Predictive AF Comparison Tests Lumix GH5, G85, Oly EM-1 Mark ll, and Nikon D500

Posted May. 14th, 2017 (1 week ago) by Daniel J. Cox

Let’s get right to the point. This post is a Predictive AF comparison test between the Lumix GH5, G85, Olympus EM-1 Mark ll, and the Nikon D500. I’ve been shooting the new Lumix GH5 for almost two months, and Panasonic’s DFD autofocus is now producing equal to or better results than any camera I’ve shot to date. 

Predictive AF Comparison

This post is not an overall review of the GH5. I’m guessing I’ll do that after several more months of using the new camera. For now, this is all about Predictive AF. For those not familiar with the differences between Predictive AF and Single AF, Predictive AF is action—fast moving action—and the ability of a camera to capture those “glory pictures” we all desperately want in our portfolio. Up until now, the Micro Four Thirds world hasn’t had a camera that could perform on a truly professional level when it comes to Predictive AF. Hold on before you start screaming about the new Olympus EM-1 Mark ll having these abilities until you see the results below. With the GH5, we finally have truly professional, predictive AF proficiency. 

So that’s the intro. Now let me give you the details of what I’ve been up to that’s convinced me we’re now in a different world for Micro Four Thirds shooters.

A screenshot that shows how it looks in Mylio with star ratings. Note all the 3 stars that show perfect, razor sharp focus.

If you read this blog, you may have read other posts showcasing my “Speeding Pooch Test.”  A speeding pooch, preferably a lab or golden retriever, coming straight at the camera charging after a thrown ball, is one of the best ways to test the Predictive AF capabilities of any camera. The downside to this test is the difficulty of finding enough fast dogs to get lots and lots of opportunities to produce the tests. To solve this problem I decided to go after something there are plenty of, speeding cars.

Single point AF pointed at the middle part of the grill of each car. This is a screenshot of the image at 100% with no sharpening added.

For the test results I’m about to share, I placed myself beside a roadside where the speed limit is 75mph/120kmh. I shot just short of 10,000 frames between four different cameras: Lumix GH5, Lumix G85, Olympus Om-D EM-1 Mark ll, and the Nikon D500. On the two Lumix bodies, I used the Leica 100-400mm set at 300mm (600mm equivalent) as well as the NEW 100-300mm lens also set to 300mm. On the Olympus, I used the Olympus 300mm F/4, and on the Nikon I used the 80-400mm set to 400mm (600mm equivalent). All lenses were set to their respective focal lengths to be as close to identical as possible.

6K Photo Mode Video Capture Sample

Here’s a video I shot in 6K Photo Mode. This gives you an idea of fast these cars are coming at the camera. That in itself is interesting, but just to whet your appetite, for a future post about 6K Photo Mode, watch this video for focus accuracy. Keep in mind this is video shot at 30FPS at a very high shutter speed. Once you hit the play button, click the same button to stop the video at any particular spot and take a look a the focus of the GMC label on the grille. Unfortunately, there is no way to extract these files once you take them from the camera, so you have select the individual stills while the 6K Photo Mode file is still in the camera. Not a perfect solution but one we can live with until software catches up with this amazing camera technology. That is going to be unbelievable.

Star Rating System

I chose to use the following star rating system for AF quality:

3 Stars = Perfect focus, razor sharp

2 Stars = Acceptably sharp, most people viewing the image at 100% would think it’s useable

1 Star = Completely out of focus and un-useable

Apple’s Numbers spreadsheet with the number of images and percentages. Test 2 was just the GH5 and EM-1 to confirm what I had seen in the first test.

AF Menu Options

All cameras were set to AF-C and each was slightly customized via the AF settings in the menu. When I say “slightly customized” I mean slightly customized. This is very important since, in my opinion, AF settings should be simple and straightforward. Each camera’s AF setting was slightly changed to adjust the AF responsiveness. All were set to be most responsive since a car heading down the roadside is not going to have any subjects moving in front of it that might affect the AF misinterpreting the actual subject

Each camera has the ability to customize AF capabilities via the menu. In theory, this could be a good thing, but in practice, I feel there are too many options on all the cameras except the G85 and the Nikon D500. It’s my opinion AF should just work without any—or at least very little—customization. Canon started the idea of ultimate customization and the others have followed. When you point the AF sensor at the subject it should just follow it. Ideally, there should be no need for selecting erratic, constant, fast, short, etc. The GH5 has a pretty complicated customization process, but I chose to ignore most of it, opting for changing only the AF Custom Settgins> AF Sensitivity to +2. Below are details of similar changes to the other cameras.

