Comparison of AF speed/accuracy on mirrorless FF

[ChrisRoss 150]

Yes it varies with the lens and with macro lenses how much magnification you crank on.  But you can always add light. 

I would also agree that avoiding adapted lenses is wise for macro - though the Canon EOS-R adapter you might hope would do a better job than a reverse engineered third party adapter??? 

I like the 60mm macro on the EM-1 II and it does a decent job much of the time, only too many particles in front of the subject really trip it up, it's not perfect but it's decent and you certainly wouldn't want to go backwards from that.

 

[Architeuthis 183]

1 hour ago, Interceptor121 said:

Yes Olympus specs are for 1.2 prime lens at not very long end so when you look at 2.8 you are at the -4.5 Ev of anybody else in single AF

Tracking uses motion vector estimation https://en.wikipedia.org/wiki/Motion_estimation this requires processing of entire frames not just a contrast point which in turns means the camera needs to have full video frames processed to analyse to see if it has lost the plot or not. This in turn means the level of light required are increased.

In the mix goes also how fast the lens is to achieve focus. Some lenses like the Olympus 60mm are pretty bad and they have a lot of breathing when you focus. So if you focus it first and then stay around that area it works much better. But the performance compared to a standard zoom lens even at the same aperture is just terrible

The ability of the camera to work at low level of illumination together with the ability of the lens to focus quickly determine how effective it is underwater. So going back to the op question assuming the test is performed with lenses of similar aperture (generally this is 2.8 for full frame not less or they may not focus at all) a camera with a rating of -6 Ev will work better than one with -3 Ev which is where this started as a starting point. Then you need to overlay focus speed and accuracy that has also to do with the lens.

For tracking there is an additional dimension which is the quality of the motion vector estimation which depends on the processing quality and power of the camera.

Sony for example has very good record of tracking in daylight but at the same time a pretty bad reputation of not focussing at all in low light and being slow with macro lenses.

If you add a focus light in the mix (obviously this is important because a lot of macro is inside crannies where there is no light and you are much lower than your desk) almost all cameras can focus (eventually) at the focus speed of the system.

I would be a bit calm before saying that underwater there is plenty of light there is not. A nudibranch on a rock sitting flat with light hitting it is not the same as trying to shoot a seahorse on a fan at an angle where light is not reflecting.

So if I had to look at performance of a lens for macro I look at the following dimensions

1. Camera ability to focus in low light (most cameras are rated -4 Ev and lower and I would say away from anything that says -3 or -2 Ev)

2. Lens time to scan for focus. Before you can even think of engaging tracking you need to pre-focus and that first operation is critical so you can continue

I do not look at how good the camera is at motion estimation and this is not a published stat so there is no way to read something and see if this camera will be better than another at tracking

Going back to @Architeuthis question. I use the MC21 adapter and metabones like he does. With adapted lenses only Single AF works reliably so tracking is not available. The two criteria above hold.

The Sony camera rating of -3 Ev is not convincing for me

Among Canon R5, R6 and R the rating is respectively -6, -6.5, -6 but obviously the R6 has a lower megapixel count 

Having discounted adapted lens if you wanted CAF or tracking you are left with the Canon 85 mm f/2 which is not known to be particularly fast.

The reality is that DSLR in this area is still way ahead when it comes to focus speed as the lenses focus really fast.

It would be worth to find someone that has tested the Canon RF 85 AND the Sony 90mm to see in real life how those work I would avoid any adapted lens for macro. Perhaps the Sony lens is better than canon and makes up for the camera being less able or you dive with a focus light and you are done with it. 

I would not invest on a system if there are no native lenses as adapted lenses are always a compromise you can take for wide angle but not for macro in my experience.

Hopefully this last post helps Wolfgang making up his mind

 

No fear Massimo - I will not decide now, I have plenty of time with the acquisition. I just start now to make up my mind, before I will invest many thousands of $ into a system, where I plan to stay for  alonger period. I do not plan to change later the system just because a new lens or camera with slightly better performance shows up from another company. We are lamenting here about shortcomes, but we must not forget that this is at a very high level of technical perfection...

I think that, when so many different people report controversial facts, likely both groups are right: it is just very difficult to find out what was the difference in the settings that resulted in the controversial final outcome...

I thank you and the others so far for this interesting and stimulating discussion. Hopefully more photographers will share their experience...

 

Wolfgang

 

[Architeuthis 183]

13 minutes ago, ChrisRoss said:

I would also agree that avoiding adapted lenses is wise for macro - though the Canon EOS-R adapter you might hope would do a better job than a reverse engineered third party adapter???

