This topic is in the 2010 archive. You can read it and like its posts here. It is read only.
Archive 20107D Sharpness Trick
p.1 #17D Sharpness Trick
I was taking advantage of some nice weather today to do some wide angle test shooting. One of the things I wanted to evaluate was whether I could use the 7D for travel and landscape shooting as a substitute for a 5D Mark II.
So I made some photographs and compared 7D images side-by-side with 5D Mark II images (captured using the same camera settings with lenses of comparable professional quality).
At first I noted that the 7D images looked quite a bit softer--even with all noise reduction turned off. This has been widely reported on the 'Net, so I chalked it up to the idea that the 7D has a stronger AA filter, or its sensor is just "outresolving" the lens.
But then I played around with the lens correction settings in DPP... and almost by accident I discovered that when I enabled both the CA and the "Color blur" corrections, 7D image sharpness substantially improved. In fact, it improved to the point where the differences between the 7D and 5D Mark II files became marginal in some cases.
Interestingly, this pixel-level sharpness improvement did not seem to occur with the 5D Mark II files. It also seems to be fairly lens-independent on the 7D. So I'm not sure if what I'm seeing is an actual improvement of inherent lens defects, or of something else in the 7D processing chain.
In any case, have a look. Hopefully JPEG compression won't obscure the results too much:
Full frame: 7D, EF-S 17-55, 17mm @ f/5.6:
http://www.vothphoto.com/files/personal/fm/7DSharpnessTrick/7DSharp1.jpg
100% crop, no lens corrections:
http://www.vothphoto.com/files/personal/fm/7DSharpnessTrick/7DSharp2.jpg
100% crop, with lens corrections:
http://www.vothphoto.com/files/personal/fm/7DSharpnessTrick/7DSharp3.jpg
Finally, a comparable 5D Mark II image, using a 24mm prime @ f/8:
http://www.vothphoto.com/files/personal/fm/7DSharpnessTrick/7DSharp4.jpg
Bottom line: if you shoot the 7D and want the best sharpness, try using these corrections in DPP.
So I made some photographs and compared 7D images side-by-side with 5D Mark II images (captured using the same camera settings with lenses of comparable professional quality).
At first I noted that the 7D images looked quite a bit softer--even with all noise reduction turned off. This has been widely reported on the 'Net, so I chalked it up to the idea that the 7D has a stronger AA filter, or its sensor is just "outresolving" the lens.
But then I played around with the lens correction settings in DPP... and almost by accident I discovered that when I enabled both the CA and the "Color blur" corrections, 7D image sharpness substantially improved. In fact, it improved to the point where the differences between the 7D and 5D Mark II files became marginal in some cases.
Interestingly, this pixel-level sharpness improvement did not seem to occur with the 5D Mark II files. It also seems to be fairly lens-independent on the 7D. So I'm not sure if what I'm seeing is an actual improvement of inherent lens defects, or of something else in the 7D processing chain.
In any case, have a look. Hopefully JPEG compression won't obscure the results too much:
Full frame: 7D, EF-S 17-55, 17mm @ f/5.6:
http://www.vothphoto.com/files/personal/fm/7DSharpnessTrick/7DSharp1.jpg
100% crop, no lens corrections:
http://www.vothphoto.com/files/personal/fm/7DSharpnessTrick/7DSharp2.jpg
100% crop, with lens corrections:
http://www.vothphoto.com/files/personal/fm/7DSharpnessTrick/7DSharp3.jpg
Finally, a comparable 5D Mark II image, using a 24mm prime @ f/8:
http://www.vothphoto.com/files/personal/fm/7DSharpnessTrick/7DSharp4.jpg
Bottom line: if you shoot the 7D and want the best sharpness, try using these corrections in DPP.
