Live view underexposure in Canon XSi

[Carlos Lindado]

Hello, I've just registered to the forums, I think this is the most comprehensive Canon-related site on the net!


Currently, I have a Canon XSi with EF-S 18-55mm f/3.5-5.6 IS and EF 50mm f/1.8 Mk I lenses, and I've noticed that when I use Live view while in Av or Tv mode, the camera underexposes the image by 2/3-1 stop. This doesn't occur all the time, but in certain lighting situations (low light, light in front of camera). Why do you think this is happening? I guess is CMOS sensor metering vs dedicated AE-meter metering, but I'm not really sure. The point is, sometimes I'm shooting on a tripod with Live view and I have to re-take the picture because I didn't account for Live view underexposing the image.


Thanks for reading!

[Scott]

Not to sure Carlos but welcome to the forum. I know one of these smart guys / galls will know what is going on. Have you got your LCD brightness set right.

[Sheiky]

Hey Carlos, welcome!


Funny that you stated this "problem". I never use liveview(at least not in AV/TV or auto mode) so I haven't experienced it yet. But I've tried it myself just yet with a Sigma 50mm 1.4 on a 5D2 and I too see the darker exposed shots when using live-view.


I tried 5 subjects inside, (only one set had the same exposure), so constant light and there's definitely some degree of difference. From 2/3 to even 2 stops. The photos don't look "badly exposed" or something. But they are a little off...most noticeable when viewing the histogram.


Interesting point and at this moment I don't know what causes it. I'll do some more testing later on...


Jan

[mikehillman89]

I've noticed this problem as well Carlos! I'm not entirely sure what it is as well. To be quite honest I just adapted myself to exposing a bit up, but in any case I've rarely used Live View... As well, (for 7D owners) I've noticed that the camera way underexposes the video in live view as well so beware!

[Carlos Lindado]

Thanks for your replies! I didn't think this was an issue on other Canon bodies too (I've used a friend's 7D, but not in live view mode). My best guess is that the CMOS sensor isn't as accurate as the dedicated AE meter, and it tries to calculate the "best" exposure its own way.


Regarding live view, I like to use it for tripod shots (awkward angles, manual focusing, more accurate framing) so I think it's a useful tool in specific situations.

[alex]

Doesn't LiveView set the exposure based on where you have the small rectangle in LiveView moved to? I use LiveView on my XSi anytime I have it on a tripod and I want to set focus to something that is not dead center in the screen. I've noticed that the exposure on the screen will vary when I move the rectangle around to place it over where I want to zoom in to set focus.


So if it's sampling from a smaller zone for metering (like Spot metering, kind of), there is the possibility of it getting a different exposure than if you didn't use LiveView and had your camera set to Evaluative or Partial Metering.


I could be completely incorrect. [:$]

[Carlos Lindado]

Hey Alex, you're completely right! I just tested moving the focusing rectangle and yes, the exposure changes drastically, however, the manual of the camera says "During Live View shooting, the metering mode will be set to evaluative metering regardless of the current metering mode setting".


Intrigued by that, I tested changing the focus point while in "normal mode" (no live view) and, as in live view mode, the exposure changed drastically, so I think Canon's evaluative metering is strongly biased toward the focus point. Nevertheless, live view calculated exposure continues to be different from the one calculated without live view.

[Sheiky]

Well that clears some things at least [:D] It's not changing anything for me personally, but I like it when a "problem" gets solved [Y] Good stuff detective Alex!

[alex]

Glad to help, but that still doesn't clear up why LiveView, which supposedly uses Evaluative Metering, would get different metering results than normal view Evaluative Metering.


Anyone have any ideas about that one?

[neuroanatomist]

Glad to help, but that still doesn't clear up why LiveView, which supposedly uses Evaluative Metering, would get different metering results than normal view Evaluative Metering.





