It is currently Thu Jun 17, 2021 12:40 am




Post new topic Reply to topic  [ 54 posts ]  Go to page 1, 2  Next
 About DSLR flicker 
Author Message
User avatar

Joined: Wed Jul 09, 2008 11:54 pm
Posts: 888
Location: Aars, Denmark
Post About DSLR flicker
What causes the flicker in DSLR timelapses? Most of us are familiar with a number of potential reasons, but it is my impression that a thorough understanding of the issue is still lacking and that we often blame the wrong factors and take the wrong measures to try preventing it. To get straight to the point, take a look at the following video.



The left half of the frame was captured with a Canon S3 (compact), the right with a 40D (DSLR). Same time, place and lighting and same camera settings: Evaluative metering, and P mode, allowing both cameras to adjust aperture and exposure time to obtain appropriate exposure. However, since it was shot at dusk both cameras used maximum aperture throughout the sequence. Manual white balance (cloudy). The viewfinder on the 40D was covered. No deflickering has been applied.

As can be seen, compact cameras (at least the models I've owned) don't have the same issues with flicker that DSLRs do. This shows that flicker is not a necessary consequence of using automatic camera settings which is good news for those hoping for a solution to the night/day transition issue. Rather, it appears to be something specific and common to DSLR designs that is causing the trouble. What in particular is it?

To go out on a limb, I doubt the lens has much to do with it. This is in disagreement with the theory that inaccurate lens diaphragms is a source of flicker. However, it is pretty easy to test if that is actually the case: Set up your camera in a room with constant lighting (plus, cover your viewfinder to be on the safe side), shoot with all manual settings and the lens stopped down. Any flicker in the resulting video? If not, your lens has a sufficiently accurate diaphragm that you can cross that off your list of potential causes. I did this test with two lenses stopped down to f10 and f22, and there was no hint of flicker to be seen. Furthermore, I have yet to experience more than mild flicker from any of my compacts even though they operate diaphragms too.

I've done a little more testing and made the following observations:
1) With full manual settings there is no flicker regardless of varying lighting (so at least the camera CAN do consistent exposures from frame to frame; it isn't a question of one parameter that is inherently inaccurate).
2) Allowing varying aperture, shutter speed or both only results in flicker during varying lighting.
3) When using automatic settings the exposure may vary from frame to frame even if EXIF info suggests that the exposure should be exactly the same. (Unfortunately, EXIF is a mess - the same photo contains several entries for aperture and shutter speed which don't always agree. I don't know how to tell which ones are true. :().

I admittedly have a hard time making sense of this. DSLRs CAN set all parameters consistent and accurate from frame to frame, and given that compact cameras can vary exposures smoothly it is hard to see why DSLRs can't. Perhaps DSLR metering systems are more sensitive, but also less accurate. Then, that still leaves the question why exposures vary when EXIF suggests they should be identical. Maybe what the camera does and what it writes to EXIF isn't always exactly the same.

Comments? Ideas?

_________________
Nikon D3s, D600, Canon 5D II, S3, SX100.
Feature film Beneath the Aurora on Vimeo.


Sun Sep 14, 2008 2:31 am
Profile
User avatar

Joined: Wed Apr 16, 2008 9:49 pm
Posts: 349
Location: Troutdale, OR http://www.vimeo.com/ac
Post Re: About DSLR flicker
People keep talking about covering the viewfinder, can someone explain to me what the deal is with that?

_________________
Andrew Curtis http://www.ffwdtime.com


Sun Sep 14, 2008 2:56 am
Profile
User avatar

Joined: Wed Jul 09, 2008 11:54 pm
Posts: 888
Location: Aars, Denmark
Post Re: About DSLR flicker
Andrew Curtis wrote:
People keep talking about covering the viewfinder, can someone explain to me what the deal is with that?

The light meter (which measures the incoming light to determine what the exposure should be) on a DSLR is exposed not only to light coming through the lens, but also light coming through the viewfinder. Mostly the amount of light entering through the viewfinder is negligible compared to what enters through the lens, but sometimes that is not at all the case. So a varying light source behind the camera can also cause inconsistent metering (especially if you're shooting in the dark) unless you cover the viewfinder.

_________________
Nikon D3s, D600, Canon 5D II, S3, SX100.
Feature film Beneath the Aurora on Vimeo.


Sun Sep 14, 2008 4:35 am
Profile
Post Re: About DSLR flicker
Q: did the 40D have it's WB locked?

I'm puzzled by your observations as well... I will make a few comments though:

-'some' lenses are prone to aperture variations. Not all and when it happens it's very very mild.
-your sample is somewhat 'unfair' one since it was a very overcast generally narrow dynamic range in the frame (you could run that through a flicker filter and get it just as consistent as the S3) I'd like to see the same test performed with a semi-cloudy sunset. The EV goes all over the place in that case and the DR is huge.
-could it be that the S3 has a much finer gradation in exposure (ie 1/4 or 1/8 EV steps) since it has an electronic shutter?
-Maybe auto ISO is the way to go and allows the camera better exposure accuracy in the engineers 'sweet spot'? to be honest I've never used auto ISO... in my entire life!
-Maybe you've found a gem in that S3 for this situation?

It's like an enigma wrapped inside an enigma! :lol: One thing IS clearly highlighted here... The DSLR blows the doors off the S3 in terms of fidelity and light sensitivity! Of course you probably had a much faster lens on it too.


Sun Sep 14, 2008 5:12 am
User avatar

Joined: Wed Jul 09, 2008 11:54 pm
Posts: 888
Location: Aars, Denmark
Post Re: About DSLR flicker
milapse wrote:
Q: did the 40D have it's WB locked?

Yes, both cameras had white balance locked (Cloudy I believe).
milapse wrote:
-your sample is somewhat 'unfair' one since it was a very overcast generally narrow dynamic range in the frame (you could run that through a flicker filter and get it just as consistent as the S3) I'd like to see the same test performed with a semi-cloudy sunset. The EV goes all over the place in that case and the DR is huge.

