Why do movie cameras photograph images upside-down?

[Jim Carlile]

I posted this on the newbie forum, but thought it might specifically interest super 8 people.

 

Anybody noticed that movie cameras photograph their images upside down?

 

I was looking at some super 8 film the other day, coming off the top of the reel, emulsion out, and really noticed this for the first time-- the images are captured, and projected, upside down.

 

Anybody know why?

 

I have a theory but am not sure it is correct-- because of the inversion properties of the numerous lens elements, and the reversal reflection-property of the screen, it's the only way it will work.

 

It must have been a bear to figure it all out years ago.

 

Any ideas?

[David Mullen ASC]

See:

http://en.wikipedia.org/wiki/Photographic_lens

 

You don't even need a lens... a pinhole causes the same phenomenon.

[Jim Carlile]

Oh, I know that the lens inverts the image, sure. But cameras, just like they do for viewfinders, could be designed to photograph the image right-side up. So why don't they?

 

This is a question I've never seen explained anywhere, or mentioned in any text. The film captures an inverted image. By all rights, when it is projected, it's in proper orientation-- the inverted image has been inverted by the projection lens. But the reflection from the screen reverses it, which means it should be reversed when we see it. But it isn't.

 

My theory is that upside-down capture is by design, because that's the only way it will work. Because each lens element or set of lens elements inverts the image, a 'straight' right-side-up image on film would result in-- at best-- a reversed image when viewed reflected from the screen.

 

Why? Because an even number of separate lens elements in a projection lens would result in a projected image that would be in proper orientation when viewed on the screen, but upside down. An odd number of lens elements would result in a right-side-up projected image, in proper orientation when it hits the screen, but it would appear reversed to the viewer when reflected.

 

Either way, it would be wrong. But if the image is projected upside down (by being photographed upside down) the variables all work out by the time the image is reflected back from the screen.

 

If you look at lens breakdown diagrams in cameras, viewfinder pathways always have an even number of lens elements or element sets, and film pathways have an odd number. The viewfinder pathway also has two mirrors, because one mirror reverses the image, so the second one reverses the reversal.

 

This makes sense, and it's obviously by design-- the odd number of total lens elements in the camera lens results in an inverted image on the film. An even total number for the viewfinder path results in an image in proper orientation. It has to be this way.

 

Does this make sense?

[Michael Waite]

Does this make sense?

I think you are way over complicating things. A camera obscura image is projected upside down & has no lens, just an opening. Our eyes focus the outside world onto the back of our eyes upside down - our brain turns things up the right way. Look at the illustration in the link that David gave - it's a simple principle of light traveling in straight lines. There is no logical reason why anyone would want a camera image to expose on the film right way up relative to the subject. You take the film out, process it & put it in the enlarger or projector so the image is the right way up. There is no issue of horizontal flipping as you seem to be suggesting, as long as the emulsion is the right way round.

[David Auner aac]

But the reflection from the screen reverses it, which means it should be reversed when we see it. But it isn't.

 

No, that's where the error lies in your line of thought. The screen DOESN'T reflect the image as a mirror would. You see the image because of the light falling on the screen.

 

Cheers, Dave

[Michael Lehnert]

Sorry if I am unable to help you well here as I can only thinly grasp the specific issue you put forward here, Jim... as I said in the other "Circuschrome" thread: after four weeks in LA, returning to Europe and London is a culture shock and I just wanna go back ... I might turn into Phil Rhodes any moment... ;)

 

Am I understanding you correctly that your basic question is why no optical engineer has settled on actually designing optical paths in such a way, that the image would be exposed "correctly upright" on the film as seen when looked at it from a camera operator's viewpoint? And hence that the film would be inserted equally "correctly upright" into a film projector instead of "on its head"?

 

I attach two images to this post, as I vaguely remember that you shoot with Nizo big-bodied cameras, like the Nizo 561. Maybe these optical and lens diagrams will be of some help for you in the quest of clarifying the optomechanical interplay of lens elements, lens groups and optical image generation...

[Mark Dunn]

Light travels in straight lines. So the top of the scene ends up at the bottom of the film. Projected, it goes back to the top. It's as simple as that. The number of lens elements has no bearing.

