1FOCUS PROBLEMS WITH DIGITAL CAMERAS
September 30, 2005
As we move further and further into the digital camera age many photographers are discovering that focus is a much more critical issue than it used to be. There are two things to understand. Why is this the case? What can photographers do to avoid or solve the problems?
Why Is Focus An Issue?
Let's define sharp. It used to be that sharpness was determined by examining a transparency with an 8X loupe and if the image looked sharp at that magnification then once it was reproduced on a magazine page at 8 ½ x 11 or smaller it would probably look sharp to a magazine reader, at a normal viewing distance. If you examined the printed picture with a magnifying glass you might see a slight softness, but you would also see gain from the film or the screen from the printing. It might have been possible to get increased apparent sharpness by using a larger piece of film but in all cases the grain would remain an issue.
With digitally created images there is no grain. It is now possible to create images with much more apparent sharpness and fine line detail than it used to be. Every hair in an eyebrow or every pore in the skin can be distinguished on close examination of an 11"x19" or larger print. If you step back to a normal viewing distance for a print this size your eyes might not distinguish this detail (at least mine can't) but it's there and prints from digitally created images tend to look crisper and sharper.
The standard for evaluating the quality of a digital image is to view it at 100% magnification on a high-quality monitor. If you're examining a 16MP file this is equivalent to looking at a poster sized print from 10 inches away. Normally, people don't view posters prints from this distance, but that's what many agencies are doing to determine if an image meets their standards for acceptance. Some think it is unfair for agencies to have such high expectations, but as the percentage of digitally produced images keeps growing compared to film produced ones buyers are coming to expect this standard of sharpness.
The fact that sharper pictures are now possible is great, but more and more photographers are discovering that even though they're sure they carefully focused on a particular point, used a fast shutter speed and mounted the camera on a tripod they are getting soft pictures and a point focus not where they intended. Most often this happens in the auto-focus mode, but some photographers report the same problem when they use manual focus.
Identifying The Problem:
When photographers produce un-sharp images the first reaction tends to be to blame themselves, and try to be more careful with their procedures the next time. But when it starts happening regularly and unpredictably it becomes a more serious issue. If a lot
of your shots look soft it may not be you or your eyes. It may be the equipment.
Photographers using all the major digital cameras are reporting focus problems, but we're hearing of the problem most frequently from Canon 1Ds Mark II users. This is probably because more professional photographers are using Mark II's than any other camera.
Still, the total number of complaints I've heard about are infinitesimal compared to the total number of MarkII's in use. And as Chuck Westfall, Director/Media & Customer Relationship of Canon's Camera Marketing Group says no manufacturing process is perfect and it is inevitable that equipment problems will sometimes occur. However, the industry experience of the people complaining, and the specifics of what seems to be happening are enough that many professional might find it wise to examine their production with a much more critical eye. Read on.
Here are some things that leading professional stock shooters have experienced.
Walter Hodges of Color Blind Images always uses a tripod and uses three Mark II bodies interchangeably on any given shoot just to make sure that if anything goes wrong with one body he'll still get some good images from the other two. He told me "The other day I started a shoot using all three bodies and everything went fine until 11:00am. Then, all three bodies went soft at the same time."
He finds that often when he uses the autofocus to focus on the model's eyes the point of critical focus appeared to be a few inches behind the eyes on the hair or ears. This slight variation becomes very critical when your only stopped down one or two stops from the maximum aperture in an effort to fuzz the background. But it doesn't only happen when you're shooting near wide open. Photographer Jon Feingersh photographed a group of athletes in a huddle at 1/200 at f/18 with the faces lit with strobe. He focused on one of the faces. All the faces are soft, not really out of focus, just not SHARP.
I've examined a number of these images taken by different photographers. Some are out of focus by anyone's definition, but a high percentage might have been acceptable for full page reproduction, or smaller, by 1995 film standards. They might be as sharp as you will be able to get if you set out to drum scan many 35mm transparencies today. But, now we're talking images created digitally, and because digital has no grain, there is the "potential" for much greater sharpness. And that level of sharpness is what agents and customers have now come to expect. It used to be that we didn't worry about each individual hair in the eyebrow being sharp and distinguishable, or being able to distinguish every pore on the face. The grain of the film was bigger than these details. Now, professionals expect this kind of sharpness. It is routine, when preparing a digital file for a stock agency, and working on a waist up photo of a model, to give that photo a digital manicure because the roughness of the cuticle is distinguishable at 100% and that is enough to get the photo rejected by the agency.
Feingersh had other problems when the lens was stopped down. In one case he was shooting a headshot of an athlete outside. The camera was on a tripod and the subject was 8 feet away. The exposure was 1/200 at f/7.l. Feingersh focused on the face. The hairline appears to be in focus, but eye (same exact distance from camera) is soft.
