New guy question about focus

[Mike Quigley]

I have a new 5D Mk3, and I just purchased a 100-400/5.6. I have been out practicing and my photos are abysmal. I set up on a tripod yesterday with the 10-400 and shot birds at the feeder. When using manual focus in live view for focusing It seems I cannot get a perfect focus no matter how hard I try. So I put on my 70-200/2.8 which is usually perfect and I experienced the same issue. Even when using AF the photos are OOF, So now I'm thinking a camera issue. Does what comes through the lens always reflect what you get in the photo? Im sure I'm not asking this properly but I hope one of you experts can help a little. Mike

[John Chardine]

Hello Mike- First a big welcome to BPN.

There are a host of possibilities for you to consider:

1. Live view focus is one-shot which means it locks in. Depending upon your depth of field even a little back and forth movement of the subject say caused by a swinging feeder or subject movement, can cause images to be OOF.

2. In my experience, some copies (maybe all) of the 100-400 are very sensitive to the presence of a filter. I have seen some supposedly good quality filters completely degrade IQ on this lens. Are you using a filter?

3. What shutter speed were you using? Was it adequate to stop camera shake for the focal length you were using, or subject movement. Remember that at 400mm there are micro movements of the tripod that will show up as blur.

4. At 400mm, I found my copy of the 100-400 was improved by stopping down a little, say 2/3-1 stop.

5. Did you have IS on? The 100-400 IS version is older and does not perform well if the tripod is locked down.

You say you got the same effects with the 70-200/2.8 (IS version or not?). There could be an AF micro-adjustment issue that you need to correct for.

However, before you blame the hardware, you need to check what IQ you are getting under optimal conditions. The way I do this is as follows:

1. Set up your camera and lens on a solid tripod indoors.
2. Set up a target like a bank note- tape one to a flat glass plate and set it up so it is parallel to the sensor and a normal shooting distances.
3. Set the lens to the desired focal length and wide open. Turn IS off and remove any filters.
4. Set up a flash to adequately illuminate the target. Set the camera to manual mode with a high shutter speed (say 1/1000 or above) so that all the light comes from the flash. This will eliminate camera shake as a factor. Set the flash to high-speed sync.
5. Turn on Live view and focus on target. This focus should be perfect because of the way LV AF works. Make a RAW image of the target. Develop RAW image on computer with no sharpening. Critically examine at 100% for sharpness.
6. Make more images using same method but at smaller apertures to see the effect of stopping down.

This should give you an idea of what your system is capable of. Once you are satisfied you are getting the best out of your rig, post an image made with the lens wide open. It would be useful to have a look.

[Mike Quigley]

Thanks John, I guess the main point of trying to use LV to focus was to see exactly what the lens saw, and if what I see in the viewfinder is not perfect to my eyes how can I expect a photo to be perfect. Example, when I look at the gorgeous photos on this site I see birds and animals so perfect that every hair can be seen, If I cant see that through the lens how can it come out in a picture? I will do what you suggested, and try to show what I mean. Thanks so much for helping. I am also taking BlueCrane classes online, but its hard to talk to a DVD

[Daniel Cadieux]

Mike, you need to use auto-focus and look through the viewfinder. For bird photography Live-view is only helpful for huge magnification on a static bird. A properly micro-adjusted lens / camera combo is very reliable when looking through the viewfinder. You even have a "beep" to tell you focus is achieved, and if you find that too annoying (like I do) you have a dot show up in the viewfinder to confirm focus. Even then I don't pay attention to either of those...I just need to trust my camera is good at what it does.

You say that you put the 100-400 on a tripod. It has a first generation IS motor that needs to be switched OFF when on a tripod or else the results will look terrible. I agree it does not really take filters all that well either.

An example with exif info would help us help you, so please do not be shy to put an example up here within this thread.

P.S. A big welcome to BPN!!!!:S3:

[Mike Quigley]

Thanks Ill get to work on the assignment and post something soon. Mike

[Mike Milicia]

4. Set up a flash to adequately illuminate the target. Set the camera to manual mode with a high shutter speed (say 1/1000 or above) so that all the light comes from the flash. This will eliminate camera shake as a factor. Set the flash to high-speed sync.


