film -> video device

[johnmeyer]

A stepper motor would be cool, but a lot of work to engineer and install. Also, if driven through belts, the mechanism would still slip.

As for " source has a varying frame rate, because is was shot with crude equipment by today's standards ..." what I am doing doesn't matter one bit whether the film was shot at 16, 18, 24, or whether it varies over time. I don't care if it was hand-cranked. Nothing in my process depends on the original frame rate. That can be adjusted later on by whatever pulldown I decide to apply. All I care about is getting one film frame onto each frame of video. When played back, without pulldown, it will look really clean, but also playback way too fast. I then apply the proper pulldown and the speed is then correct. If there is sound, that can be stretched to fit whatever pulldown I use.

[videoFred]

No need for a stepper motor, but we need an electronic regulated projector, of cource with a DC motor.
And feedback from the main axle.. the shutter is fine for this.

The sync system is available here:
http://www.laendchen.de/

It works with the time pulses from the camera, these pulses are regulating the speed of the projector. This system works only with certain electronic speed projectors like Bauers etc.

I do not have it, so I can not judge it.
Some people say it works perfect, others are complaining.....

Of cource, with this system, the result is interlaced and with a pulldown pattern on it.... And the play speed is fixed to 16,...fps. (frame rate is 25fps, of cource)

But back to Johnmeyers system:
In my case, with my 15fps machine cam, I must capture at a lower speed, right? To be sure to have all frames captured at least one time. Then remove eventual duplicates and/or blurred frames with Avisynth, right?

Fred.

[johnmeyer]

I still feel like I am sometimes talking at cross-purposes when it comes to the two independent subjects of capture speed and playback speed.

The capture speed depends on the technique you use. You can capture at full speed (16, 18, 24 fps) if you just point the video camera at the projector screen and take what comes. Some adjacent frames are averaged, but with a 1/60 (NTSC) shutter speed, you minimize the flicker. In this process, the shutter speed needs to be exactly equal to the field duration.

However, if you use a capture technique that results in one individual frame of film into one video frame, then there is no overlap, no flicker, and the result is very sharp. If you play this back using standard NTSC equipment, without doing anything else, then it will play back too fast by a factor of 30/15 or 30/18 or 30/24 (OK 29.97/15 etc.). However, if you want to play it back on your computer, all you have to do is set the frame rate flag in the AVI file, and it will playback at exactly the correct speed. Even if the projector speed varied from 1 fps to 24 fps and back again during the capture process, as long as your capture scheme captures each frame of film to one frame of video, and doesn't miss any and doesn't duplicate any, then the playback speed will be PERFECT, controlled entirely by the clock in your PC.

Since the video is progressive (regardless of whether the file header says interlaced or progressive, the actual material is progressive because there is no temporal displacement between the upper and lower field, if you captured one frame of film to each frame of video), it will look FABULOUS when played on your computer. Exactly the correct speed.

So, whether you captured the film at one frame a second, or at 24 frames a second, none of that matters during playback. The only thing that matters is the playback flag, and that is easy to set, using either AVISynth, or various header patch utilities (to change it after the fact).

[videoFred]

Hello Johnmeyer again,

I know all this:D

I capture for two years with the mouse-microswitch, just like the Workprinter works... But with my 1024x768 cam, capture speed is very low: 4fps.

I was asking about recording straight in AVI modus.
Then my camera records whatever it see at 15fps.

As far as I can see, the only good way to work like this, is by using an industrial servo-motor with a special power supply. Because these motors are giving feedback, it is possible to let them run exactly at a given speed.

Fred.

[johnmeyer]

I was asking about recording straight in AVI modus.
Then my camera records whatever it see at 15fps.
This thinking is a common mistake. Everyone assumes that because 15 is exactly half of 30 (the nominal NTSC video frame rate) that you will somehow get a better capture than if the projector is running at 16, 18, or 24 fps. It really doesn't matter. The reason is that, no matter how precisely you can control the speed of the projector, it will never be exactly the speed you think it is. It will be off by a few fractions of a percent. In addition, there is no way to sync the video camera to the projector, so you may be capturing exactly two video frames for every frame of film, but most likely you will get one video frame that has exactly one frame of film, followed by the next video frame that has that same frame of film, blended with the next film frame. I fail to see the advantage of running the projector at less than its rated speed (with the slight exception noted below). You are still going to capture blended frames.

If you want to capture video from a projector, without using the "Workprinter" technique (which DOES sync the projector and camera, as you know, since your device does this as well), or without using the technique that I have developed (which achieves the sync using an adaptation of IVTC software in post production), you simply set the shutter speed of the camera so that it EXACTLY equals the field rate of the camera. For NTSC, this is 1/60 of a second (1/30 of a second per frame, two fields per frame --> 1/60 of a second per field). This way, the shutter is "open" (the shutter is actually electronic) for the longest time possible. This lets the camera capture an average of the video from the projector.

Now, this is the part where all sorts of confusion exists. People have often recommended using a specially modified projector that has a five-bladed shutter, rather than a three-bladed shutter, in order to reduce flicker. This is true, but it has nothing to do with the original reason for adding more blades to the shutter. The original reason was to INCREASE the flicker rate from 18 or 24 fps, which caused headaches to two, three, four, or five times that number. This was done for the sole reason of reducing the perception of flicker when watching a movie in a theater. However, the extra blades have a completely different positive effect when capturing video using a 1/60 video camera second shutter.

