KaaX wrote:
This is not true, either in general, or specifically as applied to photographic images.
An easy demonstration: let me take an image and apply a simple, well-defined convolution to it, say, Gaussian blur. I will then give you the blurred image and ask you to deconvolute it. Will you be able to do it?
Yes (assuming you fully specify what the convolution kernel is).
With the caveat that computational rounding errors in the convolution process are themselves a source of noise. This noise will be magnified in the deconvolution process, resulting in some artifacts (of increasing severity as the amount of blur increases). Areas within a few radii of the image edges may also be problematic, depending on how information is lost off the edges of the image.
If I have time, I can try to post a demonstration.
In addition to the macro lenses, I would like to see more of the original versions of various L lenses on the list, including all the Supertelephotos.
And I also would like to understand if the main application for this is for improvements due to defraction effects, or if there are other kinds of distortions or effects on images that this DLO can help with.
1. No lens+extender combinations are currently supported, which is a shame considering that it's when we use the extenders that the performance of the super-telephoto primes suffer enough that it might be worth trying to correct! I've never taken a shot with my 300/2.8L IS that made me think, "oh, if only the lens aberrations/diffraction were not so prominent."
2. DLO is incompatible with the previously existing chromatic aberration correction slider. Only one or the other can be used. This indicates to me that the DLO function also corrects chromatic aberration. It may also correct astigmatism and coma to some extent, though that's just a guess.
wickerprints wrote:
Some things I read from the manual:
1. No lens+extender combinations are currently supported, which is a shame considering that it's when we use the extenders that the performance of the super-telephoto primes suffer enough that it might be worth trying to correct! I've never taken a shot with my 300/2.8L IS that made me think, "oh, if only the lens aberrations/diffraction were not so prominent."
I've not tried the super-tele's But I did have a few picts taken with the 24-70 f/2.8 and a Kenko 1.4x extender. Interestingly, the program allowed the picts to be processed. And there was still a remarkable amount of extra details showing up!
Perhaps someone with a super-tele and the Kenko 1.4x can give it a try to see if results are desirable ??
Edit:
Ok, it doesn't work. I had the same Kenko 1.4x Extender mounted on the 70-200 f/2.8 IS. It does not allow the DLO function.
wickerprints wrote:
It seems that according to the pre-release instruction manual, at the present time, only the following Canon lenses are supported (though this may change before the official release of the next version of DPP, and it does not take into account future profiles Canon might decide to create):
EF 14/2.8L II USM
EF 24/1.4L II USM
EF 35/1.4L USM
EF 50/1.4 USM
EF 50/1.2L USM
EF 85/1.2L II USM
EF 300/2.8L IS II USM
EF 400/2.8L IS II USM
EF 500/4L IS II USM
EF 600/4L IS II USM
EF 16-35/2.8L USM, II USM (both versions)
EF 17-40/4L USM
EF 24-70/2.8L USM, II USM (both versions)
EF 24-105/4L IS USM
EF 28-300/3.5-5.6L IS USM
EF 70-200/2.8L IS USM, II USM (both versions)
EF 70-200/4L USM, IS USM (either with/without IS)
EF 70-300/4-5.6 IS USM [not the L?]
EF 100-400/4.5-5.6L IS USM
EF-S 10-22/3.5-4.5 USM
EF-S 15-85/3.5-5.6 IS
EF-S 17-55/2.8 IS USM
EF-S 17-85/4-5.6 IS USM
EF-S 18-200/3.5-5.6 IS
EF-S 18-135/3.5-5.6 IS
I think the range of supported lenses suggests that Canon chose designs that (1) they felt would benefit from this algorithm, and/or (2) are especially popular lenses, and/or (3) tend to be used by photographers who wish to extract the maximum performance out of the system.
I am, however, somewhat surprised that the macro lenses were not listed. The EF 100/2.8L macro IS, EF 180/3.5L, and MP-E 65/2.8 1-5x macro are frequently used stopped down for maximum DOF. And to some extent, I could also imagine DLO being useful for the TS-E lenses, so it's a shame we don't see those in the list, either. It's entirely conceivable that we might see more Canon lenses supported in the future, though.
As for third-party lenses, I think it would be impractical for Canon to spend the resources to measure a competitor's lens and create profiles. It may not even be feasible, since for all we know Canon could be using their own ray traced simulations of the spot diagrams of their lenses to compute the profiles.
