A friend of mine just showed me some interior time lapse footage he had shot that has a rather unusual problem that I vaguely recall hearing about before and that the wizards, gurus and sages of CML sghouldbe intriged by.
The footage is an interior daylight time lapse of a library with windows on both sides of the scenne. The shot starts in pre-dawn darkness, time-lapses through full sunlight and ends at sunset. The shot is a rock steady lock off.
The problem is that in the upper right quadrant of the frame a slight moving ghost image appears with the break of day. It looks like a double exposed secondary image exposed over the library scene. The ghost image occupies about 5% of the image area. This ghost image oscillates regularly up and down over an approximately 8 frame cycle. It is definitely not a light leak and, although it looks like a flare, it is not.
At the TK session, they repositioned the ghost image and zoomed in it. It looks very much like a partial image of either the gate or the ground glass -- more exactly it looks like the lower corner of one frame and the upper corner of the next frame, including the frame line.
The camera was an Arri 35 III, standard shutter, standard door -- no video tap, 14mm Zeiss, no filter, eyepiece capped. Norris intervalometer, 3 second interval between frames, exposure 1/16 sec @ f4.0, Kodak 5245.
I seem to recall that time-lapse folks recommend removing the ground glass in certain cases. Can this problem be due to leaving the ground glass in. Does it have somethingto do with the ground glass light baffles?
Brian "memory lapsing on time lapsing and very glad that CML is back" Heller
IA 600 DP
>I seem to recall that time-lapse folks recommend removing the ground glass in >certain cases. Can this problem be due to leaving the ground glass in. Does it have >somethingto do with the ground glass light baffles?
I have seen similar problems with Arri 3's when the baffles are not installed.
Your friend's time-lapse problem could come from several sources:
#1 was there a capping shutter used? The rule of thumb is that a capping shutter SHOULD be used on time-lapse when the duration between frames [dwell time] is 5 seconds or longer. But the way you described the shot, the sun seems to be pouring into the front of the lens. With an Arri 35-3, with it's mirrored shutter, that light simply bounces all around the internal cavity of the camera, possibly creating that ghost image, more pronounced when the sun is entering the lens port, but diminishing when the sun is out of range.
#2 Maybe the mirror wasn't exactly in position when the motor was engaged onto the camera's inching knob. I've had crews rent my Norris system and call me asking how to do this..it seems not everyone is aware of how utterly important the mirror positioning is when engaging the motor.
Jeff Barklage, s.o.c.
>A friend of mine just showed me some interior time lapse footage he had shot that >has a rather unusual problem that I vaguely recall hearing about before and that the >wizards, gurus and sages of CML should be intriged by.
Well you know with time lapse it is a matter of individuals. By that I mean there are camera types and then there are the individual examples of those types which tend to have subtle differences. My reflex Mitchell used to exhibit an artifact something like the one you mention in certain situations. The rotating mirror is not a 2 dimensional object, but rather has some thickness. Until I painted the edge of the mirror with some black paint I had to live with this artifact cropping up, especially when there were some very strong highlights in top or bottom of frame, though usually top.
There may be some other idiosyncrasy of that camera system that could produce this artifact and maybe some body else can pile on here. Always keep in mind that folks who design MP cameras always think of film flowing through the gate, they never think of it like a still camera with a huge mag.
Except for rack over Mitchells that is.
>The problem is that in the upper right quadrant of the frame a slight moving ghost >image appears with the break of day
Contamination from internal reflections off the mirror? Sounds like what you get when you don't have a capping shutter, but I am by no means a timelapse maven.
Mark H. Weingartner
VFX, Photography & Lighting for Motion Pictures & Television
An Arri III with does not have a light tight shutter. Period. Though it may be possible to shoot time-lapse with a specific Arri III on a given day under particular circumstances, it is tempting fate. It's impossible to predict how long an interval between frames is acceptable without testing. There is no "rule of thumb", other than possible to say that if you roll continuously at the minimum camera speed (3 fps, I believe), you will likely not encounter a problem. So, SOP for shooting time-lapse with a camera with such a shutter (the Fries Mitchell with the spinning mirror shutter included) is to use a capping shutter in conjunction with the Norris motor. Norris makes a nice capping shutter that fits right onto the Arri III 15mm mattebox rods in front of the taking lens and plugs into the motor. The motor triggers the shutter so it is well open and clear before the camera shutter rolls over whatever number of frame specified, and then closes the capping shutter. It's a simple solution. Check with Norris about this capping shutter, and most rental houses who supply the motor should also have the capping shutter.
