Video recording rates for 32-line NBTV

[gary]

Slow_Ralph.wmv

Unfortunately, my AVIs coming out of VirtualDub and not readable by the Windows video player. Any ideas Gary?

Ralph

I am sure I can help there Ralph I just need some more information like the format of the original material, what video settings you have made, etc.

It might be better if you could send me (PM or upload here) the original material, and the avi that doesn't play. Given that I can soon sort it out (famous last words ).

I can then work out the process you need.

[Ralph]

Gary,

I have already learned a few things about the AVI file issues. If I do things like change the frame rate in VirtualDub, the AVI files produced are completely compatible with the Windows reader, just like the original files. However, if I CROP the image in VD, the resulting AVI file cannot be read! I am going to create a VERY short AVI and then crop it. The two files will be small, by the standards of video files, and I will post then in a PM. You may not be able to spot the problem, but you are far more likely to than I would be!

Ralph

[Klaas Robers]

Ralph, you write that the original file was shot at 15 fps. That cannot be possible ! If the exposure time was 1/13 or 1/11 sec., then the frame rate MUST have been lower than 13 resp. 11 fps.

These speeds can only be true if the processing time could have been zero, which is not possible. The fact that the white band moves so slowly to the right prooves that the frame speed has been close to 6.25 fps. If the speed was exactly 6.25 fps. the band would have been not moving. Yes?

And because the black band was moving faster (to the right) will say that the frame rate was higher than the frame rate of the white band. If it was the same the band should have been moving with the same speed. Yes?

So with that camera the frame rate is depending of the exposure time. I guess that the processing time is the same, so
Nipkow disc: revol'tn time = 1/12.5 sec. = 80 msec.:
white band: exposure time = 1/11 sec. = 91 msec.: overlap = 11 msec = 4.4 line (this is true, I counted them on the screeen)
black band: exposure time = 1/13 sec. = 77 msec.: missing = 3 msec = 1.2 line (this is not true, I counted 2+ on the screen)

Unhappily I cannot measure frame by frame, otherwise I could measure the displacement on the screen and could recalculate the real frames per second. This MUST be lower, much lower than 15 fps.

[gary]

Hi Ralph the output video was missing from your email.

Here is the input video arbitrarily cropped - can you check that you can play it with WMP? If so I will just go through what I did.

(it would still be a good idea to send me the cropped video that doesn't play so I can analyse the video format (I'd say that is the problem BTW - the video format).

Cheers.

(oh and I should have said that my cropped video plays on WMP on my computer)

[Ralph]

Ralph04W.wmv

(5.64 MiB) Downloaded 537 times

Klaas, you are the first in line so I will start with your comments. I thoroughly understand the points you are making. All I can do at this point is state the exposure times provided by the camera control software. Your measurements would suggest that both are very close to the claimed values. As far as frame rate is concerned, I am reporting the values indicated by the VirtualDub software. If you are puzzled by the previous videos, this one will really have you going back to your calculator.

For this video, I took the entirety of the original .WAV file, displayed it on the 32-line Nipkow monitor, and imaged with same monochrome CCD camera. What is different this time is that I used the capture function of VirtualDub to minimize the number of programs used. I captured and cropped the video in a single operation. I took a lot of time to set-up the camera and this is certainly the best-looking version to date. Because I was using VirtualDub with the camera, I could rerecord the audio track as I was making the video. So the result is both sound and video.

Now here are the numbers that you won't like:

(1) The video exposure rate was set at 1/13 s. and the width of the under-scan certainly suggests that is reasonably accurate.

(2) Original recording frame rate was 30 fps! This frame rate was set by the VirtualDub software and I made no attempt to change it as the recorded image was looking better than anything so far.

(3) Since an audio track was being laid down in real-time, there is a significant chance that any later changes in frame rate would make problems with the audio - something that VirtualDub warns against!

At the end of the processing, the final AVI was reported to have the 30 fps frame-rate, which makes sense since the audio track was unchanged.

