DSLR mediocracy or is it me or too high of expectations

[telyt]

focus fixer
Marc

Over-sharpened.

[stever]

slobodan, please explain why you are impressed by this image.  the quality of the image refllects the poor conditions under which it wa created - so what.

[ErikKaffehr]

Hi,

Yes manual focusing using live view is best. I absolutely agree and that is what I would use on my Sony Alpha SLT55, the only camera that has live view. But I still maintain that focusing on ground glass has similar issues to auto focus.

It is also my belief that viewfinder magnification is not good enough for critical focus. I have an example:

I was shooting at Oxbow Bend, and saw something that could be a boulder, a cow or something else. I simply could not see in my viewfinder. Than I made a shot using AF, and could see that it was an elk with a fine rack. No way I could see the rack in the optical viewfinder! So how can I focus on what I cannot see? I enclose full image and 1:1 crops. Now useful picture, just illustration.

Now, of course I have not perfect sight, but with the corrective glasses I think it's about 20/20. At that time the environmental conditions were also not kind, sunlight coming from the side and quite cold, causing some tearing in my eyes.

But any way, I don't think viewfinder has a similar resolution to the sensor, even with perfect sight I presume it to be less. Assumption is based on the viewfinder image being quite small. It's a projected image so it's hard to say how big it is but I'd say it corresponds to around 4x5" at 25 cm, and I normally print at 15x23". I don't think I see the same amount of detail in the smallish viewfinder image as in the big print.

Best regards
Erik

Look, guys... I am the first one to use AF most of the time... simply more practical, and in variety of situations, vastly superior to my, or average Joe's manual focusing skills. But... and that is a big but (no pun intended)... when it comes to careful, critical focusing, using live view and 10x loupe, in most situations it will beat AF, simply because of the tolerance range AF has by design. You never know what your lenses are capable of unless you do this test and see for yourself.

Eric, rangefinders have less than precise focusing? Who would have thought so?  

[Slobodan Blagojevic]

slobodan, please explain why you are impressed by this image.  the quality of the image refllects the poor conditions under which it wa created - so what.

I guess you are suggesting the quality of the image is equally poor as the conditions were? In which case we are coming to the crust of the matter: people are looking at the same thing and seeing it differently. For me, I am impressed that the image is as sharp as it is IN SPITE of the poor conditions, and you apparently see its sharpness as poor. I am impressed that I can see individual cables/ropes at the very top of the highest pole.

Remember, the questions centers on being impressed by a file. Being impressed or happy in life in general is a simple math function of achieved over expected/desired. Obviously, people who do not expect/desire much are easier to impress or be happy. In other words, as your expectations approach zero, the happiness approaches infinity . And vice versa, there are people who could never be happy/impressed by anything, as their expectations approach infinity.

So, when the OP claims he is unimpressed by his files, is it because the files are indeed poor, or is it because he is hard to please, so to speak?

P.S. Btw, we are looking at a crop at 100%, here is the whole image:

[telyt]

It is also my belief that viewfinder magnification is not good enough for critical focus. I have an example:

I was shooting at Oxbow Bend, and saw something that could be a boulder, a cow or something else. I simply could not see in my viewfinder. Than I made a shot using AF, and could see that it was an elk with a fine rack. No way I could see the rack in the optical viewfinder! So how can I focus on what I cannot see?

Viewfinders are not all the same.  IMHO it's inappropriate to claim that "manual focus" is not accurate.  It's  "manual focus with this viewfinder" that you're having problems with.

[pegelli]

The other issue I have noticed is that depth of field with todays high resolution sensors is incredibly short. With a 150 mm lens at f(8 we would definitively see difference in sharpness on an object 150 m away depending on lens being focused at 150 m or infinity.

Thanks for the test and the article. Very useful information and indeed when looking at 100% crops the depth of field is much smaller than the old fashioned dof tables that came with analog film camera's.

However I always understood the classical dof tables (or the CoC definition used to create them) were based on what still appeared "sharp" to the human eye when a well produced print was viewed from a distance of about the diagonal size of the print.

