The Importance of Mirror Lockup - Seeing is Believing! [Archive] - BirdPhotographers.net

View Full Version : The Importance of Mirror Lockup - Seeing is Believing!

Jay Gould

07-21-2009, 03:45 PM

Bouncy, bouncy, bouncy - shows why you should use mirror lockup for at least all landscapes and perhaps most macros too.

<link rel="File-List" href="file:///C:%5CUsers%5CJay%5CAppData%5CLocal%5CTemp%5Cmsohtm lclip1%5C01%5Cclip_filelist.xml"><link rel="themeData" href="file:///C:%5CUsers%5CJay%5CAppData%5CLocal%5CTemp%5Cmsohtm lclip1%5C01%5Cclip_themedata.thmx"><link rel="colorSchemeMapping" href="file:///C:%5CUsers%5CJay%5CAppData%5CLocal%5CTemp%5Cmsohtm lclip1%5C01%5Cclip_colorschememapping.xml"><style>  </style> http://www.smugmug.com/gallery/8945351_aVDzo#594124511_2CUHh-A-LB<o></o>

Doug Brown

07-21-2009, 07:29 PM

Pretty cool Jay! If only I could use mirror lockup for flight! :D

Eric Virkler

07-21-2009, 07:33 PM

If you want to see the affects that the shutter slapping have on your images, just set your camera on a tripod, turn on LiveView (If you're fortunate enough to have this feature) zoom in to 10x, and press the shutter.

That'll cure you from not using it when the situation allows it.

Eric Virkler
Faces of Nature Photography
www.ericjvirkler.com (http://www.ericjvirkler.com)

Markus Jais

07-22-2009, 01:34 AM

Great video.
In case anyone is interested I made a test some time ago and wrote about it on my website about the importance of mirror lockup including sample images:

http://markusjaisphoto.com/cameras/why-mirror-lock-up-is-an-important-feature-of-your-digital-srl-camera/

Markus

Ed Cordes

07-23-2009, 10:04 PM

Great info. Thanks to Markus for the tutorial. Now if Canon can just let us program the ever useful direct print button :confused: to enable mirror lock up we sill all be happy.

Markus Jais

07-24-2009, 01:19 AM

Great info. Thanks to Markus for the tutorial. Now if Canon can just let us program the ever useful direct print button :confused: to enable mirror lock up we sill all be happy.

That would be a great idea. With the new "My Menu" stuff in the latest cameras (like my 40D) you can put the MLU there but this is still no fun. I use MLU very often for close-up work and sometimes I forget to switch it off and when I am photographing birds and something exciting happens, it sometimes just takes too long to switch it off.
I cannot understand why Canon does not make this simpler even many, many photographers have been asking for this for years.

Markus

Ed Cordes

07-24-2009, 07:46 AM

Ha,Ha Markus. I do the same thing. After a session with MLU I often get frustrated when I press the shutter release and wonder for a second or two why the camera isn't working only to remember the MLU is enabled!! :o

I am glad I am not the only one!!

John Chardine

07-24-2009, 08:04 AM

SLRs carry a huge "overhead" for the ability of the photographer to see through the lens before an image is made. If only there was another way of viewing a quality image through a viewfinder of exactly what the lens is seeing, then the mirror, fresnel screen, and pentaprism could be dispensed with in professional cameras and mirror-slap would be a thing of the past and noise would be reduced. Also the upper-limit to FPS which I understand now is governed by how fast you can flip the mirror up and down in an SLR, would increase. Current technologies seen in P&S cameras and rangefinders don't cut it for rapid-pace photography like sports and wildlife- hence the popularity of SLRs. I read somewhere that the big camera companies are working on a new professional-level body design that was not an SLR. I think it involves a new type of HQ LCD as the viewing device. Anyone know any more about this?

Ed Cordes

07-24-2009, 08:15 AM

John, I have not heard about the HQ LCD research, but it sounds like a good, but expensive, idea. A while ago the canon rumors site had a discussion about the possibility of a pellicle mirror (semi transparent that would not have to move) in the upcoming 1 series. However remember it was on the "rumor" site.

Jason Franke

08-02-2009, 10:05 PM

Wow, this ended up a little longer than I wanted it to, but it's typed now......

Nifty videos.

----

John,

If there's will, there's a way around just about everything with some form of compromise. However, I don't think I care for an EVF camera for a number of reasons. The short of it is that I have many concerns over relying on electronics and power to do everything including letting me look though it. I have no idea if the video system can be made low enough latency to not have any lag. Then there's AF performance.

Right now the mirror feeds light to the phase detection AF system. Remove it and we're stuck with contrast detection of some form. Of course, if you could keep the mirror for the AF system but at that point the pentaprism isn't much of a reach either, you're alreayd half way there.

The other problem with FPS isn't moving the mirror but moving the data. Not only do you have to move it from the sensor (very quickly actually) to the buffer, but you have to be able to process it and get it on a card.

Sorry this is one of those sore spots that get me going. Every time someone brings up pro level EVF cameras, and I raise these kinds of questions I get a lot of hand-waving and "technology will magically make it work better than anything you can imagine" and no real answers or theories.

Granted, I think that's the way things are going to go eventually and I'll grudgingly be dragged along kicking and screaming the whole way. But ugh, just ugh.

