Lens sharpness and Photoshop Sigma 10-20mm F/3.5

[Abitconfused]

I am acquiring a Sigma 10-20mm lens. Some say it is not as sharp as this or that lens. However, shouldn't Photoshop's Unsharp mask's ability to increase apparent sharpness and the RAW tool's ability to correct aberration and distortion also be considered. Don't these tools even the lens quality playing field significantly?

[Shadowman]

What camera are you using the lens with? How are you shooting, handheld or with tripod? Manual or autofocus. What aperture do you plan to use the lens at? Where's the data that says it's not sharp? There is a lot to consider when judging a lens sharpness.

[dje]

Ed I have the Sigma 10-20mm f/4 -f/5.6 lens and have been quite satisfied with it. I'm not in a position to compare it's sharpness to other lenses but it might perhaps be more informative to compare the mtf charts rather than going on anecdotal "evidence". Sigma does publish mtf charts for it's lenses (look at the Tech Specs for a nominated lens) as does Canon and Nikon.

I guess the sharpening tools and lens correction tools in PS do level the playing field to some extent although I would think that there is no substitute for having a nice sharp lens to start with. I've found that for my Canon 600D (T3i) fitted with the Sigma 10-20mm lens, I can get an acceptably sharp RAW image in ACR without significant halos using the following sharpening settings (this is assuming proper focus when taking the shot)

Amount 60-80
Radius 0.7
Detail 100
Masking about 20 but depends on image

I look at this sharpening adjustment as a correction for the loss of sharpness during the capture (due to AA filter, de-mosaicing and lens blur).

Dave

[Stagecoach]

I am acquiring a Sigma 10-20mm lens. Some say it is not as sharp as this or that lens. However, shouldn't Photoshop's Unsharp mask's ability to increase apparent sharpness and the RAW tool's ability to correct aberration and distortion also be considered. Don't these tools even the lens quality playing field significantly?

Ed,

If you are acquiring the lens the reality is that you will own it and there's not much you can do about the sharpness of it other than use the appropriate procedures such as tripod, remote release and mirror up to get the best out of it.

I also own the 10-20 Sigma f4-f5.6 and find it acceptable for my purposes. With respect to post work specific to this lens I do nothing different than my standard procedure for my other lenses. There will always be individuals opinions and data that suggests how good a lens is but it's going to come down to whether it meets your expectations.

Here's a landscape I took with the Sigma at 14mm a while ago, not a particularly interesting image and in rather damp cloudy hills. An idea of it's capabilities in one area.



Grahame

[Colin Southern]

Hi Ed,

The short answer is "Yes" - Proper sharpening does compensate a LOT for "sub-optimal sharpness at time of capture". In fact all RAW images need multiple sharpening phases regardless. Plus. if the images are only for typical internet resolution then any inherent sharpness difference will have been long since down-sampled out (when 95% percent of image information is discarded)

[Black Pearl]

I too have the f4-5.6 version and have found it to be a very good lens, very versatile, well made and once stopped down good and sharp. I suppose the last thing is the one to remember, these sorts of lenses need to be stopped down a good few apertures to get the best from them. You are asking a lot from a lens that is so wide - and a zoom - so while they can supply the results you need to be careful with them. Wide open it is pretty grim, usable for web sizes and small prints but not great if you get critical.

I have had mine since they were launched and have loved it from day one:

[Brev00]

Sharpness is something that is to some extent in the eye of the beholder. It is one of many things that a lens contributes to an image and the acceptable nature of this one aspect is often determined by the photographer more than 'reality.' It seems common knowledge that this particular lens is not as sharp as the less expensive f 4-5.6 model. Yet, I have seen many gorgeous images shot with it. Processing skills may play a part but there are other things to consider including flare resistance (important if you shoot into or near the sun), central resolution (more important than edges depending on your subject matter), distortion (ease of correction can be so important in ultrawides), and contrast. Some programs automatically correct for some of these issues but you will lose some of your wide angle if there is a lot of distortion correction. So, don't get too hung up on sharpness--unless you are a detail junky. I do not think going into a lens purchase and use with the idea of sharpening to compensate for lens flaws is a good idea. Oversharpening is one of the worst things one can do to an image and that is often the effect I see when someone is trying to attain sharpness that is not there initially.

[Colin Southern]

I do not think going into a lens purchase and use with the idea of sharpening to compensate for lens flaws is a good idea. Oversharpening is one of the worst things one can do to an image and that is often the effect I see when someone is trying to attain sharpness that is not there initially.

Hi Larry,

It sounds correct in theory, but in practice it's not that cut and dried. It's going to depend on the resolution of the final image and other factors. A modern typical camera may spit out something along the lines of 5000 x 3500 pixels per image and yet a typical image displayed on a monitor is probably only in the region of 1000 x 650 pixels ("ballpark" figures) - so, roughly, only one in every 5 pixels along any dimension (or 1 in every 25 square pixels) contributes to the final image - and when that's taken into account, any inherent lens softness has long been down-sampled out.

With a professional-quality lens we typically need capture sharpening of around 0.3px @ 300%; with lower-quality lenses sometimes that rises to 0.4px (occasionally 0.5px) @ 300%, and usually that's pretty much the extent of it. It doesn't have any effect on content/creative sharpening nor output sharpening.

To be honest, over the years I've seen many many many folks obsessed with lens sharpness - getting the best money can buy - and then presenting unsharpened images. In reality they're "majoring in minor things" (and "minoring in major things" if there's such a term). For sure, image degradation is ultimately cumulative, but if you take that a step further, you'll find that inherent sharpness doesn't contribute a lot to the average internet-sized image.