I am (almost) convinced that the native Canon EF to RF adapter does a great job on R5/R6: have a look into the Canon R mount forum at DPreview - dozends of of photographers state that their EF lenses focus faster and better on their new R5/R6 bodies compared to their DSLRs, the lenses have been designed for (except few reports with older, outdated EF lenses)...

 

Edited March 4, 2021 by Architeuthis

[Interceptor121 843]

2 hours ago, Architeuthis said:

I am (almost) convinced that the native Canon EF to RF adapter does a great job on R5/R6: have a look into the Canon R mount forum at DPreview - dozends of of photographers state that their EF lenses focus faster and better on their new R5/R6 bodies compared to their DSLRs, the lenses have been designed for (except few reports with older, outdated EF lenses)...

 

That would be generally possible but it will very much depend on the specific lens. The Canon EF macro lenses focus fast generally they are not sluggish so this is something to consider for sure

Personally I would not invest in a Sony system for full frame for a variety of reasons. The Canon RF system instead looks very promising

 

[Phil Rudin 485]

On 3/4/2021 at 6:45 AM, Architeuthis said:

Precious information from someone who has practical experience with many different systems - thanks a lot Phil!

I have the impression that Sony offers at present a lot of advantage (size, IQ, price), would there not be the controversial issue with macro, therefore I am eager to hear as many opinions as possible, before I decide...

Since you have a lot of experience with EM1II (that I have now), may I ask you to compare AF with the Sony 90mm macro on R7iv (7c) with the EM1II and Zuiko 60mm (Panasonic 45mm macro may be even more appropriate. One may say also that the Sony 90mm should have an additional diopter to give 2:1 to be really comparable with Pana 45mm at 1:1). This would give me a reference feeling how AF works with the Sony...

What regards my personal experience, S-AF with Pana 45mm on EM5II (CAF) was good, but Zuiko 60mm a little cumbersome (when I just had EM5II, I did not regard S-AF with Zuiko 60mm cumbersome). Since I have EM1II (PDAF), C-AF, both with 45mm and 60mm is the method I prefer, not th eleast issue with 45mm, with Zuiko 60mm (probably depending on light), sometimes I have to try several times until C-AF is maintained during reframing. The example picture was (even) taken with Canon 100mm IS macro, adapted with metabones 0.71x (so 71mm in fact), C-AF to the eye and then reframing (but this really is the limit of the system, AF with the adapted 100mm is really cumbersome...):

 

Wolfgang

Hi Wolfgang, 

First let me say that I have no skin in the game and it makes no difference to me what anyone chooses to buy, I just try to report the latest news. Second I don't like many of the talking heads worry about theory and +/- this or that, I just shoot the equipment and then post my results. 

So my first advise is to say that everything in photography is a tradeoff. Smaller sensor more DOF at a given F/stop. Bigger sensor better IQ and on it goes. My testing process is simple.

Second any new macro lens I test at 1:1 using the 35mm slide mount and an assortment of C/U lenses I many have in house. To test how well a lens and camera combo focus (accuracy, speed, settings to gain the best results and more) I go to Blue Heron Bridge which is near my house. Rather than to pour over the lux and the -EV and alike I just go under the bridge in close to total darkness and shoot. Second tidbit of advise more time in the water and less on the interweb. 

Third standard I can always find fish eggs in the same spot on a sailboat keel which is under the bridge, you can see the blue bottom paint in all of the photos. Here I test how well I can focus at 1:1 and beyond with added C/U lenses.

I have also attached four images taken with the Sony A7R IV and 90mm macro, two of eggs and two of the same Jawfish. All four images are taken from the same distance the larger in full frame 35mm and the second switched to APS-C which I have set to my AF-ON button. It is faster for me to use this method which acts like a 135mm macro on FF than to add a C/U lens. I still get a 25MP image and better image quality than M43 or most APS-C cameras.

File size for FF is 9504X6336, APS-C is 6240X4160.

The slide mount images are with M43 starting at 1:1 and scaling down to a +15close up lens a very small image with very little DOF and shot at near mm's. Getting this small would require tons of magnification on FF and you would have no DOF and no working range. Up side for M43 but with 61MP files I can crop and get about the same amount of information in the frame.

