Fujifilm X-H2S Bird Subject Tracking Auto Focus Test w/XF 150-600mm

[@PhillipRPeck]

I am on firmware ver. 7

[@TheJ_G]

I really enjoy this approach to exploring the current state of fujifilm. It’s much more balanced, informative, and productive than the “pile-on” so many other videos have become from other people attempting to cover the same topics. Thanks for doing all this work and helping the conversation around it stay objective.

[@PhillipRPeck]

Thank you!

[@stanislavdrozdov8636]

X-H2s. Firmware 7. Dogs "hunting". Moving towards to me. Custom setting in C mode (first option - max, second - middle, third - auto). About 70% out of focus :(

[@JakeChambersOy]

Everything moving towards the camera is the main issue I have with AF-C on my X-T5.

[@PhillipRPeck]

Okay, I'll try to to another test of subjects moving toward the camera

[@Powerland56]

Thats too much. Sure, that you have Boost ON, AF-C in focuspriority and AF+MF not ON?

[@JakeChambersOy]

​​​@@Powerland56 and there he is back again. When will you finally acknowledge that different people shoot different things, and therefore have different hitrates?! You are probably the biggest troll in Fujifilm history without even knowing it, always assuming others don't know how to setup their cameras when they criticize the AF performance. Newsflash: my hitrate is 99% / 70% / 20-40%, depending on the subject and situation, with the settings you mentioned.

[@vinceklein4168]

@@PhillipRPeck Might as well do it now with the new 7.1 firmware :) Also, did you use focus priority with your BIF test? I did not see that setting in your video. Thanks man!

[@amermeleitor]

I think a critical comparison of portrait photo could be great. But with stricter criteria, like very fast lens, no "near in focus" allowed. Only the really eye in focus counted. That's would be great. With models moving, walking, changing position slightly, etc. For me it's the major concern when I think about going with Fuji. I still have Nikon and Canon DSLRs and want to go mirrorless

[@SummersSnaps]

I'll weigh in with my 2 cents. I use Face/Eye detection almost exclusively for my event work, rarely needing to toggle out due to some confusion/failing issue (this with the XH2S's on 5.03). I have noticed some interesting things; 1) 8fps is quite a lot more accurate than 10fps or more. 8fps gives you blackouts (annoying) but I think it is much better. I believe this was first discovered by Andrea Jump Cimini, claiming 5-8fps was giving him 80-95% but 10-15fps can be as low as 30-50%. Tbh this is in line with my experience, I believe the additional fps is just too much for the AF.C system, the difference between Focus Priority and Release begins to blur and Focus Priority behaves much more like Release when the fps get pushed hard (and yes this is without having AF+MF set to Off). But then how do you get 80% in such high fps? Well there are MANY factors that contribute here such as light (AF works better in higher contrast well let scenes), how much the subject was moving and how picky you are to calling what is in focus and what is not. I tend to rank AF on a scale, C miss focus, B in focus but could be sharper A perfect focus etc. So its very hard to come up with a definitive percentage at times. Also I think animal eye (bird) vs human eye is more complex, we have eye lashes and detailed pupils iris, sometimes the DoF is so narrow that we get sharp lashes but not eye or vice versa, it is not always easy to understand what you might want in that moment, more on the eye ball or more on the lash. For some animals its just a black dot that it needs to get right. 2) 1.03 is better, but it can't be tamed so easily, moving the Wide/Tracking AF box over the person you want and it will still be happy to choose someone else in the frame. It's good for solo work, but multiple people you lose control over the Face/Eye Detection tool. 3) When I do my bursting, I see a pattern; in focus> slightly losing focus> losing focus more> lost focus> snaps back to being in focus>repeat. When you do higher fps you collect more of those losing focus shots, whereas 5 or 8fps you get the 'required proper focus' cycle moments more. Generally this is a huge topic, and the point really is how many shots you need to take to get a perfect in focus shot. For some Sony users they might walk away with fewer shots but all in focus, whereas we tend to have to cull more on Fuji, but we'll get critically in focus shots. I'd like to see the CL burst rate become blackout free, especially if it is producing a higher hit rate. I'd like the nervous pulsing once its landed on an eye to stop (you can see and hear it with some lenses like the 56/1.2RWR even when the camera is tripoded and the subject (mannequin) stationary, why it continues to cycle AF slightly when nothing is moving is bizarre... LM needs to be a lot smoother, sometimes it steps as much as any other lens like stepping, DC or even Fringer III+adapted glass. I feel like LM is not nearly doing its potential benefits it should be. I detail a fair amount of this in one of my videos 'The Truth About Fujifilm Autofocus'.