  • Lumix GH5: Burst Rate = High, AF Custom Settings> AF Sensitivity +2, AF-C Focus/Release Priority set to Balance
  • Lumix G85:  Burst Rate = High,  Custom Settings>AF Sensitivity +2, AF-C Focus/Release Priority set to Focus
  • Olympus OM-D EM-1 Markll: Burst Rate = Low, C-AF  set to Loose +2
  • Nikon D500, Burst Rate = High, AF-C Priority Selection= Focus+Release

OK, so those were my settings and it’s here that I predict I’ll take the most heat. Why? Because there are other menu options and there will be those who will shout I didn’t do enough to explore them all. To that I say, be my guest and do it yourself. My options are a great place to start and were selected to put all cameras on as similar settings as possible so we could have an even playing field. Additionally, the settings I chose are simple and straightforward, which is how all AF systems should work. In other words, just give me great AF and I’ll be happy. Fine tuning each camera’s AF could make a difference, but the GH5’s keeper rate was so high, I’m thrilled with the results I’m getting with the basic settings. If you decide to choose one of the other cameras that didn’t do as well, you may want to do your own tests to see if digging through all the menu options can make your camera of choice more accurate than the GH5.

Download Image Samples

For those wanting to see the images for themselves and evaluate them for AF accuracy, I’ve uploaded several hundred samples from each camera to my Photo Shelter site. The links to those collections are below and you’re free to download any or all photos as long as they are kept on your own personal computer for your own personal use. Do not post these anywhere else without my written persimmon and keep in mind I retain all copyrights to the pictures I’m sharing. These galleries do not include the entire shoot from each camera. I may add more at a later day when I get somewhere with more bandwidth. I’m currently writing this from Croatia and I simply can’t get as many as 1000-1500 jpegs per camera up to my Photo Shelter account in a timely manner.

Download Lumix GH5 with 100-400mm lens samples
Password: gh5-100-400
1778 images

Download Lumix GH5 with 100-300mm lens samples
Password: gh5-100-300
455 images

Download Olympus OM-D EM-1 Mark ll with 300mm lens samples
Password: olyem-1markll300mm
1412 images

Download Nikon D500 with 80-400mm lens samples
Password: d500-80-400
416 images

Please Note: Unfortunately, PhotoShelter limits the number of photos you can download per zipped folder, so for the two galleries with 500+ images (GH5 with 100-400mm and Oly with 300mm), there are multiple zipped folders to download all of the images. Once in the gallery, click ‘Download’ in the upper right corner. Enter the password, and you will be taken to a screen that looks like this:

Select the size you want to download in the dropdown box, and then click ‘Continue.’ You will then be taken to a screen containing several links that looks like the screenshot below. Click on each link to download the corresponding zipped folder of images.

Panasonic Lumix Depth From Defocus Technology

So let’s end this conversation with a little discussion about Panasonic’s Depth From Defocus (DFD) technology. Based on these tests, Panasonic is close to my prediction of perfecting their revolutionary DFD AF system, something that virtually every magazine, website, forum, and the like have said cannot be done. Everybody, except me, has blogged, vlogged, chatted, and screamed, DFD, a Contrast Detection-based AF system will never equal the Phase Detection AF virtually every other serious camera manufacturer is using. I’ve said it time and time again, these naysayers sound just like the folks who were convinced digital would never replace film. Having the camera analyze autofocus based on the sensor itself, as DFD does, makes all the sense in the world. Phase Detection AF judges focus based on a separate AF sensor that is different from the image sensor thus making inaccuracies very possible.