Hi John,

Some good advice for Mike but your reference to the use of flash to eliminate camera shake is a bit off. Once you go to a shutter speed above the sync speed of the camera and turn on high speed sync, the flash effectively becomes a continuous light source that illuminates the subject the entire time the shutter is open. In this situation, the flash will add light but it will be of no assistance in stopping motion or reducing the effect of camera shake. If you want to take advantage of the ability of the flash to put out a very quick pulse of light and use that to stop motion or eliminate camera shake, you need to have your shutter speed at or below the sync speed and set your ambient exposure so that the flash provides the only source of light, i.e. with the flash turned off, you will just get a black frame (helps to shoot in a dark room). This can indeed be very effective since, at minimum power, the flash duration will be on the order of 1/35,000 second which gives you a much faster "effective" shutter speed than you can achieve with the physical shutter mechanism.

[John Chardine]

Hi John,

Some good advice for Mike but your reference to the use of flash to eliminate camera shake is a bit off. Once you go to a shutter speed above the sync speed of the camera and turn on high speed sync, the flash effectively becomes a continuous light source that illuminates the subject the entire time the shutter is open. In this situation, the flash will add light but it will be of no assistance in stopping motion or reducing the effect of camera shake. If you want to take advantage of the ability of the flash to put out a very quick pulse of light and use that to stop motion or eliminate camera shake, you need to have your shutter speed at or below the sync speed and set your ambient exposure so that the flash provides the only source of light, i.e. with the flash turned off, you will just get a black frame (helps to shoot in a dark room). This can indeed be very effective since, at minimum power, the flash duration will be on the order of 1/35,000 second which gives you a much faster "effective" shutter speed than you can achieve with the physical shutter mechanism.

Not really Mike. Above the sync speed, every slice of sensor will receive a very short flash, not a continuous stream of light like you suggest. The method I posted works as I've used it many many times.

[Mike Milicia]

Not really Mike. Above the sync speed, every slice of sensor will receive a very short flash, not a continuous stream of light like you suggest. The method I posted works as I've used it many many times.

That is no different than saying that every slice of the sensor will receive a very small amount of ambient light when you are not using flash. The ability to stop motion is still determined by the shutter speed. Your method will be equally effective if you simply add enough ambient light to enable a shutter speed of 1/1000. It does give you the ability to potentially use much faster shutter speeds than you would be able to do without the flash as the flash is an easy way to create that additional ambient light. But the ability to stop motion or eliminate camera shake is no different than having enough ambient light to use 1/1000 without flash. Once you decide to add flash to the equation, why not take advantage of its ability to give you extremely fast "effective" shutter speeds when you shoot at or below the sync speed? Depending on the situation, 1/1000 may indeed be enough to eliminate camera shake and your method would work fine but 1/35,000 would definitely be better.

[John Chardine]

Mike- I'm sure we are talking from the same page, but there are some practical aspects of the test that need to be considered. With a camera/lens locked down on a sturdy tripod and using a cable release and 10s shutter delay (sorry I didn't mention this in the methods) I target about 1/1000s or faster as a shutter speed. Theoretically a flash duration of 1/35,000s would be more effective at stopping motion but in this case unnecessary- we are not trying to stop the action in a hummingbird's wing or a bouncing water droplet here. Also it's important for folks to realise that flash light output is adjusted by varying the duration of the flash with larger outputs having longer durations. With the sorts of flashes we normally use, a duration of 1/35,000 would only be obtained in certain circumstances- flash set to M and the output cut back to 1/32 or 1/64 of full power. Or in ETTL, a short subject to camera distance may also give you similar, very short output durations. At the camera-subject distances I am running the tests, 1/32 or 1/64 power would not light the target adequately unless I had the flash close to the target and fired it remotely, which I don't.

For the information of readers, here's the output durations for a Nikon SB-800 flash at different power outputs (from http://www.scantips.com/speed2.html)

1/1050 sec. at M1/1 (full) output
1/1100 sec. at M1/2 output
1/2700 sec. at M1/4 output
1/5900 sec. at M1/8 output
1/10900 sec. at M1/16 output
1/17800 sec. at M1/32 output
1/32300 sec. at M1/64 output
1/41600 sec. at M1/128 output

So if you do as you suggest and use a shutter speed at or longer than the sync speed you still may only get a exposure duration of roughly 1/1000s. Flash duration is not solely dependent on whether you have h-sync set or not (implied in pane 6 above).

