Luminous Landscape Forum
Equipment & Techniques => Digital Cameras & Shooting Techniques => Topic started by: dwdallam on June 10, 2006, 03:36:34 am
-
I'm not going to dig up the links, but there are several good sources that explain where the Canon L lenses mounted on a Canon digital back, such as the 20 D, will give the sharpest image. I think I read that for a typical lens, such as a 24-70L the best image will be recorded with an aperture between 8-11. This is where the equipment delivers the sharpest image given the aperture size.
My question is this: Does this also hold true with landscape photography where things are in the distance, with nothing closer than several hundred years (think shooting a sunset off of a cliff, with the closest foreground object 300 years away). In this instance, would an aperture of say 13-18 deliver a sharper image because of the increase in DOF in relation to the closest object, which is quite far itself?
In other words, can an increase in DOF from a smaller aperture ever override the sharpest aperture that a lens can deliver? That would be something like F8 compared to F16-18. Or would it be better to shoot a landscape with a very distant foreground object (Where close DOF is not a critical) with the lens' best aperture to sharpe ratio?
Thanks again.
-
In other words, can an increase in DOF from a smaller aperture ever override the sharpest aperture that a lens can deliver? That would be something like F8 compared to F16-18. Or would it be better to shoot a landscape with a very distant foreground object (Where close DOF is not a critical) with the lens' best aperture to sharpe ratio?
Thanks again.
[{POST_SNAPBACK}][/a] (http://index.php?act=findpost&pid=67835\")
As you move from f8 to say f18 the dof that was super sharp under f8 will be come less sharp and the oof range under f8 will become sharper, but never as sharp as f8. It's a trade off. No free lunch. Remember that what constitutes dof depends on your tolerance for blur.
The aperture issue relates to diffraction, which in turn relates to pixel size. Here's a good link.
[a href=\"http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm]http://www.cambridgeincolour.com/tutorials...photography.htm[/url]
-
As you move from f8 to say f18 the dof that was super sharp under f8 will be come less sharp and the oof range under f8 will become sharper, but never as sharp as f8. It's a trade off. No free lunch. Remember that what constitutes dof depends on your tolerance for blur.
The aperture issue relates to diffraction, which in turn relates to pixel size. Here's a good link.
http://www.cambridgeincolour.com/tutorials...photography.htm (http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm)
[a href=\"index.php?act=findpost&pid=67844\"][{POST_SNAPBACK}][/a]
Tim, thanks. That is one of the links I have read on the subject.
What I'm thinking you mean is that given a specific lens length and aperture, if you have a foreground object close enough to voilate the hyperfocal distance of a specific lens length and aperture, say f8 at 50mm, then moving to f18 will bring the foreground more into focus. However, you will lose some sharpness for objects located at the point in the distance where the combination of the lens length and aperture at f8 were dead on in focus. Is this right?
If so, then given my example above, where the closest foreground object is say 300 yards and is withing the hyperfocal distance of lens length and aperture say (f8-11), then it would be best to shoot with f8-11, obviously, as no increase in sharpness will be gained by using a smaller aperture.
I'm asking because it seems like in this particular instance, the nearest foreground object is probably 250 yards away, and I can shoot at f11 at about 40mm and have the foreground object what seems to be in good focus while keeping the more distant objects very sharp. See attachment.
-
Tim, thanks. That is one of the links I have read on the subject.
What I'm thinking you mean is that given a specific lens length and aperture, if you have a foreground object close enough to voilate the hyperfocal distance of a specific lens length and aperture, say f8 at 50mm, then moving to f18 will bring the foreground more into focus. However, you will lose some sharpness for objects located at the point in the distance where the combination of the lens length and aperture at f8 were dead on in focus. Is this right?
[a href=\"index.php?act=findpost&pid=67875\"][{POST_SNAPBACK}][/a]
Yes
If so, then given my example above, where the closest foreground object is say 300 yards and is withing the hyperfocal distance of lens length and aperture say (f8-11), then it would be best to shoot with f8-11, obviously, as no increase in sharpness will be gained by using a smaller aperture.
[a href=\"index.php?act=findpost&pid=67875\"][{POST_SNAPBACK}][/a]
Right again.
I just caution against thinking about the HF distance or a DOF calculation as binary - ie in or out of focus. The plane of focus is binary - if you focus at 50' then only objects in that plane are in focus and as you look at objects closer or farther away they will become less in focus (exponentially if I recall). The calculation of Hyperfocal Distance only plugs in your ability to perceive blur and everything within that threshold is said to be within the depth of field (to say "in focus" is not really correct). You have to remember that your tolerance to blur is subjective and will depend on how you intend to disply the image.