It’s the Phase Detection system that Olympus, Nikon, and Canon use that requires they all have the ability for Micro Lens Adjustments. Micro Lens Adjustment, in the menu of the cameras I mentioned, allows the photographer to change the calibration of lens and camera that is either back or front focusing, a common problem with Phase Detection systems. The GH5 doesn’t have a Micro Lens Adjustment option since it’s never needed.  I experienced back focus issues for many years when I was shooting Nikon bodies. The issue raised its ugly head again this past February while shooting the new Olympus OM-D EM-1 Mark ll with the amazing Olympus 300m F/4. I knew the Olympus 300mm was much sharper than the images I was getting. I finally did a proper test and found the lens and camera were back focusing. A tweak of the Micro Lens Adjustment on the camera took care of the problem, but I’m happier the GH5 doesn’t need this option.

Conclusion

For those who want to see ALL the images for themselves, you can download the entire collection from my PhotoShelter account. I’ve broken them up into each camera’s own folder. These are for your own personal use and I retain all copyright to these images. Do not post them anywhere without my written permission. Go ahead and knock yourself out trying to second guess me. I think you’ll see for yourself this camera is the real deal, and to all those naysayers who are convinced Depth From Defocus is second rate technology to Phase Detection AF, this camera proves them wrong.

Speical thank you to Natural Exposures Explorer Fred Kurtz, a numbers man, who helped me get the numbers properly set up in Apple’s spreadsheet program.

Also, want to thank Marshall Lewis of Bozeman Camera for loaning me the Nikon D500 to test. If you want great service and reasonable prices give Marsha a call at +1 (406) 586-8300 or email: bozemancamera@gmail.com

Editorial Note:

Please note that I work with Panasonic as a Lumix Luminary. Some may think this will affect my integrity regarding these kinds of reviews. Nothing could be further from the truth. I worked with Nikon, unofficially, for nearly 35 years and I never received a dime. But I did so because I believed in their products. I now feel the same about Panasonic Lumix. Panasonic approached me to become a Luminary almost ten years after first purchasing one of their cameras, the Lumix GF1. Though Panasonic pays me a small stipend annually, no amount of money is worth the trust I’ve established with my readers and people who know me. I feel it’s important for my readers to know my connection to Panasonic so you can decide for yourself. 

Add Your Voice!
There are 19 comments on this post…
  1. Portrait of Jane Scott Norris

    JaneOn May. 21st, 2017 (1 day ago)

    What a terrific article, Dan. As always you bring us lots of information, information I can trust. While I am sticking with Panasonic, not considering Olympus, I am nevertheless pleased to hear you consider the GH-5 to be a great performer. We have bought the GH-5 and really like it. I believe its AF capabilities are better than the GH-4. Glad to hear more about how well the G85 performs as this is an entry level option for some family members. THANKS!!

    • Portrait of Daniel J. Cox

      Daniel J. CoxOn May. 21st, 2017 (1 day ago)

      My pleasure Jane. Thaks for adding your voice to the conversaton.

  2. Scot PerryOn May. 17th, 2017 (5 days ago)

    I have a G85 and am happy with photo focus video AFC not so much. Was surprised you mentioned adjusting the focus on the Olympus as the phase detect is on the imaging chip I would not think this is necessary?

    • Portrait of Daniel J. Cox

      Daniel J. CoxOn May. 17th, 2017 (5 days ago)

      Scot, Not sure about the Phase Detection being on the sensor. From what I understand, Olympus combines both Phase and Contrast detection and they also give the option for Micro Focus Adjustment on the camera. Thankfully it works great when needed but not excited it’s necessary.

  3. HeathOn May. 17th, 2017 (6 days ago)

    The d500 and d5 with a 70-200 f2.8 E (the latest one) would make a marked difference, it’s one of the fastest if not fastest lenses out there for autofocus.

    Also I’ve seen a d500 track from the very left side of the frame to the very right and back again with 0 issues. No camera I’ve seen has done that

    • Portrait of Daniel J. Cox

      Daniel J. CoxOn May. 17th, 2017 (5 days ago)

      Heath, you may be right but I was trying to keep all the lenses as close to the same as possible for direct comparison purposes.

  4. Rich BallOn May. 15th, 2017 (1 week ago)

    Dan – Thanks for this post it gives me more to think about. I’ll defend the Nikon a little. The 80-400 is one of their older lenses if I’m correct. I would think the AF response is slower. This is a bit of an issue with both Canon and Nikon older lens designs .

