Samuli Vahonen
Offline
• • • •
Upload & Sell: Off
|
 p.42 #6 · Official: Voigtlander 65mm f/2 Macro Apo-Lanthar | |
Fred Miranda wrote:
One way to do this is to shoot your subject at same distance (preferably on tripod) with both lenses (65 and 90mm) and in post-processing, crop the 65mm image to match the 90mm. The aperture would need to be equiv. as well, so 90/65=1.38x. The 90/2 lens would need to be set to 90mm f/2.8 in order for both lenses to show equiv. blur.
Once both images are equivalent in focal length and aperture, histograms have also have to be normalized (same exact exposure) since that affects contrast and colors. After that, we can't look at bokeh quality.
Fred, I would disagree with this method for boke quality evaluation. This leads that to situation, in which the longer lens already having "advantage"/"disadvantage" of not being cropped AND being closed down a little, which 99.99999999% of cases increases quality of boke. Method should produce same size boke highlights (=quantity of boke), but I would not use it for evaluation of quality of boke.
Based on comparing any pair of lenses with this cropping method the most common observations relate to test setup, not quality of boke:
1) longer lens vignetting is larger (most of cases as image is not cropped, of course some shorter lenses vignette so much that there is not much difference), which effects how to match exposures. If one matches center, then longer lens looks better due to vignetting, majority of people will not understand and they say "better colors" or "better contrast" - if one matches overall frame then longer lens looks bad, like having hot spot in middle of the lens [this gives nice benefit for tester to produce the results in a way her/his preference is shown as better alternative - or if one really wants to be objective a big dilemma, how to expose to avoid favoring the another]
2) longer lens corners may still have more cats eye, while shorter lens is cropped and shows less even wide open
3) due to mechanical/optical vignetting longer lens may have longer DOF in corners
4) any shorter lens harshness of boke highlights is magnified e.g. if there high light concentration to edges of boke highlights it looks worse than in real life due to magnification - or aspherical lens element "rings" in boke highlights.
5) cropped and wide open makes the shorter lens subject have less microcontrast and overall contrast, making colors etc. worse than the long lens photo - this should not affect on evaluating to boke, but there is most likely more observations of this than actual boke quality.
If one really would like evaluate boke quality, then camera would need to be moved on axis to match the FOVs to include exact same boke composition (not easy, at least I don't own >2 meter focus rail). This would allow tester to use same apertures and adjusting the boke highlight size to same by using focusing. However this method makes it impossible to include focus plane subject as focus distance is used to adjust blur disk size. Even this is the best for evaluating boke quality when boke quantity is kept the same, I would not recommend method - Most likely majority of discussion would be about test setup and lack of focus plane subject even this would enable real possibility to discuss objectively about boke quality.
Then there is naturally few "compromise" methods done by moving camera on axis to match composition. If tester makes the subject size equivalent, then boke quality cannot be evaluated as boke is different size due to different FOVs causing back-forth object size differences. And if tester makes the composition based on what is on boke, then subject size will change between images - and the amount of internet crying is gigantic due to this  . And in both cases some aperture matching is needed to equalize the boke quantity for making it possible to compare boke quality, which will lead some differences which again lead majority of observations to be setup relevant, not boke quality relevant.
For me more important than just matching exposure, is that the light is 100% the same. In cloudy day even if clouds would not block the sun (direct sunlight to focus plane subject and boke) their movement around the sky changes lightning too much (less issue in places where sun shines from really high, but when sun is not so high e.g. in Finland or Alaska, then the overall "fill light" from sky has more effect how the scene is lit). Also sun position changes very fast and difference of multiple minutes causes the light to be different. In addition if boke is formed from flexible objects (e.g. tree with leaves), there can't be any wind, which would change the angle leaves reflect light. This favors 100% artificial setup test scenes, but does that give meaningful results as most photographers shoot outdoors in natural light? I have not seen many good artificial boke test scenes, none for larger distances.
In practice it's really hard to evaluate boke quality of multiple lenses when FOV is different.
Samuli
|
