I've just posted a video tutorial of the DotTune technique. Here is the link:
DotTune: Autofocus fine tuning in under 5 minutes
DotTune: Autofocus fine tuning in under 5 minutes
DotTune: Video Tutorial for AF tuning without photographs
- Thread starter horshack
- Start date
- Status
Thanks for all the efforts - the video is a great tutorial and involved a lot of work. Much appreciated. Unfortunately, it's not working for my Series II 300mm f/2.8 and and Series III extenders on the 5D III. It's way, way out. This is bad news for me because the dot light is out of step with live view, and I know live view is correct. It;s not your method that is wrong, it is the Canon protocol. It is also very disturbing that there is such a wide span of dot focussing in manual mode. It means that you can't rely on it for manual focussing but you have to use live view.
Upvote
0
East Wind Photography said:
It's worked well for most, though I don't reasonably expect it to work for everyone or in all situations, the same as how other methods don't work in all scenarios. LOL about the 18 minute video description for a 5 minute procedure, I thought the same thing after I was done editing it. I tried to present the material as concisely as possible but it's always a juggling act between brevity and thoroughness. The sample DotTune session at the end of the video is only 3 minutes though.
Upvote
0
EXCELLENT Video ... will look forward to your follow up video. Thanks for all the hard word you put into this video and for sharing.
Upvote
0
Do you have any plans on making a Dot Tune video for zoom lenses? ... I suppose it'd be very tedious but it would be awesome if did make a video for zoom lenses.
Upvote
0
Maybe DotTune is a test for the calibration of the phase detection calibration of the camera. If the requirement for AFMA is that the camera has correct calibration and the lens needs correction, then DotTune should surely work because live view and PD will be correctly aligned. But, if the lens is correctly calibrated and it is the camera that requires calibration, then DotTune will fail. For example, if the camera's PD leaves the factory at effectively -10, then if you test a lens that is perfect using FoCal or the sloping ruler method, an AFMA of +10 is required. But, if you use DotTune, then a reading of -10 for AFMA will be found because the miscalibrated PD will be 10 units less than it should be. Similarly, if the camera is -5 and the lens is +5, FoCal or the ruler will give +10 but DotTune will give +5.
Upvote
0
Your AFMA value will only be as good as the accuracy of the AF detect. If the AF detect in your model is ineffective at determining where the AF DOF falls off then it can't accurately calculate the AFMA value. Has nothing to do with the PD calibration of the camera. In fact some have shown that the AF detect fall off is not symmetric around the ideal AFMA value. Without symmetry your setting will always be off. The lack of symmetry cant always be blamed on the camera either. PD system works more or less efficiently depending on how parallel the light rays are coming into it's sensor. Some lenses also tend to be OOF in front of the subject but have more DOF behind the subject. In those cases the symmetry will be off and an incorrect AFMA value may be calculated when using the Dot method. There are lots of factors that throw off the AF detect systems which is why this is not reliable.
AlanF said:
Upvote
0
AlanF said:
This gets into the inner workings of PDAF and how it relates to AF tune. My working theory is that AF tune addresses mis-alignments/mis-calibrations of anything to do with the full optical path of the PD sensing mechanism. This particularly includes the lens, which can have optical aberrations that lead to incorrect PD phase readings (since the split light beam on the mirror is sampled from opposite sides of the lens, any decentering, even slight, can cause the PD phase readings to be off). It also includes lens mount alignment of the body+lens, mirror and prism alignment, and the AF sensors themselves. Basically everything but the lens AF motor/servo. Since PDAF has been found to demonstrate closed-loop attributes, any AF-motor/servo specific issues should in theory be automatically corrected during the multiple iterations of the sensing+movement cycles involved in a single PDAF acquisition.
Upvote
0
DotTune works the same for zoom lenses. The only difference is that some zooms exhibit focal length-specific tune differences, which isn't unique to the tuning method used but is instead a property of the lens or lens+AF system. The typical recommendation is to tune at the focal length you use most often, or to tune at some in-between focal length as a compromise, provided that compromise produces acceptable results for you.Rienzphotoz said:
Upvote
0
Thanks East. It's clear whatever the reason that DotTune doesn't work for me. The sloping ruler method does because it is a direct readout of where the lens is focussing. My focus is spot on using it.
Upvote
0
PhotoCat said:
I agree about sharing the idea freely BUT...as Horshack, once he knew the answer to the afma question he had to scream to the dslr world....OH OH OH....OH OH OH OHH OHH OHH!!
Welcome back Kotter.
Upvote
0
It worked on my 5D2 and 85/1.8 combo! This is the first time I have any consistent results from AFMA!
A value of -10 makes my full length portrait much sharper at f1.8 at a shooting distance of 12 to 17.5 feet.
A -10 AFMA would however make any shots less then 10 feet away slightly OOF and I am not surprised as this appears to be one limitation of Canon's AFMA implementation to date IMHO. Canon's AFMA appears subject distance dependent.
For that kind of shorter distance (< 10 feet), I needed to turn off AFMA to get a sharp image.
It seems that Canon's factory lens calibration is much better than can be achieved by on-camera AFMA, as far
as distance dependence is concerned.
Now sigma has the USB lens dock for focus calibration, I hope Canon would catch up quickly by offering a Canon USB lens dock so we can perform factory quality lens calibration at home...
A value of -10 makes my full length portrait much sharper at f1.8 at a shooting distance of 12 to 17.5 feet.
A -10 AFMA would however make any shots less then 10 feet away slightly OOF and I am not surprised as this appears to be one limitation of Canon's AFMA implementation to date IMHO. Canon's AFMA appears subject distance dependent.
For that kind of shorter distance (< 10 feet), I needed to turn off AFMA to get a sharp image.
It seems that Canon's factory lens calibration is much better than can be achieved by on-camera AFMA, as far
as distance dependence is concerned.
Now sigma has the USB lens dock for focus calibration, I hope Canon would catch up quickly by offering a Canon USB lens dock so we can perform factory quality lens calibration at home...
Upvote
0
- Status