BTW, the output AVI would not play on WMP, more on that later. I ran the output AVI into Live Movie Maker, which does accept the AVIs, and outputted the file in the .WMV format. The frame rate appears to be the same as the original AVI, as supported by the integrity and sync of the audio track and the fact that 30 fps is the default output for the LWMM - hence the rapid movements is some of the earlier tapes.

How the system uses 1/13 s. exposures and creates a 30 fps video stream is quite beyond me at the moment, but it is hard to argue with the image and sound on the video. I think there is some creative buffering and formatting of the video data stream, but I have not arrived at a satisfactory explanation. Fortunately, my ignorance does not trump the reality presented by the recording.

Anyway, the original NBTV .WAV file was a much delayed response for Gary's wish to see moving video. I think I have reached the limit of the operational options for this particular CCD video camera, but I am far from unhappy with the results.

Gary,

I don't know why you only got one file, but there is a possible clue as to what is happening. Your cropped AVI, which plays on your system, does not play on mine. So, if you could send your WindowsMedia Player version, I will cross check it with mine. I would not be surprised if I had an older version. I will check that my trying to play a suspect file on a brand new laptop at this end, which should have the latest version of the media player software.

I have been slow to get back up to speed on my many projects since the death of my wife a year ago. We were married for 46 years and it has been a tough year. I must say that the last few weeks have had a very welcome impact on my mood and energy!

All the best,

Ralph

[gary]

WMP version = 12.0.7601.18840

That's very strange especially since these are stock standard AVI files - no compression and no special formatting.

I am very sorry to hear about the loss of your wife Ralph, it must be one of the hardest things in a man's life to lose one's life companion. On the other hand you were privileged to share with her 46 wonderful years. That's no little thing.

[Ralph]

Thanks Gary! It is that perspective that has made life livable. Dr. Seuss, a noted author of children's books, put it very well - "Don't cry that it has ended, but rejoice that it happened!"

Now to more trivial issues. I think I know at least one way this camera could do it's magic. I think it is actually a standards converter. This particular monochrome camera produces a 640 x 480 image. I suspect that video sensor data is clocked into a frame buffer at a rate determined by the exposure setting and is clocked out continuously at a smaller range of clock rates determined by the selected frame rate (60, 30, or 15 fps). Since write and read rates are completely uncoupled (or effectively so), there are no compatability issues with respect to exposure vs. frame rates.

Ralph

[gary]

Ralph, I have a saved post here which says, far less succinctly, exactly what you are saying. I derived that conclusion from your very first post on this thread, but refrained from submitting it lest I made a fool of myself - nothing wrong with that mind you as long has one has deliberated over the available data reasonably no one should ever be criticised of error due to paucity of data - If Newton had been aware of Maxwell's equations and the Michelson-Morley experiment I am sure he would have arrived at Einstein's General Relativity - and perhaps would have gone on to the GUT.

If I could work out how to recover that saved post I would submit it (perhaps a moderator/administrator could advise).

In any event it says little more than what you have said other than to add that the camera need only have double buffering to implement the "standards conversion" of which you infer, and to say that is exactly how Video2NBTV handles the myriad of frame rates that it is possible to provide it.

The additional thing I added, however, is that it doesn't explain the fact that the bar movement direction *should* reverse when the "capture" frame rate changes from less than 12.5 to greater than 12.5 -I am thinking, and I am sure it occurred to Klaas, of the wagon wheel effect yo see on TV and movies where the rotation seems to reverse as the wheel increases speed - I am thinking that there may be harmonics whereby that occurs at different sppeds - but my mind is tired - it is late here so will sleep on it.

[gary]

I worked it out! (how to recover a saved post that is, the content is probably rubbish)

Ralph, you write that the original file was shot at 15 fps. That cannot be possible ! If the exposure time was 1/13 or 1/11 sec., then the frame rate MUST have been lower than 13 resp. 11 fps.

These speeds can only be true if the processing time could have been zero, which is not possible. The fact that the white band moves so slowly to the right prooves that the frame speed has been close to 6.25 fps. If the speed was exactly 6.25 fps. the band would have been not moving. Yes?

And because the black band was moving faster (to the right) will say that the frame rate was higher than the frame rate of the white band. If it was the same the band should have been moving with the same speed. Yes?