    You were out in good bright light which is an advantage. Channeling my inner curmudgeon let me set up this shooting scenario. Since your a Montana guy – suppose your out on a hike in the autumn find two bull elk starting to fight over some females. A hard scene for any photographer to resist. It is also late in the afternoon and there is a heavy cloud cover. The light is pretty marginal but you are close enough to your car so that you aren’t too worried about getting out. Is the GH5 going to be able to find focus?

    Another scenario. This hike is in early summer. You find some pink lady slipper orchids. Hard to find. They are in prime condition and you have one of the very nice Panasonic macro lenses along. They are in the deep understory and again the light is marginal. Even though you don’t do flowers they are just too pretty to resist. Will the AF work? If it doesn’t have improvements been made to focus peaking so that it works in low light? With my Panasonic cameras focus peaking is useless in lower light situations. Also with macro lenses it starts the AF from infinity and often can’t find focus with subjects close to the minimum focus distance.

    I’m pretty sure my Canon 5D mkIV would be OK. I know my Panasonic cameras would struggle in both in both cases.

    In closing I realize the above are not ones that rely on the predictive capabilities of GH5. They are fairly realistic. I agree with your assessment that AF has been made too complicated. I remember reading an article a while ago that Canon had published a 160 page guide on using the AF for the EOS1DX mark II. 160 pages! Really guys!

    Thanks Rich Ball

    • Portrait of Daniel J. Cox

      Daniel J. CoxOn May. 15th, 2017 (1 week ago)

      Rich, the 80-400mm I used is the most current version as far as I know. I used it for a couple of years and was very happy with it’s AF capabilities. I guess it’s possible Nikon made have made some updates since the one I bought maybe five years ago.

      As far as the two scenarios you give, I’ve actually done both, or something very similar to both and had no issues focusing. There have been issues with Panasonic cameras in some situations where the focus is so far out, the camera won’t respond to bring the focus in. So far I’ve not experienced that with the GH5. I did have a problem in Botswana where the GH5 seemed to be doing the same thing but once I went from 49 Point AF to Single AF it had no issues at all.

  5. Dean SwartzOn May. 15th, 2017 (1 week ago)

    Dan, I just want to follow up on this AutoFocus thread. I had a chance to review about 400 images I took over the weekend of my son’s soccer game. While not as reliable as your “Speeding Pooch” testing, a field full of 16 crazed 11-year-old boys running around like maniacs is a reasonable alternative. You know how much trouble I have with “birds in flight” (no snide remarks, please). But, using the Olympus E-M1 Mk2 with the latest firmware and the 40-150mm f2.8 Pro in good light (shooting at f2.8, ISO 200, and 1/1,250-1/2,000 and “low speed silent mode” in C-AF) my 3-star rate was roughly 9 out of 10. This is by no means intended to call into question your significantly lower 3-star rate with the Oly E-M1 and 300mm Pro, but is just a different set of circumstances to consider. [Did you use the new E-M1 (1.2) and 300mm firmware (1.2)? Not sure if that would have made any difference, but they included improvements to the IS of both camera and lens.] As always, thanks for encouraging and facilitating this valuable discussion.

    • Portrait of Daniel J. Cox

      Daniel J. CoxOn May. 15th, 2017 (1 week ago)

      Hmmmm………. Dean. Interesting question about the firmware. I’m not sure. I think it’s possible I don’t have the most current firmware. Sounds like there might be more tests coming if that’s the case. I have to say I was very surprised at the low numbers myself. I won’t be able to check until I get back from my two-month shoot in Europe. That will be around July 7th. Will follow this all up at that time. Thanks for the heads up.

  6. RikardOn May. 15th, 2017 (1 week ago)

    Hi Daniel
    Thank you for this post. I really needed this.
    I am using GH4 since it came and are disappointed by the tracking on that camera. Would be nice to see the comparison between GH4 and GH5 in this test also.
    Almost all focus on the GH5 are on its video capabilities.
    I am a still photographer.
    I am considering buying the GH5. After this test I am more convinced.
    I would like to also see an ISO comparison from you with the same cameras. Maybe also include a full frame.
    I read so many different “facts” about GH5 ISO capabilities.
    Thank you
    Rikard

    • Portrait of Daniel J. Cox

      Daniel J. CoxOn May. 15th, 2017 (1 week ago)

