FLARE (a.k.a. VEILING GLARE or STRAY LIGHT):

Study of its effects

... and application to panorama photography

(sorry, no math here in this article)

 

Flare is for most photographers a dreaded phenomenon that affects more or less every photograph but it is fortunately really harmful only in some occasions. However the problem affects Panorama photography more often because bright sources of light (the sun or a sunlit window) sometimes cannot be avoided and shall be present somewhere in the e.g. 360 deg. spherical panorama.

The simultaneous and multiple causes of Flare are often not clearly distinguished and recognized by the photographer. This deficiency IMO somehow hampers the possibility to take the right measure that would be needed to possibly correct or reduce the problem.

By means of a typical example, I shall attempt to dissect the root cause of Flare (a.k.a. Veiling Glare). BTW the Opto-mechanical Designers call "Stray Light" this causal effect. It is in practice the same physical phenomenon but for them it is the name of a" disease" while the photographers just feel the" symptoms" from that disease.....

 

A pair of photos of the same scene as above was shot one hour later with the camera swiveled ~45 degrees to the left. The two images were then arranged (after annotation) in the following row of (1+ 5) images.

0- General view presentation

1- Ghost Images

2- Singly- Scattered Light

3- Multi-Scattered Light

4- Edge diffraction

7- Combination of Stray Light

Click on image to open a new page and read corresponding explanation

Note: You aloso may view: Shooting with other lenses a photograph of the same scene

What did the study results show?

1) The different causes of Stray Light that affect a photograph shot with a (D)SLR camera are in general located in the lens:

2) The present study has shown however that when a bright source of light is out of the field (but near it), the over-powering contribution to stray light may well then be the camera itself! This was AFAIK unheard of. The reader may view the effect with the mouse over (or click on) fig. 2 (title: "Singly-Scattered light") above. The general appearance of the phenomenon depends of course on the relative size and power of the bright source(s) of light. But and more important it depends also on which side of the camera the bright source(s) of light enters the lens. A strong supplementary non-uniform veiling glare is possibly accompanied by scattered glaring streak(s) of light. The latter shall be spectacular visible and perpendicular to one specific long side of the image (i.e. the side of the camera sensorthat adjoins the mirror underside).

The resulting supplementary stray light is clearly "Singly-Scattered Light" and it does not depend on the lens: it shall happen probably whatever the lens (*) is mounted on the DSLR camera, unless a specific light vane system is properly installed on the rear of that lens when it's possible (i.e. rarely **). Efficient baffling can possibly be put in front of the lens, out of the field (Light flags, the hand of the photographer shadowing the sun, etc.). The lens hoods that are available for (or are integral part of) the lens seem in fact to be generally inefficient to cope with the phenomenon:

In the course of extensive testing for Flare with about 15 different lenses (including 6 fisheye and 5 prime lenses) and with 4 different DSLR Canon cameras (including 2 APS-C cropped modes). A first lesson that I have learnt is: do not rely for sure on the lens hood when shooting "against the sun". Especially when it is out of the filed! As a matter of fact none of the hoods that I could use or that are permanently attached on the lens, protects from the adverse conditions discribed on fig 2 above for instance. Another lesson learnt is: when shooting against the sun, do not believe that WYSIWYG when viewing through the viewfinder of the DSLR. The camera chamber shall be not Flare aware the same way when the mirror wlill be flipped up: check in "Live View mode" when this is possible.

3) Knowing better the various causes of Flare should now help the panorama photographer in using the best ways to reduce veiling glare effect.

Special care shall be given when the bright source of light is just out the field: it shall be probably preferable to avoid having the bright source of light shot with critical out-of-field conditions (for instance, by reversing the camera upside-down, but there are alternative easier ways!).

Would it be the same with other lenses?

Yes and No: Most of the types of flare that were described above will exist whatever the lens with some remarkable exceptions. For instance resistance to ghost flaring has been so hugely improved in the recent years that it becomes hard to view ghost on images (even with HDR processsing!). BTW unless the circle is tangent to the long side of the image, fully circular images produced by fisheye lenses cannot suffer from bright steaking due to reflection from the backside of the DSLR mirror flap.

You may view some examples that were shot with various lenses here.

Would it be the same with another camera than the EOS 5D Mark II?

From the above analysis, one can deduce that every flare features that only or mainly depends on the lens shall also affect the images that would be shot with another camera. Of course the size of the sensor has then to be identical. The differences in the reflective and diffractive properties of the mirror box and of the sensor optical assembly shall of course modulate the aspect of the effect though.

The problem caused by mirror box concept and size (in addition to the reflectivity of its surfaces) shall of course differ from one camera to an other. It should probably do not exist with EVIL (no mirror flap) camera. Tests with different Canon DSLRs have shown a similar outcome: EOS 300D, 20D, 5D, 5D Mk II and 1Ds Mk III were reported to be prone to show the dreaded problem. I have heard that the EOS 40D does not.

Update (20 Mar 2012) : Two years ago the EOS 5D Mark II was already reported to be flawed.. George (the author of the report) answered a mail of mine. Following is an excerpt of the response:

<<... I contacted local (Greek) Canon distributor (company called Intersys) and got verification of the problem. But after that I never got an official answer from them, and unofficially the manager for Canon products commented that they cannot do anything about it, that it was a limitation of the camera's design and he said (get ready for this) "it is a cheap camera, what do you except ? Get a 1D series..".
After that I just dropped the subject because it was only giving me aggravation. I found a water soluble black paint with minimum reflection and painted the back side of the mirror. That gave me a major improvement. I now have to try a lot harder to recreate the problem (I have a post on d-preview about it).
Since then I have tested every camera that came my way, but although most show this problem under intense extremes, none was close as bad as 5DmkII. >>

I have checked again by visual inspection that the EOS 5D has exactly the same highly reflective backside of the mirror flap.

I am unfortunately lacking information from tests with gears from other camera makers....

Experimental setups for flare observation

To better observe tthe effects of flare and especially to view the differences when the bright light source comes from the underside of the camera, one can use some simple means: some are described on this other page.

 

(Nearly O.T.) remarks about the Canon EF 24-105 mm lens and about credibility of lens test review reporting:

I have searched the web for testing lenses for Ghosts and Flare. I looked then for test results and reviews concerning the very same lens that I have been reported the testing here on this page (Canon EF 24-105 mm f/4L IS USM). I have found three pertinent answers. Do they match my own finding? Hmmm... I don't think so. You may check by yourself for instance here and here. IMO it is time to set relevant Standards to better normalize the testing processes that lead to "qualitative test reviews". Anyhow, I am really intrigued by the "good" performance of a repaired (***) model that is shown on The-Digital-Picture.com website while Mikael Reichmann on the Luminous-Landscape.com website reported good as well as bad results from colleague photographers.

Notes:

(*) except when the coverage of the lens is restricted to less than the width of the sensor: fully (uncropped) circular fisheye lenses for instance: the light beam is constrained to not trespass the borders of the imaging sensor on the focal plane.

(**) except for these crazy amateurs of old photographic gears who adapt (i.e. hack) some famous medium format lenses on a 35 mm camera...

(***) A manufacturing flaw of some early production model of this lens had been reported here. Unfortunayely the link to the special announcement by Canon US is obsolete in this article by The-Digital-Picture.com. However one can read it on the French Canon Website and in English on the Canadian, the UK , the US, and the European Canon websites.

 

Michel Thoby

19 March, 2012