[@PhillipRPeck]

Thank you for the detailed reply. Many people are mentioning the pulsing forward and back, then locking on. Fujifilm definitely appears to be relying on contrast detect points to a much larger degree than other brands (at this point). And as you said, it keeps doing it on a static subject resulting in images that are front or back focused when nothing has moved. I probably could have done a more detailed analysis and ranking/tier system of focus but at this point, I wonder if I've become used to the slightly soft focus that I think I've been getting with this combo ... I really want to try out another camera and lens to compare

[@amermeleitor]

👏👏👏👏👏

[@SummersSnaps]

@@PhillipRPeck I've been through a fair amount of lenses for the X Mount system now, and it feels the same regardless of AF drive motor. XF50 (stepping), XF56RWR (direct current), XF90 (linear), XF50-140, (linear) and FringerIII, it seems to be the same with all of them, the only difference is it is harder to tell what is going on with the LM because the feedback (vibrations) and noise is much quieter (same with small stepping motors like the XF50), whereas 56RWR, FringerIII etc, its much more audible and vibrating feedback. But when you pull in LM shots into LR and study a burst, you can see the exact same pattern of pulsing. It feels like a trick somewhat, that we are fooled into thinking things are more locked on and less panicky, when in fact its still micro adjusting constantly. That's how it appears to me anyway. It could be the degree of pulsing is less with LM, but it feels like the AF system is too heavily reliant on CD, some PD would help to tell it that its got it focused as best as it can so stop hunting.

[@muppets1958]

Thanks for the great and informative video. Did you turn on image stabilization in your photos?

[@PhillipRPeck]

Thanks! Yes I did

[@rafaeluemura9430]

really appreciate your efforts on this one, looking forward for your results with the newly announced 500mm

[@PhillipRPeck]

Thanks! You sending me one? 😅

[@cameraman655]

I didn't see anything hateful about this video, just an honest assessment, kudos. BTW, I have the exact same setup as you ‘T-5’,150-600, and photographing birds, specifically Hummingbirds and I had a decent hit rate as well. Now, the birds that I typically shoot have metal wings and afterburners, so one would think that a larger target rolling down a runway, would be a cakewalk…right? Nope, my hit rate on an F-16 and other aircraft the following week was notably lower than that Hummingbird…go fig. At the end of the day, Fuji really needs to address this, hopefully, if the rumors are true a fix might be released in a few weeks, we shall see. Great video and useful, I like your sensible approach and style. Keep up the good work.

[@PhillipRPeck]

Thank you! Strange about the aircraft. I haven't tried anything like that

[@pennysteam]

Video depends on which lens, I have found no issue using the 15-45 and the 18-120 lens videoing Trains mainly but also other vintage transport. Not only that but I also zoom in and out while recoding and maintained focus, which I have only seen done with a true camcorder. Both lenses us electric zoom instead of manual, which I suspect is the key to maintaining focus while using the zoom, as the came knows what is happing while a manual zoom it has to work it put the hard way. My fuji camera is the XS-20, which I have been using to video with for about a year now. I have had issues with manual zoom lenses such as the 16-80, which was ok until Pulled out, used the zoom, while filming and it lost focus for a short while. By the way, I found the train detection of limited use as it does not recognise steam trains, which I mainly shoot.

[@benoittissier58]

Thanks for your positive contribution. I'm an outsider since I shoot Sony but I had lots of trouble with the A7iv for sports. Long story short, I ended up using 6 FPS and fixed AF spot only. I'd suggest runners coming straight towards the camera on an overcast day as the ultimate AF test.

[@PhillipRPeck]

That's surprising to hear. Thanks for the feedback

[@DFitzell]

I’ve had mixed results with birds, mainly shooting birds in flight so will always be harder to track like that no matter what system. I thought the bird subject detection worked well at times but there was also times where it just gave me the green box but was no where near in focus. I get more consistant results with zone focusing so I mainly use that for now. I have an XT5 with the 100-400 but have used the xh2s and 150-600. I preferred the burst rate on the xh2s but I was able to crop in more on the Xt5 images which I find is more useful.

[@PhillipRPeck]

Yeah, I had a couple stacks of photos that were never in focus even though I got green boxes around eyes. I will probably try turning off Bird tracking and see how it does. The detail in the X-T5 RAWs is indeed quite nice

[@benoittissier58]

Damien Bernal compared autofocus systems in a recent video. He pointed out the false positives from Fuji as a serious issue.

[@DFitzell]

@@benoittissier58 definitly an issue with that. I found that if you use back button focus, subject detection and the shutter set to focus priority but you half press the shutter the cameras will pick up the subject and turn the box green and take a picture when you press the shutter fully, all without actually focusing on it with the lens. That shows me there’s a clear missing link between the focus system and what the lens is actually doing.

