5d3 and FoCal errors

[dswtan]

This may be irrelevant to the OP, and I've not seen that error on my setup with 5D3 on Focal v1.6.0 and multiple lenses (have also used for 5D2 and 7D), but a usage tip that I often forget: you must cover the camera eyepiece as you would for a long exposure photo. If I don't do this, I get "flat" curves and test runs that don't complete. Cover it, and I get nice Gaussians.

 

[Drizzt321]

I am in the middle of a long data gathering process using FoCal, 3 Canon bodies and 10 different lens. I run each autofocus calibration test at least 3 times to assure getting the best setting. I have found that ocasionally Focal will produce one test with very different results from the others, this happens very rarely but it does happen. Generally, the results have been very consistent for any camera/lens combination at any particular distance.

I test each camera/lens combination at 4 distances: 50xFL, 100xFL, 150xFL and 200xFL. This is a lot of testing, but the results from what I thought was a "problem" lens caused me to rigorously test some of my other lenses. Then my curiosity kicked in, so now it's a project. The sheer volume of testing has forced me to use a very efficient testing procedure. The downside of this method is that right now I have a LOT of data, but no complete data for any camera/lens combination.

The results indicate that depending on the distance there can be a great difference in the calibration setting. I have an EF 24 F1.4L II when mounted on my 7D is -4 at 50FL, 4 at 100FL and 1 at 150FL. I have not run the test at 200FL yet. If I used the setting for the "normal" 50FL I'm sure I would hate this lens if I regularly used it at longer distances. I have not yet begun to think about what the proper setting for this lens should be. The bad news is that this is not the only lens that is starting to show this kind of characteristic. On the other hand, maybe I just have a bunch of bad cameras and lenses.

Part of my motivation for my testing is that I've often wondered why some people will say a lens is fantastic and others will say it is crap. Maybe, they are just using it at different distances.

Care to describe the setup you have for the rest of us?

 

[VirtualRain]

I really don't think you can beat shooting pics of an angled yard stick when it comes to MA. Your brain is bound to be much better than any software at determining where the focal plane is. Just a few shots will tell you if you have front focus or back focus issues (or both - usually confirming your lens is a Sigma) and you can narrow in quickly on a good setting you know you can trust later.

 

[sturdiva]

I bought it when it first came out a year ago, and each new version has worked better than the previous. Sure, there are a few minor bugs, but most of the time, the issues have been mine.
Correct setup, plenty of bright light, absolutely no vibration or movement, concrete floor is far better than a wooden one.
Some lenses are problem lenses, they just do not perform repeatedly or well. In some cases, its the lens design, the cheap 50mm lenses are never going to do well. The "L" lenses do very well.
I've found that the 5D MK III gives much better results than my 5D MK II's did, and that is borne out by the data sent in from testers.
If your setup is perfect - stable, bright light, viewfinder blocked, firm floor with no vibration, accurate target alignment, then AF issues are likely a problem camera or problem lens. Both can be problems, we just never had a way to test before.
Overall, it works well and is repeatable for me, but the devil is in the details of a good setup.

As far as computer issues, I've had none on my Lenovo Windows 7 64 bit laptop or old Dell XPS-420 desktop that was upgraded to Windows 7 64 bit. Having a lot of older equipment increases the probability of drivers conflicting, or software conflicts. Those just have to be worked thru, there is a infinite number of combinations of hardware and software, so no amount of testing will guarantee that it works for everyone. I'm now in the process of testing out my new Dell XPS 8500 with Windows 8 64 bit. FoCal is installed and running fine, but I've not done a full AFMA using it. I most likely will not in the near future.

I'm sure it's something I'm doing wrong, but I just can't get it to do anything.

I've tried 2 computers (windows xp (32 bit), and windows 7 (64bit)), 3 versions of FoCal (1.6, 1.7 beta, and 1.7 released). All shot with the camera on a tripod, on wood floors, in light levels from (1/25 f/2.8 iso 100 (the recommended value in the FoCal docs) up to 1/1250 f/2.8 iso 100). I've tried several lenses, various zoom settings and target distances, all with no luck. The software simply refuses to even try to focus the camera.

I've been watching it a bit closer now trying to see what the problem may be, and it really seems like a software problem to me, it will start the test, flip up the mirror, capture an image (for the live view), flip it back down, move the focus (I can tell via the distance scale) to infinity, and then do nothing for ~30 seconds before giving the "focus timeout" error.