I should point out that I'm ONLY taking about inherent lens softness; NOT focusing / camera movement / subject movement / contrast issues etc - any of those can quickly escalate an image past the point where an "invisible repair" is possible (at which point sharpening becomes a compromise).

As much as I don't particularly like Sigma gear (and I'd never buy it), I none-the-less suspect that if I shot a landscape with that lens I really doubt anyone could tell the difference at normal internet sizes.

In Ed's case I'd suggest that good technique, good tripod & gear, and good post-processing will have a far far far greater impact on the final image.

[Brev00]

Hi Larry,

It sounds correct in theory, but in practice it's not that cut and dried. It's going to depend on the resolution of the final image and other factors. A modern typical camera may spit out something along the lines of 5000 x 3500 pixels per image and yet a typical image displayed on a monitor is probably only in the region of 1000 x 650 pixels ("ballpark" figures) - so, roughly, only one in every 5 pixels along any dimension (or 1 in every 25 square pixels) contributes to the final image - and when that's taken into account, any inherent lens softness has long been down-sampled out.

With a professional-quality lens we typically need capture sharpening of around 0.3px @ 300%; with lower-quality lenses sometimes that rises to 0.4px (occasionally 0.5px) @ 300%, and usually that's pretty much the extent of it. It doesn't have any effect on content/creative sharpening nor output sharpening.

To be honest, over the years I've seen many many many folks obsessed with lens sharpness - getting the best money can buy - and then presenting unsharpened images. In reality they're "majoring in minor things" (and "minoring in major things" if there's such a term). For sure, image degradation is ultimately cumulative, but if you take that a step further, you'll find that inherent sharpness doesn't contribute a lot to the average internet-sized image.

I should point out that I'm ONLY taking about inherent lens softness; NOT focusing / camera movement / subject movement / contrast issues etc - any of those can quickly escalate an image past the point where an "invisible repair" is possible (at which point sharpening becomes a compromise).

As much as I don't particularly like Sigma gear (and I'd never buy it), I none-the-less suspect that if I shot a landscape with that lens I really doubt anyone could tell the difference at normal internet sizes.

In Ed's case I'd suggest that good technique, good tripod & gear, and good post-processing will have a far far far greater impact on the final image.

I definitely agree, particularly with the final sentence, but what you are saying is different in approach than what the op suggested--I think. He basically asked if sharpening after the fact can compensate for poor sharpness at capture. Your response seems to be that essentially it does not matter because the flaws will not be apparent in the image as seen on the internet. I don't think the op was necessarily limiting the presentation of his images to the internet. He may be printing large for wall framing as far as I know. I would not want to buy a lens that could only perform in such a limited way. I really do not have your basis in knowledge especially about the ins and outs of sharpening. What I do know is that there are a lot of people who oversharpen their images and the flaws of doing this are apparent in internet sharing. So, I was more warning about this danger--of improper pp practices from overcompensating for inherent flaws.

I also have a sense that many people are much more concerned about ultimate sharpness than I am or others may be and maybe the Sigma 10-20 3.5 will suit his needs to a tee. Not only will it be hard to see the softness at internet levels but it may be even at normal to 100x viewing levels (if not 200-400x). I would buy it with a good return window. Test it out. And, then judge. I actually returned a Sigma 10-20 f4-5.6 when it did not pass my testing protocol (only time that has ever happened). I considered the 3.5 but ultimately went with the Tokina 12-24 with price being a factor along with wanting to give Tokina a shot.

[Colin Southern]

I definitely agree, particularly with the final sentence, but what you are saying is different in approach than what the op suggested--I think. He basically asked if sharpening after the fact can compensate for poor sharpness at capture. Your response seems to be that essentially it does not matter because the flaws will not be apparent in the image as seen on the internet. I don't think the op was necessarily limiting the presentation of his images to the internet. He may be printing large for wall framing as far as I know. I would not want to buy a lens that could only perform in such a limited way. I really do not have your basis in knowledge especially about the ins and outs of sharpening. What I do know is that there are a lot of people who oversharpen their images and the flaws of doing this are apparent in internet sharing. So, I was more warning about this danger--of improper pp practices from overcompensating for inherent flaws.

I also have a sense that many people are much more concerned about ultimate sharpness than I am or others may be and maybe the Sigma 10-20 3.5 will suit his needs to a tee. Not only will it be hard to see the softness at internet levels but it may be even at normal to 100x viewing levels (if not 200-400x). I would buy it with a good return window. Test it out. And, then judge. I actually returned a Sigma 10-20 f4-5.6 when it did not pass my testing protocol (only time that has ever happened). I considered the 3.5 but ultimately went with the Tokina 12-24 with price being a factor along with wanting to give Tokina a shot.

I think it really comes down to how good/bad the Sigma lens is, and to be honest, I have no idea as I've never used it. If it's as bad as the kit lenses of a few years ago then for sure, compensation in post-processing will only go so far if printing large and/or pixel-peeping, but if it's just an average performer then I honestly doubt that anyone would see the difference in real-world results, even if they were doing big prints.

First and foremost what I'm NOT saying is "don't worry about the capture, we'll just fix it in Photoshop" - and I'm also NOT saying that even with all the right techniques that certain lens deficiencies can be fixed with impunity - but what I AM saying is that for some characteristics such as pixel-level sharpness, it's generally not a big issue in the grand scheme of things; The example I like to use is the Bugatti Veyron or a mere Ferrari Testarossa to get a pint of milk from the corner dairy -- in theory the Veyron will get you there and back faster, but in real-world terms there are far more significant factors influencing the result. Same here I think. The best result will from the best lens with the best sharpening - the next best (and very close) will be an average lens with the best sharpening. Waaaay down the list will be a good lens with bad/no sharpening right next to an average lens with bad/no sharpening.