I am not sure I understand what you mean by 'unfair' here. Conditions are exactly the same for both cameras. I uploaded an S3 sequence with a semi-cloudy sunset here. There may be a hint of flicker, but nothing like what you get with a DSLR. SOMETIMES a deflicker filter can save the DSLR sequence, other times not, and in any case it is a relief in post processing not to have to worry about it (in my case a big relief).
milapse wrote:
-could it be that the S3 has a much finer gradation in exposure (ie 1/4 or 1/8 EV steps) since it has an electronic shutter?

This is the most plausible explanation I can think of currently. I am trying to find some info on this.
milapse wrote:
-Maybe auto ISO is the way to go and allows the camera better exposure accuracy in the engineers 'sweet spot'? to be honest I've never used auto ISO... in my entire life!

I actually doubt it helps any for this purpose. It isn't meant to help exposure accuracy; it just cranks the ISO up as lighting gets worse in an attempt to avoid camera shake. The reason I used it is that I think higher ISO allows the S3 to keep up a little longer before everything becomes dark. I haven't tested thoroughly and may be wrong about that though.
milapse wrote:
-Maybe you've found a gem in that S3 for this situation?

My S2 and SX100 perform just as well. I'd like to know how other brands of compacts behave.

milapse wrote:
One thing IS clearly highlighted here... The DSLR blows the doors off the S3 in terms of fidelity and light sensitivity! Of course you probably had a much faster lens on it too.

The metering systems on my compacts simply aren't up to the task once the dark falls, so in that respect they clearly lose out to the DSLRs (btw, apertures were constant f3.2 for the S3 and f4.6 for the 40D, so actually faster on the S3). I am anxiously waiting for SLR-like cameras with large sensors and electronic shutters which might be able to capture day/night transitions just fine... depending on what the issue really is!

_________________
Nikon D3s, D600, Canon 5D II, S3, SX100.
Feature film Beneath the Aurora on Vimeo.


Sun Sep 14, 2008 7:02 am
Profile
Post Re: About DSLR flicker
flyvholm wrote:
I am not sure I understand what you mean by 'unfair' here. Conditions are exactly the same for both cameras. I uploaded an S3 sequence with a semi-cloudy sunset here. There may be a hint of flicker, but nothing like what you get with a DSLR. SOMETIMES a deflicker filter can save the DSLR sequence, other times not, and in any case it is a relief in post processing not to have to worry about it (in my case a big relief).


Again that sample has a narrow dynamic range as well.. unfair may not be the right description. What I'm trying to say is that a dslr on auto would probably handle that situation just fine with a little post help. The tough situation is from late day to sunset to dusk to night with ample sun clouds. This situation can see 14+ Ev (20+ if you count the sun itself) in the frame a good 6-7 EV wider than the camera can handle. Thus the light sensor 'serches' or 'bobs' in that 6-7 EV range.


Sun Sep 14, 2008 9:31 am

Joined: Sun Jun 08, 2008 10:21 pm
Posts: 159
Location: Salt Lake City, Utah
Post Re: About DSLR flicker
is anyone noticing a decrease in flicker based on the time between frames? I have settled on shooting 1 fps and have seen my flicker decrease dramatically...regardless of shooting mode or aperature setting.

_________________
Michael Slade
Homepage
Book Project


Sun Sep 14, 2008 12:51 pm
Profile
User avatar

Joined: Wed Jul 09, 2008 11:54 pm
Posts: 888
Location: Aars, Denmark
Post Re: About DSLR flicker
Michael Slade wrote:
is anyone noticing a decrease in flicker based on the time between frames? I have settled on shooting 1 fps and have seen my flicker decrease dramatically...regardless of shooting mode or aperature setting.

Yes, shorter frame intervals can be expected to decrease flicker. For the testing I did I found that my DSLRs don't flicker at all when lighting is constant, and the shorter interval you use, the less the lighting is likely to change from frame to frame.

Regarding available shutter speeds for electronic vs. physical shutters: I checked the cameras, and they have exactly the same shutter speeds available (within the 1/3200 to 15" range of the S3) in shutter priority mode. I also checked the EXIF info for both image sequences in above video, and the shutter speed values there are the same as those available in shutter priority mode. Also, it turns out that both aperture and ISO was constant throughout the sequence for both cameras, so it is only the shutter speed that varies. According to the EXIF, the 40D is consistent when increasing the exposure time whereas the S3 may jump between values a couple of times before settling at a new exposure time. This suggests that the metering on the 40D is indeed accurate and consistent, but that the resulting camera exposures are much less consistent for some reason while for the S3 it is exactly the other way around. Really strange. Since the EXIF data is obviously unreliable (for at least the 40D) it is hard to get to the bottom of this since we just can't know what the cameras are actually doing. :-(

_________________
Nikon D3s, D600, Canon 5D II, S3, SX100.
Feature film Beneath the Aurora on Vimeo.


Sun Sep 14, 2008 6:32 pm
Profile

Joined: Sun Jun 08, 2008 10:21 pm
Posts: 159
Location: Salt Lake City, Utah
Post Re: About DSLR flicker
After reading your posts I decided to go out and shoot a sequence tonight and try something I have been wondering. I have a lens from a Holga plastic camera that I've modified to put on my D2x. It gives a nice soft-focus (and I mean REALLY soft) effect. Constant aperature - wide open.

So, I shot on manual and adjusted the shutter speed. Since the lens can't 'talk' to the camera, I was unable to use Aperature Priority shooting mode (which would let the camera adjust the shutter as the sun went down...).

Anyway, I did not see ANY flicker. NONE. I am pretty much convinced (again), that flicker is a function of aperature blades not closing *exactly* the same shot-to-shot.

Lemme process a short clip and get it up to Vimeo tonight...


_________________
Michael Slade
Homepage
Book Project


Sun Sep 14, 2008 7:27 pm
Profile
User avatar

Joined: Tue Apr 29, 2008 2:48 pm
Posts: 1144
Post Re: About DSLR flicker
DING - DING - DING - I think you've hit the nail on the head - aperture blades not closing exactly to the same place - I have invested in a whole host manual lenses to use on my time lapse and my flicker is gone - In order to see if I could continue to use my automatic lenses I even tried using a rubber band to keep the "DOF check" button down so that the aperture was locked in place, however when you shoot a picture the aperture opens for a fraction of a second and then close again. check out manualfocus.org - for a forum devoted to the wonder of manual lenses.

timt

_________________
Please check out how to embed a Vimeo link.
Tim T


Sun Sep 14, 2008 8:26 pm
Profile

Joined: Sun Jun 08, 2008 10:21 pm
Posts: 159
Location: Salt Lake City, Utah
Post Re: About DSLR flicker
I'm going to call Nikon tomorrow and see about having a body modified so that it won't do the stop-down lever opening trick. I'd rather modify a body than re-purchase a slew of lenses.