So indeed when you look at dof in 100% crops digital greatly reduced it, however in a practical sense (looking at a print) I think there is no difference between the past (analog) and current (digital) pictures. In other words the right/top dollar bill would still look sharp when the whole frame was skillfully printed and then viewed from the proper distance.

Is this also your understanding, or am I missing something?

[michaelnotar]

i have always noticed that AF accuracy seems to greatly reduce with distance, esp 50ft+. wide angles seem to be worse, telephotos better. experiences within 50ft are good (with any length lens).

well i havent had a 1 series for a while, i was selling my 1d3 early this year. so i am getting used the beast again on my 1d4, but have used the 1 series since 2001.

how is it i can take 2 shots in a few seconds, completely refocus each shot, but focus on the same subject/AF point/same competition; and one is good and the other is not.

[ErikKaffehr]

OK,

To rephrase the question. How large is the viewfinder image at 25 cm (which is normally regarded close limit for eye sharpness)? To achieve correct focus the viewfinder image needs to be the same size as your final print. In addition ground glass, mirror, sensor and lens must be adjusted within 0.01 mm.

I'd suggest that correct focus can be estimated by turning focusing ring back and forth and finding midway far and short focus, but for that you need a lens with long focusing rotation, AF lenses tend not to be built that way. I'd also say that some photographers are better at focusing manually than others.

Folks with much more experience than me have looked at this. Erwin Puts who seems to be an authority on Leica lenses claimed that critical focus can only be achieved using focus bracketing. One of the problems is that many lenses shift focus when stopped down. So a lens would not have the same focus at f/1.4 as at f/4.0, telephoto lenses may not have this problem.

That said, I think I'm familiar with your bird pictures and I'm impressed by their sharpness.

Best regards
Erik

Viewfinders are not all the same.  IMHO it's inappropriate to claim that "manual focus" is not accurate.  It's  "manual focus with this viewfinder" that you're having problems with.

[David Sutton]

i have always noticed that AF accuracy seems to greatly reduce with distance, esp 50ft+. wide angles seem to be worse, telephotos better. experiences within 50ft are good (with any length lens).

well i havent had a 1 series for a while, i was selling my 1d3 early this year. so i am getting used the beast again on my 1d4, but have used the 1 series since 2001.

how is it i can take 2 shots in a few seconds, completely refocus each shot, but focus on the same subject/AF point/same competition; and one is good and the other is not.

Just some random thoughts....
If it was the camera (for example the 1D4) you would expect better results from the other model. If there are focussing errors with both then it is logical to assume that either it is user error or the cameras are not up to it.
It could be an either/or but often it is a combination of factors and at least probably not user error.
Canon cameras do have a history of focussing errors. However a friend who uses the 7D gets sharp images on the whole, but reports back that the choice of focussing mode is important and often using a single focussing point in real life doesn't work at all. Also, there is not a lot of depth of field up close at f5.6, and it is possible that one shot hits the eyelashes but the next one gets the eyebrow and the eye goes out of focus. Especially if the lens is slightly front focussing. I see you've calibrated the lens but I've found using LensAlign that if I get the lens accurate at say 70mm, then it is slightly out at 200mm, so I set it up at an average and keep in mind how much it will be out at the two extremes.

[ErikKaffehr]

Hi,

You are right in your reasoning, but I have a couple of objections:

The major one is that we print larger and viewing distance increases but not proportionally to image diagonal.

The other factor is that we do a lot of pixel peeping with digital. Perhaps bad but still a fact.

Three personal reflections:

When I switched from 135 film to 120 film I was dissatisfied with the sharpness of my images. The only reason I switched to 120 was to get better sharpness and not equal sharpness. I started to reconsider depth of field and lived happy.

The other reflection is that there is enjoyment looking at fine detail in a good and large print. That is pixel peeping, but also one of the main reasons to print big.

The third one is that we cannot get everything into focus in high res digital images. So instead of trying to get everything sharp using extended DOF we can try to use selective focus as an artistic tool.