-
Doesn't the 20D start to hit diffraction around f/11?
If I was really obsessed with front to back sharpness I would definitely want to get one of the TS-E lenses. Set the lens to the sweet spot around f/5.6-f/8 and then use the tilt function to get the depth of field you want.
-
Tim, yes I was just using HFD as a term, not to indicate "focus" in reality. What I should have said is although both distant and close objects are in focus as well as the lens can focus, the sharpness of f18 will be less than f8 where both are focused on the same object. So if you focused on the same point, and the objects is far enough away, f8 will produce superior sharpness than will f18. However, as you point out, sometimes you will need to use f18 if you want to get the foreground objects sharper. However, as you stated also, you will lose some sharpness in the distance--even though the entire image will be sharper percentage wise and even though you did not change the focal point when changing from f8 to f18. In other words, your hyperfocal distance will increase, which you may need, but the distant objects will be less sharp compared to f8.
So in summary, always use the lens' best aperture unless it cannot be avoided, for reasons exaplained above. And this now seems obvious to me.
Yes
Right again.
I just caution against thinking about the HF distance or a DOF calculation as binary - ie in or out of focus. The plane of focus is binary - if you focus at 50' then only objects in that plane are in focus and as you look at objects closer or farther away they will become less in focus (exponentially if I recall). The calculation of Hyperfocal Distance only plugs in your ability to perceive blur and everything within that threshold is said to be within the depth of field (to say "in focus" is not really correct). You have to remember that your tolerance to blur is subjective and will depend on how you intend to display the image.
[a href=\"index.php?act=findpost&pid=67880\"][{POST_SNAPBACK}][/a]
-
Doesn't the 20D start to hit diffraction around f/11?
If I was really obsessed with front to back sharpness I would definitely want to get one of the TS-E lenses. Set the lens to the sweet spot around f/5.6-f/8 and then use the tilt function to get the depth of field you want.
[a href=\"index.php?act=findpost&pid=67915\"][{POST_SNAPBACK}][/a]
Well, that's how medium and large format landscape cameras work. I'm not sure the tilt lenses for 35mm digital backs work as well though. But it's a good point. At this point, after buying the Canon 17-35L, the 24-70L and the 70-200L IS, I'm not looking forward to yet another lens.
And by the way, the more you do full scape shots, the more you will become obsessed with front to back sharpness because you will become more obsessed with balance and synchronicity in your images, which in turn means you need foreground objects midway objects, and distant objects. Almost all top notch landscape images I've seen have this element to them. It's just the rule for me now. I mean even when you are shooting sand, sea and horizon, well, there you go--the three balancing objects--sand, leading the eye to the sea and the sea and reflection from the sun leading the eye to the horizon and probably the setting sun--just as an example.
But this is yet another topic, but you brought up a very interesting subject. Maybe you could start a thread? Perhaps I am dead wrong in this practice? And for sure, there are times when you can always break these rules and get an outstanding image. But tehn I would not know. I have yet to get an outstanding image--lol.
-
because you will become more obsessed with balance and synchronicity in your images, which in turn means you need foreground objects midway objects, and distant objects. Almost all top notch landscape images I've seen have this element to them. It's just the rule for me now.
Yes, this is a good rule of thumb for landscape photos although sometimes there are exceptions of course, like this shot (http://www.lashier.com/home.cfm?dir_cat=35192) where getting a foreground just wasn't possible. Other times it can be difficult, the most notable example being shots from mountaintops, although usually I manage. (http://www.lashier.com/home.cfm?dir_cat=35174&sr=1) It's amazing how dull a typical summit shot looks compared to actually being there.
- DL
-
Hey Don,
Thanks for sharing the mountain pix. They're great!
The only places you show that I've been to personally are the Athabasca and Saskatchewan Glaciers (sigh!).
Eric
-
What we really need is a combined laser/pointer/ distance calculator with built-in DoF calculator. For landscapes it is generally impractical to take a measuring tape to the nearsest object in the scene and then the object at the hyperfocal distance for focussing purposes. The distance markings on lenses are often not accurate enough, or detailed enough, so one usually has to rely upon guesswork.
