A high % of my images look soft when viewed at 100%. What to blame? AF? Technique? Lens?

Discussion in 'Canon EOS' started by nicolasraddatz, Feb 4, 2009.

  1. Hi all!
    I was recently comissioned my biggest job to date, a set of 50 photos that depict the work of a local FM radio. Some pictures are photojournalistic in style, others are more formal portraits. The budget was tight, so I had to work with available light in the radio station. No strobes, no flash. Just available light, since it was impossible for me to interrupt the normal flow of the radio work.
    Since most of my work has been used on the internet, I never really had to worry too much on image sharpness. Most *decent* images when shrinked to normal screen viewing resolution (smaller than 1800 x 1600) look flawless.
    BUT, now some of my work may be used for printing, so I needed to check much more carefully the images, and I'm really shocked at what I found. Many of the images have soft focus. For instance, I was looking at a series of portraits taken at something like 2 meters, just beside a well lit window, f:4.0 at 1/80s with Canon EF 50mm f:1.8 (a lens well known for its good optics). A lot of them (and I mean it) had soft focus, I don't think it's motion blur since speeds around 1/80 (which should be enough to freeze normal human movement in such a small place as a radio booth).
    Could it be something regarding autofocus accuracy? Right behind the main subject there was a wall, but the main subject filled something like 70% of the frame, so I guess it should still focus correctly. And with a 50mm lens (equivalent to 80mm on APS-C) working at 1/80 I guess I'm on the safe zone regarding body movement when taking the picture.
    I'm puzzled about this. What are your opinions?
    PS:
    Let me tell you a little about me. I've been taking pictures with some *serious* interest for the last 4 or 5 years. I started with a Canon S2IS P&S camera, moved to an EOS 400D (Rebel XTi) DSLR and since last year I've been working with film in B&W (rollei 35, EOS film body, etc). I've bought several photography books, joined a photo club, studied B&W techniques (development, printing), taken photography lessons, read, read an read more on the internet and forums, etc. Some of my pictures have been published on internet magazines, and artist's flyers. So, you get the idea regarding my background and knowledge.
     
  2. [[BUT, now some of my work may be used for printing, so I needed to check much more carefully the images, and I'm really shocked at what I found.]]
    Does the softness exist in the prints?
     
  3. It could be any of the things you mentioned. You will have to do some testing to determine the cause.
    1. 1/80 isn't all that great freezing subject motion, even if the subject motion is small, and particularly if the motion is across the frame and not forward/backward. Your handholding technique may not be up to snuff.
    2. It could be the lens/camera combo or camera or lens themselves. Some cameras and combinations have bad tendencies to back focus (typical Canon lens reaction is back focus). It could be the lens actually has a back/front focus problem. It could be your autofocus technique. It could be that you're using AI Servo on static subjects inside with dim light.
    The first thing to do is to look at your images and determine if there IS a point of sharp focus in them. If there isn't, the tendency would be to look at item 1 issues. If there is, note where the sharp focus point is--on the background, foreground, etc.
    Test your lenses using one of the many focus test targets availabe online to determine if they are really back/front focusing. If so, you need to consider calibration by Canon.
    Research good autofocusing technique. There are many previous threads about this. I'd employ these anyway, even if you find the real culprit is one of the above.
     
  4. The problem is you. Images viewed at 100% look soft. If you look at a print under a microscope it will look soft. If you look closely enough at anything you will see flaws.
    Unless your technique is bad or your focus is off, try unsharp masking and/or looking at the prints, not 100% screen displays.
    If the XTi can't produce files with enough detail for large prints, you'll need a 5D MkII
     
  5. While I can certainly see Bob's point, I have to say that when I view my 100 percent images in Lightroom, I can tell if the focus was right on or not, even without USM applied. However, if you were looking at JPEGS that were taken directly from your camera, Bob's point might very well be the case. If so, and you set parameters, where was the sharpness set?
     
  6. I agree with Bob. No point in judging your images at pixel level unless you're planning to print murals for close viewing. Check out any sharp print, neg or slide from yesteryear at 24X and they look pretty soft. Small prints are typically to be viewed from 18 to 20", not 4 or 5"...
     
  7. It may be a combination of pixel peeping at 100% and shooting at 1/80th sec. Try getting a print made and see how it looks and do your prints appear sharp in normal viewing mode on your comuter ?
    Ross
     
  8. I've routinely printed 17x22 prints from an XTi. It can certainly produce excellent prints at larger sizes. So I don't think you can blame the camera itself.
     
  9. Just checked the 100% view and as usual Bob is right - even the 5DII looks slightly soft at 100%. However, you should be OK at F4 for minor focus issues. I use an 85mm F1.2 at F1.2 for some portraits and even with this shallow depth of field I can get a good image at shutter speeds below 1/125 but above 1/60. I suspect you are either too critical (the impact of a 100% view). If the photo does not look sharp as a 16x20 print however then you have a bigger issue. The issue could be camera shake or focus problems - it is unlikely to be subject movement as for a static person in a portrait I have never found that subject movement is a problem at speeds above 1/60 some one trying to keep still is really only able to move less than 2-3mm in 1/60 of a second.
     
