Live-View, Legacy Lenses ... and CoC!

[Manfred M]

I refer to the size, 995KB and 1310KB. Info in the pictures.

George

Ask the people that designed the JPEG compression algorithm. This type of swing is file size (given that the images are NOT identical) is quite common, in my experience.

[Stagecoach]

Yep, you guessed it; the left was what I got from the stupid green light and the right was from using my eyeball!

So, it looks like the Sigma AF sensor is easily fooled - or my simpleton's view of what a cross-type sensor can do is just plain wrong. I'm sure "most of us" know better.

How many times did you repeat this test, and is this the best or the worst example out of the total?

[xpatUSA]

How many times did you repeat this test, and is this the best or the worst example out of the total?

I get your drift, Grahame. Yes, with the 28mm it was indeed a sample of one test. Statistically insignificant, of course. I was so gob-smacked, I posted it, anyway ..

Had tried it two or three times previously with the 24mm Takumar with similar, but not as clear-cut, results.

Things yet to do:

Same outside where there's more light (10EV in the shade at noon today, good old Gulf Coast Texas).

Try Brian's idea with a loupe if I can handle it without dropping the camera.

[xpatUSA]

Things yet to do:

Same outside where there's more light (10EV in the shade at noon today, good old Gulf Coast Texas).

Sometimes it beeped, sometimes not. Beginning to lose faith in the Sigma AF sensor and to maintain my faith in Live View ...

Try Brian's idea with a loupe if I can handle it without dropping the camera.

Didn't work at all, neither on low diopter nor on high. Optical theory is probably not that simple, eh?

[pschlute]

My experience with using green signal/ audible beep to confirm focus with a manual lens is it is not reliable. Try focusing the same shot once starting from the infinity end of the lens and then secondly starting from the nearest focus end. Often you will get different results similar to your findings. Rather than the signal being given at a single focus position there seems to be a small but noticeable range where focus is confirmed

I think the point is that PDAF is not the 100% accurate tool we assume it to be. It is a 40 year old technology that we are able to examine now in great detail with our umpteenth MP cameras. It is no coincidence that all major DSLR manufacturers now include a autofocus fine-adjustment menu in their cameras now to correct for front or back focus issues

[xpatUSA]

My experience with using green signal/ audible beep to confirm focus with a manual lens is it is not reliable. Try focusing the same shot once starting from the infinity end of the lens and then secondly starting from the nearest focus end. Often you will get different results similar to your findings. Rather than the signal being given at a single focus position there seems to be a small but noticeable range where focus is confirmed

Yes, that is my experience too, in that there is what is called "hysteresis" forcing one to find a middle ground in what is, in any case, a highly non-linear adjustment. I mean, just look at the stupid lens marks which go from infinity to about 10 feet in maybe 5 degrees of rotation!

I think the point is that PDAF is not the 100% accurate tool we assume it to be. It is a 40 year old technology that we are able to examine now in great detail with our umpteenth MP cameras. It is no coincidence that all major DSLR manufacturers now include a autofocus fine-adjustment menu in their cameras now to correct for front or back focus issues

Yep, I actually prefer CDAF ... nice and simple.

[Manfred M]

Try Brian's idea with a loupe if I can handle it without dropping the camera.

That's how the folks that shoot with a view camera focus.

[Stagecoach]

If we consider a simple control circuit it could very well be that the PDAF is telling a lens to 'drive' in one direction until the PDAF measures/determines that things are acceptably in focus. At this point it may be that the 'green' light comes on and the drive motor ramps down to a stop, slightly overshooting the physical position when the light was activated.

This may have been designed in the system as a means to bring things to a standstill around the centre of the 'dead band' that any system has to have to alleviate hunting.

Lots of 'ifs' and 'maybes', all logical, but we will never know how the PDAF control circuit has been designed or the accuracy of any user testing.

Edit, being typed whilst last two comments added.

[rpcrowe]

I find it fairly difficult to use live view enlargement on my 6D2 to nail manual focus unless I am using a tripod or some other camera support. Hand holding using the enlarged live view as a means to focus is difficult, especially with a longer focal length since the camera jiggles. The lens IS doesn't seem to kick in when I am focusing this way. At least I don't notice the IS kicking in.

OTOH using the eye level EVF of my Sony A6500 it seems easier for me to nail the focus. Add Focus Peaking into that equation and it is almost as easy for me to nail focus (on a still subject) as it is when I am using an auto focus lens. Now, moving subjects are quite different. AF wins hands down in that case.

DP Review mentions this about focus peaking with my Canon 6D Mark ii, "Jun 29, 2017 - Although the 6D Mark II doesn't offer focus peaking, the magnification feature in live view mode provides a detailed enough on-screen image for accurate manual focus. ... And of course Dual Pixel AF is so good that there's less need for manual focus anyway." I will admit that the dual pixel autofocus is the cat's meow However, it doesn't do a darn bit of good when using a lens that is manual focus only