badlydrawnboy wrote:
Here's a clear example of what I'm talking about.
After doing the MFA as I described above, I dialed in +18 in camera because that was the value (or within 1 of the value) for 12", 16" and 26". (It was also the value for 50x focal length.) I then took this picture about 2 feet away from the subject:
Obvious back-focus at this distance with this MFA value. I have a number of other similar examples.
This is of course what the dock is supposed to be for, but FoCal was giving me errors when I tried to calibrate it at longer distances—perhaps because of some issue with the lens. I don't know.
I think I will send this in to Sigma. I'm not quite ready to give up, and as Ernie said they should be able to either make it work well or replace it. I guess my fear is getting stuck with a lemon if they don't. Does anyone have experience with Sigma customer service?
She wasn't moving and my other lenses don't have this problem. The DOF of the 50L and 85/1.4 is even shallower with shots like this and they nail it every time (provided the subject isn't moving). This was a CONSISTENT effect; I just posted one example.
I'm interested in how a lens works in real-world shooting conditions, not just shooting a focus target on a tripod. As I said, this is not a problem I have with any other lenses--including my Sigma 85/1.4.
I love the dock and love what it can do. I love microfocus adjustment and love what it can do.
I'm also aware that autofocus is a shotgun: change nothing and take 10 pictures in a row on a test target in a carefully controlled environment and you're going to get a range of focus points. That's the reality of phase detection.
All of these tools can move the center of the range to its proper location, but it can't make the range narrower.
It seems you are experiencing the same issues I had with my S35. The inconsistencies of its necessary focus adjustments made it impossible to correct them all. I agree that the USB Dock is fantastic for properly-behaving lenses.
I also found that my Canon lenses had a much tighter group of scores using the Focal AF Consistency tool. 7% was the maximum deviation for any of them, where 30% was the norm for the Sigma. This makes it harder to determine the correct MFA, and less likely to get a good shot even if that MFA is correct.
I think there is just too much sample variation with this lens. People that get a good one straight outta the box LOVE it. People that get a good one that the USB Dock can easily correct LOVE it. People that get one like I had, or three as it seems BDB has tried, simply can't live with the results. It's so hard to kick them to the curb because of the intermittent reinforcement. Just like a slot machine, we mentally disregard all those losses when that one perfectly focused shot comes in. Jackpot!
RCicala wrote:
I love the dock and love what it can do. I love microfocus adjustment and love what it can do.
I'm also aware that autofocus is a shotgun: change nothing and take 10 pictures in a row on a test target in a carefully controlled environment and you're going to get a range of focus points. That's the reality of phase detection.
All of these tools can move the center of the range to its proper location, but it can't make the range narrower.
It should just take 15-30 min to calibrate the sigma with the USB port. Yard stick, something to focus on (tilted and/or perpendicular) and your computer. I did it 3 times and all 3 times, I get the same MA adjustment numbers. I tried Dot Tune, but that method didn't produce good results for me.
With that said, my AF consistency is about 85-90%. Nothing far off like your pictures, just a tick off. Yes the 35L is better at AF consistency and slightly better in low low light.
I had two lens that had bad AF consistency. the sig 50 and the 50 1.4. Both I sent for service, both came back great.
I am using a 5diii.
Looks like a few here have sig35s with bad AF consistency issues. Not sure if its the lens or maybe it could be FoCal, since those using FoCal has the focus adjustment issues. I never used FoCal, so don't flame me, just putting it up on the table too.
myron lee wrote:
It should just take 15-30 min to calibrate the sigma with the USB port. Yard stick, something to focus on (tilted and/or perpendicular) and your computer. I did it 3 times and all 3 times, I get the same MA adjustment numbers. I tried Dot Tune, but that method didn't produce good results for me.
With that said, my AF consistency is about 85-90%. Nothing far off like your pictures, just a tick off. Yes the 35L is better at AF consistency and slightly better in low low light.
I had two lens that had bad AF consistency. the sig 50 and the 50 1.4. Both I sent for service, both came back great.
I am using a 5diii.
Looks like a few here have sig35s with bad AF consistency issues. Not sure if its the lens or maybe it could be FoCal, since those using FoCal has the focus adjustment issues. I never used FoCal, so don't flame me, just putting it up on the table too....Show more →
Just curious how you're doing the MFA if not with FoCal or DotTune. Are you simply taking pictures at a number of different values and comparing the results? If so, how can that only take 15 minutes?
I did the process using TurboCal, which is FoCal's version of DotTune (that has worked well for me with my other lenses), and it still took well over 30 minutes to 1) do the initial calibration in camera at all 4 shooting distances, 2) enter those values into the lens and save, 3) test them out with the camera MFA set to zero.