[@dannybaresi]

If there are micro adjustments within the plane of focus, would it not be better to shoot at a much lower frame rate, to allow the computational analysis made by the camera to keep pace? Do these camera's have a focus bracketing feature? If missed focus can be a hair shorter or longer, I would have thought this would be much preferred to huge bust rates.

[@PhillipRPeck]

You may be right. Another commenter has similar advice (much slower bursts). My focus (heh) here wasn't trying to put the camera in the best scenario to succeed, but using it in a reasonable manner as it's intended (fast action, fast burst, fast focus) and see how it did. It would be interesting to try some of the methods you mention

[@dannybaresi]

@@PhillipRPeck I dont really know what auto focus is, I'm just imagining if it was a human brain

[@aj-zk1wx]

Great video! I bought this combo a year and a half ago and I want to love it, but I'm on the verge of picking up an OM-1 II (would love a Z8 but it's hard to justify the price and I like having a lightweight kit). I can deal with the autofocus for birds that are stationary or moving around in trees, but the camera's autofocus just can't handle fast action - birds in flight, herons fishing, my dog running around the backyard, etc. it just can't keep up. Shooting 40fps or 30fps and having precapture sounds awesome, but what's the point if the autofocus can't keep up? Sure, I still get some good photos in these scenarios, but I'm usually settling for any in-focus photos I get instead of being able to choose the ones I want. And I never get a full sequence of shots where the subject is tracked accurately throughout with most of them being in focus.

[@evtimstefanov8377]

I was having a similar if not higher hit rate years ago with with my old Pentax K70 DSLR back then.

[@CrnaInBela]

much appreciated, it's nice to see this kind of videos about fujifilm.

[@grahamniven]

Thanks for this valuable contribution to the Fuji AF debate. Love the Vespa!

[@matusmisik_photography]

Well done, finally someone shows wildlife photography focus with fujifilm system .

[@PhillipRPeck]

There used to be more YTers who shoot wildlife w/ Fuji, but unfortunately some switched systems 🫠 Wild & Free Victoria is still going strong

[@matusmisik_photography]

@PhillipRPeck yes I know. Thank you.I am waiting for a new firmware update in November .So after firmware update, if AF will be bad, I will sell wildlife gear (XH2S & 150-600 mm + 100&400) and jump on the nikon system.

[@matusmisik_photography]

I think lots of people will make the same decision .Fujifilm is good for landscape photography, so 100 per cent I will keep Xh2 and lenses for landscape.

[@matusmisik_photography]

@PhillipRPeck try make video for wildlife video with your setup you will see it is horrible. I need to use manual focus is well.

[@PhillipRPeck]

@@matusmisik_photography I've shot some wildlife video with X-H2S and 150-600 and think it's pretty good. The biggest problem for me is the IBIS/OIS stickiness when trying to pan the camera

[@NickErickson]

I’d say 80+% is pretty good, especially at 20-30 FPS. Like you said, you’d get plenty of keepers from that hit rate. Other systems might be more, but what does that really get you?

[@cy9nvs]

The thing is, those aren't always evenly spaced, there are times when you'll absolutely miss the shot, because the camera just won't properly focus. And I also doubt that those 83% were actually perfectly in focus, at least my hit rate with the X-H2 (non-S) was way lower than that. Most shots are slightly soft, with a couple of tack sharp ones in between, as if the focus is constantly pulsing.

[@NickErickson]

@@cy9nvs You make a very good point that I didn’t consider! Great observation

[@JakeChambersOy]

During your testing, did you ever had a bird fly towards you on your photos? Because that's the main problem with AF-C in my opinion, moving subjects which move towards or away from the camera. On events, my hitrate is pretty high, but once people start moving towards me it can go down to 30-50%.

[@PhillipRPeck]

I only got the one sequence of the eagle flying toward me, but it was pretty far away. In that seqeunce, I counted 55/62 photos in focus, but again, the bird was far away so it's hard to judge details. I'll maybe do a short update if I get a better sequence to show.

[@Unrealsensor]

i'm loving your videos on fuji AF, this is really quality content there are some users suggesting that fujifilm is using contrast AF instead of Phase detect, could you do a test compared to sony ZV-E1 with high contrast vs low contrast to test them.