I also happen to have an article on expanding DOF using different techniques:

http://echophoto.dnsalias.net/ekr/index.php/photoarticles/29-handling-the-dof-trap

Best regards
Erik

Thanks for the test and the article. Very useful information and indeed when looking at 100% crops the depth of field is much smaller than the old fashioned dof tables that came with analog film camera's.

However I always understood the classical dof tables (or the CoC definition used to create them) were based on what still appeared "sharp" to the human eye when a well produced print was viewed from a distance of about the diagonal size of the print.

So indeed when you look at dof in 100% crops digital greatly reduced it, however in a practical sense (looking at a print) I think there is no difference between the past (analog) and current (digital) pictures. In other words the right/top dollar bill would still look sharp when the whole frame was skillfully printed and then viewed from the proper distance.

Is this also your understanding, or am I missing something?

[Paulo Bizarro]

Hi,

Why do you think manual focus is better calibrated than autofocus?

Best regards
Erik

Assuming I have perfect eyesight, two things can happen when I use manual focus:

1. I focus the lens on the intended subject, and I can see in the viewfinder (or live view camera LCD, or thethered comptuer screen for that matter) that the image is indeed focused;

2. Or, using any of the above, I can not get the image focused. If this is the case, then indeed there may be something wrong or misaligned inside my camera.

The bottomline for me, after using SLRs for 20 years, I will trust manual focus more than autofocus for critical work. Of course, a proper focusing screen for manual focus is mandatory. As is a lens that has a good focus throw and "feel" for manual focusing (not that many in Canon land).

With DSLRs, and digital in general, people blow up their images to 100% and complain things are not sharp... combine that with lack of knowledge/experience on how to use the gazillion options for autofocus in modern day cameras, and that gives a powerful recipe for failure...

[telyt]

To rephrase the question. How large is the viewfinder image at 25 cm (which is normally regarded close limit for eye sharpness)? To achieve correct focus the viewfinder image needs to be the same size as your final print. In addition ground glass, mirror, sensor and lens must be adjusted within 0.01 mm.

You don't have to see as much detail on the viewscreen as you would in the final print.  Human eyes are much better at detecting contrast than they are at detecting detail.  Photojournalists in the 1960s and 1970s knew this, they focussed by maximizing contrast on their Nikon E viewscreens.  It's the technique I'm using for my photos.  100% of the photos on my website (no exaggeration) were made using this technique, and my eyes are now 59 years old.

AF systems are subject to the same "ground glass", mirror, lens mount and sensor alignment issues (substituting AF sensor for "ground glass") with the additional issues of discrepancies between actual and nominal focal length and the granularity of the AF motors.

As an aside, I could get on my soap box but instead I'll make a very brief comment about the absurdity of using AF for landscape photos, however I recognize the camera makers have been providing little else and aspiring photographers who prefer to use a modern digital camera have little choice in the matter.

[pegelli]

Hi,

You are right in your reasoning, but I have a couple of objections:

The major one is that we print larger and viewing distance increases but not proportionally to image diagonal.

The other factor is that we do a lot of pixel peeping with digital. Perhaps bad but still a fact.

Three personal reflections:

When I switched from 135 film to 120 film I was dissatisfied with the sharpness of my images. The only reason I switched to 120 was to get better sharpness and not equal sharpness. I started to reconsider depth of field and lived happy.

The other reflection is that there is enjoyment looking at fine detail in a good and large print. That is pixel peeping, but also one of the main reasons to print big.

The third one is that we cannot get everything into focus in high res digital images. So instead of trying to get everything sharp using extended DOF we can try to use selective focus as an artistic tool.

I also happen to have an article on expanding DOF using different techniques:

http://echophoto.dnsalias.net/ekr/index.php/photoarticles/29-handling-the-dof-trap

Best regards
Erik

Thanks Erik, no objections to your reasoning. It shows sharpness (and its perception) is highly depending on the assumptions you make. So it's not only digital that makes dof seem smaller, but also the fact we print larger and look closer (to a print or pixel peeping).

