Strange overspec metering results with Canon 5D Mark III

[Teemu76]

Hi,

This is my first post even if I have followed this forum for long time.

I was shooting today bird time lapse sequence with Canon 5D mark III + 70-200 F2.8 IS II. Sequence ended just after sunset. I took one exposure in every three seconds. Exposure settings were M mode, F2.8, 1.6 seconds and auto ISO. I took more than one thousand pictures. First picture was taken with ISO 160 and last with ISO 4000. ND 400 filter was used to get long exposures.

It was however strange that almost 400 pictures at the end were shot with ISO 4000. Pictures were more and more underexposed towards to end. First ISO 4000 picture was correctly exposed. I got an idea to calculate exposure value of last pictures. It is EV -3 !!! According to specifications almost all Canon DSLRs are metering between 1-20 EV. I should not get -3 EV exposure metered but somehow I did

I made some further testing. I managed to meter -5 EV with Sigma 85mm F1.4 (1s, F1.4, auto ISO 6400). Also other -5 EV exposure parameter combinations were possible. I did not use ND filter with these additional tests. So it seems that 5D mark III meters up to -6 EV with F1.0 lens. Every stop of lens brigtness drops this result by one stop.

My friend tested Canon 30D metering with EF-S 17-85 zoom. I understood that he tried metering with full zoom in Av mode and lens cap on. Exposure paramters were F5.6 and ISO 100. Resulted exposure time was ten seconds. That gives exposure value of one and two thirds of a stop (1.7 EV).

Does anyone have explanation to my findings?

-Teemu

 

[Teemu76]

One addition. Both 30D and Canon 5D mark II should meter 1-20 EV. That value is probably measured with Canon 50mm F1.8. Even 30D would probably have measured negative exposure values if brighter lens had been used.

 

[Teemu76]

Time lapse metering was done with live view and later tests without.

 

[Lichtgestalt]

i never use auto iso myself but maybe auto iso was set to go no further then iso 4000?

 

[Loswr]

You're outside the metering spec. Maybe it'll work sometimes, but you should not expect it to function reliably.

 

[Teemu76]

i never use auto iso myself but maybe auto iso was set to go no further then iso 4000?

Auto ISO limits use only full stops. My limit was ISO 12800.

 

[Teemu76]

You're outside the metering spec. Maybe it'll work sometimes, but you should not expect it to function reliably.

I will continue using it for this kind of shooting until I find out some problems with it. If I had used F.1.4 lens with F2.8 aperture then auto ISO would have worked until ISO 16000

 

[Teemu76]

i never use auto iso myself but maybe auto iso was set to go no further then iso 4000?

I used auto ISO because I wanted to keep exposure time same for all of the shots. There is almost five stops difference between first picture and last correctly exposed picture. I would have needed shutter speeds of about 1/15 seconds for first shot to get 1,6 seconds for last shot. Look of the motion in video would changed too much between beginning and end.

 

[Lichtgestalt]

i guess neuro is right... ND400 is basically a 9 stop (rounded) filter, right?

i found that my canon cameras have trouble metering when i attach my 10 stop filter.
the metered exposure varys a lot.

ps: i think this is the wrong forum for this post. as it is not about eos body rumors.

 

[Teemu76]

i guess neuro is right... ND400 is basically a 9 stop (rounded) filter, right?

i found that my canon cameras have trouble metering when i attach my 10 stop filter.
the metered exposure varys a lot.

ps: i think this is the wrong forum for this post. as it is not about eos body rumors.

I am sorry about not reading full path of the forum structure. I just saw label "EOS bodies" and got confused.

I think my finding does not have anything to do about using ND filter. That was just way to get -3EV exposure result in real life situation. I replicated the phenomenon in dark room with 17-40L and Sigma 85mm F1.4 and without ND filter. Histogram confirmed that accuracy was good enough in my case. Maybe that is not always the case.

 

[Lichtgestalt]

i guess neuro is right... ND400 is basically a 9 stop (rounded) filter, right?

i found that my canon cameras have trouble metering when i attach my 10 stop filter.
the metered exposure varys a lot.

ps: i think this is the wrong forum for this post. as it is not about eos body rumors.

I am sorry about not reading full path of the forum structure. I just saw label "EOS bodies" and got confused.

I think my finding does not have anything to do about using ND filter. That was just way to get -3EV exposure result in real life situation. I replicated the phenomenon in dark room with 17-40L and Sigma 85mm F1.4 and without ND filter. Histogram confirmed that accuracy was good enough in my case. Maybe that is not always the case.

I did not use ND filter with these additional tests

did you use the ND filter for this tests?
i can only say with a 10 stop filter, on a bright day, my cameras have troubles to meter 100% correctly.

im just completely guessing here, but maybe my ND filter absorbs some light frequency and fools the meter? i mean they tend to make images look blueish or redish.... so it´s not a complete crap idea... right?

 

[Teemu76]

i guess neuro is right... ND400 is basically a 9 stop (rounded) filter, right?

i found that my canon cameras have trouble metering when i attach my 10 stop filter.
the metered exposure varys a lot.

ps: i think this is the wrong forum for this post. as it is not about eos body rumors.

I am sorry about not reading full path of the forum structure. I just saw label "EOS bodies" and got confused.

I think my finding does not have anything to do about using ND filter. That was just way to get -3EV exposure result in real life situation. I replicated the phenomenon in dark room with 17-40L and Sigma 85mm F1.4 and without ND filter. Histogram confirmed that accuracy was good enough in my case. Maybe that is not always the case.

I did not use ND filter with these additional tests

did you use the ND filter for this tests?
i can only say with a 10 stop filter, on a bright day, my cameras have troubles to meter 100% correctly.

