That last shot looks very familiar to what my 7DII + 400DOII + 1.4ex will typically do. In strong overhead lighting that is coming from an slight angle (slight back lighting) it is even worse. It is one of the reasons why I want mirrorless to succeed for birding and wildlife. PDAF is simply an indirect focussing concept that is prone to failure in certain lighting and certainly with a lens aperture from f8 to higher, but even f5.6 is already above ideal.
Sensor based AF on the other hand has much
More potential for highly accurate and consistent focussing. After 10 years of PDAF from several brands, I have given up on the concept, and am waiting for mirrorless to grow up...
armd wrote:
Care to elucidate? I found the 7d2 to be unreliable in AI servo and it tracked BIF poorly. Likewise, in single shot if I focused, defocused and refocused on a static subject the resulting AF were wildly inconsistent.
I've tried every case with the 5dmkiv including custom cases recommended by Art Morris and others and have never been able to achieve consistent tracking for BIF with the 5dmkIV with large stable primes including the 500 and 600 f/f lenses. At times it simply fails miserably and it is much worse with a 1.4x iii tc despite calibration. Here is an example of a series of successive shots with a 500 f/4 with a 1.4x iii tc where I was shooting a frog in the shadows (thus the higher ISO) when an egret popped up and lazily flew across my field of view. Straight OOC, default RAW conversion to .jpg no sharpening/adjustments applied other than crop of 25%. Perhaps there were a total of 10 shots in the burst and not one was in focus.
Also, I included another example of the wonky 5dmkiv af of a largely static subject a green heron (cropped 25%) where again despite appropriate calibration, with strong overcast light, the camera consistently FF'ed.
AmbientMike wrote:
I still have to turn off NR in DPP, if I don't want it, even though I try to get rid of the in camera NR as much as possible. Does DPP turn up the NR settings depending on in camera settings?
Yes, if you change the NR settings in-camera (e.g. change High ISO speed NR to High) this is used by DPP for rendering the jpeg, just like (almost?) any other in-camera setting for the image.
I'd have been happy with 10fps, maybe 9, but when you can't get it to fire above 7, then that doesn't work.
--------------------------------------
As mentioned by another poster you must turn off everything. Lens correction is on by default and doesn't help anyone shooting RAW. I was out today with my 90D and believe I am getting 9fps in servo. This camera is great, with real tracking unlike many croppers in the past.
Andrew J wrote:
I was out today with my 90D and believe I am getting 9fps in servo. This camera is great, with real tracking unlike many croppers in the past.
Thanks, ~9 FPS in servo is not too bad. And the tracking sounds great. These are the two key features that would set it apart from my 80D.
lighthound wrote:
Brad, no "aiming" intended towards you or any specific person, just a general observation / thought. And yes, I hear you on the consistent 7 fps findings. This does seem odd to be that far off from the stated frame rates. That coupled with almost spot on frame rates in LV compared to the stated rate makes this even more confusing.
Perhaps I missed this but had you looked at your exif timestamps of a burst to count the fps? If so, did this show the same results?
Time stamps in live action showed it was less. When I tested it with Canon, time stamp indicated 7fps. We checked to make sure everything that should be turned off was and all checked out there (most of it I turn off upon starting the camera the first time).
I guess I got a bad apple. Still, it's a sour taste in my mouth that Canon touts a camera as shooting 10 fps, but then conditions it like they do. Can you image Canon coming out and saying that the 1Dx3 shoots 16fps and then adds in if you are shooting under ISO 400, in jpeg with a shutter speed of 1/2000 or higher when the ambient temperature is 75.3 degrees F and the battery is at 64%? I don't recall ever having a Canon camera where the frame rate they touted as highest was so conditioned like the 90D and EOS R are (the EOS R was reported to have 8 FPS, which is true, if you are shooting in One Shot. Who needs that?). They've always stated their frame rates, which were easily attainable and then gave some conditions when it might not hit that rate. The last 2 seem different.
AmbientMike wrote:
More of an impression, I haven'treally tested it, but it seems like DPP has better NR on 24mp than 18. So it makes sense it may be even better on 32.5mp.
That's probably mostly because the 18MP sensors (7D, 60D etc) have blotchy fixed pattern noise that DPP has problems removing (LR does a much better job). The 24MP sensor generation (80D&co) have vastly improved FPN characteristics so naturally DPP fares better.
bipock wrote:
Time stamps in live action showed it was less. When I tested it with Canon, time stamp indicated 7fps. We checked to make sure everything that should be turned off was and all checked out there (most of it I turn off upon starting the camera the first time).