 

  

 

Edited March 5, 2021 by Phil Rudin

[Interceptor121 843]

Hi Wolfgang, 
First let me say that I have no skin in the game and it makes no difference to me what anyone chooses to buy, I just try to report the latest news. Second I don't like many of the talking heads worry about theory and +/- this or that, I just shoot the equipment and then post my results. 
So my first advise is to say that everything in photography is a tradeoff. Smaller sensor more DOF at a given F/stop. Bigger sensor better IQ and on it goes. My testing process is simple.
First any new macro lens I test at 1:1 using the 35mm slide mount and an assortment of C/U lenses I many have in house. Second to test how well a lens and camera combo focus (accuracy, speed, settings to gain the best results and more) I go to Blue Heron Bridge which is near my house. Rather than to pour over the lux and the -EV and alike I just go under the bridge in close to total darkness and shoot. Second tidbit of advise more time in the water and less on the interweb. 
Third standard I can always find fish eegs in the same spot on a sailboat keel which is under the bridge, you can see the blue bottom paint in all of the photos. Here I test how well I can focus at 1:1 and beyond with added C/U lenses.
I have also attached four images taken with the Sony A7R IV and 90mm macro, two of eegs and two of the same Jawfish. All four images are taken from the same distance the larger in full frame 35mm and the second switched to APS-C which I have set to my AF-ON button. It is faster for me to use this method which acts like a 135mm macro on FF than to add a C/U lens. I still get a 25MP image and better image quality than M43 or most APS-C cameras.
File size for FF is 9504X6336, APS-C is 6240X4160.
The slide mount images are with M43 starting at 1:1 and scaling down to a +15close up lens a very small image with very little DOF and shot at near mm's. Getting this small would require tons of magnification on FF and you would have no DOF and no working range. Up side for M43 but with 61MP files I can crop and get about the same amount of information in the frame.
 
  
 

Not everybody lives 5’ from blue heron bridge
Understanding how things work helps maximising your chances of success especially now that all pools are closed and in essence there is no diving other than your bathtub
Sure going and testing is the right thing to do but to make dismissive statements spend more time in the water denotes a lack of understanding and empathy for majority of people that do not have the opportunities that you have and don’t do this as their day job
Sorry that was really poor


Sent from my iPhone using Tapatalk

[Gudge 59]

Wolfgang sent me a PM and asked me to comment on my experience with the AF performance of the Sony A7RIV and Sony 90 macro and the Canon 100 macro on Canon dSLR.

I was a long term Canon dSLR user (20D, 40D, 50D, 7D, 7DMkII and 5DMkIII) all with the Canon EF 100mm f/2.8 USM Macro.  Just over a year ago I changed to Sony A7RIV.  Initially I used the Canon 100 macro on the Sony with a Metabones adapter but changed to the Sony 90 macro after 3 months.  My wife still uses my 7DMkII with the Canon 100 macro so I was able to compare all combinations of camera/macro lens possible just now.  Room was dimly lit.  Results varied based on what focus point setup I used on each camera/lens.

With all focus points in play and cameras set to AI Servo (Canon) AF-C (Sony) the winner in grabbing focus quickly was the 7DMkII/Canon 100 macro with the A7RIV/Sony 90 macro a very close second (it was occasionally faster than the Canon).  The A7RIV/Metabones/Canon 100 macro was a distant third (Which is why I got the 90 macro 12 months ago when I did the same test).

When I  changed to my preferred mode of shooting with each camera - AI Servo with Expand AF (Canon) and AF-C and Tracking Flexible Spot (Sony) - things changed.  The A7RIV/Sony 90 macro was first grabbing focus fastest and holding it better than the 7DMkII which came a close second.  The A7RIV/Metabones/Canon 100 macro came third.

Points to consider:

• Before I switched to Sony the 7DMkII was by far the best low light focussing camera I had ever used.
• Depending on how you set up the autofocus system the A7RIV is better than the 7DMkII with their respective native macro lenses
• The effect of the Metabones adapter on focus performance will differ from lens to lens.  While the Canon 100 macro is slower on the Metabones adapter my Canon 8-15 fisheye (for which there is no Sony equivalent) focusses much faster with the Metabones adapter on the A7RIV than it did on the 7DMkII.

 

Edited March 5, 2021 by Gudge

[Architeuthis 183]

2 hours ago, Gudge said:

Wolfgang sent me a PM and asked me to comment on my experience with the AF performance of the Sony A7RIV and Sony 90 macro and the Canon 100 macro on Canon dSLR.

I was a long term Canon dSLR user (20D, 40D, 50D, 7D, 7DMkII and 5DMkIII) all with the Canon EF 100mm f/2.8 USM Macro.  Just over a year ago I changed to Sony A7RIV.  Initially I used the Canon 100 macro on the Sony with a Metabones adapter but changed to the Sony 90 macro after 3 months.  My wife still uses my 7DMkII with the Canon 100 macro so I was able to compare all combinations of camera/macro lens possible just now.  Room was dimly lit.  Results varied based on what focus point setup I used on each camera/lens.

