Create one image from overlaying several frames?

[sincostan]

OK, I know that there are nearly 1,000 different ways to take a single frame out of a video and save it as an image file.

However, what I would like to do is take a range of frames and overlay the images to create a single image. (I'm sure this can be done in something like Adobe Photoshop if all of the frames are saved seperately.)

So what I am looking for is software that will open a video file and then let the user see all frames, select a range of frames (say something like 20-34), and then overlay them into a single image.

You may ask why I would want to do this.

The story behind this is that I am a physics student that does research for one of my professors and he asked me to help him find a solution to the following situation, hopefully using a digital camera.

When he teaches his freshmen level class, one of the experiments they do is to let a golf ball roll off a table while continuously taking a picture of it with an old Polaroid camera with a shutter wheel in front of it that allows the camera take an image every 1/30th of a second. There are metersticks that are positioned such that, when combined with the camera position, they show up on the side and the top of the picture. Between the metersticks and the ball being in a different position in the picture very 1/30th of a second, this allows the students to determine the x and y components of the ball's velocity.

That's all fine and good, but the film that is used in this old Polaroid camera is no longer in production, so he's looking to me and digital cameras for something to replace this.

I've already tried using my digital camera (with my camera's shutter speed reduced down to 2 seconds) in place of the old Polaroid (behind the spinning shutter wheel), but my camera's Image Stabilization removes any trace of the golf ball in the picture.

So...that leaves video mode, which conveniently happens to record at 30 frames per second in most cameras. If there is a way to easily overlay several frames of video into an image, then all that would have to be done is to select the range of frames the golf ball shows up in the video and then slap them together into a single image.***

It should also be noted that about 10-15 frames would be put together since I found the golf ball to travel through the range of the camera in about 0.35-0.50 seconds.

***I am worried that even if I do find software that will do this, the golf ball won't be able to be seen. Say 12 frames need to be overlayed, wouldn't only 1/12 of the image of the golf ball be able to be seen at each progression? Or is there software that only overlays the changes in each frame when each frame is almost identical except for small movements?

And one more thing, the reason I said I would like this to be done in one program is because we're talking about a professor and some college freshman using it on state-owned University computers, not me. So, that's why I also said I would like this program/method to be simple.

I mean sure, I could always copy the individual frames in VirtualDub, save them using Adobe PhotoShop, and then turn around and mess with overlaying them in PhotoShop on my own computer, but...this isn't me.

I know that this is kind of a tough problem and because of that I apperciate any help I can get. Thanks in advance!

[Des]

With the old Polaroid camera is the shutter locked opn on a 'B' or 'T' setting? I ask this bcause it sounds like the wheel is actual allowing and producing the exposure at 1/30 of a second. (Curious as to which Polaroid alowed that, Model 110 perhaps...). Basically the wheel is acting as a shutter.

If this is the case then you should be able to do this with a digital camera with the shutter locked opn on 'B' (Bulb) and then using the shutter wheel to create the exposure in the same way. Which camera do you have?

You can turn off the Image stabilization on your digital camera if that is getting in the way.

ISO setting on the camera might have something to do with the results, Polaroid film (assuming this is the peel apart kind as that is now nearly completely defunct) came in ISO 80, 100, 600 and 1000 if memory serves right. If you can find out the type of film (or at least camera) I can cross reference the ISO for you.
Anyway you probably want to set the ISO to match the Polaroid film and the lens opening to match the polaroid lens.

After that it should work as expected. You can probably even modify it in some ways pending the results.
It's really just simple multiple exposures on a frame.

Not sure how this would work on video as I would think overlaying the frames really hides the frame beneath it (think layers in Photoshop). Perhaps on video a stop motion or strobe effect could be used, but I would think that photographically is simpler.

--dES

[sincostan]

I couldn't tell you much about the old Polaroid camera - I didn't look at it that closely, but I will tomorrow when I see my professor. The only thing I can tell you at this point is that two bright lights are shown at the meterstick/golf ball area where the experiment takes place. These lights are located on both sides of the old camera. And the shutter wheel that spins is actually not part of the camera. It's attached to a support and it spins right in front of the camera's lens.

So, yeah, tomorrow I can check into those details and post back here.

