06-14-2010, 11:24 PM
[quote name='mst' date='14 June 2010 - 02:23 PM' timestamp='1276521801' post='496']
Well, at 1m that's hardly surprising, since wide open isn't reported as f/2.0 anymore (probably f/2.2 instead) and consequently f/2.8 isn't a full stop away. If this is how the test was performed, than it was doomed to fail because of a faulty setup.
[/quote]
Hi, I'm the one who started the mentioned dpreview thread.
The computation I did took the f/2.2 into account. You can judge yourself. Pictures from bokeh test are here:
http://jtra.cz/foto/tamron-underexposure-vs-bokeh/
The first picture shows target. Then I set manual focus and set 1m, I stayed at the same distance from target.
The second is f/2.2, the third f/2.8, the fourth is f/4, the last is f/8.
Disregard the blueish tint on the f/2.2 picture, at this aperture, the speed was 1/200s which was interfering with multi-color cycle of the compact fluorescent lights.
Circles of bokeh in this setup should be in size relative to aperture. This is valid for f/4 compared to f/8. I measure about 128 pixels diameter for f/4, while about 63 pixels diameter for f/8.
I measure about 180-190 pixels diameter for f/2.8. That is ok too, compared to f/4, the expected value is 4/2.8*128=183. But for f/2.2 I measure 190 pixels while it should be about 4/2.2*128=233.
In other shop which has had better lighting, I measured light intensity in the center of picture (should not be affected by vignetting) for corresponding combinations of increasing aperture and decreasing time in manual mode (which is not affected by camera metering). Result was the widest effective aperture was f/2.7. See here:
http://forums.dpreview.com/forums/read.asp?forum=1030&message=34305325
After three month without any better lens (I tried about 6 over three months) I decided to go with different lens.
I do not know what was the reason for the behavior. Canon users of the lens were happy indeed. Only Nikon's were not. That lead me to speculate about nature of the problem. It might have been caused by aperture level interface the Nikon uses but Canon does not. The lens looked it opened fully via DOF preview, but perhaps at high shutter speed, the things may be different. Other reason might be the lens design is restrictive so that wide open aperture blades to not contribute fully to light path, but Canon users would be affected by that too.
Main attraction to Tamron was bokeh for me, not macro. The background bokeh was very pleasant (foreground bokeh is often opposite in pleasantness to background bokeh as explained at http://www.sony.ca/html/uf/microsite/Capturing_Depth_Minisite/2.html in section "Aberration and the Defocused Image").
In comparison current Nikons 50/1.4 AF-S is awful for background bokeh. This photozone's picture shows it clearly: http://photozone.smugmug.com/photos/457146842_X447C-O.jpg
In background, the bokeh has pronouced circle edges (moreover they are green) - in foreground, the bokeh is pleasant. I wanted a lens that has good background bokeh, or at least neutral.
Nikon has such lens, it is 85/1.4 AF-D from what I have seen, but that was out of my price range.
So I choose Sigma 50/1.4 which is the only fast and current 50mm known to me that has pleasant background bokeh. Bokeh is still slightly colored (green in background, magenta in foreground as usual for fast non-apochromatic lenses).
I'm quite happy with that lens. Only issue is a bit unreliable AF accuracy at large apertures on my D90.
Other issue is size, it is quite big and heavy among 50mm, which intimidates some subjects.
Well, at 1m that's hardly surprising, since wide open isn't reported as f/2.0 anymore (probably f/2.2 instead) and consequently f/2.8 isn't a full stop away. If this is how the test was performed, than it was doomed to fail because of a faulty setup.
[/quote]
Hi, I'm the one who started the mentioned dpreview thread.
The computation I did took the f/2.2 into account. You can judge yourself. Pictures from bokeh test are here:
http://jtra.cz/foto/tamron-underexposure-vs-bokeh/
The first picture shows target. Then I set manual focus and set 1m, I stayed at the same distance from target.
The second is f/2.2, the third f/2.8, the fourth is f/4, the last is f/8.
Disregard the blueish tint on the f/2.2 picture, at this aperture, the speed was 1/200s which was interfering with multi-color cycle of the compact fluorescent lights.
Circles of bokeh in this setup should be in size relative to aperture. This is valid for f/4 compared to f/8. I measure about 128 pixels diameter for f/4, while about 63 pixels diameter for f/8.
I measure about 180-190 pixels diameter for f/2.8. That is ok too, compared to f/4, the expected value is 4/2.8*128=183. But for f/2.2 I measure 190 pixels while it should be about 4/2.2*128=233.
In other shop which has had better lighting, I measured light intensity in the center of picture (should not be affected by vignetting) for corresponding combinations of increasing aperture and decreasing time in manual mode (which is not affected by camera metering). Result was the widest effective aperture was f/2.7. See here:
http://forums.dpreview.com/forums/read.asp?forum=1030&message=34305325
After three month without any better lens (I tried about 6 over three months) I decided to go with different lens.
I do not know what was the reason for the behavior. Canon users of the lens were happy indeed. Only Nikon's were not. That lead me to speculate about nature of the problem. It might have been caused by aperture level interface the Nikon uses but Canon does not. The lens looked it opened fully via DOF preview, but perhaps at high shutter speed, the things may be different. Other reason might be the lens design is restrictive so that wide open aperture blades to not contribute fully to light path, but Canon users would be affected by that too.
Main attraction to Tamron was bokeh for me, not macro. The background bokeh was very pleasant (foreground bokeh is often opposite in pleasantness to background bokeh as explained at http://www.sony.ca/html/uf/microsite/Capturing_Depth_Minisite/2.html in section "Aberration and the Defocused Image").
In comparison current Nikons 50/1.4 AF-S is awful for background bokeh. This photozone's picture shows it clearly: http://photozone.smugmug.com/photos/457146842_X447C-O.jpg
In background, the bokeh has pronouced circle edges (moreover they are green) - in foreground, the bokeh is pleasant. I wanted a lens that has good background bokeh, or at least neutral.
Nikon has such lens, it is 85/1.4 AF-D from what I have seen, but that was out of my price range.
So I choose Sigma 50/1.4 which is the only fast and current 50mm known to me that has pleasant background bokeh. Bokeh is still slightly colored (green in background, magenta in foreground as usual for fast non-apochromatic lenses).
I'm quite happy with that lens. Only issue is a bit unreliable AF accuracy at large apertures on my D90.
Other issue is size, it is quite big and heavy among 50mm, which intimidates some subjects.