One of the strangest things is that in many cases there seems to be absolutely no depth of field, even when shooting at f5.6 or f8. I have seen pictures of Feingersh's that were shot from 8 feet with an 85mm lens where there appears to be a point of sharpness, but an inch closer to the lens or an inch further away is not sharp. It's hard to believe that there isn't a wider range of sharpness at those f-stops. I've also seen Feingersh portraits taken at other times with the same camera that are absolutely needle sharp throughout the image.
In still another studio portrait Feingersh used the camera on a tripod with a Canon 135mm lens at 1/200 at f/16. The subject was lit with an HMI to provide plenty of light for focusing. The image was then exposed using a ring-flash. This resulting image was sharp enough to see that the model was wearing contacts, but not really "sharp"
When Canon was asked about the problem at first the technicians insisted that they didn't know of any general focus problem with the camera. Finally, Chuck Westfall said, "in some cases, it is certainly true that EOS-1Ds Mark II owners are experiencing equipment problems resulting in mis-focused images. But the fact of the matter is that when a focusing problem can be demonstrated, it can be fixed."
In Europe Canon France and Germany has admitted that there are focus problems on some 10D cameras. They said, "the range of tolerance of an autofocus is larger when working with film than when holding picture information on a chip." Thus, they say that if the point of sharpness lies before or behind the ideal point of focus the tolerance will be smaller with digital cameras. This is particularly a problem when using a long focal length lens.
Jessica Spruill, assistant manager for Jack Hollingsworth studios says Jack has two Mark II's and two 1Ds' and the studio has out-of-focus problems with all bodies. She says that 50% to 70% of the frames Jack shoots are un-sharp. The cameras also tend to go in and out of focus rather than once they are out of focus being consistently soft.
Feingersh also had a situation where the subject was 8 feet from the camera and the background 15 feet. He focused on the subject, but the subject was out of focus and the background was sharp. One of the strangest things is that in some cases part of the image was sharp, but other parts on the same plane appear out of focus.
Other photographers have observed that the Mark II's does not seem to focus on all colors equally well and that there seems to be greater problems when focusing on the color red.
This summer Hodges got to the point where in order to do any production with confidence he needed to bring multiple laptops hooked to 20 inch monitors onto the set and checked each frame for sharpness before he moved on to a new situation. As soon as he filled a flash card, he would pass it to an assistant who would check all the images on the high-resolution monitor at 200%. Hodges believes that in a month he lost over 4,000 frames due to sharpness and says, "Often we have shot set ups multiple times in an effort to get one frame in focus." For the last two weeks of August he closed down all normal production shooting and went to full time testing in an effort to find a solution to the problem.
Feingersh also immediately downloads all images shot in studio or on location for checking. If he is in the middle of a session, and wants to be doubly-sure of the results, he will keep the models on the set and download the card long before it is finished while everyone waits to be sure the images are good. He says, "this is a ridiculous and expensive exercise."
Both Feingersh and Hodges recently sent their cameras to Canon's Factory Service Center and they came back fixed and performing as they were designed to perform. Hodges says, "I shot about 500 frames using all three bodies and every frame was sharp." However, neither photographer is confident that they will stay that way for long because the cameras had been to the service center before. In addition Canon provided no explanation as to what might have caused the problem of whether there is anything the photographers can do to prevent the problem from re-occurring in the future.
Westfall says, "Canon is quite sensitive to the concerns of its customers. We certainly understand that customers want their Canon equipment to perform as well as possible. That is one of the main reasons we produced a PDF document entitled "Camera Handling & Maximum Image Quality," which is available for download via http://www.photoworkshop.com in the Canon Digital Learning Center section of that web site. In this document, we identify many of the best practices that are essential to achieve top quality images with EOS Digital SLRs including the 1Ds Mark II. In the 10+ months since this document has been posted, it has been downloaded more than 100,000 times. And as an ancillary result, our repair volume on EOS Digital SLRs has also gone down considerably even as sales continue to increase."
While this 35 page document provides a lot of detail and is probably must reading for anyone using a Mark II, it doesn't seem that there is anything recommended in it that the photographers above have not tried. Another document that may be useful can be found on Fred Miranda's site at www.fredmiranda.com/forum/topic/241524. It is a long article by RDKirk based on material in Canon's "Lens Work III," the description in their US patent application, and remarks by Chuck Westfall.
In this report a possible, if not necessarily desirable, solution to the problem is suggested. Kirk said that the center focus square in the viewfinder has both a horizontal and vertical sensor, while all the other focus squares look at the subject on only one axis, (according to the shape of the marks) either horizontal or vertical. Consequently, the center square can focus just as well on either vertical or horizontal lines of detail. In all cases the sensor focus best on lines of detail that are perpendicular to its orientation.