Hmmmm....

_________________
Michael Slade
Homepage
Book Project


Sun Sep 14, 2008 8:56 pm
Profile
User avatar

Joined: Wed Jul 09, 2008 11:54 pm
Posts: 888
Location: Aars, Denmark
Post Re: About DSLR flicker
pixelbot wrote:
DING - DING - DING - I think you've hit the nail on the head - aperture blades not closing exactly to the same place

It doesn't appear to be quite that simple. Here's are my reasons:

1) I've done tests shooting sequences with two of my lenses stopped down, constant lighting. Result: No flicker whatsoever. Conclusion: Diaphragm is not the source of inaccurate exposures.
2) Cheap compact cameras (at least my models) operate diaphragms too without displaying the heavy flicker problems of DSLRs.
3) I have numerous examples (like my video above!) with sequences where the lens was wide open; no diaphragm operation. Still heavy flicker.

Rather, I believe the key is using full manual settings (which I believe you end up doing when using a manual aperture lens). Not because of the lens, but because of the camera itself. For some odd reason, whenever the camera is not on full manual settings it will expose quite inconsistently from frame to frame even though it should be using the exact same exposure parameters, and according to reason #1 and #3 above it is not because of the diaphragm. Something else is at play.

_________________
Nikon D3s, D600, Canon 5D II, S3, SX100.
Feature film Beneath the Aurora on Vimeo.


Mon Sep 15, 2008 10:58 am
Profile
Post Re: About DSLR flicker
Michael Slade wrote:
I'm going to call Nikon tomorrow and see about having a body modified so that it won't do the stop-down lever opening trick. I'd rather modify a body than re-purchase a slew of lenses.


It would be at least nice to set the body to 'manual lens' and be able to adjust the aperture ring. If I'm not mistaken my D200 body will not kick into 'non-cpu' lens mode unless there is an ol' lens on the body...


Mon Sep 15, 2008 2:38 pm
Post Re: About DSLR flicker
I think the answer to your question lies not in the mechanics of the camera, but in the software...

That is, when your camera spits out a jpeg, it means it has completely processed, interpreted, and compressed that image. A lot of times people complain, when first moving to a higher-end dSLR from their point-and-shoot how "flat" the images look. Most better dSLRs try to do less processing by default, as they're designed to let the photographer have more control in post. I've found that a lot of point-n-shoots will sometimes result in the _wrong_ exposure given the particular settings, e.g. using a zone metaphor, if I meter off my highlight, and then set exposure to bring the highlight into zone 8, it ends up either in zone 6 or blown out in zone 11 more often than not. Whereas my dSLR gives consistent exposure, bringing my metered element into the right zone each time.

I'm thinking that what you're probably seeing is the algorithm for the camera, by default, tries to process the images to give them a certain look, that will automatically adjust out sight variations in exposure.

Do both cameras shoot in RAW? Without shooting them both in RAW, you're including a lot of image processing in the formula, which means you're going to have some serious variation in results as you add more cameras to the mix.

Also note, that not all "manual" lenses are truly manual in their aperture. For example, using any KA lens (K-mount lens that supports AV mode on a camera), even using the manual aperture ring results in it not being stopped down until fired when placed on a camera that supports reading the aperture. All of the manual lenses I have are either KA, M42, or Tamron adaptall - of these, only the M42s and Tamrons can actually be "fully" manual on my body.

!c


Mon Sep 15, 2008 4:12 pm
User avatar

Joined: Tue Jul 08, 2008 12:36 pm
Posts: 1787
Location: Antarctica/California/New Zealand
Post Re: About DSLR flicker
I'm not familiar with the Nikon lenses, but I was having a chat with a cinematographer a while back who told me he will use an old Nikon lens with manual aperture control, and he will cut off the control tab on the lens so the camera cannot control it between shots. This will give him good smooth shots on manual settings.
Don't ask me exactly which part to cut off, but if you are willing to give it a go... :shock:

_________________
Anthony Powell
"Antarctica: A Year On Ice" Feature Film
Vimeo Youtube Photos
Twitter Instagram Google+
Facebook Facebook Movie Page


Mon Sep 15, 2008 5:11 pm
Profile
User avatar

Joined: Wed Jul 09, 2008 11:54 pm
Posts: 888
Location: Aars, Denmark
Post Re: About DSLR flicker
shutterdrone wrote:
Do both cameras shoot in RAW? Without shooting them both in RAW, you're including a lot of image processing in the formula, which means you're going to have some serious variation in results as you add more cameras to the mix.

Neither camera was shooting raw. To keep the cameras in sync I was controlling them both from the same laptop, and that way the S3 can only shoot jpg; just the way Canon intended it. But thanks to a bunch of camera hackers it is possible to do RAW with the S3 when using CHDK. :-) I will try to do another test, shooting RAW with both cameras to see if it makes any difference.

I will be surprised if that's it. Even dedicated deflicker software often have difficulties removing DSLR flicker, so it would be amazing if Canon has software built-in that can do a better job of equalizing exposures (in spite of only analyzing one frame at a time).

BTW, I edited the original post to link to a video that has EXIF info superposed.

_________________
Nikon D3s, D600, Canon 5D II, S3, SX100.
Feature film Beneath the Aurora on Vimeo.


Mon Sep 15, 2008 6:21 pm
Profile
User avatar

Joined: Tue Apr 29, 2008 2:48 pm
Posts: 1144
Post Re: About DSLR flicker
Quote:
It doesn't appear to be quite that simple. Here's are my reasons:

1) I've done tests shooting sequences with two of my lenses stopped down, constant lighting. Result: No flicker whatsoever. Conclusion: Diaphragm is not the source of inaccurate exposures.
2) Cheap compact cameras (at least my models) operate diaphragms too without displaying the heavy flicker problems of DSLRs.
3) I have numerous examples (like my video above!) with sequences where the lens was wide open; no diaphragm operation. Still heavy flicker.