Mar 22, 2010 at 04:45 PM
p.1 #27D Sharpness Trick
So are you just saying use 0 luminance and chroma noise reduction? What sharpening setting, if any? (I see your 5DII settings, but not for the 7D)
Mar 22, 2010 at 04:58 PM
p.1 #37D Sharpness Trick
timpdx wrote:
So are you just saying use 0 luminance and chroma noise reduction? What sharpening setting, if any? (I see your 5DII settings, but not for the 7D)
Not exactly. At ISO 100 you must explicitly turn off NR in DPP or you will get some reduction in sharpness, so that is a given. However, my point is that the CA and color blur lens corrections also improve sharpness in the 7D files.
All test images were shot with Standard Picture Style, Contrast -1 (my default in-camera settings), so these examples all have Sharpness set to 3.
So are you just saying use 0 luminance and chroma noise reduction? What sharpening setting, if any? (I see your 5DII settings, but not for the 7D)
Not exactly. At ISO 100 you must explicitly turn off NR in DPP or you will get some reduction in sharpness, so that is a given. However, my point is that the CA and color blur lens corrections also improve sharpness in the 7D files.
All test images were shot with Standard Picture Style, Contrast -1 (my default in-camera settings), so these examples all have Sharpness set to 3.
Mar 22, 2010 at 05:02 PM
p.1 #57D Sharpness Trick
I just tried this with a shot from my 7D with 500L + 1.4x and couldn't tell a difference after I applied these settings. This is certainly interesting, though, and worth looking into more. I'll try this with shots from my 17-55 and 100 2.0.
Justin
Justin
Mar 22, 2010 at 05:08 PM
p.1 #67D Sharpness Trick
timpdx wrote:
So are you just saying use 0 luminance and chroma noise reduction? What sharpening setting, if any? (I see your 5DII settings, but not for the 7D)
That's not the point the OP was making. He seems to be saying that there's a relationship between the sharpness of the 7D's RAW output and two apparently unrelated post-processing algorithms (CA + color blur).
If true, this is very puzzling indeed. In fact, the 100% crops are taken from the center, where the CA adjustment shouldn't have any visible effect. However, seeing the screen caps I too agree that the second 7D output does appear slightly improved. This is complete speculation, but perhaps there is something wrong with DPP handling of the 7D RAWs?
So are you just saying use 0 luminance and chroma noise reduction? What sharpening setting, if any? (I see your 5DII settings, but not for the 7D)
That's not the point the OP was making. He seems to be saying that there's a relationship between the sharpness of the 7D's RAW output and two apparently unrelated post-processing algorithms (CA + color blur).
If true, this is very puzzling indeed. In fact, the 100% crops are taken from the center, where the CA adjustment shouldn't have any visible effect. However, seeing the screen caps I too agree that the second 7D output does appear slightly improved. This is complete speculation, but perhaps there is something wrong with DPP handling of the 7D RAWs?
Mar 22, 2010 at 05:09 PM
p.1 #77D Sharpness Trick
Hm, I've tried it on a couple of images without success (70-200 MkII @ 200mm, f/2.8):
 |
||
| http://peltarion.eu/img/sharpen7d.jpg | ||
|
Mar 22, 2010 at 05:27 PM
p.1 #87D Sharpness Trick
wickerprints wrote:
That's not the point the OP was making. He seems to be saying that there's a relationship between the sharpness of the 7D's RAW output and two apparently unrelated post-processing algorithms (CA + color blur).
If true, this is very puzzling indeed. In fact, the 100% crops are taken from the center, where the CA adjustment shouldn't have any visible effect. However, seeing the screen caps I too agree that the second 7D output does appear slightly improved. This is complete speculation, but perhaps there is something wrong with DPP handling of the 7D RAWs?
Agreed. I'm not claiming this effect will be seen with all lenses, but the few that I had on hand for this test seemed to show some difference.