The only cameras with truly light-tight shutters are those with focal-plane shutters like Panaflexes, or Mitchells, or Bell&Howell 2709's (did I get that number right Jeff?), animation/optical cameras like Oxberry and Acme (that I know of). Care must be take when using a Mitchell S35R, because it's possible to get light leaks through the pulldown claws and registration pins, unless proper baffles are in the camera.
Another contribution to the problem is that it sounds like the camera shutter was not squarely and fully closed in-between frames, resulting in the problem occurring in the area and shape you indicated. The causes of the shutter not being squarely closed range from a mis-connection of the Norris's drive coupling (either internal to the Norris, or in its connection to the Arri's drive shaft) to an overly tight film magazine clutch causing the film to "pull" the movement slightly past it's proper end position.
Now, there is something else in your post that leaps out at me.... You mention an exposure time of 1/16 second. Generally, this time is not used as an actual exposure time, due to the tremendous acceleration of the movement necessary. The inertia of the camera movement can cause an inconsistancy in the exposure time. An error at that brief exposure can be considerable, in terms of exposure to the film. For example, a 1/64th second error is a 1/4 stop at that rate, and that is definitely noticeable as flicker. The same error at a 1 second exposure would be a 64th of a stop error, and insignificant. In my experience, the 1/16th second setting is usually only used as a "half/stop" adjustment of exposure times, like to shoot between 1/8 and 1/4 second. You would flip the 1/8 and 1/16 dip switches to get an exposure time of 3/16. The only time I've know people to get consistently successful results at the 1/16 setting was with old rack-over Mitchells which had their flywheels removed to lessen the inertia of the movement.
Don "been there, done that" Canfield
Gear+Rose Motion Control
Guess we've established single-frame with mirror shutters is dicey.
If you don't need to stop motion (to produce crispy, pixalated people in the shot as opposed to blurry, ghost-like people) much better to stick on an ND and set a half-second exposure or longer. No reciprocity failure (exposure correction) up to one second. Eastman says the correction for 5246, the high speed, is 2/3 of a stop at ten seconds, no info on 45.
A no-brainer is to snag an "n-twelve" (ND 1.2), meter the shot at sound speed, and set 1/4 second, or 3/8 if you can do it. Since the mirror-shutter light leaks seem to show up around three or four seconds at 1/15, it seems you should keep the "closed" intervals shorter than 1:50 of the exposure. Ergo, 1/4 sec. should be okay as slow as three frames a minute.
Ten second timelapse (shutter open ten secs, closed one) of moving traffic at dusk or night is very tasty, very client-impressive; always try to grab one whenever I get my hands on an intervelometer... kind of ties up the camera for a while.
>I have seen similar problems with Arri 3's when the baffles are not installed.
I'm thinking it may have something to do with the baffles. I'm checking to see if they were still in place on the camera that was used for the shot, or if they had been removed.
Jeff Barklage wrote :
>Your friend's time-lapse problem could come from several sources:
>#1 was there a capping shutter used?
No capping shutter, but it was not necessary.
>The rule of thumb is that a capping shutter SHOULD be used on time-lapse when >the duration between frames [dwell time] is 5 seconds or longer.
The duration between frames in this case was 3 seconds.
>But the way you described the shot, the sun seems to be pouring into the front of the >lens.
Perhaps I didn't describe the shot clearly enough. The windows are on the sides of the scene/frame. There was no direct sunlight entering the lens at any time.
>#2 Maybe the mirror wasn't exactly in position when the motor was engaged onto the >camera's inching knob.
The intervalometer was installed correctly. However, I do think it the problem has something to do with the mirror. The ghost image cycles up and down slightly, and the mirror is the only reflective thing moving that could account for this movement. As I have said, the principal image is rock steady.
Further analysis of the shot and conversations with the shooter revealed that he changed the battery midway through the shot, i.e., after 6 hours.
Immediately after the battery change, the ghost image jumps to a new vertical position slightly below the original and then gradually returns to its previous position. We are hypothesizing that the slight change in voltage caused the mirror stopping point to move a little further when opening.
Mark Weingartner wrote :
>Contamination from internal reflections off the mirror?
That's what I think, because of the very slight variation in the position of the stopped mirror in the open postion might account for the movement of the ghost image. The scene itself is rock steady. (see above reply).
>Sounds like what you get when you don't have a capping shutter, but I am by no >means a timelapse maven.
Nor am I, but this problem is beginning to bug me.
Lack of a capping shutter in cases where one is needed would be seen as a light leak around the stopped shutter in the closed position. However, in this case, it is an image definitely photograhed onto the negative within the gate area, so it must have happened when the shutter was open, and when this ghost image was enlarged on the TK, the image of what appears to be the corners of two frames and the frame line is clearly visible. At no point does the ghost image extend beyond the gate.