Rather, I believe the key is using full manual settings (which I believe you end up doing when using a manual aperture lens). Not because of the lens, but because of the camera itself. For some odd reason, whenever the camera is not on full manual settings it will expose quite inconsistently from frame to frame even though it should be using the exact same exposure parameters, and according to reason #1 and #3 above it is not because of the diaphragm. Something else is at play.


I also have done test where I have shot with automatic lenses (in full manual mode - manual everything exposure/shutter speed/WB/Focus) and I have flicker even wide open (some lenses are better than other and some aperture settings are better than others) - I think there is some movement in the diaphragms even at wide open. Once I switched to manual lenses (the manual setting are exactly the same) and I have had no flicker (I was shooting indoors in a closed room with constant light)- I would also guess that cheap compact cameras don't open and close their diaphragm on every shot like an SLR does

on a side note I have a shoot that had a flicker when I was using a manual lens but that was because I had a heater on the same circuit and everytime it turned on the light would dim :D

For me I'll stick with my manual lenses and no flicker.

timt

_________________
Please check out how to embed a Vimeo link.
Tim T


Mon Sep 15, 2008 7:56 pm
Profile
User avatar

Joined: Wed Jul 09, 2008 11:54 pm
Posts: 888
Location: Aars, Denmark
Post Re: About DSLR flicker
pixelbot wrote:
I also have done test where I have shot with automatic lenses (in full manual mode - manual everything exposure/shutter speed/WB/Focus) and I have flicker even wide open (some lenses are better than other and some aperture settings are better than others) - I think there is some movement in the diaphragms even at wide open. Once I switched to manual lenses (the manual setting are exactly the same) and I have had no flicker (I was shooting indoors in a closed room with constant light)- I would also guess that cheap compact cameras don't open and close their diaphragm on every shot like an SLR does


It was my impression that flicker doesn't occur in full manual mode, automatic lenses or not. You say it happened wide open but varied with aperture - is it possible that the 50 or 60Hz AC frequency can have been at play (so it only happens for faster shutter speeds)? Can others confirm getting flicker in full manual mode?

Using the DOF (depth-of-field) preview button you can both hear and see the diaphragm operate for all aperture settings except wide open. That OUGHT to mean that the diaphragm doesn't go anywhere during actual exposure either (when using max aperture).

Compacts do close and open the diaphragm on every shot (verified by looking into the lens while shooting).

_________________
Nikon D3s, D600, Canon 5D II, S3, SX100.
Feature film Beneath the Aurora on Vimeo.


Mon Sep 15, 2008 9:51 pm
Profile
User avatar

Joined: Tue Apr 29, 2008 2:48 pm
Posts: 1144
Post Re: About DSLR flicker
Quote:
It was my impression that flicker doesn't occur in full manual mode, automatic lenses or not. You say it happened wide open but varied with aperture - is it possible that the 50 or 60Hz AC frequency can have been at play (so it only happens for faster shutter speeds)? Can others confirm getting flicker in full manual mode?


I have gotten flicker in full manual mode (I only shoot manual mode for my time lapse as I don't want anything to change - in a small studio - 7'x3' totally enclosed - constant light ) - I was shooting a longer shutter speed (between .25 and 1 second) because I was shooting at f22 (to get maximum DOF) so it wasn't the 60Hz frequency. you can check out my shots at http://www.artbeats.com/collections/744 all those were shot with manual lenses - some of my earlier shots were with automatic lenses and I had to use deflicker plug-ins to clean those up.

Quote:
Using the DOF (depth-of-field) preview button you can both hear and see the diaphragm operate for all aperture settings except wide open. That OUGHT to mean that the diaphragm doesn't go anywhere during actual exposure either (when using max aperture).


honestly it's been a couple of years like early 2006 when was testing for flicker I don't remember exactly how wide open I was - partially because totally wide open wouldn't have helped my DOF (my plant shots wouldn't have turn out with shallow DOF) so I may not have been shooting full open - I can't remember.

Quote:
Compacts do close and open the diaphragm on every shot (verified by looking into the lens while shooting).


I just checked a old point and shoot and you are right there the diaphragm does in fact open and close everytime. so I don't know what is happening there - I can't shoot time lapse with my point and shoot so testing is moot for me.

_________________
Please check out how to embed a Vimeo link.
Tim T


Mon Sep 15, 2008 10:26 pm
Profile
User avatar

Joined: Wed Aug 13, 2008 4:10 pm
Posts: 85
Post Re: About DSLR flicker
I did hundreds of TL shootings with Canon DSLR, and from my point of view, flicker I may get can only be explained by either iris system and/or shutter speed variations.
When extreme settings were used (mainly mistakes made by noob shooter, like 2.8 wide opened iris, with a 1/4000th second exp time) the result was jittery. In such extreme case, it's obvious that shutter speed variations are cause of jitter.
At the opposite side, lens really quite closed and a long time exposure (let's say 1/10s or longer), we also get jitter. In this case, shutter speed variations can't be guilty, as I guess shutter speed jitter are quite absolute and don't depend on exposure time, so can be negligeable compared to exposition time. So it should be an iris position variation.

Years ago I tried to shoot with Nikon D70, and result was jiterry no matter what I could do. Everything locked to manual still led to quite important jitter. When analysing files, not only brightness was flickering, but also black levels, even with raw shots. This can't be caused by either iris or time variation, as black level can't be caused by overall exposure variation, but only by obscure hidden process that is not linear and predictable. So my conclusion was that Nikon DSLR were not suitable to shoot TL. At that time I've read that Nikon raw files were not really raw, because of such hidden process, but I have no more information about it...

So I think flicker can be caused by at least three different reasons. I assume shutter variation can't be fixed, but using lens with continuous iris, like the ones on broadcast or film camera could at least fix iris jitter.