Timelapse was shot with ND filter and further testing was made without ND filter. I think camera does not know difference between metering high exposure value with ND filter in good light and without ND filter in low light. I think it is just about amount of light through the lense during measuring period. For me this was positive finding. Even if measuring with ND is not 100 percent accurate this is positive finding for me. There are means to batch process timelapse pictures with suitable varying exposure compensation.

 

[Drizzt321]

i guess neuro is right... ND400 is basically a 9 stop (rounded) filter, right?

i found that my canon cameras have trouble metering when i attach my 10 stop filter.
the metered exposure varys a lot.

ps: i think this is the wrong forum for this post. as it is not about eos body rumors.

I am sorry about not reading full path of the forum structure. I just saw label "EOS bodies" and got confused.

I think my finding does not have anything to do about using ND filter. That was just way to get -3EV exposure result in real life situation. I replicated the phenomenon in dark room with 17-40L and Sigma 85mm F1.4 and without ND filter. Histogram confirmed that accuracy was good enough in my case. Maybe that is not always the case.

I did not use ND filter with these additional tests

did you use the ND filter for this tests?
i can only say with a 10 stop filter, on a bright day, my cameras have troubles to meter 100% correctly.

Timelapse was shot with ND filter and further testing was made without ND filter. I think camera does not know difference between metering high exposure value with ND filter in good light and without ND filter in low light. I think it is just about amount of light through the lense during measuring period. For me this was positive finding. Even if measuring with ND is not 100 percent accurate this is positive finding for me. There are means to batch process timelapse pictures with suitable varying exposure compensation.

Correct, the camera doesn't have any idea when there is an ND filter on or not. It just sees the light that comes through the lens.

However, with a heavy ND filter (say, the 9/10-stop 'black glass') there is too little light coming in to the lens for the metering system to correctly meter. You're better off metering w/out the ND filter and adjusting X stops for the ND and then putting it on or just taking educated guesses and then adjusting based off the histogram on the back of the camera.

 

[Lichtgestalt]

However, with a heavy ND filter (say, the 9/10-stop 'black glass') there is too little light coming in to the lens for the metering system to correctly meter. You're better off metering w/out the ND filter and adjusting X stops for the ND and then putting it on or just taking educated guesses and then adjusting based off the histogram on the back of the camera.

thought kind of complicated for him when doing timelapse.

but my question still stands if the ND filter filters out frequencys and that might fool the metering too?

 

[Mt Spokane Photography]

However, with a heavy ND filter (say, the 9/10-stop 'black glass') there is too little light coming in to the lens for the metering system to correctly meter. You're better off metering w/out the ND filter and adjusting X stops for the ND and then putting it on or just taking educated guesses and then adjusting based off the histogram on the back of the camera.

thought kind of complicated for him when doing timelapse.

but my question still stands if the ND filter filters out frequencys and that might fool the metering too?

Not only that, but the colors of the light change as the sun goes down, so different frequencies would filter differently, its a possibility.

In any event, after sundown with a 10 stop ND filter is going to be a big problem, the camera metering might be totalally unreliable.

 

[Teemu76]

However, with a heavy ND filter (say, the 9/10-stop 'black glass') there is too little light coming in to the lens for the metering system to correctly meter. You're better off metering w/out the ND filter and adjusting X stops for the ND and then putting it on or just taking educated guesses and then adjusting based off the histogram on the back of the camera.

thought kind of complicated for him when doing timelapse.

but my question still stands if the ND filter filters out frequencys and that might fool the metering too?

Not only that, but the colors of the light change as the sun goes down, so different frequencies would filter differently, its a possibility.

In any event, after sundown with a 10 stop ND filter is going to be a big problem, the camera metering might be totalally unreliable.

Proportion of the red decreases and proportion of the blue increases towards end of the series. Filter itself should be quite neutral. I have B+W 6 stop filter which lets too much red through.

 

[Drizzt321]

However, with a heavy ND filter (say, the 9/10-stop 'black glass') there is too little light coming in to the lens for the metering system to correctly meter. You're better off metering w/out the ND filter and adjusting X stops for the ND and then putting it on or just taking educated guesses and then adjusting based off the histogram on the back of the camera.

thought kind of complicated for him when doing timelapse.

but my question still stands if the ND filter filters out frequencys and that might fool the metering too?

Not only that, but the colors of the light change as the sun goes down, so different frequencies would filter differently, its a possibility.

In any event, after sundown with a 10 stop ND filter is going to be a big problem, the camera metering might be totalally unreliable.

Proportion of the red decreases and proportion of the blue increases towards end of the series. Filter itself should be quite neutral. I have B+W 6 stop filter which lets too much red through.

I'd think that makes sense, once the sun is truly set and you're well into dusk the sky goes more blue and the sun isn't around to add the red back in via reflections. Think of walking through the middle of sky scrapers towards the evening. Even though the sun is still technically up, you tend to get very cool (towards blue) tones when not in direct sunlight.

 

[Teemu76]

However, with a heavy ND filter (say, the 9/10-stop 'black glass') there is too little light coming in to the lens for the metering system to correctly meter. You're better off metering w/out the ND filter and adjusting X stops for the ND and then putting it on or just taking educated guesses and then adjusting based off the histogram on the back of the camera.

thought kind of complicated for him when doing timelapse.

Very I am spending all of my spare time to code metadata editor to be used with Lightroom. Then I can use aperture priority and auto iso of manual mode without flickering. Atleast i hope so.

 

[Busted Knuckles]

Did you set the Auto ISO limit at 4000? I didn't see anyone ask this question.

 

[Loswr]

Did you set the Auto ISO limit at 4000? I didn't see anyone ask this question.

Someone did ask, but I don't think that's the issue - the min/max for the ISO range is set in whole-stop increments, ISO 3200 or 6400, but 'tweener' stops can't be set.