With all focus points in play and cameras set to AI Servo (Canon) AF-C (Sony) the winner in grabbing focus quickly was the 7DMkII/Canon 100 macro with the A7RIV/Sony 90 macro a very close second (it was occasionally faster than the Canon).  The A7RIV/Metabones/Canon 100 macro was a distant third (Which is why I got the 90 macro 12 months ago when I did the same test).

When I  changed to my preferred mode of shooting with each camera - AI Servo with Expand AF (Canon) and AF-C and Tracking Flexible Spot (Sony) - things changed.  The A7RIV/Sony 90 macro was first grabbing focus fastest and holding it better than the 7DMkII which came a close second.  The A7RIV/Metabones/Canon 100 macro came third.

Points to consider:

• Before I switched to Sony the 7DMkII was by far the best low light focussing camera I had ever used.
• Depending on how you set up the autofocus system the A7RIV is better than the 7DMkII with their respective native macro lenses
• The effect of the Metabones adapter on focus performance will differ from lens to lens.  While the Canon 100 macro is slower on the Metabones adapter my Canon 8-15 fisheye (for which there is no Sony equivalent) focusses much faster with the Metabones adapter on the A7RIV than it did on the 7DMkII.

 

Great additional info, Grudge - Thank You a lot!

 

Wolfgang

[Interceptor121 843]

4 hours ago, Gudge said:

Wolfgang sent me a PM and asked me to comment on my experience with the AF performance of the Sony A7RIV and Sony 90 macro and the Canon 100 macro on Canon dSLR.
I was a long term Canon dSLR user (20D, 40D, 50D, 7D, 7DMkII and 5DMkIII) all with the Canon EF 100mm f/2.8 USM Macro.  Just over a year ago I changed to Sony A7RIV.  Initially I used the Canon 100 macro on the Sony with a Metabones adapter but changed to the Sony 90 macro after 3 months.  My wife still uses my 7DMkII with the Canon 100 macro so I was able to compare all combinations of camera/macro lens possible just now.  Room was dimly lit.  Results varied based on what focus point setup I used on each camera/lens.
With all focus points in play and cameras set to AI Servo (Canon) AF-C (Sony) the winner in grabbing focus quickly was the 7DMkII/Canon 100 macro with the A7RIV/Sony 90 macro a very close second (it was occasionally faster than the Canon).  The A7RIV/Metabones/Canon 100 macro was a distant third (Which is why I got the 90 macro 12 months ago when I did the same test).
When I  changed to my preferred mode of shooting with each camera - AI Servo with Expand AF (Canon) and AF-C and Tracking Flexible Spot (Sony) - things changed.  The A7RIV/Sony 90 macro was first grabbing focus fastest and holding it better than the 7DMkII which came a close second.  The A7RIV/Metabones/Canon 100 macro came third.
Points to consider:

• Before I switched to Sony the 7DMkII was by far the best low light focussing camera I had ever used.
• Depending on how you set up the autofocus system the A7RIV is better than the 7DMkII with their respective native macro lenses
• The effect of the Metabones adapter on focus performance will differ from lens to lens.  While the Canon 100 macro is slower on the Metabones adapter my Canon 8-15 fisheye (for which there is no Sony equivalent) focusses much faster with the Metabones adapter on the A7RIV than it did on the 7DMkII.

 

So low light matters
And the - ev of both those cameras is -3 ev
Which means they are very very close
So the extrapolation of this metric out of specs in this case is a valid proxy
Tracking however does need specific testing


Sent from my iPhone using Tapatalk

Edited March 5, 2021 by Interceptor121

[Phil Rudin 485]

18 hours ago, Interceptor121 said:

Not everybody lives 5’ from blue heron bridge
Understanding how things work helps maximising your chances of success especially now that all pools are closed and in essence there is no diving other than your bathtub
Sure going and testing is the right thing to do but to make dismissive statements spend more time in the water denotes a lack of understanding and empathy for majority of people that do not have the opportunities that you have and don’t do this as their day job
Sorry that was really poor


Sent from my iPhone using Tapatalk

First I chose to move to Palm Beach County over fifty years ago specifically for the diving and I have been diving  at the bridge for the entire 50 years, well before it became internationally known. Second I have been teaching U/W photography most of those fifty years and have always stressed spending time in the water as one of the best ways to improve your photo skills. I also stress to my students who don't get a chance to dive as much as I do that the techniques can be learned by shooting on land as well as in the water. I am very aware of the covid issues both in the UK and around the world, I work for a UK magazine. The point I believe I was making is that using your camera is always going to be of a larger benefit than talking about how it may work. 