Tue Sep 16, 2008 12:07 am
Profile
User avatar

Joined: Wed Jul 09, 2008 11:54 pm
Posts: 888
Location: Aars, Denmark
Post Re: About DSLR flicker
pixelbot wrote:
I have gotten flicker in full manual mode (I only shoot manual mode for my time lapse as I don't want anything to change - in a small studio - 7'x3' totally enclosed - constant light ) - I was shooting a longer shutter speed (between .25 and 1 second) because I was shooting at f22 (to get maximum DOF) so it wasn't the 60Hz frequency. you can check out my shots at http://www.artbeats.com/collections/744 all those were shot with manual lenses - some of my earlier shots were with automatic lenses and I had to use deflicker plug-ins to clean those up.

Hmmm. So even in full manual mode using an automatic lens *may* give flicker while manual lenses do not. Maybe for some lenses and low apertures diaphragm operation is inaccurate, but I don't think that's the source of flicker in the majority of cases. At least it isn't for cameras/lenses where flicker can be eliminated by using full manual. Also, it is usually when lighting varies that the exposure starts jumping around.
ridindave wrote:
Years ago I tried to shoot with Nikon D70, and result was jiterry no matter what I could do. Everything locked to manual still led to quite important jitter.

Interesting; perhaps there is significant variation from brand to brand. For my Canons (300D, 350D, 40D) changing to full manual mode eliminates flicker while the exposure will jump around if any metering is involved, even during intervals where the camera is supposedly using fixed settings - look how it is still jumping around after the exposure time has settled at 4 sec in the video in original post. Also, the lens was wide open and exposure time long, so the cause in this case isn't diaphragm or shutter inaccuracy.

_________________
Nikon D3s, D600, Canon 5D II, S3, SX100.
Feature film Beneath the Aurora on Vimeo.


Tue Sep 16, 2008 1:00 am
Profile
User avatar

Joined: Wed Jul 09, 2008 11:54 pm
Posts: 888
Location: Aars, Denmark
Post Re: About DSLR flicker
To test Shutterdrone's suggestion that in-camera jpg conversion may be smoothing exposures for compact cameras I did a new test shoot, this time with both cameras shooting RAW. The RAW sequences were then converted to jpg with ACDSee batch converter (which I assume treats the sequences equally). Also, using a much faster lens allowed the DSLR (300D in this case) to vary the aperture this time, giving more room for a smoother exposure transition. Result: Hardly any difference. The sequence from the compact is still much smoother. Take a look:



Another interesting point is that the S3 stops at 1 sec exposures again in spite of larger aperture and higher ISO. This makes me suspect that it may not be the metering that can't keep up, but that perhaps Canon implemented a 1 sec limit when using automatic settings. It is not unlike Canon to cripple their hardware by software limitations.

_________________
Nikon D3s, D600, Canon 5D II, S3, SX100.
Feature film Beneath the Aurora on Vimeo.


Tue Sep 16, 2008 3:21 am
Profile
Post Re: About DSLR flicker
Man the DSLR is bouncing all over the place. It's like the meter is "too smart"... It needs a not so smarty pants mode. This is why some sort of external meter pointing at a grey card or only metering ambient is the solution for the overly sophisticated DSLR metering system.

I'm quite impressed with that S3's consistency.


Tue Sep 16, 2008 6:56 am

Joined: Sun Jun 08, 2008 10:21 pm
Posts: 159
Location: Salt Lake City, Utah
Post Re: About DSLR flicker
Now that would be a great function to market. Just imagine a manufacturer touting their latest development in metering technology. The 'dumb' meter. You could customize and dial in the amount of 'dumb-ness' into the meter.

Hmmm........

_________________
Michael Slade
Homepage
Book Project


Tue Sep 16, 2008 2:21 pm
Profile
Post Re: About DSLR flicker
Ok, so we can check off jpeg conversion as playing any factor. However, I'm having a hard time with the assertion that stopping your lens all the way down is a good test, or that using M-mode in your camera will reduce flicker.

I'm getting noticeable flicker using m-mode on my camera body (a K10D). M-mode should produce the exact same exposure every time (although I sincerely doubt it could, given the number of factors that come into play with the much larger shutters of the dSLR vs. a compact), however it doesn't provide the _exact_ same every time. More like a reasonable approximation of a given time.

But, real quick, back to why I don't agree with stopping one's lens all the way down is a good test. It seems, on the surface to be a good test, considering that "the aperture blades have the furthest to move." However, I have concern in that the camera can overshoot as much as it can undershoot on aperture blade placement. Moving to the maximum position may always overshoot for a given lens, producing a consistent stop-down. I might even speculate that we see more over-shooting on aperture selection than under-shooting (my videos seem to flicker darker, not brighter). Given that both over-shooting (stopping down a bit too far) and under-shooting are possible, one might choose a different test - perhaps the exact mid aperture, or even better, an aperture pattern test. That is, find a good night scene with a few specular highlights (a street light or two), take a shot from the widest aperture, and then continue up in aperture until you start to see the diffraction patterns of your aperture blades most clearly (but not perfectly) in the specular highlights, find an aperture value where it's just slightly less distinct than the following one. Now, set that aperture and exposure setting, and then fire off a hundred shots or so in M-mode.

You could then crop down the image to that particular highlight, and run it at 1s per image. Watch it -- if the shape of the pattern changes _at all_, then your aperture has changed. If it doesn't change at all, you're left with a shutter issue.

Now, I get almost no flicker at exposure times >= 1.5s. But, I get lots of flicker at exposure times less than 1/100 of a second. Knowing that our dSLRs generally use a vertical metal leaf shutter mechanism, that involves two metal shutters being moved at slightly different times, yet in conjunction, we can conclude that any imperfection in this mechanism could lead to a slight exposure variation. A "perfect shutter" is also why people spend thousands of dollars to this day on old Hasselblad's =) I would bet you that variance gets minimized as you go further up the "prosumer" lines to "pro" lines. The faster you want a "perfect shutter" to move, the more you're going to have to pay for it.

!c


Tue Sep 16, 2008 6:36 pm
Post Re: About DSLR flicker
And here's a test video showing the diffraction difference between f/5.6 and f/6.7 on my Sigma 17-70mm. Notice the few occasions where more rays suddenly appear, and once it gets dangerously close to f/6.7 -- this would be a more appropriate test for aperture variation than simply looking for flicker in the final video.