If you have very few blades, you have to draw out the timing diagram, but what happens is that you can have some captures where the exposure includes the entire time the shutter is closed, with the remaining time being the exposure of the next frame of film. The actual exposure of that film will be much darker, because the total light hitting the sensor is accumulated during the time the video camera shutter is open. If, during most of that time, there is no light, and then, for just a short moment, the video camera sees the frame of film, that frame will be quite dark. However, the next frame of film might not have the projector shutter closed at all, which will result in a much brighter image. This light/dark image change results in pretty nasty flicker. However, if you had a theoretical (doesn't actually exist) 100-bladed projector, then almost all the captures would have 30 or 40 shutter closures. Some might have 35, others would have 37. The percentage difference in exposure would be imperceptible. For those familiar with the term, the percentage difference in "duty cycle" becomes far less as you add more blades to the shutter.

While 100-blades is impractical, 3- and 5-bladed shutters are fairly easy to find, and these greatly reduce the difference in exposure from one frame to the next.

You can remove some flicker using software in post production. Also, if you can vary the speed of the projector slightly so that the "beat" pattern of when the exposure is the greatest and when it is the least varies slowly over time, rather than repeating every few frames.

So, a 15 fps projector isn't going to help much, and in fact is going to be a pain in the neck, because you are going to have to adjust the speed in post production, and unlike the Workprinter captures where you are dealing with discrete film frames and therefore can use standard pulldown to change speed, the results here are going to be not so good, because your original captured frames are sometimes going to be blends of adjacent frames.

Finally, if you do capture discrete frames, here is a pulldown script. I tried over half a dozen scripts, some using doubleweave, and some using changefps. The all produce the same thing in the end. I like SelectEvery, because it lets me see exactly what I'm doing.
AVISource("d:\frameserver.avi")
AssumeFrameBased

separatefields()

# Pulldown for 24 fps
# SelectEvery(8, 0,1, 2,3, 2,5, 4,7, 6,7)

# Pulldown for 18 fps using: normal-repeat-normal-weave-normal
# Might be a little sharper, but also a little "jerkier"
#selectevery(6, 0,1, 0,1, 2,3, 2,5, 4,5)

# Pulldown for 18 fps using: normal-weave-normal-weave-normal
SelectEvery(6, 0,1, 0,3, 2,3, 2,5, 4,5)

weave()
AssumeFPS(29.97, true)

[dokworm]

I think we all need to clear ourselves up.
Videofred and I understand everything you said, we both know that the speed of the original film is irrelevant, we want to end up with 1 frame of film on one frame of video (or computer image)

Fred has a camera that can *only* grab 15fps in progressive scan.
I think his question was how can he use that camera to utilise your method. i.e. How he might be able to capture faster than his current system which tops out at about 4fps.

He uses the workprinter method where the projector triggers the camera to grab a frame, so ends up with exactly one frame in the computer for each frame of film.

This is also what I currently do.

I had tried realtime methods but the film movement between fields caused me headaches, but you seem to have solved this.

My projector has a drive shaft that protrudes through the back of the projector, so it easy to attach an external motor.

Stepper motors are very easy to control, even if you have no electronics skills, you can buy a kit to interface to your PC for less than $50.

I figured if I had the *camera* trigger the projector stepper motor then I would end up driving the projector at exactly the speed of the camera, i.e. one frame of film per video frame (25fps for PAL).

Then you could bypass any need for frame culling later.

Also if you remove the shutter on the projector, and pick the right shutter speed on the camera, you should get only the clean unmoving film frame, and no flicker/exposure issues. (If using manual exposure settings)

[videoFred]

Fred has a camera that can *only* grab 15fps in progressive scan.

You got it ;)

Fred.

[johnmeyer]

Thanks, I understand now. I was stuck thinking he was using an NTSC DV camera for capture.

[videoFred]

Now, about the stepper motor:
A stepper motor does not give feedback.
If it is missing a few steps, you have a problem.

A servo motor gives feedback: you can use this feedback to fine adjust the real speed it is running on.

But I do not know how accurate this all can be.

There is also the frame changing problem. Suppose the projector is running at 100% the frame rate of your camera. Then suppose the camera is capturing a frame every time at the moment it changes... Then you get a complete blurred video file.

So we need some kind of regulation for this, too.

I think I'm gonna try the 640x480 modus from my 1024x768 camera. Then I might be able to capture frame by frame at higher speeds, maybe 8fps or something.

My Avisynth scripts will run faster with 640x480 files, too, and I will get much smaller files.

Fred.

[dokworm]

The accuracy isn't a problem if you remove the shutter. The film is stationary for a long time, and only moving for a short time, also if using the timing pulse from the camera itself, then you will not have it take an image of moving film, only when the film is still.

The feedback from the stepper isn't really a problem when rotating them in 360 degree increments.

I already built this exact setup with a stepper motor synchronised to a video camera for a 3D TV project many years ago, so I know it works perfectly re synchronisation.

I just need now to get a bigger motor and build it into the projector.

[johnmeyer]

also if using the timing pulse from the camera itself, then you will not have it take an image of moving film, only when the film is still.
I can see how this would work if you can get a timing pulse from the camera. But, how do you get such a pulse? Hmmm ... I guess one could make a simple circuit that would trigger off one of the sync signals in the composite video and then output that. Is that what you are planning to do?

[videoFred]

I already built this exact setup with a stepper motor synchronised to a video camera for a 3D TV project many years ago, so I know it works perfectly re synchronisation.

I just need now to get a bigger motor and build it into the projector.

This is very very interesting!
Could this work with my fire wire machine camera, too?
This kind of cam is using another protocol....

Fred.

[dokworm]

Don't know, you would need to be able to get a sync pulse from it somewhere.