The camera I have is an Olympus Camedia C-4000 ZOOM. Yeah, I know it's older than the hills but it was pretty good back in the day. (And does still take very good pictures, even at 4MP )

At any rate, I used this camera as an experiment to see if digital cameras would work in place of the old Polaroid camera in the same setup. I was (and still am) thinking like you - all it is is just simple multiple exposures on a frame. Therefore, if I set the camera to hold the shutter open the whole time, why wouldn't it work? Well, it didn't because there was no trace of the golf ball in the picture. (I'm sure my camera's Image Stabilization screwed it up.)

And for the record, I totally agree with about setting the ISO to match the old Polaroid film's.

Now, I looked at my camera and there is no "B" setting, nor is there a way to turn off image stabilization. Keep in mind that if I can find a solution using a digital camera, my professor will get a new one, since the one I'm using now is mine, not the University's.

With that being said, I'm sure there are a lot more features on the newer cameras, which will hopefully include a feature to disable Image Stabilization and a "B" mode. So if he gets one, we'll have to make sure that it has the option to change the setting to make this work.

My roommate has a camera and also works for the same professor. He's not in right now, but I can look on his camera to see if I can't disable image stabilization. And if so, we can try it out and see what happens.

Yeah, taking a video of it and then trying to combine the frames would be a huge pain in the neck. And I think when the frames are combined, you'll first get half of each, then a third of each, then a quarter of each, etc. So if you combine 10 frames, you'll only see 1/10 of the golf ball's brightness at each interval. That's why I asked for software that would only overlay the changes between each picture instead of the whole picture. But if we can just take a regular picture of the experiment, we won't have to worry about it.

[Des]

Your camera predates image stabilization which is why you have no facility to turn it off. I agree though it is an excellent camera, unfortunatley the biggest stumbling block with that camera is the smart media cards. Frankly 4 mps is plenty for most people, good quality lab prints at up to 11x14.

Your images didn't come out because your exposure was wrong.

If the wheel turns at 1/30th of a second then this is your shutter speed. Ideally you want to set your fstop to make an exposure at 1/30th of a second in the lighting your experiment is in.

Longest shutter speed your camera can do is 16 seconds, I don't remember if your camera has a manual mode but I am pretty sure it has a shutter priority mode ('S' on the command dial). Set the camera there and then set the shutter speed to something like 4 seconds, this will give you about 12 exposures with the wheel moving at 1/30th (1 second equals 3.33 exposures at 1/30th of a second).
Probably best to let the camera set the ISO.

You want the wheel to block out all light if possible from the lens (cover the lens) except for the slit or aperture that makes the exposure.

Experiment a little. Please let me know how it goes and I can help you fine tune it more.

--dES

[sincostan]

I followed almost everything that you said, however, I'm not sure what you mean by "set your fstop to make an exposure at 1/30th of a second" when the shutter speed is to be set at 2-4 seconds.

Remember that I did set the shutter speed to 2 seconds when I tried this the first time, and yes my camera does have a Shutter Priority mode where I made it have a shutter speed of 2 seconds.

Now with that being said, I don't see where you got these numbers from for setting the shutter speed on 4 seconds with a wheel that exposes the lens of the camera 30 times a second:

"this will give you about 12 exposures with the wheel moving at 1/30th (1 second equals 3.33 exposures at 1/30th of a second)"

Um, wouldn't 1 second equal 30 exposures at 1/30th of a second? And how did you arrive at 12 exposures?

At any rate, I'll experiment with it tomorrow and let you know how it goes.

[Des]

Um, wouldn't 1 second equal 30 exposures at 1/30th of a second? And how did you arrive at 12 exposures?

My bad, I was thinking 1/3rd of a second not 1/30th of a second. You are right.

As for the f-stop, exposure for film or digital is made by two facotrs: 1) How Much light enters the lens -This is controlled by the f-stop or aperture. 2) How much time that light is allowed to strike the film or the sensor - this is controlled by the shutter speed.

EDIT: In this situation the wheel is acting as shutter not the camera's built in shutter which is just acting to 'open the box' to the sensor. So if the opening in the wheel is moving at 1/30th of second then in this situation your shutter speed is 1/30th of a second and you adjust the f-stop of the lens accordingly for exposure. It's important to limit stray light as much as possible by getting as close to the wheel as you can.