!c


Tue Sep 16, 2008 7:33 pm
Post Re: About DSLR flicker
milapse wrote:
Man the DSLR is bouncing all over the place. It's like the meter is "too smart"... It needs a not so smarty pants mode. This is why some sort of external meter pointing at a grey card or only metering ambient is the solution for the overly sophisticated DSLR metering system.

I'm quite impressed with that S3's consistency.


Oddly enough, I tried a couple of sessions a few nights ago using spot and matrix metering. Unfortunately, my spot element was a dark brown incinerator, and the shot suffered horribly. When using matrix metering, there was only one section with big jumping -- and that was right at dusk, it suddenly reduced the exposure quite a bit and left it that way. Like an idiot though, I had a bright wall nearby (lit up by a floodlight), and forgot to put the viewfinder cover over. However, I did not get wildly varying exposure times - they were fairly consistent over each ten-fifteen minute time frame.

!c


Tue Sep 16, 2008 7:36 pm
User avatar

Joined: Wed Jul 09, 2008 11:54 pm
Posts: 888
Location: Aars, Denmark
Post Re: About DSLR flicker
shutterdrone wrote:
Ok, so we can check off jpeg conversion as playing any factor. However, I'm having a hard time with the assertion that stopping your lens all the way down is a good test, or that using M-mode in your camera will reduce flicker.

I did another two tests to show these things. First, here's a video with 300D in full manual and f10 vs. the 350D in aperture priority. There's a hint of flicker, but not really significant (no challenge for a deflicker filter), for the 300D in full manual. This shows both that full manual works to prevent flicker on the 300D and that variation in lens diaphragm operation isn't much of a source (this was the Canon 24mm f1.4L).



Second is a test with 350D in full manual and 300D in shutter priority. As can be seen, full manual effectively eliminates the flicker problem for the 350D as well while the 300D suffers from flicker this time. The consistency of 350D exposures shows that the Sigma 18-125mm used (which is a relatively poor lens, horrible autofocus) doesn't have a problem with diaphragm inaccuracy either.


shutterdrone wrote:
I'm getting noticeable flicker using m-mode on my camera body (a K10D).

Ridindave reported problems with Nikons in any mode too. Maybe the lack of flicker in full manual is brand specific. Can anybody else tell of their experiences in this regard?
shutterdrone wrote:
M-mode should produce the exact same exposure every time (although I sincerely doubt it could, given the number of factors that come into play with the much larger shutters of the dSLR vs. a compact), however it doesn't provide the _exact_ same every time. More like a reasonable approximation of a given time.

As Ridindave pointed out, these shutters can do 1/4000s (some 1/8000s), so the inaccuracy at the shutter speeds we're dealing with here really should be negligible. However, the physical shutter on DSLRs vs. the electronic ones on compacts may be a key. If the mechanical shutters are locked to certain values so they can only move in 1/3 ev steps, then that could explain why the exposure jumps around instead of changing smoothly. Perhaps the electronic shutters on compacts can do much finer steps. Not that there's much reason why the physical shutters shouldn't be able to, but perhaps that's just the way they are implemented. I am planning to do more tests with the compacts to look into how they work.

shutterdrone wrote:
Now, I get almost no flicker at exposure times >= 1.5s.

I do. A lot. Here's an example, shot in aperture priority mode and lens wide open.

BTW, I have a deflickered version of this on vimeo as well, but "deflickering" involved having to go through individual frames to correct exposure :evil:

_________________
Nikon D3s, D600, Canon 5D II, S3, SX100.
Feature film Beneath the Aurora on Vimeo.


Wed Sep 17, 2008 9:11 pm
Profile

Joined: Wed Aug 06, 2008 8:22 am
Posts: 52
Post Re: About DSLR flicker
That is a great video!
Anyways, I use a Ricoh GX100 compact camera for timelapse and have noticed flicker on alot of sequences where the aperture is not wide open. The shutter speed doesn't seem to matter(I always shoot full manual). I have the dreaded "dust on sensor" problem so usually I use a 64x ND filter and shoot wide open. The filter also limits the damage due to shooting into the sun.


Thu Sep 18, 2008 3:08 am
Profile
Post Re: About DSLR flicker
flyvholm wrote:
I did another two tests to show these things. First, here's a video with 300D in full manual and f10 vs. the 350D in aperture priority. There's a hint of flicker, but not really significant (no challenge for a deflicker filter), for the 300D in full manual. This shows both that full manual works to prevent flicker on the 300D and that variation in lens diaphragm operation isn't much of a source (this was the Canon 24mm f1.4L).


What do you mean by "full manual"? Obviously, relying on metering is a problem. I don't use metering in the camera for exposure timing in TL shots. I shoot in 'M' mode (i.e. I dial in aperture, iso, and shutter speed, and it uses that) and with my Sigma 17-70 that I shot for the test above, I get flicker. If I shoot with my m42 lenses, I don't get flicker. The test above showed that the flicker was caused by the aperture going smaller than it was intended to - you can tell by the increase in rays around the specular highlight.

So, I'm not convinced it has anything to do with anything other than the camera-controlled diaphragm, as I get no flicker with a full-manual diaphragm in M mode. Obviously, if you're going to rely on metering in-camera on each shot, you're going to get variations. I have some stuff to do this evening, but tomorrow I'll show you a test comparing a KA 50/2 to a Super-Takumar 50/1.2 (both, say, at f/5.6) and you'll see flicker in one and not the other.

Now, as to why your dSLR varies its meter differently than the compact; or to be more exact, it finds and adjusts for small EV changes more often - it has a better meter than the compact.

!c


Thu Sep 18, 2008 8:13 am
Post Re: About DSLR flicker
flyvholm wrote:
shutterdrone wrote:
Now, I get almost no flicker at exposure times >= 1.5s.

I do. A lot. Here's an example, shot in aperture priority mode and lens wide open.


Again, you're relying on the camera to do the metering. AV mode just lets you set the f/stop and iso, and then it picks an exposure.