I guess I got a bad apple. Still, it's a sour taste in my mouth that Canon touts a camera as shooting 10 fps, but then conditions it like they do. Can you image Canon coming out and saying that the 1Dx3 shoots 16fps and then adds in if you are shooting under ISO 400, in jpeg with a shutter speed of 1/2000 or higher when the ambient temperature is 75.3 degrees F and the battery is at 64%? I don't recall ever having a Canon camera where the frame rate they touted as highest was so conditioned like the 90D and EOS R are (the EOS R was reported to have 8 FPS, which is true, if you are shooting in One Shot. Who needs that?). They've always stated their frame rates, which were easily attainable and then gave some conditions when it might not hit that rate. The last 2 seem different....Show more →
C.Fn II - 4
"Release" on far left
C.fn II - 5
AI Seervo 2nd image priority
"speed" on far left
With any other settings on those two you will never get 9 fps.
Andrew J wrote:
C.Fn II - 4
"Release" on far left
C.fn II - 5
AI Seervo 2nd image priority
"speed" on far left
With any other settings on those two you will never get 9 fps.
Yes, I get that and it makes technical sense. However if you think about it logically, what you are telling Camera is take the picture with no regard to Fuchs. Why would you want to do that? Maybe more importantly, why would a camera manufacturer, regardless of brand, want to allow you to do that? That’s what leads to “my camera has a focusing issue” statements when really it’s not true - your camera has a problem tracking a moving object while keeping a higher FPS.
Just my opinion though on the choice of wording from Canon.
Well, you may consider timing more important than exact focus, in that case you might use these settings. It doesn't mean the camera won't focus between frames, but rather that it will prioritize getting the shots at the specified frame rate. If you stop down the lens to get some depth of field, you may prefer the outcome (in some situations).
AmbientMike wrote:
I still have to turn off NR in DPP, if I don't want it, even though I try to get rid of the in camera NR as much as possible. Does DPP turn up the NR settings depending on in camera settings?
DPP doesn't do anything to the slider values, they are either determined by the camera right out of the gate, or you can override those settings by telling DPP to ignore those settings, or always override them with your values.
My workflow was always to use standard in camera, then take the lumeniscent slider down about 3 or 4 notches, make sure white balance is corrected, reduce sharpening, bring up shadows a bit if needed, lower highlights a bit if needed, and export.
Then I would run a NR step very mild on the overall image, then manage each of the 3 color channels independently, with some final RGB channel filters to sharpen and add contrast. That is the basic recipe, but each model needed more or less of these steps with different settings. That is where I spent my time up front when I bought a new model, and then I could reap those benefits from that point forward.
armd wrote:
Care to elucidate? I found the 7d2 to be unreliable in AI servo and it tracked BIF poorly. Likewise, in single shot if I focused, defocused and refocused on a static subject the resulting AF were wildly inconsistent.
I've tried every case with the 5dmkiv including custom cases recommended by Art Morris and others and have never been able to achieve consistent tracking for BIF with the 5dmkIV with large stable primes including the 500 and 600 f/f lenses. At times it simply fails miserably and it is much worse with a 1.4x iii tc despite calibration. Here is an example of a series of successive shots with a 500 f/4 with a 1.4x iii tc where I was shooting a frog in the shadows (thus the higher ISO) when an egret popped up and lazily flew across my field of view. Straight OOC, default RAW conversion to .jpg no sharpening/adjustments applied other than crop of 25%. Perhaps there were a total of 10 shots in the burst and not one was in focus.
Also, I included another example of the wonky 5dmkiv af of a largely static subject a green heron (cropped 25%) where again despite appropriate calibration, with strong overcast light, the camera consistently FF'ed.
If you ran a series of AFMA steps to really nail down AF accuracy at the typical distances you would shoot the lens (or lens and TC combo, two different sets of calibrations), and then set up your cases so that it doesn't act skittish with AF changes if objects get in the way or objects change, and set your focus priority to focus and not burst, then these results have to be an issue with the camera.
Some copies do indeed have issues. Many that have these issues are pretty vocal as well as frustration would lead them to research and find out how to fix this. Completely understandable but adds to the implication that the entire model line is inaccurate. However there are others with no such issues, or lesser misses anyways. The AF system being a separate system to the sensor just lends itself to these inaccuracies by copy I feel, but a sensor that records the image that it itself is responsible for focusing on should always be accurate without these misses.