I suspect Carlos was on the right track:



<div>



I guess is CMOS sensor metering vs dedicated AE-meter metering


Both are 'evaluative' in the sense that the camera is using (undisclosed) algorithms to compare the luminance of subject and the surrounding area in determining autoexposure. But there are several differences:




'Standard' AE and Live View AE use different sensors (dedicated AE sensor vs. imaging sensor)
The areal coverage of the sensors is different (standard AE uses 35 zones on the XSi, Live View uses the entire frame, including the edges that extend beyond the coverage of the dedicated AE sensor)
AF points are different - evaluative metering determines the subject based on which AF point is selected; for standard AE that's restricted to the 9 AF points, whereas for Live View the AF point can be anywhere in the frame.
The algorithms used to determine exposure likely differ between the two modes (and of course, the inputs to those algorithms also differ).




All of this will add up to situational differences in the way exposure is determined.


I tried a little experiment with my 7D, comparing exposures in standard shooting vs. Live View. I shot some trees this morning, backlit by the near-dawn sky. Changing from standard to Live View shooting, I could see the underexposure that has been mentioned - the Live View shot was about 1/2 stop darker (visually and based on the histogram). Conversely, last night I had shot some indoor test shots in the living room. A torchire lamp in the corner of the room was a bright spot at the edge of the frame. For that composition, the Live View was about 2/3 of a stop overexposed compared to the standard VF shot. So, it seems not to be a consistent, systematic underexposure with Live View, but rather simply a difference in the way exposures are calculated, which is impacted by composition and lighting.


Just as a side note that I uncovered while reading up on these issues - evaluative metering is linked to one and only one AF point. So, if you're shooting with automatic AF point selection and several AF points 'light up', that simply means that all of those points are within the depth of focus (determined by the AF system). However, only one of them will be used as the basis for the evaluative metering and unfortunately, there's no way to know which of the 'lit' AF points the camera is using as the basis for the metering evaluation.


I suppose all of this just reinforces the need to check the histogram as you shoot!
</div>

[alex]

Thanks neuro, I knew I could count on you for a great response! Good info, I find the evaluative metering using only one AF point very interesting. Sounds kind of like roulette...you never know which point you're gonna get for the exposure. [:O]

[Richard Lane]

The point is, sometimes I'm shooting on a tripod with Live view and I have to re-take the picture because I didn't account for Live view underexposing the image.


Thanks for reading!






Hi Alex,


Could the problem be due to light entering the viewfinder from the rear of the camera?


When you use LiveMode, you obviously have to take your face away from the viewfinder to view the LCD Screen. This lets more light in through the back of the viewfinder and therefore the camera thinks it has more light then it really does have in front of the lens, so it under exposes. Try your experiment again with the tripod, but use the viewfinder cover on your neck strap and test it again.


I know it sounds very simple, but I noticed this happens when I take a test shot with my tripod looking through the viewfinder and then I step away and use a remote switch and the exposures are always different.


Rich

[neuroanatomist]

Could the problem be due to light entering the viewfinder from the rear of the camera?


I know it sounds very simple, but I noticed this happens when I take a test shot with my tripod looking through the viewfinder and then I step away and use a remote switch and the exposures are always different.





Good thought, but no. You're correct in that it's a good idea to cover the eyepiece when shooting in 'normal mode' on a tripod, due to light entering the VF and affecting the metering. That's why Canon provides the little rubber eyepiece cover (on the strap), although personally I just hang the lens cap over the eyepiece since that saves me having to remove (and lose?) the eyecup.


But, in Live View mode the mirror is flipped up, blocking light entering through the VF.






<div>



I find the evaluative metering using only one AF point very interesting. Sounds kind of like roulette...you never know which point you're gonna get for the exposure.


Yep. That's one reason I'm glad I almost always manually select an AF point (even though the 7D offers an in-between option - Zone AF - where you select a smaller cluster of AF points, I still prefer to choose one myself).
</div>

[Carlos Lindado]

Hi Richard, I guess your reply was addressed to me, so: I always use the viewfinder cover when shooting with a tripod (normal or live view), nevertheless, sometimes the shots are underexposed (or overexposed) with live view depending on the lighting conditions. As neuroanatomist states, it's not a systematic error but kind of a random one (unless one could be able to figure out the live view exposure calculation algorithm and know ahead if the shot will be underexposed or overexposed).