An ideal device would be a laser pointer of the type used by golfers. However, having checked on the specs of such devices, they do not seem to be accurate enough. The types used by surveyors and architects are much more accurate, but also expensive and bulky. I'd say there is a niche in the market for such a device, aimed at photographers and preferrably including a DoF calculator.
Better still, a DSLR that could give a precise distance reading in the viewfinder, in relation to whatever the lens is focussed on.
-
I haven't seen any evidence to point to the various Canon (And Nikon has one too) Tilt/Shift lenses not getting the job done.
A tilt is a tilt regardless of the format. I haven't seen the Nikon one but the only problem with the Canon ones is they don't have the full range of movements that a bellows has. But for depth of field and basic perspective correction they seem to be fine.
It does suck that they all cost ~$1000 but I'll probably be getting one in the next year. Just a question of which one to get.
Canon would have to come up with a DSLR that could magically fill my wallet with money, allowing me to quit my job to get me to upgrade the body now that I have the 5D. So eventually I will have some money to get one of those TS-E lenses.
-
What we really need is a combined laser/pointer/ distance calculator with built-in DoF calculator. For landscapes it is generally impractical to take a measuring tape to the nearsest object in the scene and then the object at the hyperfocal distance for focussing purposes. The distance markings on lenses are often not accurate enough, or detailed enough, so one usually has to rely upon guesswork.
An ideal device would be a laser pointer of the type used by golfers. However, having checked on the specs of such devices, they do not seem to be accurate enough. The types used by surveyors and architects are much more accurate, but also expensive and bulky. I'd say there is a niche in the market for such a device, aimed at photographers and preferrably including a DoF calculator.
Better still, a DSLR that could give a precise distance reading in the viewfinder, in relation to whatever the lens is focussed on.
[a href=\"index.php?act=findpost&pid=67958\"][{POST_SNAPBACK}][/a]
Couldn't you just keep it simple and use two legs and a map?
For large distances a reasonable walking map should give accurate enough distances (at least to the nearest 10m or so).
For short distances one pace, two pace, three pace...or a good guess.
Obviously, if you do need to be very very accurate then perhaps you could dig out a GPS and walk between the two points, take readings, etc...but why rely on technology when people have been achieving decent results with a hole and eggy paper for years.
-
Yah KISS should get you through the day. If you can't get it done with a hyperfocal chart something is wrong.
It's only hard with very close subjects & wides, chances are with a tele you either a) Cannot physically do it Everything if far from the camera and it's easy.
-
You guys are very slack. If my DoF calculator tells me the hyperfocal distance is 50 metres for everything to be sharp between 20 metres and infinity (at f8), how do I find the 50 metre mark which might well be on the other side of a river?
-
Well other people have certainly managed to get the job done without any fancy gadgets.
Personally I don't think I have any lenses which actually have anything out to 50m marked on the lens anyway.. 50m = infinity for lots of lenses.
-
Isn't it true how dull most things seem to really being there? True enough. It's our job to try and bring out "some" of the interest. Rules help, but as you know, better to break them than to get nothing at all.
Yes, this is a good rule of thumb for landscape photos although sometimes there are exceptions of course, like this shot (http://www.lashier.com/home.cfm?dir_cat=35192) where getting a foreground just wasn't possible. Other times it can be difficult, the most notable example being shots from mountaintops, although usually I manage. (http://www.lashier.com/home.cfm?dir_cat=35174&sr=1) It's amazing how dull a typical summit shot looks compared to actually being there.
- DL
[a href=\"index.php?act=findpost&pid=67952\"][{POST_SNAPBACK}][/a]
-
Ben,
Now you got me curious. I'm going to check out those lenes. I still really want the 5D also. Hopefully in the next two months I can upgrade and gtet the tilt lens.
I haven't seen any evidence to point to the various Canon (And Nikon has one too) Tilt/Shift lenses not getting the job done.
A tilt is a tilt regardless of the format. I haven't seen the Nikon one but the only problem with the Canon ones is they don't have the full range of movements that a bellows has. But for depth of field and basic perspective correction they seem to be fine.
It does suck that they all cost ~$1000 but I'll probably be getting one in the next year. Just a question of which one to get.
Canon would have to come up with a DSLR that could magically fill my wallet with money, allowing me to quit my job to get me to upgrade the body now that I have the 5D. So eventually I will have some money to get one of those TS-E lenses.
[a href=\"index.php?act=findpost&pid=67984\"][{POST_SNAPBACK}][/a]
-
You can buy simply mechanical rangefinders. They cost about 10 bucks and are from what I read, very accurate.