For example, If you are shooting a picture and the lighting conditions allow a shutter speed of 1/60th of a second and an f stop of f5.6 to create a perfect exposure then any change to eitther shutter speed or f-stop will cause the image to be either over or under exposed. But, if you change the f-stop you can also change the shutter speed to compensate for it. So, if you wanted to shoot 1/30th to show motion or some other reason then in order to keep the same exposure (written as Exposure Value = EV) then because you are letting light strike the sensor for a longer period of time you have to limit the light coming into the camera. You would do this by using the next smallest f-stop, or f8. Because you have changed the shutter speed one number slower you need to block light by one number from the lens. It's a 1:1 ratio, for each click one (shutter speed or f-stop) goes up the other has to go down.

The thing is, this was all predetermined with the Polaroid set-up, the pain in the butt is recreating it from scratch without the Polaroid.


--dES

[GTRBudda]

This may not be helpful at this point but going back to the original question, there's a tool called iMerge that seems to do what you were looking for, import an avi and merge a range of frames. It's meant as an astronomy tool, like a comparator, but might do what you need.
http://www.geocities.com/jgroveuk/iMerge.html
@des: Your avatar Rules! Sorry, I'm a fan

[Des]

I just tried a little experiment to make sure I am leading you in the right direction.

The set up I used is very crude but does work:


I used a camera set to 10 seconds and used a black card to block the lens. I moved the quarter each time I replaced the black card in front of the lens. Exposure was about 1 second (probably more) and I used an fstop of 22 to compensate for the bright lights (and my guess of 1 second) I have over-head.

One thing I hadn't thought about till this is that I needed to turn off the autofocus as the camera wanted to refocus on the black card when I pushed the button.

GTRBudda

@des: Your avatar Rules! Sorry, I'm a fan

Thanks! it's from an old concert shirt. The shirt doesn't fit anymore as it seems to have shrunk around the belly over 20 years time

--dES

[sincostan]

Alright, I didn't understand what you were talking about at first with the aperture setting, but after it sunk it, it made perfect sense. I don't know why I didn't realize it before. (And I call myself a physics major... one who's taken an optics class to boot. )

Well, here's how everything went:

EDIT: What the heck? Why are the thumbnail links wrong? And I also find it interesting that Imageshack detected what type of camera I used to take these pictures with. (Look at the bottom of one of the pages.)

(All the linked pictures are in full 4MP size, in case you're on dial-up.)

First, this is the setup the Polaroid camera used to be in. The camera is actually turned onto its side to the side of the tripod (and facing the wrong way) so I could have room to hold my digital camera in front of that shutter wheel.



This is the setup we try to take a picture of:

(Obviously, we roll the golf ball off that rail.)



Here's what happened when I tried to take the picture the first time:



And after I tried the advice you gave me:



My professor wanted me to print out the image, and this is what it looked like after it printed out. He said that just taking the picture and printing out out directly from hooking the camera up was at least five times faster than the previous method of messing with Polaroid film and all of that.:



So, THANK YOU VERY MUCH! I really appreciate it! I let my professor know that a guy by the username of "Des" on the "videohelp.com forums" helped me out, and he said that it's really interesting that I seek the help of of people I don't even know over the Internet.

At any rate, if you want to know, this is the model of Polaroid camera:



And these are some specs I found on its side:



GTRBudda, thanks for that link. Even though I may not use it at school, I may have need of something like this in the future. Thanks!

[Des]

Hey Sincostan, glad it worked out. I was hoping to look at the picture of the photo rig and say 'Oh that's a Polaroid P-S 150' or something like that but instead I thought, 'Wow, cool Frankenstein!'

Anyway, glad it worked out. I enjoyed giving the photography lesson.

Image shack read the EXIF data stored with the jpeg file from your Olympus C 4000 which is how it knew your equipment. EXIF data tells you wonderfull stuff about an images photographic details. And, they don't overwrite it if you d/l the image from Image Shack. You shot at f2.8 for 1.6 seconds, ISO 100.

For what it's worth the Polaroid Camera can use Polaroid type 669 film which is still available. It's about $18 for 16 pictures so going digital is certainly more cost effective, as well as faster.

I let my professor know that a guy by the username of "Des" on the "videohelp.com forums" helped me out, and he said that it's really interesting that I seek the help of of people I don't even know over the Internet.

Knowing when and then where to find assistance is a key to success. You should have a bright future ahead of you.

BTW, what was the velocity of the ball?

Take care.

--dES

[sincostan]

Hehe, my professor's all about getting by with as little as possible for everything, so it doesn't surprise me that that camera's really old.

Interesting, I never knew that there was a tiny amount of extra data stored in digital camera photos that gave info on how the picture was shot!