[@PhillipRPeck]

I think it's a hybrid phase/contrast system, which is pretty typical. But it definitely looks like Fuji is relying on contrast points much more and more visibly (obvious to see in video) than Sony and the other brands. A low contrast scene would be worth trying out. I'll see. Thanks for the kind words

[@beepneldo]

Wonder what the hitrate would be for continuous shooting with you walking towards the camera in a straight line at 33mm f1.4 or similar. In the start position you fill the hole frame with your full body, in the end position you fill the frame from head to chest. Normal walking speed. Would be awesome to have a reference for that scenario. And then repeat it with the announced November update.

[@INFOXlive]

This kind of exam is demonstrated in a number of videos. In my opinion, though, that is not the most effective method for determining a camera's usefulness. My recommendation is to put it through its paces in a real-world setting, such as a large gathering of people who are constantly on the go and turning heads. Testing can be done even on crowded streets. Your camera can do better on these static tests than it would in real life. By taking part in these tests, you can gauge how fast your camera becomes unusable or how reliable it actually is. For event video and photography, I must say that the a6400 does a respectable job. However, the a6700 is superior because it has an AI chip and IBIS and much better sensor. For APS-C, there aren't a tonne of choices. In addition to their other limitations, Canon does not offer a wide selection of lenses. Your final decision will be between a Sony or Fujifilm camera. Unless your project specifically calls for full-frame capabilities, APS-C is the way to go from an economic and ergonomic perspective. The good news is that Sony has introduced a variety of APS-C cameras to meet a variety of demands. At this time, I believe Sony is the better investment option. After a few years of use, your lenses will still be functional, and I have no doubt that future cameras of Sony will include even better autofocus. Consequently, Fujifilm will fall behind Sony for the time being. I really hope Fujifilm improved their cameras, but I know that hoping won't help.

[@dannybaresi]

Can we get some kind of end-user guide on what super tele work is possible on any system handheld. Half frame 400 being 800 compared to r5 12 bit 2x crop or/and everything in between? I've stuck with pocket 4k and thought 400mm 6.3 is about as good as....but i'd be interested to know if there are any other config's that might be good for the birdies and aeroplsanes? Nice one.

[@PhillipRPeck]

Camera Conspiracies has a pretty extensive library of videos using various systems for urban bird and wildlife videography. I don't know if he's organized them into a playlist though

[@trym2121]

I think Fuji should simplify the AF, if they already have eyeAF for many subjects, the AF behaviour menu should be simplified to just track speed and sensitivity, not many options like now

[@cy9nvs]

What I noticed, when I had my X-H2 + XF 100 400, is that the focus always seems to move back and forth slightly, even while standing still and shooting static subjects. I didn't notice it in the viewfinder, but when reviewing the images, they're always every so slightly soft, then you get 1-3 that are tack sharp, and they start to get slightly soft again. I don't know how you rated the images, but I'd be surprised if those 83% were actually perfectly in focus, my hit rate was way lower than that, and I don't even wanna talk about moving subjects.

[@PhillipRPeck]

I think it's the contrast detection pulsing front to back. I tried to only include images I thought were tack sharp so I definitely know what you're saying with the slightly soft images

[@jan-nilss.680]

Really enjoying your videos, I don't use Fuji but always thinking about getting one.

[@PhillipRPeck]

Thank you. I appreciate the kind words. What camera/s do you use?

[@jan-nilss.680]

@@PhillipRPeck I used the A6400 for 5 years and got the A6700 a few weeks ago, but the X-T5 looks so compelling to me.

[@PhillipRPeck]

@@jan-nilss.680 It's pretty sweet, but the a6700 is really great, too

[@lionheart4424]

Bro I beg of you... PLEASE do this test but on the X-T5 if you own one. NO ONE has done the test specifically for birds with the XF 150-600MM on the X-T5 and I would love to have a benchmark before the rumored firmware comes. Just today I ordered a second hand XF 150-600MM specifically for bird photography. I do not own an X-H2s and I know the X-T5 is not a camera for shooting action but I still would love for someone to help gauge the performance for bird pbotography. I have done bird photography with past firmwares and I know false positives are still there. Great job and thanks for sharing!

[@PhillipRPeck]

I'll do it. I might not take as many photos but ... I'll do my best

[@lionheart4424]

@@PhillipRPeck yeah no need for that many photos I hope 😅 looking forward to the video

[@yourtallness]

My experience with the X-T5 and the 150-600mm at an airshow was very disappointing.

[@PhillipRPeck]

That's interesting. I haven't tried subjects like that. Did you use the airplane tracking or did you use box or zone?

[@yourtallness]

@@PhillipRPeck used airplane tracking. The keeper rate was very low and I had to sharpen with Topaz in order to get a passable end result. Also considerable asymmetrical vignetting on my 150-600mm.