Only thing maybe to add is that in the digital age sharpening is also easier applied then in the old film days, USM and deconvolution sharpening are relatively easy software manipulations, while in the darkroom they were either much harder to do or impossible. So the perception of sharpness when viewing from a certain distance can be improved much easier, however detail that is not there cannot be regained.

[ErikKaffehr]

Hi,

Just a comment. For some reason we have an obsession with image quality. That is the reason we chase more megapixels and some of us even buy quite expensive MF digital backs. I think it's much about satisfaction.

Regarding sharpening, it's an interesting area. I just hung an exhibition print today with a couple of coyotes I shot (with camera!) in the Grand Teton NP. The image is not really sharp, due to movement, panning and slow shutter time. The coyote in the center is pretty sharp, tough and I sharpened it selectively using Topaz InFocus. I much like the result.

http://echophoto.dnsalias.net/ekr/images/LLVJ/coyotes.jpg

Best regards
Erk

Thanks Erik, no objections to your reasoning. It shows sharpness (and its perception) is highly depending on the assumptions you make. So it's not only digital that makes dof seem smaller, but also the fact we print larger and look closer (to a print or pixel peeping).

Only thing maybe to add is that in the digital age sharpening is also easier applied then in the old film days, USM and deconvolution sharpening are relatively easy software manipulations, while in the darkroom they were either much harder to do or impossible. So the perception of sharpness when viewing from a certain distance can be improved much easier, however detail that is not there cannot be regained.

[marcmccalmont]

Over-sharpened.

give me a full size RAW!

[BernardLanguillier]

is any one truely impressed by an slr file at 100%, it always seems mediocre. MF sees easier to focus due to a longer focus throw and bigger viewfinder but focus is very critical, more so for stabilization of the camera really.

Hum... Are DSLRs the same? I am not sure why you generalize what appears to be a focusing issue with the copies of the bodies you own to be a generic DSLR issue relative to MF.

To start, you could:
- try using live view
- send your bodies back to Canon for inspection
- try equipment from other DSLR brands with a better reputation in terms of focusing

I can feel frustration, but I am not sure it is directed at the right target.

Cheers,
Bernard

[ErikKaffehr]

Hi,

The longer focusing throw certainly helps, because it helps finding focus by rocking back and forth and finding mid point, which probably has optimum sharpness. Depth of focus is equal on near and far side, so midpoint is best focus.

On the other hand I doubt that focusing screens are perfectly aligned, and cannot normally be individually adjusted. DSLRs have focus adjustment and more importantly live view. LV uses the actual sensor image for focusing, so flange distance tolerances are not an issue. On the other hand LV does in no way guarantee that lens in properly aligned with sensor. But it can guarantee correct focus at the point of focus.

Best regards
Erik

Hum... Are DSLRs the same? I am not sure why you generalize what appears to be a focusing issue with the copies of the bodies you own to be a generic DSLR issue relative to MF.

To start, you could:
- try using live view
- send your bodies back to Canon for inspection
- try equipment from other DSLR brands with a better reputation in terms of focusing

I can feel frustration, but I am not sure it is directed at the right target.

Cheers,
Bernard

[aman74]

Subscribed.

[torger]

My experience is that current digital SLRs has so high resolution that it is unfeasible to get maximum sharpness hand-held in most situations. Few images get to be as sharp as the maximum the lens and sensor allows, but most get to be "acceptable sharp" when not pixel-peeped. Small errors in focus is also easier detected due to this high resolution. Hand-held 10-12 megapixels or so seems to be "the limit" of what is feasible, higher resolution than that requires tripod, mirror-up, cable release etc.

I hope future DSLRs will have say 40 megapixels (for still life photo) and a 4-in-1 pixel-binning mode of 10 megapixels suitable for hand-held action.

[pegelli]

I hope future DSLRs will have say 40 megapixels (for still life photo) and a 4-in-1 pixel-binning mode of 10 megapixels suitable for hand-held action.

Since you can do that in PP (with a choice of different algorithms for different situations) I would not care so much for that in cameras. Only advantage would be the smaller size of your image files, but other than that I see no real advantage. Or am I missing something else?