Couldn't you just keep it simple and use two legs and a map?
For large distances a reasonable walking map should give accurate enough distances (at least to the nearest 10m or so).
For short distances one pace, two pace, three pace...or a good guess.
Obviously, if you do need to be very very accurate then perhaps you could dig out a GPS and walk between the two points, take readings, etc...but why rely on technology when people have been achieving decent results with a hole and eggy paper for years.
[a href=\"index.php?act=findpost&pid=67985\"][{POST_SNAPBACK}][/a]
-
You make your best guess and then take several pictures focusing on points more near and further waway than you think is teh distance you need. In other words, you bracket with distance.
You guys are very slack. If my DoF calculator tells me the hyperfocal distance is 50 metres for everything to be sharp between 20 metres and infinity (at f8), how do I find the 50 metre mark which might well be on the other side of a river?
[a href=\"index.php?act=findpost&pid=67996\"][{POST_SNAPBACK}][/a]
-
Well other people have certainly managed to get the job done without any fancy gadgets.
I do get the job done, but usually with the aid of a very fancy gadget ... the camera .
Personally I don't think I have any lenses which actually have anything out to 50m marked on the lens anyway..
This is precisely the point I am making. Distance markings on lenses are often useless. With a cropped format camera and 100mm lens at f11, using a CoC of 0.02mm, the hyperfocal distance is slightly less than 50m, actually 44.3m. Focussing on the 44m mark in the scene would give me an acceptably sharp image from 22m to infinity. If I get this slightly wrong and focus on a point say 60m away, then the foreground at 22m is likely to be slightly fuzzy.
If I'm using a 200mm lens at f11, the distances are greater and any markings on the lens even more useless. I'd have to focus on a point 177m away to get everything sharp from 88m to infinity.
Of course, if I'm using a wide-angle lens, say 24mm, there's much more latitude. Focussing on something as close as 2.5 metres gives a sharp result from about 4ft to infinity. If I get that slightly wrong and focus on a point 3 metres away, I've changed the nearest 'acceptably sharp' point by just a few inches, which is why I do not bother using the tilt on my TS-E 24mm for big picture landscapes.
-
Ray,
You are correct. There is no way to get an exact distance in the field in some instances. Bracket with distance.
I do get the job done, but usually with the aid of a very fancy gadget ... the camera .
This is precisely the point I am making. Distance markings on lenses are often useless. With a cropped format camera and 100mm lens at f11, using a CoC of 0.02mm, the hyperfocal distance is slightly less than 50m, actually 44.3m. Focussing on the 44m mark in the scene would give me an acceptably sharp image from 22m to infinity. If I get this slightly wrong and focus on a point say 60m away, then the foreground at 22m is likely to be slightly fuzzy.
If I'm using a 200mm lens at f11, the distances are greater and any markings on the lens even more useless. I'd have to focus on a point 177m away to get everything sharp from 88m to infinity.
Of course, if I'm using a wide-angle lens, say 24mm, there's much more latitude. Focussing on something as close as 2.5 metres gives a sharp result from about 4ft to infinity. If I get that slightly wrong and focus on a point 3 metres away, I've changed the nearest 'acceptably sharp' point by just a few inches, which is why I do not bother using the tilt on my TS-E 24mm for big picture landscapes.
[a href=\"index.php?act=findpost&pid=68047\"][{POST_SNAPBACK}][/a]
-
Ray,
You are correct. There is no way to get an exact distance in the field in some instances. Bracket with distance.
[a href=\"index.php?act=findpost&pid=68097\"][{POST_SNAPBACK}][/a]
Yes. Bracketing is easy with digital cameras. I bracket because I'm not exactly sure about the best exposure for ETTR. I bracket again at perhaps f11, f16 and f22 because I want maximum DoF and I want to examine the trade-off between maximum resolution at the point of focus compared with resolution at the extremities of the DoF range. I bracket a third time because I'm not sure of the best point in the scene to focus on, in order to maximize DoF. And maybe it is sometimes useful to bracket at different ISO settings, just to be sure I am using an adequate shutter speed to freeze movement of the subject.
According to my maths, that's at least 27 shots. Now that's fine. Pressing that shutter button doesn't cost anything, except the additional time I have to spend later sitting at the computer. I'm not sure I want to do that. When I return from a shooting trip I usually have about 5,000 shots to process including a modest number of bracketed shots.