It is possible that the pixel density of the 7D is revealing some color blur defects in even very good lenses (e.g., 17-55) that may not be visible with other bodies---but it is also possible that there is some glitch in Canon's RAW decoding algorithms that is "fixed" by this alternative path through the processing pipeline. I'd want to do a lot more testing before coming to that conclusion, but had never seen this discussed or demonstrated, and thought it was worth sharing.
timpdx wrote
So are you just saying use 0 luminance and chroma noise reduction? What sharpening setting, if any? (I see your 5DII settings, but not for the 7D)If true, this is very puzzling indeed. In fact, the 100% crops are taken from the center, where the CA adjustment shouldn't have any visible effect. However, seeing the screen caps I too agree that the second 7D output does appear slightly improved. This is complete speculation, but perhaps there is something wrong with DPP handling of the 7D RAWs?
Agreed. I'm not claiming this effect will be seen with all lenses, but the few that I had on hand for this test seemed to show some difference.
It is possible that the pixel density of the 7D is revealing some color blur defects in even very good lenses (e.g., 17-55) that may not be visible with other bodies---but it is also possible that there is some glitch in Canon's RAW decoding algorithms that is "fixed" by this alternative path through the processing pipeline. I'd want to do a lot more testing before coming to that conclusion, but had never seen this discussed or demonstrated, and thought it was worth sharing.
Mar 22, 2010 at 05:33 PM
p.1 #97D Sharpness Trick
denoir wrote:
Hm, I've tried it on a couple of images without success (70-200 MkII @ 200mm, f/2.8):
http://peltarion.eu/img/sharpen7d.jpg
I think Canon's best telephotos are very well corrected for CA--so maybe this effect is something that is indeed limited to specific lenses. (Even if it only works with certain wide angle lenses it would be helpful for landscape shooters.)
It is also probably best to test this using a tripod, mirror up, self-timer, etc. (as I have done) to eliminate other sources of sharpness loss.
Hm, I've tried it on a couple of images without success (70-200 MkII @ 200mm, f/2.8):
http://peltarion.eu/img/sharpen7d.jpg
I think Canon's best telephotos are very well corrected for CA--so maybe this effect is something that is indeed limited to specific lenses. (Even if it only works with certain wide angle lenses it would be helpful for landscape shooters.)
It is also probably best to test this using a tripod, mirror up, self-timer, etc. (as I have done) to eliminate other sources of sharpness loss.
Mar 22, 2010 at 05:36 PM
p.1 #107D Sharpness Trick
Whoa, be careful with that CA correction:
Tokina 11-16 @ 11mm, f/2.8:
Tokina 11-16 @ 11mm, f/2.8:
|
||
| http://peltarion.eu/img/sharpen7d_2.jpg | ||
|
Mar 22, 2010 at 05:56 PM
p.1 #117D Sharpness Trick
CA and color blur lens corrections will improve sharpness on nearly any photo where these lens abberations are present.
Mar 22, 2010 at 06:04 PM
p.1 #127D Sharpness Trick
Correcting lateral chromatic aberration should indeed make the image sharper, but not in the centre (on-axis). I don't know what "Color blur" does in DPP so I won't guess the effect it might have on sharpness. The Canon 17-55 mm f/2.8 is a decent lens but it has plenty of problems at 17 mm.
However I do notice more noise in the version with the lens corrections switched on. Along with the increased noise is improved low-contrast detail in the brickwork in shade (and other places). Noise reduction normally hits low-contrast detail first, so it's possible that DPP is applying more noise reduction to the version without lens corrections (though it might be applying some noise reduction to both images; I don't know this software).
Still, I wouldn't want to make any grand proclamations from these JPEGs. Do you see this difference in noise on your screen, garyvot?
However I do notice more noise in the version with the lens corrections switched on. Along with the increased noise is improved low-contrast detail in the brickwork in shade (and other places). Noise reduction normally hits low-contrast detail first, so it's possible that DPP is applying more noise reduction to the version without lens corrections (though it might be applying some noise reduction to both images; I don't know this software).