----

Ed,

God I hope they don't do a EOS 1N RS pellicle mirror again. It was a novel solution to making a 10 FPS camera at the time it was developed, but it's clear we can do that now with a movable mirror. However the big downside is the 2/3 stop hit in film speed and that means more noise. In an RS style camera, setting the camera to ISO 800 would have the effect of setting the sensor to the same gain as ISO 1200<sup>1</sup> in a camera with a movable mirror. That might not make much of a difference at ISO 200, but at ISO 6400 you're way out in noise land. Even if the hypothetical sensor had a stop less noise compared to the 1D. Would you rather have the whole stop or 1/3rd of it?

Oh yeah, the EOS 1N RS still had a secondary half mirror that moved that drove the AF system. Vibration, while smaller, doesn't completely go away anyway.

-----

1) Assuming the 63/37 split the EOS 1N RS used.

John Chardine

08-03-2009, 03:12 PM

That's a very good point about AF Jason but I wonder how much the modern AF system was built around the fact that the mirror was there already rather than it being an optimal engineering solution?

Not that I doubt it (well I do actually!), are there hard data showing that electronic throughput from sensor to memory is the bottleneck in FPS rather than the physical constraints of flapping a big mirror up and down? I should have thought with Canon's Digic 4 processor (and whatever other companies offer), and fast memory up to UDMA, that data throughput is pretty fast.

Michael Pancier

08-03-2009, 03:42 PM

FWIW, by accident I discovered that if you use the Live View function (on my 5d Mark II and 50d), and you set the image for auto-bracket (which I do for HDR) .... keeps the mirror up the entire time....

the a/f is slow though.

Jason Franke

08-03-2009, 07:34 PM

John,

The alternative is already there, it's an external coupled rangefinder a la the Leica M.

To do through the lens focusing, AFAIK, you have two choices; you either have a beam splitter (mirror or prism) so you can put a specialized sensor in the optical path with out putting it in the middle of the imaging sensor. Or you need to do the AF with the imaging sensor.

If the sensor is film, the choice is made for you. However, the phase detection system that evolved out of film has some nice properties that I don't think can be replicated in a contrast based system. The AF system knows immediately which way to focus most of the time (unless the image is severely defocused) and computationally it's much simpler. The other advantage is that the AF sensor can be tuned to be very sensitive relative to the imaging sensor.

I think we would have ended up with a beam splitter and a phase detection system any way you cut it. The alternatives are simply worse one way or another.

Not that I doubt it (well I do actually!), are there hard data showing that electronic throughput from sensor to memory is the bottleneck in FPS rather than the physical constraints of flapping a big mirror up and down? I should have thought with Canon's Digic 4 processor (and whatever other companies offer), and fast memory up to UDMA, that data throughput is pretty fast.The best I can do is point to the D3. In FX mode it tops out at 9 FPS, in DX mode that goes up to 11. Granted this isn't perfect as the D3 clearly has other limiting factors that cause losses in functionality (AF and sometimes AE) at 10 and 11 FPS. However the most significant change in the is the data throughput<sup>1</sup><sup></sup>(18.8-24.7MB/frame v. 8.1-10.7MB/frame). If the mirror was the limiting factor in the D3, it would seem reasonable to expect the FX and DX frame rates to be the same.

A bit more of a reach is the 1D and 1Ds, as far as I understand it they are completely identical except for the sensor. The 1D can do 10 FPS (179MB/s from the sensor, ~130MB/s of data to the card) the 1Ds can only do 5 (183MB/s from the sensor, ~125MB/s to the card).

Incidentally the 1D-III/1Ds-III are on the low end of things now as well. The D3 can churn out 222MB/s of data in 14-bit FX C<sub>H</sub> mode. The D3x can push 195MB/s in 12-bit C<sub>H</sub> mode.

You're right though, throughput is very fast as it stands, at least from the sensor to the buffer. Even better is that naively Digic 4 is about 50% faster than Digic III<sup>2</sup>. The card is a bit more problematic, the best I've seen to top end cards get is around 35MB/s (go look tough Rob Galbraith's database). I really can't say I expect to see that doubling any time soon, at least not realistically. I certainly don't think CFast cards are going to debut at 2x the speed of top end CF cards. And given their track records, I don't expect to see dual card interleaving either, though that would be nice even now.

-----

1) Well there's another possibility that the sensor can't be completely read out at 10 or 11 FPS, but without knowing if the readout changes between DX and FX modes, I'm going to assume that's not the case. It also seems rather odd, since it's merely an engineering problem and it shouldn't be an insurmountable one at that.

2) Assuming the Digic 4 in the 5D-II is close to topping out at 3.9 FPS and the Dual Digic 3 in the 1D/1Ds is also close to topping out at 10/5 FPS respectively.

John Chardine

08-03-2009, 07:53 PM

Makes sense. Thanks Jason. I'll go and have a think and a read now!

Jason Franke

08-05-2009, 02:47 PM

Speak of the devil, I just saw this on Nikon Rumors. They are saying (I haven't read the patent yet, so I'm not sure) that Nikon is applying for a patent on embedding AF pixels in the imaging sensor. That would seem to solve the beam splitter problem and the slowness of contrast AF. The only problem I see is now you'd need to interpolate the area where you're no longer imaging.

For anyone who's interested, it's US patent number 20090167927.