As for me, I have 3 lenses in that list--4 if you count ones I've since parted with but from which I still have images. ...Show more →
I doubt they will be making profiles for TS-E lenses. The amount of tilt / shift used is not recorded by the camera. ...unless they make profiles for the lenses without any movements which would defeat the purpose.
wickerprints wrote:
Some things I read from the manual:
1. No lens+extender combinations are currently supported, which is a shame considering that it's when we use the extenders that the performance of the super-telephoto primes suffer enough that it might be worth trying to correct! I've never taken a shot with my 300/2.8L IS that made me think, "oh, if only the lens aberrations/diffraction were not so prominent."
2. DLO is incompatible with the previously existing chromatic aberration correction slider. Only one or the other can be used. This indicates to me that the DLO function also corrects chromatic aberration. It may also correct astigmatism and coma to some extent, though that's just a guess....Show more →
According to Chuck Westfall, the algorithm addresses many optical flaws:
"A new feature called Digital Lens Optimizer processes RAW images to achieve ideal optical characteristics for all types of optical aberration or diffraction, effects of a low-pass filter in front of a CMOS sensor, etc. This function improves image quality particularly in the image periphery in addition to the image center. This function is made possible because the entire design-through-manufacture process, for camera, CMOS sensor, EF lens, and DPP, is carried out entirely at Canon. Images are processed optimally using lens information in the image files (focal length, subject distance, and aperture) and lens data specially for the Digital Lens Optimizer. (However, the size of a .CR2 file will be two to three times larger after applying the Digital Lens Optimizer.) Adjustments are made for such aspects as spherical aberration, chromatic aberration, astigmatism, curvature of field, sagittal halo, chromatic aberration of magnification, axial chromatic aberration, diffraction, and the effects of a low-pass filter in front of the CMOS sensor. DPP’s Digital Lens Optimizer will be usable with any of 29 compatible lenses initially. It works with .CR2 files from EOS models released since 2006 (EOS 30D and forward)."
wickerprints wrote:
I am, however, somewhat surprised that the macro lenses were not listed. The EF 100/2.8L macro IS, EF 180/3.5L, and MP-E 65/2.8 1-5x macro are frequently used stopped down for maximum DOF.
Roland W wrote:
And I also would like to understand if the main application for this is for improvements due to defraction effects, or if there are other kinds of distortions or effects on images that this DLO can help with.
One could test for this by correcting an f4 image at the 24 end of the 24-105. See if it corrects for the barrel distortion and the vignetting.
Why bother? LR-4 is pretty already and now they've added the "Remove CA" so all you need to do is check the box and you are good to go. For local correction/tweaks you can always rely on NIK and CS-5 or CS-6 Beta.
AGeoJO wrote:
I am now considering using DPP as a part of my workflow, in addition to LR but I am not looking forward to the additional time to do so . I need to figure out a way to streamline that process.
Stargazer 78, do you know whether this feature will work on TIF files or do they have to be RAW?
Fred Miranda wrote:
According to Chuck Westfall, the algorithm addresses many optical flaws...
If what Chuck says is true, then (i) we may not see similar offerings from Adobe and Nikon (particularly when sensors are not designed entirely by Nikon in the latter case) (ii) Canon is strongly discouraging landscape photographers from adapting Nikon's renowned 14-24 f/2.8 lens on Canon bodies.
bobbytan wrote:
Why bother? LR-4 is pretty already and now they've added the "Remove CA" so all you need to do is check the box and you are good to go. For local correction/tweaks you can always rely on NIK and CS-5 or CS-6 Beta.
Bobby, I am referring to the removal of diffraction effect and to increase the overall sharpness without or with less halo that can be apparent using the usual sharpening tool.
Got it ... but I would just as well avoid stopping down too much i.e. do it the old-fashion way.
AGeoJO wrote:
Bobby, I am referring to the removal of diffraction effect and to increase the overall sharpness without or with less halo that can be apparent using the usual sharpening tool.
I understood that the DLO data will be restored in the .CR2 files and the new .CR2 files' size will increase 3X than the original because Canon uses the debayering process to apply the DLO adjustments. Although it is a bummer alright but it is not unexpected. I don't like the file size increase but I like the idea of being able to do the post processing in the development module of LR.
I am waiting for Aravind (thedigitalbean) to get back to me on this.... . Anyone else out there can give me any input, please?