Updated: neuro makes a good point about the mirror being flipped when live view is active: no light from the viewfinder should reach the sensor as the mirror blocks it.

[Richard Lane]

OK Guys, Thanks for the clarification about the mirror being up in Live Mode.


Rich

[JJphoto]

But, in Live View mode the mirror is flipped up, blocking light entering through the VF.


Neuroanatomist, when you say "mirror is flipped up", is it as same as mirror lock up?, the reason I 'm asking is that your post actually reminds me of something about live view and mirror lock up from Juza's web site ("http://www.juzaphoto.com/eng/articles/live_view_mirror_lock_up.htm), here is what it says






<span style="color: #6daa37;"]"Different types of
Live View and MLU


Nearly all cameras have Live View, but
there are four main different types of Live
View.



A) The most advanced Live View
implementations (currently found in the
Canon 7D and 5DII) allow to shoot
using the electronic first curtain; in other
words, when you press the shutter the camera
reset electronically the sensor to begin the
exposure, then it closes the physical
shutter curtain to end the exposure. With
this kind of live view, you no longer need
MLU.

B) Another implementation of
live view is found into older Canon cameras
(1DIII, 1DsIII). These cameras don't have
the electronic first curtain; when you press
the shutter the camera closes the physical
curtain to reset the sensor, it opens the
curtain to begin the exposure and it closes
again the curtain the end the exposure. With
this kind of live view, you still need MLU."





I'm not a tech guy, I don't even 100% understand what the above is saying, I got really confused. my question is that for rebel XSi, when use live view mode, is the mirror locked up or it's just flipped up but not locked? and how it effects the exposure? thanks.

[neuroanatomist]

Neuroanatomist, when you say "mirror is flipped up", is it as same as mirror lock up?


Yes, when the mirror is up for Live View, that's equivalent to mirror lock up. In fact, since mirror lock up is somewhat buried in the menu options, in situations where you'd want MLU it's often easier to simply activate Live View. So yes, on the XSi in Live View mode, the mirror is locked up and will stay locked up until the end of the exposure, unless you're using 'Quick AF' or you're using a flash with E-TTL (see below).


To counter Juza's misinformation, curtain = shutter. Curtain &ne; mirror. Witholder implementations of Live View that Juza mentions (in cameras without an electronic first curtain), the camera has to close the shutter to start the exposure. In that case, MLU is still not needed, since the mirror is already up. It's likely Juza is confusing the issue by thinking of the autofocus modes in Live View - in 'Quick AF', the mirror is flipped down to use the camera's AF system for focusing. The metering is still performed using the image sensor rather than the dedicated AE sensor. In 'Live AF' (aka contrast-detect AF), the mirror is never flipped down at all, meaning MLU is not needed then, either. On Nikon cameras, apparently, the mirror always flips down then up during Live View shooting - that would make that mode rather handicapped, IMO, since there'd be no way to use Live View and avoid the shake from mirror-slap that can affect exposures of certain lengths (which is the purpose of MLU).


It's also worth noting that even on the 7D, the electronic first curtain (EFC) is not used by default. In Live View, that feature is only usedwith one of the silent shooting modes. Otherwise, the first curtain is mechanical just like on models lacking an EFC. Unlike the 1DIII and 1DsIII that Juza mentions, which don't have EFC, on the XSi in Live View mode, an EFC is used (and is the only option). With the Rebel series, the mirror will cycle after the exposure, since a single motor drives both the mirror and the shutter, so the mirror must cycle to reset the shutter curtains (the 50D/7D/5DII and 1-series bodies have separate motors for mirror and shutter).