My point is that I spent time fiddling with settings, AF mode, priority, and shutter speeds to get NBA games, and other sports such that focus rarely was off the subjects I was tracking even with others getting in the way, with the copies I had. My keeper rate was probably around 85-90% for various seasons, I rarely had to go to the raw to correct for exposure (using auto ISO in manual) and rarely threw away images due to missed focus, but most tossed were simply due to this person's inability to make each capture one that was interesting.
I don't do a ton of BIF, as most of the birds in my area aren't that exciting, we have some osprey and a ton of vultures, all pretty boring. However we did have some starlings at one point, and here is a sample of my setting up the 5D4 to track and not come off, as the bird flew past a maple tree that lay between me and the bird. Terrible picture, I only keep it to show tracking results.
We do have some gulls, but they are pretty slow and easy to capture, hardly an issue for AF. This was with the 7D2.
These look very nice!
Not as superb as many of your shots with the 1DXII, which really stand out i.m.o.
The cardinal shot at iso 1250 looks very good. What are your thoughts on this new 32,5mp sensor?
Chris
RobAmy wrote:
Here is a couple with the 90D with the 400mm f2.8 Mk III and some with the 1.4x. Fun little camera, it is no 1dx by no means but works fairly well.
Chris you hit on the head, the files are not as good as the 1dxII. The files get noisy real quick. I like the 32.5 sensor but I feel to me cropping and upsizing my 1dxII, the file looks better. My examples with the 90D were almost full frame and the squirrel was. I really have not played with it much, just yesterday afternoon a bit. The AF was not consistent with the 1.4x on it, the strait lens did much better. With my 1dxII, I do not even notice the 1.4x on it. However with that said, at this price point, I think it works well. The flip screen is nice and very responsive. I never used a 7D II so I can not compare. I need to tweak it a bit more and play with it a bit more. At this quick point, it will make a nice lite back up camera but I am spoiled with the 1d series. One thing I really do not like is the main wheel, the outer wheel has ridges that actually hurt your thumb after a bit of use. I always like the 1.3x crop bodies the best, I think they are the sweet spot between the 1.6 crop and full frame.
ChrisMak wrote:
These look very nice!
Not as superb as many of your shots with the 1DXII, which really stand out i.m.o.
The cardinal shot at iso 1250 looks very good. What are your thoughts on this new 32,5mp sensor?
lighthound wrote:
Nice shots Rob. I heard you on the AF slowing down with the 1.4x, but what are your thoughts on BIF AF lock and tracking straight from lens?
I will test it but not sure when, hopefully this weekend. That job thing gets in the way
RobAmy wrote:
Chris you hit on the head, the files are not as good as the 1dxII. The files get noisy real quick. I like the 32.5 sensor but I feel to me cropping and upsizing my 1dxII, the file looks better.
This is how I felt all the time comparing my 1DX and 1DX2 files to the 7D2 and 80D files.
arbitrage wrote:
This is how I felt all the time comparing my 1DX and 1DX2 files to the 7D2 and 80D files.
The 1DX2 seems to have a roughly one third to half a stop high ISO lead over the 80D on top of the 1⅓ stop advantage admitted by the 2.5x sensor area difference. But the 80D, and especially the 90D, should still resolve more detail even with stronger noise reduction applied.
It was to be expected of course. Pixel density on the 7DII that I currently use is already quite high. There is no real substitution for a longer c.q. brighter lens, coupled with a larger sensor
I asked because I was asking myself if I should add the 90D to the 7DII, but have doubts about that really. Besides, despite all the bad that I read about the 7DII before getting into the Canon system, my experiences with the 7DII (with 400DOII + 1.4TC) are by large very positive. Perhaps I have a good one, I don't know. My next stop, a fair bit into the distance, will likely be the 500/4.
Chris
RobAmy wrote:
Chris you hit on the head, the files are not as good as the 1dxII. The files get noisy real quick. I like the 32.5 sensor but I feel to me cropping and upsizing my 1dxII, the file looks better. My examples with the 90D were almost full frame and the squirrel was. I really have not played with it much, just yesterday afternoon a bit. The AF was not consistent with the 1.4x on it, the strait lens did much better. With my 1dxII, I do not even notice the 1.4x on it. However with that said, at this price point, I think it works well. The flip screen is nice and very responsive. I never used a 7D II so I can not compare. I need to tweak it a bit more and play with it a bit more. At this quick point, it will make a nice lite back up camera but I am spoiled with the 1d series. One thing I really do not like is the main wheel, the outer wheel has ridges that actually hurt your thumb after a bit of use. I always like the 1.3x crop bodies the best, I think they are the sweet spot between the 1.6 crop and full frame.