I'm now in the process of trying to return the software for a refund, but I still can't even get a response from the company. Quite frustrating.

 

[Loswr]

I really don't think you can beat shooting pics of an angled yard stick when it comes to MA.

Well, you can't beat it...if you like bad results. I think this sort of thing is the reason Canon says in their manuals that AMFA may cause problems - if done wrong, it sure can. An angled yard stick can work, provided that's not what you're focusing on. You may think you're focusing on the thin horizontal line by the number 18, but the AF system doesn't know that. The AF point is larger than it's little representative box in the VF, and it's going to lock onto the highest contrast subject at the right orientation under that point. So, for your angled yardstick, that's most likely the edge of the yardstick itself, which is running at an angle through the DoF.

Here's an example, where the focal plane is clearly different in the two shots, but as far as the AF system is concerned, both shots are correctly focused.

If you're going to use a DIY setup for AFMA, do it right - use a flat focus target oriented in the same plane as the image sensor, aligned and level, and if you use a yardstick, use it only to judge the DoF, not as the focus target. There's a very good reason why all of the commercial solutions use a flat target perpendicular to the lens axis. Here's a viable DIY setup that mimics the commercial tools.

 

[RLPhoto]

My method of Focal is shooting a barcode over and over and over to determine the lenses behavior of either front or back focusing. Make an adjustment and repeat. It just tuned my primes last night and now they're even better. ;D

 

[VirtualRain]

I really don't think you can beat shooting pics of an angled yard stick when it comes to MA.

If you're going to use a DIY setup for AFMA, do it right - use a flat focus target oriented in the same plane as the image sensor, aligned and level, and if you use a yardstick, use it only to judge the DoF, not as the focus target. There's a very good reason why all of the commercial solutions use a flat target perpendicular to the lens axis. Here's a viable DIY setup that mimics the commercial tools.

That's exactly what I meant.

 

[al2]

This thread has been stagnant for a while, but Drizzt321 asked about my test set up. I haven’t answered because my 5D3 arrived a few days ago and I’ve been busy shooting the newness off of it.

I have to do my testing at my office after hours or during the weekend. It’s the only place I have access to with enough space to let me test at 131 feet (40 meters). When I start with tele-convertors or lenses longer than 200mm I will have to move to the parking garage in the basement. Besides the available space, the office building has much more rigid floors to minimize vibration, but even so, I sit down and don't move while the tests are running.

I enlarged the standard FoCal 50FL target 2x, 3x and 4x to be appropriate with the 100FL, 150FL and 200FL test distances. The targets are mounted on pieces of foam-core board, with the center of each target located at the same place on each board. I made a target stand that lets me quickly change the targets without disturbing the stand or the lights. After a target change only slight adjustments must be made to where the camera is aimed.
The camera is mounted on a tripod with the legs at their shortest length. The tripod is anchored using a very taunt bungee cord running from the hook on the tripod to a 40 lb (18.2 kg) weight resting on the floor.

I use two 1000 watt halogen lights to illuminate the target. These are construction lights so I have to use my Expodisc to correct white balance. It seems to work OK. I get between 9.0 and 10.0 EV on the target, depending on how I can arrange the lights. The office furniture limits me sometimes. Mostly, I can get between 9.5 and 10.0 EV.

The following is an example of the process I use:

1. Set up target stand and lights at 16.5 feet (5 meters) from the camera. This is approximately 200FL for 24mm, 100FL for 50mm and 50FL for 105mm.
2. Slid 200FL target into stand and run AFMA tests, autofocus consistency test and aperture sharpness test for EF 24 F1.4L II.
3. Mount EF 24-105 F4.0 on camera and test 24mm focal length.
4. Change to 50FL target and test 105mm focal length.
5. Mount Sigma 50 F1.4 on camera, change to 100FL target and run tests.
6. At this point, if I have enough time I will change the distance and run another series of tests; or if I have another camera body available I will rerun the tests using the second body.

This may all seem like overkill, but I am interested to see if I can find any pattern to how the autofocus setting will vary with a change in distance.

Also I wonder if some lens manufacturers are optimizing their lens designs to perform well at the standard test distances, but at the cost of performance at other distances.