You don't seem to be testing the problem. You're testing metering on your camera and saying flicker isn't caused by aperture, but you need to be in M-mode to say flicker is caused by anything other than aperture, because you have no control over exposure time in other modes. I can watch my camera hunt around exposure values in any mode other than M - that would definitely cause flickering, but the source here is the hunting meter, and would likely _combine_ with any aperture variations.

Have you run that diffraction pattern test in M-mode yet to see if your aperture is always perfectly resting on the same spot*?

* - admittedly, over-exposing a specular highlight can also result in diffraction patterns, but exposure variation enough to result in the diffraction pattern between two aperture values would have to be equivalent to the difference in exposure from two different aperture values.

!c


Thu Sep 18, 2008 8:19 am
User avatar

Joined: Wed Jul 09, 2008 11:54 pm
Posts: 888
Location: Aars, Denmark
Post Re: About DSLR flicker
bert wrote:
Anyways, I use a Ricoh GX100 compact camera for timelapse and have noticed flicker on alot of sequences where the aperture is not wide open. The shutter speed doesn't seem to matter(I always shoot full manual).

Strange. Someone in the DPReview forums reported getting "surprisingly consistent results" with that particular model. His post is here. Perhaps you could get in touch with him to see if he has any tricks to share.

Milapse, what is your experience with your Ricohs?

_________________
Nikon D3s, D600, Canon 5D II, S3, SX100.
Feature film Beneath the Aurora on Vimeo.


Thu Sep 18, 2008 1:27 pm
Profile
User avatar

Joined: Wed Jul 09, 2008 11:54 pm
Posts: 888
Location: Aars, Denmark
Post Re: About DSLR flicker
shutterdrone wrote:
What do you mean by "full manual"?

By full manual I mean the mode where you dial in aperture, exposure and ISO manually, thus locking the camera exposure (it's called "M" on the Canons too).

shutterdrone wrote:
Obviously, relying on metering is a problem.

The videos I've posted show that on a cheap compact you CAN rely on the metering system - it gives consistent exposures with smooth variations from frame to frame on automatic settings. Then, why is it a problem on DSLRs? That's the question...

shutterdrone wrote:
So, I'm not convinced it has anything to do with anything other than the camera-controlled diaphragm, as I get no flicker with a full-manual diaphragm in M mode.

And you are not alone. One of the reasons I started the thread was to disprove this. Note, I am not saying that diaphragms are never at fault; I can only say that in the tests I have done so far, the diaphragm has proven to be sufficiently accurate (that is, accurate enough that I can shoot a time lapse with hardly any flicker as my last two test videos show). On the other hand, for the exact same DSLRs/lenses/apertures where I have verified that the diaphragm is sufficiently accurate I still get flicker if the camera is not in full manual mode, showing that there is another significant source of flicker - specifically, a source that is absent on my compacts which do fine on automatic settings. That's the source I am trying to track down.

When you get flicker on the K10D in M mode with automatic, but not manual lenses, it does suggest that the lens diaphragm is the problem. That's what I would conclude too. Have you experienced getting flicker with automatic lenses even at max aperture and the camera in M mode?

shutterdrone wrote:
Now, as to why your dSLR varies its meter differently than the compact; or to be more exact, it finds and adjusts for small EV changes more often - it has a better meter than the compact.

Maybe you are right that the meter on the DSLR is more sensitive/accurate. That begs the question why the DSLR makes big jumps in exposures when only a small adjustment is needed. The compact is capable of making smoother adjustments which could come down to much finer steps in shutter speeds. I haven't been able to verify this by exif info, but I have another test in mind to see if it's the case - I will return with this later. :-)

shutterdrone wrote:
Have you run that diffraction pattern test in M-mode yet to see if your aperture is always perfectly resting on the same spot*?

I am quite sure that diaphragms are never 100% accurate. :-) What I am interested in testing is whether they are sufficiently accurate that they are not a significant source of time lapse flicker. The diffraction pattern test may reveal fine variations, but as long as the overall image brightness doesn't change noticeably it isn't much of a problem.

_________________
Nikon D3s, D600, Canon 5D II, S3, SX100.
Feature film Beneath the Aurora on Vimeo.


Thu Sep 18, 2008 7:11 pm
Profile
User avatar

Joined: Wed Jul 09, 2008 11:54 pm
Posts: 888
Location: Aars, Denmark
Post Yet another test
Ok, here's a test of the theory that the electronic shutters on my compacts use smaller than 1/3 ev steps (even though they only write 1/3 ev steps to exif). I took the SX100 in P mode, letting it step exposure times automatically. Next to it I had my S3 in full manual, and then I manually adjusted the exposure time in 1/3 ev steps along the way.



As can be seen, the 1/3 ev steps are quite visible, so these compacts must be using significantly smaller steps in shutter speed to achieve the much smoother transitions on automatic settings. This means that neither the compacts nor DSLRs write the actual exposure settings to exif which is pretty annoying. :evil: It makes it really difficult to figure out the inner workings of these cameras - which may just be the purpose.

So, if physical shutters on DSLRs can only do 1/3 ev steps it is no wonder they exhibit flicker as soon as the lighting varies. The sad thing is, I see no reason why the physical shutters should be constrained to 1/3 ev steps when electronic shutters on cheap compacts allow much finer gradations. You'd think DSLR owners would appreciate accurate exposures more so than compact owners...

Michael Slade, I am surprised by your American Flag time lapse. I assume you are fading by manually going down in 1/3 ev steps; no post processing applied? I find it strange that your 1/3 ev steps appear much smoother than they do in my test above. How many frames are in the video for each 1/3 ev step?

_________________
Nikon D3s, D600, Canon 5D II, S3, SX100.
Feature film Beneath the Aurora on Vimeo.


Thu Sep 18, 2008 11:25 pm
Profile
Post Re: About DSLR flicker
flyvholm wrote:
Milapse, what is your experience with your Ricohs?


I've seen very very slight variations to none at all with my GRs (manual). Very easy to take out in post. I've yet to try this camera in full auto to see how well it performs.... Seeing how your S3 seems to go below 1/3eV, I'll give it a try this weekend and see if they can pull it off too. This doesn't give me much confidence though

One thing that has been yet to be discussed here is that flicker is not only born 'in camera' it's also a result of the fact that the illumination levels in a scene can vary quite a bit from frame to frame... It's hard to perceive though our own eyes but haze/clouds/variations in atmospheric can let more light in at one moment and less at the next. It can 'look' consistent to us because as it changes our eyes/brains are fluidly adjusting to it... A camera on manual can 'reveal' these variations as 'steps'... Know what I mean?