Still, I wouldn't want to make any grand proclamations from these JPEGs. Do you see this difference in noise on your screen, garyvot?
Mar 22, 2010 at 06:15 PM
p.1 #137D Sharpness Trick
denoir wrote:
Whoa, be careful with that CA correction:
Tokina 11-16 @ 11mm, f/2.8:
DPP's CA correction requires manual tuning for non-EF lenses. Often it still requires manual tuning even when EF lenses are used. You don't just click "yes" and expect it to magically work. Why would you even think that with a Tokina lens?
Whoa, be careful with that CA correction:
Tokina 11-16 @ 11mm, f/2.8:
DPP's CA correction requires manual tuning for non-EF lenses. Often it still requires manual tuning even when EF lenses are used. You don't just click "yes" and expect it to magically work. Why would you even think that with a Tokina lens?
Mar 22, 2010 at 06:17 PM
p.1 #147D Sharpness Trick
FWIW, "Color Blur" to the best of my understanding reduces the purple fringing effect, which is especially visible when photographing specular highlights with certain wide-aperture lenses.
Mar 22, 2010 at 06:20 PM
p.1 #157D Sharpness Trick
wickerprints wrote:
Tokina 11-16 @ 11mm, f/2.8:DPP's CA correction requires manual tuning for non-EF lenses. Often it still requires manual tuning even when EF lenses are used. You don't just click "yes" and expect it to magically work. Why would you even think that with a Tokina lens?
Actually, I had a very good reason. When I tried the first seagull shot the tuning options were disabled. Only after I upgraded to a newer version of DPP that had the new 70-200 in the lens database could I set any parameters. With the Tokina image everything was enabled - so it meant that DPP had it in the database. As the distortion and peripheral illumination parameters were set correctly it was perfectly reasonable to expect that the CA parameters would be set as well.
denoir wrote
Whoa, be careful with that CA correction:Tokina 11-16 @ 11mm, f/2.8:
Actually, I had a very good reason. When I tried the first seagull shot the tuning options were disabled. Only after I upgraded to a newer version of DPP that had the new 70-200 in the lens database could I set any parameters. With the Tokina image everything was enabled - so it meant that DPP had it in the database. As the distortion and peripheral illumination parameters were set correctly it was perfectly reasonable to expect that the CA parameters would be set as well.
Mar 22, 2010 at 06:27 PM
p.1 #167D Sharpness Trick
denoir wrote:
With the Tokina image everything was enabled - so it meant that DPP had it in the database. As the distortion and peripheral illumination parameters were set correctly it was perfectly reasonable to expect that the CA parameters would be set as well.
Tokina's lens IDs often trample on the lens IDs of brand lenses. Obviously when you apply corrections for a lens of completely different optical design (i.e., Canon's), the results aren't what you expected.
With the Tokina image everything was enabled - so it meant that DPP had it in the database. As the distortion and peripheral illumination parameters were set correctly it was perfectly reasonable to expect that the CA parameters would be set as well.
Tokina's lens IDs often trample on the lens IDs of brand lenses. Obviously when you apply corrections for a lens of completely different optical design (i.e., Canon's), the results aren't what you expected.
Mar 22, 2010 at 07:07 PM
p.1 #177D Sharpness Trick
colinm wrote:
Tokina's lens IDs often trample on the lens IDs of brand lenses. Obviously when you apply corrections for a lens of completely different optical design (i.e., Canon's), the results aren't what you expected.
+1
denoir wrote
With the Tokina image everything was enabled - so it meant that DPP had it in the database. As the distortion and peripheral illumination parameters were set correctly it was perfectly reasonable to expect that the CA parameters would be set as well.+1
Mar 22, 2010 at 07:12 PM
p.1 #187D Sharpness Trick
I noticed that issue a few years ago with various cameras. It depends mainly on the lens and how much CA correction is applied. I turn it off in DPP and use PS to remove CA.
EBH
EBH
Mar 22, 2010 at 07:18 PM