 

The viewfinder inverts the image for convenience- it's difficult to follow action upside-down. If you've ever used a large-format camera, you'll have seen the image inverted on the focusing screen, just as the lens produced it- there's no need to turn it right way up and you get used to seeing it that way.

 

The screen isn't a mirror, it's a reflective surface. If the viewfinder is a window on the scene, so is the screen. The orientation of the image you see is the same.

 

Have a look at the link. Maybe you've thought too hard about this- go back to first principles over a nice cup of tea.

[Jim Carlile]

No, the screen reverses the image that is cast upon it by the projector. Why doesn't it? It has to, it's reflecting the projected image.

 

I probably complicated this all by talking about lens elements and such-- I should have mentioned prisms that are in the viewfinder pathway, that split off the image and reverse it at the same time.

 

But it's an interesting topic. Camera lenses are just compound lenses that could be designed to cast an 'erect' image on the film plane, rather than an inverted one (reversed and upside-down.) But they aren't.

 

I think the easiest way to look at is that an inverted image is cast upon the film plane. Then, when projected (which is upside down, by the way-- just like the camera), the image is cast upside down but in proper orientation 'upon' the rear lens element.

 

The projector lens then casts a complete inversion of this image, which is right-side up but now reversed, upon the screen, which then reverses the image to the viewer in proper orientation.

 

What I have noticed is that camera and projector lenses always have an odd number of convex lens elements in them. It is the convex lenses that cast inverted images upon a plane. Maybe there's nothing to this but it seems to make sense.

 

When you look at cross-sections of viewfinder pathways, you will always find an even number of convex lens elements, along with prisms which reverse the image without inverting it.

 

Clearly, an inverted image is the easiest one to keep in proper orientation on the screen, which is the point of it all. I probably answered my own question, but it's something that most people don't think about, or even notice, all this light-pathway trickery to the art.

[Hal Smith]

 

The optical path image does not show the correct ray tracing. There should be a nodal point in the viewfinder path above the first prism at the same distance as the nodal point behind the prism in the optical path to the film plane. (Boy do I love to catch a German engineer with his pants down.) ;)

[Brian Dzyak]

If you turned the camera upside down, then re-inverted the viewfinder image, then the image would be "right side up" on the film! :P Or we could just turn the set over. :blink:

[Paul Bruening]

On your question about inverting the image before it hits the film: Lens elements flip both X and Y axis. Mirrors flip only one axis in 45 degree applications. All other axis changes have to be done with prisms.

 

http://en.wikipedia.org/wiki/Prism_(optics)

 

The prism at the top of an SLR is a pentaprism (roof prism). These prism do right the image. If a mirror was used instead, the image would still be upside down by the time you see it.

 

I went round and round with Bruce McNaughton when he was solving my XL2/video tap issue on Frankenmitchell. We finally dumped all of the prisms and used a TFT monitor with image flip built into it.

 

The problem with prisms is that they eat too much light, or they did back in the old days, back when cine cameras were being worked out. It was way better to just leave everything upside down and keep prisms out of the chain.

[Sam Wells]

No, the screen reverses the image that is cast upon it by the projector.

 

You can't make the statement "Reverses." With respect to where ? Go behind a screen which has any degree of transparency and take a look :)

 

-Sam

[Richard Baines]

Wha? :blink:

 

I really, really, think you're overcomplicating this one.

 

 

A lens inverts the image, but this is cancelled out by projector.

 

Why would you want it to be photographed the right way round? There would be no point to doing so.

[David Auner aac]

The optical path image does not show the correct ray tracing. There should be a nodal point in the viewfinder path above the first prism at the same distance as the nodal point behind the prism in the optical path to the film plane. (Boy do I love to catch a German engineer with his pants down.) ;)

 

Hi Hal,

 

yeah, you're right. But that's probably to blame on marketing! Or some illustrator. I bet you'Re like the first person to notice!

 

Cheers, Dave

[Jim Carlile]

Wha? :blink:

 

I really, really, think you're overcomplicating this one.

 

 

A lens inverts the image, but this is cancelled out by projector.