There is one situation in which the mirror is flipped down for exposure in Live View, in all Canon cameras offering Live View - that's when you're using an external flash with E-TTL (I or II) metering enabled. For E-TTL metering, a pre-flash detectedthrough-the-lens (thus, TTL)is used to determine the proper flash exposure, and the dedicated AE sensor is needed to measure that pre-flash. But even in that case, for Live View shooting the image exposure determination is still made by the CMOS imaging sensor, not the AE sensor (the AE sensor is used only for determining flash exposure, in that case). Page 26 of the 1DIII White Paper ("http://www.usa.canon.com/uploadedimages/FCK/Image/White%20Papers/EOS-1Ds_MarkIII_WP.pdf) has a nice flow chart showing the specific steps, including when the mirror goes up, and which sensors are used when, for normal vs. Live View shooting with and without flash. Note that on that chart, mirror lockup just means the mirror going up, not the MLU custom function (but, since all MLU does is result in a longer interval between the mirror going up and the exposure being taken, with Live View it amounts to the same thing).


Sorry...that was probably far more than you wanted to read!



Juza's web site...says...


Many of Juza's facts are inaccurate. For example, in his review of the 85L and 85/1.8 lenses ("http://www.juzaphoto.com/eng/articles/canon_85mm_f1-2_vs_85mm_f1-8.htm), he originally stated that the 85L was weather-sealed, and he also stated that it has micro-USM which is false (it uses a focus-by-wire system as opposed to mechanical coupling to actuate the ring-USM motor). At least he reads the comments on his reviews, so he edited the review to state that he 'originally thought it had micro-USM' - but he doesn't admit that he originally wrote it had micro-USM, and the correction to no weather-sealing in the specs table was a stealth edit. In fact, in his review of the 135L, he still states, "The built quality is excellent and, unlike the 85mm f/1.2 that I reviewed few weeks ago, it had the true ring type AF motor..."


Juza takes nice pictures. But I look to him for technical information like I look to Greece for fine wine or the Pacific Northwest for historic architecture...that is to say, not at all.


--John

[JJphoto]

John, thank you very much for taking the time to answer my question, really appreciate it.






Sorry...that was probably far more than you wanted to read!








like I said, I'm not a tech guy, it will take me long time to digest it. the bottom line is that as long as I know I don't need to go to menu to set MLU when shoot with live view on XSi(I'v been missing my stolen XSi and thinking buy it back since the price is getting cheaper).






But I look to him for technical information like I look to Greece for fine wine or...


I don't drink(I drink water only), but I can understand what you mean, I'm pretty sure you have a great taste of it.[:)]


thanks again!

[Carlos Lindado]

...on the XSi in Live View mode, an EFC is used (and is the only option).


Hey neuro, look what I got from verifying that!


/cfs-file.ashx/__key/CommunityServer.Components.UserFiles/00.00.00.45.65/IMG_5F00_5075.JPG


/cfs-file.ashx/__key/CommunityServer.Components.UserFiles/00.00.00.45.65/IMG_5F00_5076.JPG


/cfs-file.ashx/__key/CommunityServer.Components.UserFiles/00.00.00.45.65/IMG_5F00_5077.JPG


/cfs-file.ashx/__key/CommunityServer.Components.UserFiles/00.00.00.45.65/IMG_5F00_5078.JPG


I was stupid enough to take the lens apart and activate live view in my not-so-clean bedroom, and I ended with that big speck. Thankfully, Canon's sensor cleaning function worked this time! (after three tries, actually). As a side note, I know my sensor is not precisely "clean", but I don't think it's worth fully-cleaning it yet (I've had bad experiences with air blowers, so I use the "wet" method: E2 and sensor swabs)


BTW, you're right neuro!, Canon XSi uses an EFC (bad way to figure it out, though).

[neuroanatomist]

...on the XSi in Live View mode, an EFC is used (and is the only option).


Hey neuro, look what I got from verifying that!


BTW, you're right neuro!, Canon XSi uses an EFC (bad way to figure it out, though).





Might have been easier to Google "xsi electronic first curtain"... [8-|]


The second link that comes up is this one ("http://krebsmicro.com/Canon_EFSC/index.html), and if you scroll down past the images and commentary, you'll find a pasted email from Chuck Westfall (Canon's technical guru extraordinaire, although that's not his official title) which provides confirmation, along with all the juicy details...and the statement that EFC on Rebels is an undocumented feature.


But, there's nothing like confirming something for yourself...kudos to you!