 

[Drizzt321]

This thread has been stagnant for a while, but Drizzt321 asked about my test set up. I haven’t answered because my 5D3 arrived a few days ago and I’ve been busy shooting the newness off of it.

I have to do my testing at my office after hours or during the weekend. It’s the only place I have access to with enough space to let me test at 131 feet (40 meters). When I start with tele-convertors or lenses longer than 200mm I will have to move to the parking garage in the basement. Besides the available space, the office building has much more rigid floors to minimize vibration, but even so, I sit down and don't move while the tests are running.

I enlarged the standard FoCal 50FL target 2x, 3x and 4x to be appropriate with the 100FL, 150FL and 200FL test distances. The targets are mounted on pieces of foam-core board, with the center of each target located at the same place on each board. I made a target stand that lets me quickly change the targets without disturbing the stand or the lights. After a target change only slight adjustments must be made to where the camera is aimed.
The camera is mounted on a tripod with the legs at their shortest length. The tripod is anchored using a very taunt bungee cord running from the hook on the tripod to a 40 lb (18.2 kg) weight resting on the floor.

I use two 1000 watt halogen lights to illuminate the target. These are construction lights so I have to use my Expodisc to correct white balance. It seems to work OK. I get between 9.0 and 10.0 EV on the target, depending on how I can arrange the lights. The office furniture limits me sometimes. Mostly, I can get between 9.5 and 10.0 EV.

The following is an example of the process I use:

1. Set up target stand and lights at 16.5 feet (5 meters) from the camera. This is approximately 200FL for 24mm, 100FL for 50mm and 50FL for 105mm.
2. Slid 200FL target into stand and run AFMA tests, autofocus consistency test and aperture sharpness test for EF 24 F1.4L II.
3. Mount EF 24-105 F4.0 on camera and test 24mm focal length.
4. Change to 50FL target and test 105mm focal length.
5. Mount Sigma 50 F1.4 on camera, change to 100FL target and run tests.
6. At this point, if I have enough time I will change the distance and run another series of tests; or if I have another camera body available I will rerun the tests using the second body.

This may all seem like overkill, but I am interested to see if I can find any pattern to how the autofocus setting will vary with a change in distance.

Also I wonder if some lens manufacturers are optimizing their lens designs to perform well at the standard test distances, but at the cost of performance at other distances.

Wow. Fantastic, thanks! I never would have thought of using the office before/after hours. Mine has a great looong hall, probably 100' or so, maybe longer. And printing out at additional FL test distances seems like a good idea. I'll have to get some foam board, and find a good place that can print up the targets for me.

 

[al2]

Hey Drizzl321, if you do start to test at multiple distances, if you don't mind I would really like to see a summary of the results. I'm trying to collect enough data to see if there are any patterns to the change in AFMA as the distance changes. More data from different lenses and cameras would be better.

At some point I will post my results on the CR Forum.

 

[Drizzt321]

Hey Drizzl321, if you do start to test at multiple distances, if you don't mind I would really like to see a summary of the results. I'm trying to collect enough data to see if there are any patterns to the change in AFMA as the distance changes. More data from different lenses and cameras would be better.

At some point I will post my results on the CR Forum.

Hmm...do you know if FoCal can export the results as a CSV? I know it does a full PDF report, but if we can get the raw data, that could start making for an interesting database of lens results.

 

[al2]

Great idea. I have been saving the csv files along with the PDF files. When you save a report, a menu pops up and one of the options is to save csv file. At least there is in the Pro version.

 

[Drizzt321]

Great idea. I have been saving the csv files along with the PDF files. When you save a report, a menu pops up and one of the options is to save csv file. At least there is in the Pro version.

Oh, how interesting. I never really bothered looking. Next time I do AFMA I'll have to take a look. I do have the Pro version, I figured it wasn't that much more and you never know when you might need some additional features

 

[sturdiva]

Well, I finally figured out my FoCal issues w/the 5d3. It's snowing here in the Baltimore area and that gave me a bit of down time to spend playing with it. Hopefully this will help others out if they are having a similar problem.

It seems that FoCal will not perform the phase detect autofocus step if the AF-ON button is not configured to "Metering and AF start". I had mine set to AE lock (hold).

I've e-mailed their support, and hopefully this will be added to the configuration changes the program makes before it runs it's tests.