Pages: [1]   Go Down

Author Topic: Measuring vignetting: trouble to achieve an evenly lit surface  (Read 2272 times)

Guillermo Luijk

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2005
    • http://www.guillermoluijk.com

I have been trying to measure vignetting for several Canon FDn lenses at max aperture:






But the asymmetry of the obtained profiles denotes a not 100% evenly lit measuring surface, which de-centres the vignetting pattern:



The luminosity gradient introduces an error in the measured profile. Any tips to minimise this issue?.

The calculations were easily done in R for anyone interested:

http://www.guillermoluijk.com/elmomentodecisivo/vignetting.R

Regards
« Last Edit: July 20, 2016, 04:10:32 pm by Guillermo Luijk »
Logged

EricV

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 270
Re: Measuring vignetting: trouble to achieve an evenly lit surface
« Reply #1 on: July 20, 2016, 04:37:50 pm »

One simple trick would be to rotate the camera 180 degrees, capture another image of the same target with the same framing, then average the two images in your processing.  Not a complete solution, but in practice this should greatly reduce the effect of uneven illumination.  You could also construct a difference image to evaluate just how much asymmetry there is in your illumination.
Logged

EricV

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 270
Re: Measuring vignetting: trouble to achieve an evenly lit surface
« Reply #2 on: July 20, 2016, 04:41:42 pm »

Another technique would be to capture an additional image with the lens stopped way down, to minimize vignetting.  Use this reference image to normalize the original test image.
Logged

Guillermo Luijk

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2005
    • http://www.guillermoluijk.com
Re: Measuring vignetting: trouble to achieve an evenly lit surface
« Reply #3 on: July 20, 2016, 06:04:49 pm »

One simple trick would be to rotate the camera 180 degrees, capture another image of the same target with the same framing, then average the two images in your processing.

Thanks Eric. What about rotating the single capture itself? in fact three rotations (left-right, upside down, left-right + upside down) generating four final images to be averaged, should provide a perfectly centred maximum (or nearly).

If the gradient is sufficiently linearly uniform it should work.

Regards.



www.guillermoluijk.com

NancyP

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2513
Re: Measuring vignetting: trouble to achieve an evenly lit surface
« Reply #4 on: July 20, 2016, 07:24:12 pm »

Why? I have never tried to make my own Adobe Lens Profile, but for making astro program positive control frames, I just get a big piece of quite translucent 1/8" thick "lucite" off an old radiology light box, point the camera more or less at the sun, hold the lucite fairly close to the lens so the FOV of lens contains only lucite, and expose x several, later averaging the frames. (done this way because I have the ingredients already and don't have to buy anything. Or the simpler way might be a good diffusion filter in your filter-holder. I don't know how even the white-balance lens caps plastic is, but that might be good.
Logged

EricV

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 270
Re: Measuring vignetting: trouble to achieve an evenly lit surface
« Reply #5 on: July 20, 2016, 09:19:45 pm »

Rotating a single captured image will of course force the averaged image to be symmetric and centered, but that is probably not what you want.  You want to simulate a symmetric illumination source, not a lens with symmetric vignetting.  Imagine for example that the illumination is uniform but one corner of your lens vignettes badly.  Symmetrizing that one image will give you the wrong answer.  If lens vignetting is known to be symmetric, your single image method will work.

As Nancy suggests, these are just mathematical work-arounds, when it may be better to put the effort into creating a more uniform illumination source.  But where is the fun in that :)

Nancy, I trust that your translucent illumination is radially symmetric, but why would it not be brighter in the center, where the scattering angle is smallest?  If you place the sun off-center, do you still get the same result?
Logged

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8913
Re: Measuring vignetting: trouble to achieve an evenly lit surface
« Reply #6 on: July 21, 2016, 04:08:43 am »

As Nancy suggests, these are just mathematical work-arounds, when it may be better to put the effort into creating a more uniform illumination source.  But where is the fun in that :)

Ah, but the rewards ...

Quote
Nancy, I trust that your translucent illumination is radially symmetric, but why would it not be brighter in the center, where the scattering angle is smallest?  If you place the sun off-center, do you still get the same result?

A diffuser like Nancy describes is a more or less Lambertian diffuser. Personally, I use 2 sheets of opaline glass taped together at the edges, to avoid yellowing after a long time of light exposure and temperature changes. As long as the light that reaches the surface of the diffuser is relatively uniform, the out of focus (infinity focused lens) image of that diffuser, held flush to the lens barrel/hood, is very uniform.

That should leave the lens vignetting and light fall-off as the only variable, then modified a bit by the sensor's pixel shading, but that's mostly constant for lenses with a similar exit pupil position.

I use such a diffuser plate for creating a Lens Cast Correction (LCC) file for Capture-One, which then corrects the slightly different sensor response for different colors, and the vignetting and light fall-off, as the angle of incidence gets more oblique towards the corners of the sensor.

Cheers,
Bart
Logged
== If you do what you did, you'll get what you got. ==
Pages: [1]   Go Up