As a result Hodges made a B&W checkerboard with squares about 3 inches in size when using an 85mm lens. (The squares need to be larger when using a shorter focal length lens.) An assistant places the checkerboard at the point of focus and Hodges focuses on it, not the subject. For example, if he is shooting a portrait of a woman and he wants to focus on the eye the checkerboard is placed by the side of the face on the exact plane of the eye. Using the center sensor (the only one he now uses) he focuses on the checkerboard, locks the focus and then rotates the camera on the tripod to center the frame. This system seemed to produce consistently sharp pictures, but it certainly doesn't solve the problem of shooting a moving subject. It also slows down a shoot dramatically, and may increase the number of people needed on a set. As I said earlier, not an ideal solution, but one that at least produces sharp pictures.
Kirk also indicated that the "actual focus sensor arrays are three times larger than the viewfinder marks. A user could put the intended subject in the mark, but if there is a strong detail just outside the mark (but within the sensor area), the camera would focus on the strong detail." Canon claims this is not the case. Other photographers insist that from their tests it is the case. I'm not sure which is right but it is certainly something to consider. I've seen soft shots where even if the area of focus was three times the size of the sensor it would not have touched any strong competing detail so I'm inclined to think that this has little or nothing to do with the focus problems.
Solving this problem has recently become even more critical for those trying to make their living producing stock images because now Corbis will only accept digital submissions and Getty Images has announced that starting in the spring of 2006 they will do the same. Thus, if you want to sell stock it is absolutely critical to be able to offer the agencies files that were shot digitally, and yet if you go out to shoot digital there is no guarantee that on any given day you'll come back with sharp images.
Photographer Suggested Solutions
The following are solutions that various people have been suggested, but there is no indication that any of them work consistently:
1 - Have the camera adjusted to factory specs, but these photographers have tried that and it doesn't seem to solve the problem for very long.
2 - Install the latest firmware, but photographers have tried that and it usually seems to make little difference.
3 - Always use prime lens instead of zoom lenses.
4 - Always shoot at least two stops below the maximum aperture of the lens. But many photographers want to use those maximum apertures in order to minimize depth of field (but that doesn't mean they want no depth of field at all).
5 - Use only the center focus point of the 9 to 1 focus option because it is the only focus point that can resolve on both horizontal and vertical lines. All the other focus points resolve on either horizontal or vertical lines, but not both.
6 - Use the Canon ST-E2 Speedlight transmitter to help with focusing, but some who have tried it have had inconsistent results.
7 - Don't use UV filters.
8 - Change camera bodies frequently during a shoot.
9 - Turn sharpness off on the camera.
10 - One theory I've heard is that even though the batteries have enough power to drive the camera they may not be strong enough to drive the autofocus sensor. I've not been able to find any evidence that anyone has tested this theory carefully but if it does have something to do with the problem changing batteries frequently might help.
11 - Some have theorized that the low pass filter in front of the imaging chip that is designed to help eliminate moiré is reducing sharpness. But, if that's the case why do photographers get some sharp images while using the filter and unsharp images at other times?
The simple fact is that photographers don't seem to have a clear idea of what might be causing the problem. Canon evidently knows because they can fix the problem "sometimes". But they are not explaining the cause to their customers - the photographers. Photographers are left to guess whether it is:
something they are doing in the way they use the equipment,
a problem in the basic design of the equipment,
a mistake the photographer made when focusing and taking the picture,
a problem caused by the photographer cleaning the sensor, or
a gremlin beyond human control.
Westfall says, "although it is regrettable, the fact that equipment problems sometimes do occur is inevitable. Just like human beings, no individual piece of equipment is perfect. Fortunately, though, malfunctioning equipment is repairable or replaceable. When equipment performance problems are brought to our attention, our long-standing policy is to resolve those problems to the best of our ability, on a case-by-case basis. Canon is conscientiously doing its best to help its customers whenever we are given the opportunity."
Canon certainly tries to minimize the problem and points out that only a relatively few photographers (maybe 100 out of the tens of thousands who own the MarkII) have had problems. And with any manufacturing process it is inevitable that some problems will occur. Skeptic that I am, I tend to believe that there may be a lot more than 100 problem cameras out there, and that many of the photographers may not have examined their results as closely as the photographers named in this article have done. Or if they did they blamed themselves for the problem rather than considering that it might be the equipment.
It seems very strange to me, if there are only 100 or so cameras out there with this problem, that every one that Jon Feingersh or Walter Hodges or Jack Hollingsworth buys - at different times and different places from different manufacturing batches - has the problem at one time of another. To me that indicates that the problem might be larger than Canon would like us to believe. On the other hand maybe gremlins are chasing these particular photographers!!
There are photographers out there who paid $8,000 a copy for their cameras and who may be losing tens of thousands of dollars in production costs on a blown shoot. Some may also be losing jobs and clients. I think Canon owes them a detailed explanation of what is causing this problem.