(And yes, that is how I took the picture.)

LOL, if I tell my professor how expensive that film is and how much the digital camera will save, he'll be glad he went digital.

The ball's velocity? Now you've got me curious.

First, the ball doesn't have a constant velocity because it's under a constant state of acceleration due to the pull of gravity and it's parabolic trajectory.

So what I'll do is figure out the velocity of the ball when it hits the ground.

The velocity in the y-direction is 9.8*t meters/sec where t is the time from when the ball begins to drop.

Or, to make things easier (trying to "eyeball" the time from each position of the golf ball is both difficult and inaccurate), we could make the velocity a function of distance instead of time.

That is (v^2)=2*a*(yfinal - yinitial)

[a=(9.8m)/(sec^2) is the acceleration due to gravity. The initial y position is 85, since the meterstick reads 15cm, but is 100cm and is flipped upside down, actually making the height 85cm. Of course the final y position is 0cm since the ball hits the ground. 85cm = 0.85m Also, I omitted the initial velocity variable since it is only an added constant that is 0.]

(v^2)=2*[(9.8m)/(sec^2)]*(0-0.85)

(v^2)=-16.66m^2/sec^2

v=-4.08m/s

(Technically, you can't take the square root of a negative number, but that's something that has to be ignored for this type of problem. :P Also, the velocity is negative because this ball is dropping, therefore it's moving in the negative-y direction. However, that's getting too nitpicky with this type of problem, so we might as well just say the velocity is 4.08m/s.)

For the x-position, if you look at the picture, the ball travels about 6cm every 1/30th of a second. 1/30 is about 0.033333.

0.06/0.03333 = 1.8m/s

So, the velocity in the x-direction is about 1.8m/s.

Using Pythagorean's theorem of a^2 + b^2 = c^2, we can get the resultant velocity.

1.8^2 + 4.08^2 = c^2

3.24 + 16.66 = c^2

19.9 = c^2

4.46 = c = 4.46m/s

What about the angle? We can set the velocities up like a triangle and figure the angle between the bottom and the hypotenuse using sine, cosine or tangent. Let's use tangent. (Being able to draw this out on paper would help, but I can't do that here.)

tan(theta) = opposite/adjacent

(The y-velocity is the opposite side of the triangle, and the adjacent side, the bottom of the triangle, is the x-velocity.)

tan(theta) = 4.08/1.8

tan(theta) = 3.7778

Inverse tan (tan(theta)) = inverse tan (3.7778)

theta = 75.17 degrees (If you look at the image, this seems about right.)

So right when the ball hits the ground, it's incident angle is 75.17 degrees measured from the ground up traveling at 4.46 meters per second.

Of course when the ball just leaves the rail, it's traveling at 1.8 meters per second with an angle of 0 degrees from the ground since gravity hasn't started to pull on it yet.

Heh, that probably more like a scalping when you were asking for a shave, but I enjoy this kind of stuff. :P

Anyway, thanks for all of your help Des.

[Des]

Thanks for the physics/math lesson, never my strong point.

Enjoy!

--dES

[olyteddy]

Wouldn't a 'Disco Era' strobelight have made this a lot simpler?

[Midzuki]

If there is a way to easily overlay several frames of video into an image, then all that would have to be done is to select the range of frames the golf ball shows up in the video and then slap them together into a single image.

With Avisynth, perhaps:

-- selective "Trim"s;

-- selective "Crop"s;

-- StackHorizontal/Vertical;

-- ImageWriter;

Wouldn't a 'Disco Era' strobelight have made this a lot simpler?

Or a multi-vibrator driven white LED, who knows?

+++++++++++++++++++

[Des]

All are probably good, but considering the set up already exists and the camera is the only thing needing replacing, then why re-invent the wheel here.

I like the disco strobe though

But it would have to be a continuous 30 bursts per second.

--dES

[thecoalman]

You can use virtualdub to select a range of frames to export Load the video into virtualdub, on the bottom move the slider to where you want to start exporting, select the second to last button on the left on the bottom which will set the mark-in point. Move the slider to tha last frame and select the last button on the bottom to select the mark-out point. Go up to file>save image sequence

Load the images into photoshop or other image editing application and import the first image as your canvas, import the second image and use the object eraser to remove the unwanted portion of the image, repeat for each image. It's a little tedious but you'll get perfect images. You could also use the magic wand to select the ball or even use a circular selection tool... I'd give you specifics but I don't use photoshop.