Another practical aspect is that it would be most inconvenient and downright ineffective to perform the test I described in conditions so dark that you would expose virtually no ambient light at the sync speed or less. You need enough ambient light to allow the AF system to work effectively and repeatably. I guess if I had a set of lights to illuminate the target enough to allow a shutter speed of 1/1000s or above I would use it but I don't have such equipment (I'm a wildlife photographer). All I was trying to emphasise was to use a relatively short shutter speed so the ambient light is severely underexposed and all the light comes from the flash. If I set a shutter speed of say 1/1000s and use a flash set to h-sync I'll obviously get a shutter speed of 1/1000s, which is what you want. I wasn't trying to imply anything else when I recommended using a flash as the main source of light.

The bottom line as I mentioned earlier is that I done these tests 100s of times and the method works.

I hope this clears up any misunderstandings.

[Mike Milicia]

Hi John, Yes, sounds like we are on the same page. As long as 1/1000 second is a fast enough shutter speed to eliminate all concerns with potential movement and vibration, flash on hi-speed sync will work just fine. I just wanted to make it clear that the flash in this case is just being used as a light source and it not increasing your ability to eliminate movement and vibration beyond what you would get with 1/1000 shutter speed without the flash if you had enough ambient light.

[Diane Miller]

Good discussion about flash, and I may have read things too quickly, but Mike says "even when using AF things are OOF." If the subject is moving toward or away from the sensor even a little, One Shot AF will be locked on one plane and the subject may have moved to another. AI Servo may be necessary to keep up with even slight movements. (Don't think he said which he was using.)

I have the 5D Mk III and the AF is simply amazing. Select one point and get it on the part of the subject you want in focus, and let AI Servo do its thing. Shoot a burst. Some will be a little off, but most will be great.

Apologies if I've missed the point -- I'm a bit rushed.

[John Chardine]

Hi Mike (Quigley) and Diane- I see that Mike Q. was using manual focus in Live view. I misread this. Mike- were you focusing with the image twice magnified in LV? If not, critical focus would be hard to acquire by hand. In LV you can autofocus very accurately, which leaves the guesswork out of the equation. Focusing by hand is essentially "one shot" because it would be very difficult to track the focus of a moving subject in this way.

[Mike Quigley]

Hi John, I was magnified at2x in LV. But my question was meant to be, when I look at a subject in this way and get it as sharp as i can with my eye, is that what I get on the photo end. Is it reliable? Because sometimes I swear it never gets as tack as i want. But I have eyes that are 67 years old also. Hope im not confusing you. mq

[John Chardine]

Hi Mike- When you focus using Live View, either manually or automatically, you are seeing exactly what the sensor is seeing so yes, this will be what you get in the image. Compare that to the diverted light path when looking through the viewfinder and AF using the phase-detect system. Contrast-detect AF in Live View should be the most accurate AF available to you and if done carefully, manual AF in Live View should be good as well.

However, there are factors that will make the image appear softer than you see in Live View. Most digital cameras have an anti-aliasing filter that tends to blur your image slightly. It would not be enough to make your images look "abysmal" though. It would be really useful if you could post an example. Make sure you don't sharpen the image at all (shoot RAW and turn off any capture sharpening) and crop the image to 1024 x 800 px or less but don't do any resampling to get there. That way the image will be displayed on BPN at 100%. I hope that all makes sense to you. If not just ask.

[Daniel Cadieux]

According to the first post Mike is photographing birds around a feeder and this is where to problem is IMO. Live View is good for static images where the bird is sitting still, but birds around feeders hop actively from branch to branch, and on and off the feeders. I would have difficulties getting sharp images of fast active birds manually (or auto) focusing with Live View as I'm not certain I'd be able to properly track their movements...even manually focusing through the viewfinder would be a challenge. For this type of photography using auto-focus (set to AI Servo) and using the viewfinder is the way to go.

[Sam Maggio]

When using a tripod and pre-focusing on a feeder you have to disable AF so the camera won't try to focus again when pushing the shutter button. If you're using a cable release, the camera will try to AF when the button is pushed unless you switch the lens to MF or disable the AF feature on the shutter button in custom functions menus and only use the back focus button.

Using a UV filter will cause images to be soft, and having IS on while on a tripod can cause blurred images.

We have had problems with our new 100-400 and 400 5.6. We sent them in to Canon and they are spot on now. You might want to send in your 5D3 and the 100-400 and have them calibrated together...