Originally posted by BigMackCam I assume you used an official lens profile rather than manually correcting the vignetting yourself? If so, it looks like whoever developed the lens correction profile didn't test it for accuracy
So far as my limited understanding goes, f/4
at any focal length will let in one stop less of light than f/2.8 at any focal length. Or have I misunderstood your point?
photoptimist was much quicker on the draw, and provided a much better explanation!!!
Originally posted by photoptimist That rule about apertures collecting the same amount of light regardless of focal length is true for diffuse or area-illuminated subjects (sky glow, walls, etc.).
But point sources like stars are different. A 30 mm lens at f/4 (7.5 mm absolute aperture) collects one stop more of starlight than a 15 mm lens at f/2.8 (5.4 mm absolute aperture).
For diffuse or area-illuminated subjects, the larger absolute aperture of a longer focal length lens also collects more light but then the magnification effect of the lens just spreads that light over more pixels cancelling the effect. For point sources on a dark background the larger absolute aperture of a longer focal length lens collects more light but the magnification does not cancel the light gathering effect because the star is a point.
P.S. I don't think this panel effect is vignetting. It's too linear and of constant width from top-to-bottom. This looks more like a resampling artifact.
Morning - it's 4am here, well now 5. I got up to get a drink of water, and had a nagging suspicion that perhaps another stitcher would do a better job - and perhaps what was appearing was an artifact of the stitching process, re-sampling error. Sure enough, PTGui has a much better result, no resampling problems, better color transmission through the process, among others. See the new example below.
Also, thanks!!! - an excellent description of the aperture effect with point sources.
Originally posted by BigMackCam Very interesting! Thanks for the clarification. So... dumb question... does this mean at 30mm f/4, sky-glow would be captured darker than at 15mm f/2.8, but stars (as point light sources) would be captured brighter?
I did wonder if it might be something other than vignetting. You'd expect to see rounded shading in the corners of each panel for that (right?), and it's not evident here... just vertical lines.
- Yes - I would think that the stars would be brighter.
Yes, again you would think that the corners would be darker, however due to how the light falls up at the very top of the resulting image the sky is substantially darker, due to the lack of stars (Milky Way being lower in the frame), so there is a natural fall off of light. Having said that, re-running the stitching in PTGui, shows a substantial improvement in the stitching, lack of artifacts (resampling errors, etc.). If I remember correctly, PTGui in the past (several years ago) had a major update in terms of how it handled astro imaging just due to producing similar results. I have been on the edge for at least 5 to 7 years on what stitcher to buy, PTGui or Autopano. This is pushing me off the mark to move away from ICE and make a decision.
Originally posted by PJ1 Vignetting or not, you have a fabulous photo location.
It's a very nice and picturesque location. It has a lot of appeal, sunsets are spectacular here and the scenery alone just makes the location. The state park occupies the the absolute perfect vantage point. As time passes, the population growth from Phoenix just continues to grow, putting pressure on the outlying areas. The light pollution here is just getting worse by the year.
I'm going back to bed..... Thanks, for solving the re-sampling artifact faster than what I was doing in my sleep.
_________________________
Just one more thought. The Sigma 18-35/f1.8 lens does cover the full frame sensor starting around 28mm all the way to 35mm. At 35mm f1.8, there is at least another stop of light to be recovered. I had tested that lens on the K1 before getting the 15-30 and run in to the same "patch work quilt" problem - re-sampling error with ICE. So, now with this, I have another excellent lens to use.