 

Why would you want it to be photographed the right way round? There would be no point to doing so.

 

You're totally right, and that's the answer to my question-- because it's easier.

 

Oh, and I was totally wrong about the screen reflecting and thus "reversing" the image. The screen doesn't reflect the image-- all it is is the plane upon which the image is projected, in proper erect orientation. All we do as observers is just cast our eyes on the plane-- it's not like a mirror.

 

And the image on the film is cast by the light source as inverted, and then inverted again by the projection lens, resulting in the correct image.

 

Neverrrmind.....

 

But, it's interesting the lens breakdowns. I'd love to see the complete ray-angles of each element, as to what each does within the complete compound lens. Prisms, too. I know that not all lens elements flip the X and Y axis(a)-- convex lenses do that...the others bend the rays in certain ways. But just how and for what purpose?

Edited September 2, 2008 by Jim Carlile

[Jim Carlile]

The optical path image does not show the correct ray tracing. There should be a nodal point in the viewfinder path above the first prism at the same distance as the nodal point behind the prism in the optical path to the film plane. (Boy do I love to catch a German engineer with his pants down.) ;)

 

Are you saying that the nodal points should be in the same place on the prism in the diagram, where the rays split off-- that one nodal point is indicated, but not the other?

[David Auner aac]

You're totally right, and that's the answer to my question-- because it's easier.

 

Not only that. Image quality would suffer too, if you were to introduce yet another element into both taking and projection lenses to flip the image (twice!)... Seems like you got your mind around it now! Optics can be very confusing at times!

 

Cheers, Dave

[Hal Smith]

Are you saying that the nodal points should be in the same place on the prism in the diagram, where the rays split off-- that one nodal point is indicated, but not the other?

 

Yes, there are three rays shown striking the reflecting/transmitting diagonal surface in the prism behind the camera lens. Therefore there should be three rays tracing through the prism as well as three rays reflecting vertically within the prism. Then following each set of rays they should be converging at two identical nodal points. It's not a huge deal but the mistake happened to catch my eye. My academic training was in Physics (I hold an MS in the Teaching of Physics). I both took and taught some optics courses giving me a somewhat educated eye looking at optical diagrams.

[Paul Bruening]

My academic training was in Physics (I hold an MS in the Teaching of Physics). I both took and taught some optics courses giving me a somewhat educated eye looking at optical diagrams.

 

Wow, Hal. I didn't know. It's obvious from your postings that you are an articulate, knowledgeable person. But, I didn't know that you were edumacated'nalldatshiznet.

 

Optics can be confusing. I recall bending my brain trying to figure out my systems. I learned a lot in the process. Much of it still eludes me.

 

Have you caught my madman's threads on reducing SLR lens images to Techniscope framing? Cooke, Zeiss and Vantage have treated me like I have leprosy. That tells me that I'm either: 1) A dumb-ass, 2) On to something.

 

Since bigshots don't waste their time on punks like me, I'll dig at this one a little on my own. I've got some teleconverters coming from Ebay. I'm going to set them up in-line on my bench with my SLR lenses and a piece of cardboard that I use as a screen. I want to see what a teleconverter facing backwards does. I'm guessing that the faces won't bend light the way I want. But, it's worth a try.

 

My question with you is: Will the teleconverter elements behind the normal lens alter DoF or is that locked-in in the SLR lens' group?

 

As always, Hal, thanks for your input,

Paul

[Jim Carlile]

Great discussions-- optics is something people don't think much about these days. The older film making books (20's and 30's) are obsessed with it.

 

One of these days I'll bring up a new question, about whether 35mm movie film-- or even super 8-- has superior depth of field qualities, that give it a meaningful aesthetic advantage over digital. And why that is, optically.

 

It's something I've noticed that video people worry a lot about, but film people don't mention much-- and they should, I think.

[Stephen Williams]

It's something I've noticed that video people worry a lot about, but film people don't mention much-- and they should, I think.

 

Hi,

 

Just video people are desperately trying to get a film look, they mistakenly think that they do not lean how to control light, compose images & select the most appropriate lenses for the task in hand.

 

Stephen