Fri Sep 19, 2008 4:28 am

Joined: Mon Sep 08, 2008 1:24 pm
Posts: 6
Location: Worcester, UK
Post Re: About DSLR flicker
My timer remote finally came today so after weeks of reading posts I can finally make a reply. I've been trying a few very short timelapses in Manual mode this afternoon just to see how changes affect the amount of flicker this is what I've found.
Canon 30D, 50mm f/1.8, 1/250 @ f/10, Cloudy WB, manual focus - flicker is worse than the 300D and 350D in manual from the earlier posts but just about bearable.
All the following had no effect.
Changing to a Kelvin WB - I wondered if Cloudy WB might still be slightly adaptive.
Changing to leaving the lens wide open.
Changing to a L lens (28-300), open or f/10
Covering the viewfinder - though I was shooting out the window so behind camera illumination was quite constant anyway.

So I've found nothing but learnt a bit.

One thing I did discover (that might help with these program mode tests) is in the exif there is a Measured EV value that is recorded even when in manual and it does vary. It is in the Canon Maker Note and the only tool I have that will read it is the Firefox Exif Viewer. It might be worth adding that to a video to see if it corresponds with the flicker as that might help show if it is a metering/electronic issue or a mechanical one.

For the shots I took that value was normally 292 but some shots show it between 280 and 296 (it seems to go up in steps of 4) so it looks like it might be a finer measure than using the 1/3 stop exposure info - the flicker I saw was definatel less than 1/3rd stop.


Fri Sep 19, 2008 9:46 am
Profile
User avatar

Joined: Wed Jul 09, 2008 11:54 pm
Posts: 888
Location: Aars, Denmark
Post Re: About DSLR flicker
milapse wrote:
I've seen very very slight variations to none at all with my GRs (manual). Very easy to take out in post. I've yet to try this camera in full auto to see how well it performs.... Seeing how your S3 seems to go below 1/3eV, I'll give it a try this weekend and see if they can pull it off too. This doesn't give me much confidence though

Yeah, looks like they are not doing any better than DSLRs. Perhaps the graduated shutter speed is unique to these Canon compacts. Add CHDK and it makes them very attractive for time lapse shooting.

ZiPP wrote:
One thing I did discover (that might help with these program mode tests) is in the exif there is a Measured EV value that is recorded even when in manual and it does vary.

That helps a lot, thank you very much! This prompted me to dig deeper with the exif data, and I have now SEEN the measured EV data and even the exact exposure times that verify the much smaller steps! It's there... I can see the values using a GUI for Exiftool, but I'll be darned if I can make Exiftool print them. I've spent hours now trying Exiftool in different ways, via command line or Perl, feeding it with different options and trying to tell it in different ways exactly the tag I want... and I just can't get to the MeasuredEV tag. Anybody in here familar with Exiftool/Perl?

This is really useful. If the exif info I'm finding here is as reliable as it appears it will tell a whole lot more about what's going on.

_________________
Nikon D3s, D600, Canon 5D II, S3, SX100.
Feature film Beneath the Aurora on Vimeo.


Fri Sep 19, 2008 3:10 pm
Profile
Post Re: About DSLR flicker
flyvholm wrote:
I've spent hours now trying Exiftool in different ways, via command line or Perl, feeding it with different options and trying to tell it in different ways exactly the tag I want... and I just can't get to the MeasuredEV tag. Anybody in here familar with Exiftool/Perl?


I'm familiar with perl (been coding in perl for 13 years =) but not exiftool, did a quickscan of the docs, did you try this?:

Code:
#!/usr/bin/perl

use Image::ExifTool;
use Data::Dumper;

my $file = shift;

if( ! defined($file) ) {
   die("Usage: $0 <file>");
}

my $ex_obj = Image::ExifTool->new();

$ex_obj->Options( 'Unknown' => 2);

my $info_hRef = $ex_obj->ImageInfo($file);

print Dumper($info_hRef);



That should dump every single tag/value pair from the file, no matter if it occurs in the header or elsewhere. (according to the docs, I don't have exiftool installed locally to test it.)

!c


Sat Sep 20, 2008 6:12 am

Joined: Sun Jun 08, 2008 10:21 pm
Posts: 159
Location: Salt Lake City, Utah
Post Re: Yet another test
flyvholm wrote:
Michael Slade, I am surprised by your American Flag time lapse. I assume you are fading by manually going down in 1/3 ev steps; no post processing applied? I find it strange that your 1/3 ev steps appear much smoother than they do in my test above. How many frames are in the video for each 1/3 ev step?


The 'fade' in that test is from the sun going down and the physical light level dimming. I didn't alter the exposure at all from start to finish. No post other than stretching the blacks out and a slight color-balance in Lightroom.

I'll do a quick test today...shifting light levels, lots of clouds moving in and out, I'll shoot on fully manual, lens wide open, constant shutterspeed.

We'll see...

_________________
Michael Slade
Homepage
Book Project


Sat Sep 20, 2008 7:31 am
Profile
Site Admin
User avatar

Joined: Thu Mar 27, 2008 2:15 pm
Posts: 1696
Post Re: About DSLR flicker
flyvholm wrote:

shutterdrone wrote:
Now, I get almost no flicker at exposure times >= 1.5s.


I do. A lot. Here's an example, shot in aperture priority mode and lens wide open.


BTW, I have a deflickered version of this on vimeo as well, but "deflickering" involved having to go through individual frames to correct exposure :evil:


But why are you shooting at night with any type of auto settings? You should be full manual.


Mon Sep 22, 2008 10:51 am
Profile
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 54 posts ]  Go to page 1, 2  Next


Who is online

Users browsing this forum: No registered users and 3 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
cron
Powered by phpBB © phpBB Group.
Designed by Vjacheslav Trushkin for Free Forums/DivisionCore. pozycjonowanie