[@Powerland56]

Thank you for this information. My experiences are similar. About 80-85% keeper for BiF. Thats ok, the 95-99% of a Sony A1 etc. is better, but this rate is absolute solid.

[@INFOXlive]

Thank you for effort! I considered Fujifilm for events (video & photo), but Sony is far better for humans. I suppose that Sony and Canon are the best for nature. You really need sharp eyes for events and cannot afford missed shots. Never tested birds photography, as not my cup of tea. But I don’t think that ~80% hit rate is top notch category. For me top notch is at least 95%. Please note that commonly accepted margin of error is ±3%. Obviously there are certain areas where you can survive with Fujifilm cameras, for example landscape and architecture. It’s really complicated to switch entire ecosystem if you have already heavily invested into. Camera is the cheapest element compared to lenses. That’s why really honest camera reviews are so important. Sadly there are too much marketing bs and blind hype by fun boys. Thank you for your effort.

[@stuartsilverman3797]

Very well done! Woukd you guess that the xt5 would do as well as 83%?

[@PhillipRPeck]

I don't know. I guess I could find out 🤔

[@jacquesbrunelle9352]

​@PhillipRPeck my guess: +- 50%

[@PhillipRPeck]

@@jacquesbrunelle9352 Wait, you're saying it could be 50% better or 50% worse?

[@jacquesbrunelle9352]

​@PhillipRPeck 50% on focus. I think speed is a factor in that equation.

[@stuartsilverman3797]

@@PhillipRPeck your guess. Of 50% is good enough for me. Thanks for the speedy response

[@chunhangtang]

Great video and great effort.

[@PhillipRPeck]

Many thanks!

[@markwiemels]

Hit rate seems reasonable enough.

[@benoittissier58]

I think we are very tolerant about AF because it's new technology and we compare it to manual focus. But if a car started 83 % of times after stopping at a red light, would it be considered as a reasonable performer ?

[@markwiemels]

@@benoittissier58 haha... But it's a wildlife photo, a very different task than a car starting at a red light. I don't think they are comparable.

[@markwiemels]

@@djstuc he's alright, just offering his thoughts. I mean, in time, maybe Autofocus will be as reliable as a car starting. Who knows, I think that may have been his point, not so much an attack on my comment.

[@grahamniven]

For stills, yes it seems ok. But... for someone like me looking at the possibility of upgrading from an X-T3 to an X-H2S, the achieved hit rate here doesn't seem better than I'm getting with the X-T3 As I see things video is a different situation, you can't always edit out the bits with missed focus and that's what's really giving me pause. What I'm looking for is a body which will reliably do slow-mo video and stills of wildlife with long lenses hand-held. (XF150-600 OIS is really good) So ideally crop sensor with large viewfinder, good stability and reliable auto-focus. And that isn't any of Fujifilm's current cameras, so where next.....? Rather than blunder around in a system swap minefield I'm sticking with the X-T3.

[@PhillipRPeck]

Panasonic G9ii

[@oliverlentzen3371]

Doesn‘t work satisfactorily on my XH2S with the 50-140mm. I earn my money with dogs photography. Doesn’t work here. I need an reliable animal eye detection! And this doesn’t work on my Fuji XH2s. And even when the eye detection box hits an eye, the focus! is too often not the eye. This is for me the main problem. I am in direct contact with Fuji Film Germany, to whom I have sent a lot of examples of misfocusing and incorrect auto-focus boxes. I completely renewed and adjusted my camera settings with a Fujifilm employee to fit my type of photography. I need a wide aperture for subject isolation. With fast movements, the camera fails 60% of the time.

[@PhillipRPeck]

Completely understand and I find this to be very similar to the video autofocus performance which I just made a video on. It's not good.

[@benoittissier58]

I'm an outsider here since I shoot Sony, but I'd like to know if you can see the real focus area used by the camera. It is possible with Sony’s proprietary software and with A7iv pics the results were ...surprising - very different from what was shown in the EVF - and that explained a lot of misfocused shots.

[@josseR_74]

which firmware are you using now? was it better with the previous firmwares?

[@PhillipRPeck]

@@benoittissier58 You mean while reviewing the shots in the camera? Yes, you can

[@benoittissier58]

@@PhillipRPeck I mean reviewing the shots on the computer with an indication of the focus point. Sony Imaging Edge shows the focus box that was actually used.

[@Rich-y8n]

Good test. Kudos. 8 out of 10 is not bad for non-professional use. But not great or exceptional if you are a professional and your making a living depends on your hit rate. I imagine that Canon or Sony would be better. But hey, there are other reasons for owning a Fuji. Right?

[@PhillipRPeck]

I think there are other reasons for owning a Fuji--handling, colors, good lenses ...