSI sensors. Perhaps it arises from a macro-scale asymmetry in sensor architecture (not an individual pixel-scale or Bayer filter element-scale or micro lens-scale asymmetry)?

For long-exposure captures taken from my Nikon, D810 there was a magenta cast on the bottom edge that I was told arose from “amp glow” (a misnomer, since in modern CMOS sensors I don’t think there are any unintegrated amps) from an asymmetric feature of the circuitry near the sensor that generated a bit of hit or IR radiation that was picked up at very long exposures by the near edge of the sensor. I’m sure the LCC asymmetry is caused by a different phenomenon, but I wonder if the basic concept—that some large-scale asymmetry in a feature of sensor architecture can cause different regions of the sensor to pick up different color aberrations—is also responsible.
 

MGrayson

Subscriber and Workshop Member
Could be. It would have to be temperature effects that target photons at crazy Schneider obtuse angles, not straight-on Rodenstock photons. Hmm. I wonder how temperature affects micro lenses or sensor well walls.

Dave in the Weeds
I missed the part about lens dependency. Never mind.

M
 
Top