Some of the first images I show in my macro class that could be done in your backyard. Learning how to get a black background and a balanced ambient light background. Macro shots of marine life are like shooting people, eye contact and slightly upward angle. 

Also just to clarify my "day job" for over thirty years was as a Professional Police Officer and Public Safety Diver. I am blessed to be in good enough health to pursue my passion for diving and U/W photography, this is not a job for me.  

Edited March 5, 2021 by Phil Rudin

[Nicool 30]

On 3/5/2021 at 9:11 PM, Gudge said:

Wolfgang sent me a PM and asked me to comment on my experience with the AF performance of the Sony A7RIV and Sony 90 macro and the Canon 100 macro on Canon dSLR.

I was a long term Canon dSLR user (20D, 40D, 50D, 7D, 7DMkII and 5DMkIII) all with the Canon EF 100mm f/2.8 USM Macro.  Just over a year ago I changed to Sony A7RIV.  Initially I used the Canon 100 macro on the Sony with a Metabones adapter but changed to the Sony 90 macro after 3 months.  My wife still uses my 7DMkII with the Canon 100 macro so I was able to compare all combinations of camera/macro lens possible just now.  Room was dimly lit.  Results varied based on what focus point setup I used on each camera/lens.

With all focus points in play and cameras set to AI Servo (Canon) AF-C (Sony) the winner in grabbing focus quickly was the 7DMkII/Canon 100 macro with the A7RIV/Sony 90 macro a very close second (it was occasionally faster than the Canon).  The A7RIV/Metabones/Canon 100 macro was a distant third (Which is why I got the 90 macro 12 months ago when I did the same test).

When I  changed to my preferred mode of shooting with each camera - AI Servo with Expand AF (Canon) and AF-C and Tracking Flexible Spot (Sony) - things changed.  The A7RIV/Sony 90 macro was first grabbing focus fastest and holding it better than the 7DMkII which came a close second.  The A7RIV/Metabones/Canon 100 macro came third.

Points to consider:

• Before I switched to Sony the 7DMkII was by far the best low light focussing camera I had ever used.
• Depending on how you set up the autofocus system the A7RIV is better than the 7DMkII with their respective native macro lenses
• The effect of the Metabones adapter on focus performance will differ from lens to lens.  While the Canon 100 macro is slower on the Metabones adapter my Canon 8-15 fisheye (for which there is no Sony equivalent) focusses much faster with the Metabones adapter on the A7RIV than it did on the 7DMkII.

 

Super useful real-life test Gudge, thanks for taking the time!

 

I watched this thread with curiosity as whether it would change my mind on Mirrorless Autofocus being a showstopper for me to use one of these cameras.

I started UW photo with a Nikon D300, striking AF, and then moved to a mirrorless Olympus (2013) only to rush back to Nikon DSLRs in 2014, after missing too many shots due to disappointing AF. 

In 2016 I bought one of the first Nikon D500 out there and enjoyed shooting it since then.

Specs sheets aside, my shooting style means I challenge autofocus to its limits: some silty dives, with current and sometimes surge bouncing left & right, I like to shoot at dusk or under rocks, or at the entrance of ocean caves. In those dark places, I turn OFF my focus light (more used as a dive light), as I noticed countless situation where, even in red mode, it would impact how close I get to a subject, and how comfortable the subject is to display a behavior. 

Oh, and I use ALL the time continuous AF tracking. It compensates my own movement in current/surge, as well of movements of my photo subjects, brilliantly (Nikon D500 + Nikon 60mm AF-S macro, or Nikon D500 + Tokina 10-17mm fisheye). Not sure if it was designed for my usage, but it surely works well, and makes a big difference in bringing that memorable shot home. 

 

My observations and frustrations over the years, when it comes to researching the mirrorless offering:

-the test benches that reputable photomags would use to compare autofocus is just irrelevant: I stopped reading these when I realized a few years back all modern cameras would nearly get top mark

-testing AF in a store is useless: those are brightly lit spaces to make sure all cameras look & perform great

-limited "field" observation shared by users of those new cameras, at least in situations comparable to my shooting style: tracking a bird in bright daylight isn't the same game, and snapping a few fish portraits in a bright tropical reef with 20 meters viz isn't the same game neither.

So, thanks again for sharing your experience. I think in the next few months I'll try to borrow/loan a recent mirrorless setup to try and see for myself, if it matches or beats my D500 in my shooting style. 

Edited March 17, 2021 by Nicool