Parameters for filter tests

[rpcrowe]

If I were to continue my testing on the UV filter/no filter question using a Canon 17-55mm f/2.8 IS lens... How should I post process and display the images.

I planned on just shooting two versions of the image: one with a B&W UV and one without. Then I plan to post two versions of each image one unsharpened and one sharpened with USM.

I plan to post the full size images on smugmug and then link to those postings.

Should I post any croppings of the images? And how should I go about doing it.

Do you think it is necessary to shoot several sets of images at different f/stops or do you think that shooting a single set of each image at say f/8 would be satisfactory.

I plan to add USM at 250% at .3 pixels. Would that be O.K. This will be pictures of the side of my house showing the brick chimney.

I just want to see for my own benefit if in real life shooting, a quality UV filter will make a noticable difference.

I haqd post processing problems when I did the test at first because the paameters were too complicated.

BTW: If someone has previously posted this type of test, please direct me to it. I don't want to reinvent the wheel!

[Dave Humphries]

Hi Richard,

OK, I'll leave aside "why?"

... and ask what artefacts you are expecting to find? - since that should influence what you are shooting.

From the mention of 'brick chimney', I guess you are considering sharpness mostly?

I would post 700px square crop from the centre of the image, so we all see it at 1:1 for assessment of sharpness - I would not capture sharpen it at all if sharpness is what you want to assess, since that will have a larger effect and likely hide any small image difference.

You might want to photograph a scene at night with a few pin-pricks of bright light to look for reflections between sensor and filter surfaces and flare.

Cheers,

[Black Pearl]

I will kind of leave aside 'why' and ask if you show a pair of 100% crop images with no post sharpening that shows a resolution difference will you then sharpen them and print the resulting files. If these have a noticeable difference I'd be interested in hearing back.

[Colin Southern]

Personally, I'd suggest that we'd need to see ALL of the image (who says any 'degradation' would only affect the centre portion?). Personally, I would capture sharpen the images at the settings you specified, and then (for internet display) adjust them as you see fit - so long as the same settings are used for both images.

The big variable is lighting though - if you have extreme contrast scenes (eg naked sun or lights at night) then filters will make things worse (I add to emphasize WORSE; you'll still get flare and ghosting even without a UV filter in those circumstances).

As for the other question: "has this already been done" then the answer is yes -- a zillion times. Almost invariably anti-filter camp chooses worst-case scenes to "validate their position" and the pro-filter camp chooses general scenes that confirm their position.

[William W]

I think that F/2.8, F/5.6 and F/11 are all necessary.
Also FL = 17mm; 24mm; 35mm and 55mm.
That accounts for 12 images.
I would also post FF crop and also bottom or top corner at 100%.
That accounts for 24 images.
If you post sharpened and unsharpened that accounts for 48 images (if my maths is correct)


I have done some testing, but NOT of the brick wall of the house, but rather shooting into the light.

I don't think that shooting a bland, front lit Subject is "testing" very much or is very "testing" especially as you will be using very good quality Filters.

The objective of my tests was to get to get some basic feel for how much the Flare (additional Veiling Flare) is introduced by the UV Filters, as that is my main concern.

I have also done some tests along the same lines indoors to get a feel for Ghosting.

But my tests were primarily aimed toward me being able to choose between different (Prime) Lenses, so I was more concerned with learning about which lens (e.g. my 50 or 85) would be safer, for example, to shoot the Full Length Shot, with sun as a Low Backlight in outdoor Portraiture.

So I did not run a scientific battery of tests but rather tested the lenses as I would most typically use them and to find each lens's limitations and also to establish IF the Filter was having much effect or not.

Regarding my tests and the effect of the Filters - in the end I decided that the best approach (for me) was to toss the filter if I had any doubt whatsoever – and that means basically the UV Filter comes off inside and it comes off when shooting into the sun (maybe except at the windy beach or cliff side).


WW

[herbert]

One filter will probably not be noticeable. However 50 filters will be. Have a look at this article where they stack 50 filters on a lens and take a shot:

http://www.lensrentals.com/blog/2011...th-bad-filters

It seems that some filters are worse than others. However adding one to your lens will probably not matter.

On a more anecdotal point I recently micro focus adjusted (MFA) my camera and lens using a new automated software tool. Basically you plug the camera into your computer, point it at a test target and press go. The software takes a lot of photos using different MFA settings and picks the best one. With my B&W UV filter on the software could not get consistent results and failed to converge. I took it off and the lens calibrated.

So perhaps even if we cannot see any difference between photos the camera autofocus might. The amount of light that passes through the translucent mirror section and is reflected back to the autofocus sensor is very small. Tiny differences in light could cause problems for autofocus. Maybe I had a dirty filter. Again it was nothing I could see. However that does not mean it was not there. But if I leave my filter off I know I will get a dirty lens. I am happy cleaning my filter. I don't want to constantly clean my lens front element with the same casual technique.

Alex

[Dave Humphries]

Whether you capture sharpen or not depends whether;
you want to try to detect a tiny difference in sharpness or
prove that it doesn't matter - because I suspect you won't detect a difference after capture sharpening

The bigger problem, but only under certain lighting situations, is likely to be flare though.