In any event, I'm getting somewhat better results after entering +18, +19, +18 and +15 into the lens. +15 was a guess on my part, as I couldn't get reliable results with FoCal or DotTune at infinity. It's still pretty inconsistent, though, so I'm going to send it in to Sigma along with my 5DIII body.
When it hits it's an incredible lens, so I'm going to give it one last try.
badlydrawnboy wrote:
Here's a clear example of what I'm talking about.
After doing the MFA as I described above, I dialed in +18 in camera because that was the value (or within 1 of the value) for 12", 16" and 26". (It was also the value for 50x focal length.) I then took this picture about 2 feet away from the subject:
Obvious back-focus at this distance with this MFA value. I have a number of other similar examples.
This is of course what the dock is supposed to be for, but FoCal was giving me errors when I tried to calibrate it at longer distances—perhaps because of some issue with the lens. I don't know.
I think I will send this in to Sigma. I'm not quite ready to give up, and as Ernie said they should be able to either make it work well or replace it. I guess my fear is getting stuck with a lemon if they don't. Does anyone have experience with Sigma customer service?
With these pictures, you don't need FoCal. Firstly, find your MA for infinity. Shoot anything at 100 ft, or just 20 feet away just to get close. Then try shooting at your problem 4 feet mark again to see if it improves.
At 4 feet, DOF is two inches in front and two inches in back. So you don't need to be precise, just close enough.
But if you in fact get poor AF consistency , send it in if it is a sharp copy.
badlydrawnboy wrote:
Just curious how you're doing the MFA if not with FoCal or DotTune. Are you simply taking pictures at a number of different values and comparing the results? If so, how can that only take 15 minutes?
I did the process using TurboCal, which is FoCal's version of DotTune (that has worked well for me with my other lenses), and it still took well over 30 minutes to 1) do the initial calibration in camera at all 4 shooting distances, 2) enter those values into the lens and save, 3) test them out with the camera MFA set to zero.
In any event, I'm getting somewhat better results after entering +18, +19, +18 and +15 into the lens. +15 was a guess on my part, as I couldn't get reliable results with FoCal or DotTune at infinity. It's still pretty inconsistent, though, so I'm going to send it in to Sigma along with my 5DIII body.
When it hits it's an incredible lens, so I'm going to give it one last try....Show more →
I have a piece of paper with type text all over it. I get a big black marker and draw a "thick" horizontal line in the middle of the paper for my focus point. Lay the paper flat on the floor. Grab a yard stick in one hand and touch the middle of the paper (focus point) with the end of the yard stick. Use my other hand, line up the back of the camera to the 12",16 and", 26" distance, aim and shoot (about a 45 degree tilt). Adjust for back/front focus for each point. For infinity, shoot anything with a good contrast-y edge far away. 20 feet if that is as far as you can go in your house. Do it with Live view 10x and see if the infinity shoot is sharp like the live view shot. Adjust till sharpness match. Input the numbers via USB port software. Done. Go test it out like you did in your post.
PS. make sure you are aligned with the paper. tilting up and down is OK. Tilting in the right to left direction is NOT OK.
Also all lens are elliptical in design. So that is why we use +50x focal length (as a rule of thumb) to micro-adjust. Anything else will not be optimal focus if you are using just one MA adjustment. I am not sure if FoCal has this rule implied as well. Thus if it does, FoCal probally needs to verify that other distances will produce good results.
p.5 #11 · Sigma USB Dock calibration data for 35mm
RCicala wrote:
I'm also aware that autofocus is a shotgun: change nothing and take 10 pictures in a row on a test target in a carefully controlled environment and you're going to get a range of focus points. That's the reality of phase detection.
All of these tools can move the center of the range to its proper location, but it can't make the range narrower.
This is exactly how I imagine focus accuracy and focus consistency issues. Queue 'if you can't explain it simply, you don't understand it well enough' quotes .
I'm still slightly afraid to 'dot tune' my lenses because I'm pretty sure I'll find that nothing works perfectly; but once I'm done moving they're getting the crap beat out of them.
And even Sigma's tunable lenses lack real specificity in tuning ranges, it seems, and it sounds like they'd be better off allowing users to define 'curves' for the phase-AF calibration adjustments. Still better than Canon's 'one adjustment to rule them all' method for primes and 'one adjustment for each end' rule for zooms.
p.5 #12 · Sigma USB Dock calibration data for 35mm
Okay, I got my S35 back from Sigma. It's better, for the most part. But I'm still having some issues where it misses focus consistently in certain situations. For example, in this series of shots I focused on her eye, but it focused on her hood each time instead. (I was using a non-center point with the 5DIII).
I haven't noticed this kind of thing with my Sigma 85, 50L or 135L.
Another issue is that the AF seems to be more accurate indoors than outdoors. I can't really explain that, other than the fact that I calibrated it inside. I think I recall other people mentioning this before.
.