The results are "fragile"; if you look closely, you can see checkerboarding artifacts caused by noise added by rounding errors in the computation. If I try to deconvolve a more strongly blurred image, or add just a tiny bit of extra noise to the blurred image, the deconvolution falls apart completely into random noise. If I add just +/- 1 bit of random noise to the blurred (8-bit) image, I can only get away with a considerably weaker level of de-blurring, to get: http://praetoriusphoto.images.s3.amazonaws.com/fmforums/20120322_deconvolution/deconvolved_noised.jpg
which, even at this weaker level of de-blurring shows a lot of noise-induced artifacts.
The calculation needed by Canon's program is far more difficult, because the shape of the blur varies across the image and is more complicated than a simple gaussian. In addition, they are always fighting against noise added in the blurred image.
speedmaster20d wrote:
All EF lenses including tele-convertors will be eventually supported. Older manual focus lenses without electronic focus distance readout will not be supported because the focus distance cannot be determined.
Canon are doing actual measurements to calibrate each and every lens as opposed to using their simulated design data. Here is an example of correcting spherical aberration (field curvature) in the extreme border of EF 70-200 f/2.8II with EOS 5D MarkII which has resulted in corner softness.
The calculation is performed pre-demosaic and it takes about 10 seconds on my 4.2GHZ quad core system with 24GB RAM. So it's not simply sharpening the output.
The reason for file size increase is that when the correction is performed, the corrected pixel values are added to the CR2 file (as opposed to overwriting the original values). It is faster to load the new values from disk as opposed to re-calculating them each time the RAW file is rendered. The latter will be very slow, especially on lower end machines/laptops so Canon's approach was the best.
Great, thank you very much! Like posted by the OP, I can tell the difference between the two. Because of the file increase and the time required to incorporate the DLO, I will reserve this feature for selected images only then.
Monito wrote:
The image alone is not sufficient. You need to know the operator as explained above. That would be the lens data detrmined by Canon. Even that is imperfect, as explained above.
Further, the deconvolving is senstive to noise, as explained above.
Which lens? I'll take an image, let's say a purely computer-generated one with no lens or camera involved. No noise either. I'll apply Gaussian blur with known and disclosed parameters -- that is your operator, fully defined. I'll give you the blurred image -- will you give me back the original one?
speedmaster20d wrote:
Canon are doing actual measurements to calibrate each and every lens as opposed to using their simulated design data.
As we well know, there's significant copy variation in lenses. The best Canon can do -- if they do it properly -- is to measure an unbiased sample of production lenses and calibrate to the mean. Your particular lens may or may not be a good match.
mpmendenhall wrote:
If I try to deconvolve a more strongly blurred image, or add just a tiny bit of extra noise to the blurred image, the deconvolution falls apart completely into random noise.
Yep.
Convolutions are not necessarily invertible. Given, e.g., blurring, there are many source images which will blur to the same blurred image. The more blur there is, the wider is the universe of possible source images. A deconvolution algorithm does not and cannot know which one of them is the correct source image.
KaaX wrote:
As we well know, there's significant copy variation in lenses. The best Canon can do -- if they do it properly -- is to measure an unbiased sample of production lenses and calibrate to the mean. Your particular lens may or may not be a good match.
Who is we? In my 10+ years of owning and shooting with Canon gear from 16mm to 1200mm I yet have to observe the significant variation internet experts on these forums talk about
Many people struggle to make a sharp photograph, then blame the gear. Many times I have met people in the field who complain their lens is bad or camera doesn't AF. What I do is I borrow their gear and give them my own gear. At the end of the day we switch our cards. Suddenly their gear becomes sharp and AF starts to work and at the same time, my gear becomes defective, go figure
Convolutions are not necessarily invertible. Given, e.g., blurring, there are many source images which will blur to the same blurred image. The more blur there is, the wider is the universe of possible source images. A deconvolution algorithm does not and cannot know which one of them is the correct source image.
Given a convolution with no zeros in the fourier transform of the kernel (e.g. a gaussian blur), in theory there is exactly one unique source image that will produce the convolved one. In the limit of a computer that can carry out calculations to infinite precision, the original image can be perfectly restored.
This theory breaks down in real life with the introduction of noise, both the noise from real camera images and quantization/rounding noise in computer calculations. The real-life limitation to deconvolution is not the non-uniqueness of deconvolution, but the non-uniqueness of de-noising. The inevitable addition of noise, rather than the blurring process, is what actually destroys information and prevents return to the initial image. Blurring greatly exacerbates the problem by magnifying the impact of noise.
The real-life usefulness of de-blurring is going to be most strongly limited by the noise quality of images coming from the camera. Low-ISO noise levels previously considered completely irrelevant will become highly important when aggressive deconvolution is applied to images.