      Rikard, I do feel the GH5 offers substantial AF capabilities over the GH4 but that’s based more on gut feeling rather than any comparison tests. Not sure I will do any comparison between the cameras since I’m so pleased with the GH5, I’ll only be using my older GH4s for some video work. Regarding the ISO capabilities, again I have no actual comparisons on this but I’m very happy with the GH5. There is an improvement in higher ISOs with the GH5, how much? I’m not sure. All I know is that MFT cameras don’t compete with full frame Nikon, Canon and Sony cameras. However, with good noise reduction software, I use DXO Optics Pro 11 for noise reduction, I’m very happy with what I’m able to get out of my GH4 and even happier with the GH5.

      Here’s an image I recently shot in Namibia of the night sky that I used a slight amount of noise reduction on.

  7. Marko KoskenojaOn May. 14th, 2017 (1 week ago)

    Interesting results – technology marches forward with M43 cameras and lenses

  8. AndrewOn May. 14th, 2017 (1 week ago)

    Should be noted that according to the G85 manual (pg 116) it does not perform predictive focus when you you set release priority to focus, therefore under the context of your title it was not tested correctly.

    • Portrait of Daniel J. Cox

      Daniel J. CoxOn May. 14th, 2017 (1 week ago)

      Andrew, you are correct. This may actually improve the G85’s results which were pretty good for an inexpensive camera already. I’ll rerun this test with the G85 later this summer when I get back from Europe. This minor change still keeps within my parameters of AF changes needing to be relatively simple and easy to figure out. Thanks for catching this.

  9. Dean SwartzOn May. 14th, 2017 (1 week ago)

    I know you were trying to use lenses designed for the respective cameras, hence the PL 100-400 with the GH5, and the Oly 300mm with the E-M1 Mk2, but compared to those great glass, we’re you saddling the D500 with inferior glass? The Nikkor 80-400 isn’t in the same league with the others. Just wondering if you tried out the Olympus 40-150mm f2.8 Pro with your comparison. As you have written before, the 40-150mm is a fantastic lens. Do you think being able to shoot wide open at f2.8 on the GH5 and E-M1 Mk2 would have made a significant difference? Perhaps not, because even the GH4 had better focus tracking than the E-M1 Mk1. Looks like Lumix still has the lead on Olympus. I was just surprised that the D500 did not do better than it did, but that may have been more the lens than the camera.

    • Portrait of Daniel J. Cox

      Daniel J. CoxOn May. 14th, 2017 (1 week ago)

      I was also surprised the Nikon didn’t do better Dean but I used the 80-400mm for several years and was thrilled with its performance on much older Nikon bodies than the D500. Additionally, I felt it was important to have lenses that were all similar in focal length but also similar aperture range. That was the case except for the Olympus 300mm F/4. No test will ever be perfect but based on these tests and literally 40 years of shooting fast moving subjects, I can tell you, the GH5 is the real deal. On my recent trip to Botswana, I was getting photos of birds in flight, opportunities I’ve tried to get with my Nikons in the past, that 1-100 were ever in focus. With the GH5 those numbers were dramatically higher for similar situations. It’s hard to explain and part of the reason people have to know and trust the folks who are doing these kinds of tests. I’m hopeful I can be that person for my readers.

  10. Dean SwartzOn May. 14th, 2017 (1 week ago)

    You have, once again, provided the photo community with invaluable information with which to make equipment selection choices. I hope Panasonic and Olympus keep pushing each other to perfect their autofocus systems. Let’s hope that the next tweaks by Lumix and Oly get that 3-star rating capture rate closer to 100%. (Seems like Sony’s A9 has a sensor that will move things in that direction sooner than later.)

    What you have accomplished with this latest “research” is to make perfectly clear (tack sharp?) that MFT is a force with which to be reckoned. Way to go, Bro’.

    • Portrait of Daniel J. Cox

      Daniel J. CoxOn May. 14th, 2017 (1 week ago)

      Thanks, Dean. I recently updated this post with a video showing the speed of these cars coming towards the camera. It adds a great deal of additional perspective.

Add your voice to this conversation

Your email address will not be published. Required fields are marked *

In an effort to combat spam, your comment may be held for a brief moderation period.

Natural Exposures, Inc.