I feel that I need a personal assistant who is a whiz at using Photoshop, who enjoys sitting in front of a computer all day and who will work for nothing. I could then concentrate on on just taking photos .
-
That's because you still need to know what you are doing or you do end up with 5, 000 shots. I use to do that also, and the images would go unprocessed for months just because it would take me eight hours a day to look through them all.
Now, when I shoot, say a sunset, I usually have about 30 images for a set up that takes around an hour. It's hard to be selective, but if you don't, it just takes too much time to go through them.
Yes. Bracketing is easy with digital cameras. I bracket because I'm not exactly sure about the best exposure for ETTR. I bracket again at perhaps f11, f16 and f22 because I want maximum DoF and I want to examine the trade-off between maximum resolution at the point of focus compared with resolution at the extremities of the DoF range. I bracket a third time because I'm not sure of the best point in the scene to focus on, in order to maximize DoF. And maybe it is sometimes useful to bracket at different ISO settings, just to be sure I am using an adequate shutter speed to freeze movement of the subject.
According to my maths, that's at least 27 shots. Now that's fine. Pressing that shutter button doesn't cost anything, except the additional time I have to spend later sitting at the computer. I'm not sure I want to do that. When I return from a shooting trip I usually have about 5,000 shots to process including a modest number of bracketed shots.
I feel that I need a personal assistant who is a whiz at using Photoshop, who enjoys sitting in front of a computer all day and who will work for nothing. I could then concentrate on on just taking photos .
[a href=\"index.php?act=findpost&pid=68123\"][{POST_SNAPBACK}][/a]
-
Perhaps if you're doing this it's best to invest in some older manual focus lenses.
When I looked at my lenses the other night the only one which even has any marks beyond 5m is my 300mm lens. Honestly I don't think I've ever even tried to do a hyperfocal shot with that lens... it's largely pointless considering how hard it is to stabilize and how much it needs to be stopped down to get any depth of field. The only landscape use I have for it is a telephoto shot at infinity of mountains, etc.. very far away.
All the other lenses I have I think only go out to 3m or so on the focus scale. My memory fails me but I could have sworn the old mechanical lenses were much better in this respect. It's not like autofocus is all that useful for this particular job.
-
Perhaps if you're doing this it's best to invest in some older manual focus lenses.
[a href=\"index.php?act=findpost&pid=68164\"][{POST_SNAPBACK}][/a]
No point in going backwards, Ben. It's not a major issue but something that might be addressed as technology advances. I'm actually surprised that we can't get a read-out of focussing distance in the viewfinder. Some stitching programs require distance to subject information. As I recall when I trialled the DXO lens correction software, it needed a distance to subject to do a thorough job, but EXIF doesn't seem to carry such information.
Perhaps a future camera will have an optional autofocus feature which searches for the hyperfocal distance, whatever the focal length. If you know you want the horizon (or infinity) sharp, there's no point in focussing at infinity if you are also striving to get maximum DoF.
-
I haven't seen any evidence to point to the various Canon (And Nikon has one too) Tilt/Shift lenses not getting the job done.
[a href=\"index.php?act=findpost&pid=67984\"][{POST_SNAPBACK}][/a]
could someone please direct me to a forum/topic or site where I could learn about how to use tilt lenses to correct perspective or increase the dof?
-
could someone please direct me to a forum/topic or site where I could learn about how to use tilt lenses to correct perspective or increase the dof?
[a href=\"index.php?act=findpost&pid=68176\"][{POST_SNAPBACK}][/a]
There'll be lots of examples if you do a Google search. Basically it's the shift that corrects for perspective and the tilt that skews the plane of focus resulting in an 'apparent' increase in DoF. DoF actually remains the same, depending on f stop and distance to subject. However, the parts of the image that are in or out of focus will change. This can have unwanted consequences. For example, if you wanted to take a close-up of an erect person whilst keeping an interesting background sharp, out to infinity, you might find the person's feet were sharp but head and shoulders not sharp. It sometimes requires a bit of messing around to get a good result and the job is made more difficult with a small viewfinder.
But don't let me put you off. They certainly 'work'. I use mine mainly for easy stitching of close subjects with little parallax error (using the shift of course). You see, I'm attracted to easy solutions. Complex pano heads with various adjustments in relation to focal length and distance to the subject are too fussy for me, especially when I can only make a wild guess as to distance to subject. I'm not even sure if a dedicated pano head would be able to do as good a job as a shift lens for really close subjects, say 3 or 4 metres away.