  10. A few thoughts...
    Just as with viewing small jpgs, what matters is how the photo looks as a finished print, etc - not how it looks at 100%.
    Are you shooting RAW? Are you sharpening in post? If so, how. If not, do! RAW files will not look good until you sharpen, in fact they will often look worse that jpg files from the camera which have been subject to default camera sharpening... and other processes.
    AF problems are a possibility, but I'd also want to eliminate shooting problems as a possibility first. It is easy to get the AF point on something other than the subject, or to have one of the peripheral AF points pick up focus rather than the point on your subject. You might try using only a single AF point and making sure it is on the subject.
    Camera motion is also still a possibility at the shutter speeds you mention, especially if you use a cropped sensor camera. 1/80 second is the "rule of thumb" minimum for a 50mm lens on crop. (1/50x1.6) If you make a bunch of 1/80 second exposures w/50mm on crop, there is a very good chance that you'll see motion blur, etc. in a good percentage of them at 100% magnification.
    Posting a sample or two would take the guesswork out of our responses.
     
  11. Can you post a sample pic or two? Dozens of eyes are better than two!
     
  12. zml

    zml

    Blame it on the habit of viewing at 100% on a fuzzy monitor with your eyeballs 3 inches from the screen :) Then, of course, the possibilities of booboos are endless: bad shooting technique (1/80s handheld seldom delivers pin-sharp results...), not using simple things like a reflector (be it a big sheet of paper...) to brighten up the scene, etc. Also - bad postprocessing technique (sharpening...) may be to blame.
     
  13. The larger the print the better the original has to be.
    I have 40"x30" prints in my studio, all were set up with off camera lighting, camera mounted on a tripod, manual focus, manual shutter & f stop. Used a light meter. Looked at the histogram. I used a cable remote. Once I got the couple into the scene, I had everything ready for them. I made one test shot, then, take a deep breath, relax, got them to smile with their eyes, bang, image done!
    Simple for the client once I had the bases covered before I got them involved with the scene.
    Hard to do this with the camera in auto focus, hand held, relying on in camera light meter, on camera flash unless you want a snapshot with a small print.
     
  14. Also depends upon your idea of "soft". If that means saturation, presence, impact - you will get better results with a flash at f4 than without one. At f8 & flash, better yet.

    Many people also confuse "softness" with "dynamic range". An image with greater dynamic range will appear "more sharp". High ISO = Low Dynamic Range, that's just the way things work with present technology. What ISO was used?
     
  15. I had a whole series of prints done from 40D (10mp) files at 84" x 48" and one done at 120" x 72". They all looked AWESOME. Even when i started looking closer. What you see at 100% on your screen doesn't necessarily translate to bad prints. But you won't know until you get some prints done...


    So I say get some prints done then come back here!
     
  16. Just another analytical tool: you could go into DPP or ZoomBrowser and check where your active focus points were for each shot--unless you used the "focus and recompose" method.
     
  17. If you were just firing away at 1/80th, then, yes I would expect some motion blur. If you were trying to minimise this and conciously holding as steady as possible then the blur should be much reduced. I find in good conditions the 100% views with my 50 1.8, 85 1.8 and L zoom are sharp. But if there is poor light, and I've auto'ed everything the full size view is not so good.
    I support the comments made regarding 1/80th not being sufficient for reliable sharpness, and the AF system being not perfect.
    Finally the comments regarding the end result being what's important are true, I made a mistake and shot a good subject slightly out of focus. At A4 size and some PP sharpening the mistake is bearly visible, and although nobody else can see this mistake the disconcerting bit is I know the shot is not perfect.
     
  18. I'd add that I agree that 1/80s with a 50mm lens on an APS-C DSLR is not a recipe for the sharpest images. For most people that would be right on the edge of being able to hand hold the camera steady enough for images that were going to be inspected closely.
     
  19. I'm pretty sure what you're describing is due to the lack of a tripod (necessarily given the situation you describe, but nonetheless, still a factor).
    I've compared shots from my lowly Canon G7 handheld vs. tripod-mounted and it was no comparison. In fact, I bet the tripod mounted shots would impress many DSLR users.
    People tend to underestimate the contribution of hand shake to image degradation, even when obeying the 1/focal length shutter speed "rule".
    Try a little experiment with your current set up: two identical shots, same settings etc, one hand-held, one tripod-mounted with self-timer or cable release.
     
  20. Also remember that the formula for todays cameras should be 1/(focal length x crop factor). In this case 50mm lens works out at 1/80th. As Bob says, you are on the borderline at this. I would suggest about 1/125 at least or use a monopod if a tripod is unwieldy in a confined space.
     
  21. 1/80s is definitely not "in the safe zone" with regards to pictures of people. I would set 1/200s if I can in such a situation, at 1/125s I start to see the effects of subject movement on parts of the body.
     