I have spoken with Dan Norris (the inventor of the Norris Intervalometer) who recalls similar incidents from the past. However, he cannot remember the exact details as to the cause, but believes it to be a reflection from the back side of the ground glass. When feasable he recommends removing the ground glass.
Does this ring a bell with anyone?
Brian "baffled again" Heller
IA 600 DP
I'd agree that it's probably light kicking off the mirror. Of course, the source of light may be out-of-frame, but it would still be available unless the lens is very well masked off.
I had a similar problem with an optical printer at a lab I won't name -- a contrasty frameline (i.e. dark at the bottom of 16mm frame, bright at the top) caused out-of-frame bright light to kick off the edge of the aperture itself. Only happened in some scenes, and certain lab personnel claimed it didn't exist.
I think Brian's got it with it being a kick off the mirror, given the fluctuation. Was the lens well-matted? It's quite possible that it was a compound bounce involving both the groundglass and the shutter edge That'd be my guess. Have you played around with the camera and a flashlight?
Jeff "gonna invent a black groundglass (matte finish!) for just this problem" Kreines
>Double image flicker on timelapse shot in Library
Its an interesting problem.
In the gate of the Arri 3, the upper right corner of the image is the lowerleft corner of the gate. The image of the upper windows of the library should be upside down in the lower left portion of the gate. If memory serves me, the Arri 3 shutter moves from the upper right corner of the gate to the lower left corner of the gate, where it parks when closed. The ground glass is in the left side of the gate.
What I find interesting is the appearance of a "frame line" in the ghost image. That's a tough one, but one that should make it easy to find the bug. When the mirror parks over the gate, the film is part way through the pull down cycle, since the Norris may not be parking the mirror in the exact spot each time, it might be the source of the movement.
Here's what I think: When the Norris parks for the 3 second interval, it parks in such a position that the film image is almost completed its pull down, but not quite. The "frame line" you see is a frameline created by the film still pulling down. The light from the Library windows is kicking off the ground glass behind the mirror just in the one portion of the film frame onto the almost parked film, which parks completely as soon as the Norris takes the next picture.
I suggest you pull the pressure plate and shine a flash lite into the lens in front of the camera, move it around in the upper right hand portion of the image, put a dark bag/cloth over the camera and let your eyes ajust to the dark. You will be frightened by how much light you see kicking around, and with the shutter closed, you should see if the ground glass has the ability to kick light past the mirror. If you want to see an image, take a piece of clear leader and sand it with 220 grit sand paper and thread it into the gate with the sanded surface pointed toward the lens. With a Dentist's mirror and 7x Lupe, and the shutter closed, you should see where the image is forming on the film in the parked position.
In most cases, it is not advised to do intervals longer than 5 seconds on cameras that do not have shutters unless you use a capping shutter. And this varies from camera to camera. The Norris has a 3 second minimum interval with his capping shutter, so anything up to 3 second interval you have to do without. Three seconds is a border line case, which would indicate a light leak test prior to shooting on any camera without a separate shutter, or by doing the above to see how bad a mirror leaks light.
Are you 100% sure it was an Arri 35 III ? (The original message never got to me because of Lyris problems). This is exactly the problem that John Duclos has described to me as being from the 2C.
Focus Puller and Listpig
Yes, it definitely was an Arri III, standard shutter not variable, standard door, not video.
IA 600 DP
Most of the remedies have already been discussed, including removing the ground glass. This I don't recommend. It is better to black out the end of the ground glass facing the film chamber with a black marking pen. Removing the glass leaves a great big hole (amongst the many other gaps) for light to travel into the film chamber. I remember being very puzzled by the apparent 'ghost frameline' when I first used my Norris kit, until I realised it was the only part of the frame not fogged.
I recently posted a message about a problem a friend of mine had with a time lapse shoot :
The problem was that in the upper right quadrant of the frame a slight moving ghost image appeared with the break of day. It looks like a double exposed secondary image exposed over the library scene. The ghost image occupies about 5% of the image area. This ghost image oscillates regularly up and down over an approximately 8 frame cycle. It is definitely not a light leak and, although it looks like a flare, it is not.
To all who responded, Thank you very much.
It turns out that the shooter was able to duplicate the problem using the same set-up, and was able to eliminate the problem by removing the ground glass. The ghost image was definitely a reflection of the ground glass.
It also turns out that the spurious reflection could be easily eliminated in the transfer process by means of the recursive filter function on the Spirit machine, since the shot was a lock off. Ain't technology wonderful.
IA 600 DP