So with that camera the frame rate is depending of the exposure time. I guess that the processing time is the same, so
Nipkow disc: revol'tn time = 1/12.5 sec. = 80 msec.:
white band: exposure time = 1/11 sec. = 91 msec.: overlap = 11 msec = 4.4 line (this is true, I counted them on the screeen)
black band: exposure time = 1/13 sec. = 77 msec.: missing = 3 msec = 1.2 line (this is not true, I counted 2+ on the screen)

Unhappily I cannot measure frame by frame, otherwise I could measure the displacement on the screen and could recalculate the real frames per second. This MUST be lower, much lower than 15 fps.

Klaas, going back to Ralph's first post the implication is that, to change exposure time the camera internally by changing it's "capture" frame rate, i.e. equivalent to changing the shutter rate, so, indeed, assuming this is so, the frame rate IS slower than 12.5 for the white bar example, and IS higher than 12.5 for the black bar example - the output frame rate can be any frame rate you like PROVIDING there is double buffering of the image within the camera (such that the only critical processing time is a memory to memory transfer which is very quick in comparison the 1/fps involved)- in fact this is exactly how Video2NBTV handles the multitude of frame rates presented to it. The problem is though, in this scenario, the bars should be reversed in direction, and the 15 fps output would not have any effect on that at all.

The fact that the bar direction ISN'T reversed implies that, while the above explanation is "possible", and maybe even plausible, it can't be correct and something else must be going on - at least that's how I see it. It will be interesting to see if Ralph can dig up the technical information on this camera so we can get to the bottom of this curious but fascinating puzzle.

BTW, why 11 fps and not 12 fps ?

[Ralph]

Gary,

WMP on my work computer and on my newest laptop play ALL my AVIs without difficulty. Although I cannot find version numbers to check, that is almost certainly the issue with WMP and some AVI files.

Ralph

[Klaas Robers]

Gary, the shooting speed is much slower than 12.5 fps. This is because I know how the sensors work.

- There are two analogue memory arrays:
- One is photo sensitive,
- the other is just a storage array.

- The photo sensitive array is exposed to the picture,
- The capacitor cells are initially charged,
- by illumination the charge leaks away.

- Then very fast the charge of all cells is transferred to the second array.
- Then the cells are shifted out (analogue shift register)
- and the output is converted from analogue to digital.

- During this process the photo sensitive cells are charged to a reference voltage.
- When this ends the new exposure starts.

For a CCD camera it is impossible to shoot continuously,
it always takes individual snap shots
and there is a certain "dead" time in between the exposures.

The shooting speed depends on the exposure time + the reading / recharging time.
When the exposure time = 91 msec (1/11 sec, bright bar) and the reading /recharging time = 69 msec then each frame is taken in 160 msec = 2 NBTV frames. Then the film is shot at 6.25 frames per second. The bright bar will be stationary (not moving).
If the reading / recharging time is lower, e.g. 65 msec, the new exposure starts earlier, the bright bar will be moving to the right.
If the exposure time is 77 msec (1/13 sec, black bar) and the reading / recharging time is still 65 msec (why not?), the black bar will be moving faster to the right.
This is what we have seen.
I wonder that the read / recharge time is so long.

[McGee2021]

Going back to a previous post, where did you get the motor for the daily express monitor?

[Ralph]

The unit is a permanent magnet DC motor made by Bison Gear and Engineering. The specs for this motor are available online at:

http://www.bisongear.com/051-206-4005.html

Ralph

[gary]

The unit is a permanent magnet DC motor made by Bison Gear and Engineering. The specs for this motor are available online at:

http://www.bisongear.com/051-206-4005.html

Ralph

Gulp...

[Ralph]

I feel you pain Gary! Actually, it cost me about half the current price when I got it few years ago (and put it on the shelf). It is my personal view that the disc and the motor are two of the most important components in a classic Nipkow monitor. The motor is powerful enough to ignore the disc load and thus does not draw a lot of current. I can also expect the bearings and brushes to last longer than I am likely to be around.

It has worked so well in this monitor that if I do start polishing the mirrors for a mirror screw, I will get another. If so, I better do it soon......

Ralph