  22. "Also remember that the formula for todays cameras should be..."
    The 1/f "formula" was instituted in the days of grainy b/w films blown up to 3"x5" prints. 1/80 is *way* too short to guarantee pixel-sharp images every time on a multi-megapixel sensor.
     
  23. And, as my vague wording acknowledged, it is only a "rule of thumb," and not a "rule." Sometimes I can get sharp handheld shots at this or slower equivalent shutter speeds, though the success rate gets lower. But given the choice I would usually select a higher speed. And each photographer will have different success at a given speed - not everyone is equally steady and not everyone is equally careful.
     
  24. G Dan, you're "right on" with your observation IMO. I know I will get jumped on for this, but I've noticed that more experienced photogs seem to do better at hand held exposures. There is some aspect of "learning" to be motionless during the exposure. Many of my newer students have a big problem with this. I grew up in the country and "Rifle Training" was almost mandatory. Shooting an exposure is much like shooting a rifle. Squeeeeeeeze the trigger/shutter firmly, be very still...

    Over the years, I've moved my "threshhold" Tv down to 1/60 because physically, I can still support that Tv. For an experienced photo "athlete" in good physical condition, 1/80 s/b a piece o' cake up to at least 70mm...possibly even 85.
     
  25. Thanks for that, Christopher.
    You can definitely train yourself to shoot a slower shutter speeds. Most of the time when I shoot handheld I'll use an IS lens, but sometimes I go out and shoot with a prime. When I do so and I want to get a shot that is not significantly affected by camera motion blur I have to concentrate more on what I'm doing. Basically, if I just "point and squeeze" the shutter release I need to use higher shutter speeds than when I think about how I hold the camera, how I breath as I make the shot, and how I release the shutter.
    Of course, I may do all of this even when I'm using an IS lens - just as general good practice, but also because I may be pushing the shutter speeds even lower in those cases.
    Dan
     
  26. Chris is right, it's truly a learned skill. With my 5Dmk2 / 24-70 2.8 I can easily shoot 1/80 and with non-moving subjects, 1/50 -> 1/60 isn't too much of a stretch. Thats a far cry from a few years ago when 1/125 would result in blurry shots.
     
  27. Thanks to all for the replies. I've been in a rush, so couldn't get back to reply to you earlier, but some of your suggestions did help my second shooting session.

    @ Rob Bernhard

    "Does the softness exist in the prints?"


    No, the softness is visible on-screen when zooming into the picture (and not necessarily at 100%).


    @ Nadine Ohara - SF Bay Area/CA


    "Your handholding technique may not be up to snuff."


    I generally get quite decent shots at low speeds, so I'm not sure it's that. Besides, the unsharpness doesn't look like motion blur, but rather as soft focus


    "It could be that you're using AI Servo on static subjects inside with dim light."


    Bingo. I've now switched to one shot for static portraits, and despite the fact that I'm still not getting all the pics correctly, I think I'm now getting a much better good/bad ratio. I had somehow *assumed* AF was something not to think too much on. I was wrong. There's a whole world of things to take into account when it comes to AF.


    "Test your lenses using one of the many focus test targets availabe online to determine if they are really back/front focusing. If so, you need to consider calibration by Canon."


    I'll definitely get into that as soon as possible.


    @Bob Atkins

    "The problem is you. Images viewed at 100% look soft. If you look at a print under a microscope it will look soft. If you look closely enough at anything you will see flaws."

    Well, the problem is probably me, but not for the reasons you suggest. I know RAW images look soft, but I also know when something is in focus or not. In some of the images I don't even have to pixel peep to notice they are soft focused. I can notice just by looking at the picture shrinked to fit the window in lightroom.


    @ Philip Wilson

    "I suspect you are either too critical (the impact of a 100% view). If the photo does not look sharp as a 16x20 print however then you have a bigger issue. The issue could be camera shake or focus problems - it is unlikely to be subject movement as for a static person in a portrait I have never found that subject movement is a problem at speeds above 1/60 some one trying to keep still is really only able to move less than 2-3mm in 1/60 of a second."


    I don't have a printer at home, but it seems like a good exercise to go out and have a print made to be able to compare what I'm seeing with a printed copy. For the same reasons you mention, I think the problem is focus, not movement (either camera or subject movement).


    @ G Dan Mitchell

    "AF problems are a possibility, but I'd also want to eliminate shooting problems as a possibility first. It is easy to get the AF point on something other than the subject, or to have one of the peripheral AF points pick up focus rather than the point on your subject. You might try using only a single AF point and making sure it is on the subject."


    I did exactly that for my second shooting session, using central AF point and recomposing, and my results were much better, if not 100% perfect, much, much better!


    @ Sarah Fox

    "Can you post a sample pic or two? Dozens of eyes are better than two!"

    As soon as I can!


    @ Michael Liczbanski

    "Also - bad postprocessing technique (sharpening...) may be to blame."

    No protprocessing at all.


    @ Christopher Hartt
    "What ISO was used?"

    ISO 400. Not too much. Not too little.
     

Share This Page