Jump to content

How to replicate 'multiple exposure' effect?


Recommended Posts

<p>I've touched on this subject before, however, I pretty much ended up nowhere. I want to attack this bastard from a different angle.<br /> There's this very nice feature in my Nikon D800 camera. It's called 'Multiple exposure' and the applications of which are countless.<br /> What I need is to 'merge exposures'. I will now illustrate my point.<br /> Let's say that a certain (static) scene dictates the exposure of 8 seconds. We capture the image, name it "Picture A" and set it aside. Then we put the camera into the multiple-exposure mode and set the number of exposures to: 8. We change the shutter speed to 1 second and make 8 exposures. Upon completion, the camera does its thing and we get a single photo, combined from those 8 exposures, which we call "Picture B".<br /> Upon close inspection, Photo A and Photo B look almost identical. Any variations are trivial at the most. Apparently this works just as good as it used to with film. Obviously there are several issues with this technique. One would be:<br /> If the camera is moved ever so slightly between the exposures, the resulting image (merged from 8 different exposures) will register this misalignment.<br /> In a real world scenario, especially when the tripod is on unstable ground (such as sand), every mirror flip and every shutter click will offset the image plane thus each of the 8 images will be recorded from 8 different camera positions.<br /> Nikon D800 does NOT align the images taken in the multiple-exposure mode. The solution to that would be to take 8 separate (RAW), underexposed photos and merge them outside the camera.<br /> To my surprise, Photoshop does a marvelous job at aligning such images automatically, at what appears to be sub-pixel level.<br /> You go to Files>Scripts>Load files into stack You then select the photos and check the "Attempt to Automatically Align Source Images" box. I also check the "Create Smart Object after Loading Layers".<br /> Once it's ready I go to Layer>Smart Objects>Stack Mode>Summation.<br /> That's the closest I'm able to get to the results obtained 'in camera'. Unfortunately, the resulting image is overexposed and all the tones look wrong, although shadow detail appears to be all there. So I figured this was not the right way to go about merging exposures.<br /> I've tried using the blend modes such as the "Linear Dodge" and "Screen" but the picture still doesn't look right.<br /> Previously I made a serious attempt with DeepSkyStacker, which is a software designed for astrophotography. Needless to say I got nowhere with it.<br /> Once I realized that PS can properly align several images I figured that it should have the proper tools for merging exposures. Maybe the "Smart object" and "blend modes" route is not the right way to go about this. Can anyone shed some light on the subject? I'm not married to Adobe, so if there's some other software that can merge exposures, I'd be glad to hear about it! Although I'd much rather stick to what I'm familiar with.<br /> Thanks in advance!</p>
Link to comment
Share on other sites

<p>There might be a software hack for this, tho I'd prefer to do this in the camera. Get few pieces of plywood, so the tripod legs can remain more solid and even hang something heavy from the mid section....preferably touching the ground. OK, you are not necessarily off the hook yet, you still have to calculate the multiple exposures....so the whole thing works properly.</p>

<p>Having said that, your tripod needs to be of decent quality (sturdy), since a flaky type could be a problem on its own. Also, wifi or cable release should be used to prevent any minute shakes from touching the camera.</p>

<p>Les</p>

<p> </p>

Link to comment
Share on other sites

<p>It is not at all clear what you are trying to accomplish with a multiple exposure technique. Assuming you can overcome the problems of camera movement and image alignment, how does all this extra effort improve the image you get from a single properly metered exposure?</p>
Link to comment
Share on other sites

<p>Assuming you want short exposure times to keep your ISO on an acceptable level, my take would be to use a single raw file, and then use the same approach you took with photoshop. Why shoot 8 different ones if you're looking to use 1 identical shot 8 times?<br>

I'm not too sure about the idea that 8 shots of 1 sec have the same exposure as one shot of 8 seconds. A 1 sec. shot is 3 stops underexposed, so somehow my brain itches that you'd only need 4 shots to get the same exposure. But I haven't tried, and I might be rambling.</p>

<p>Otherwise, I'm much with what Mike said: do whatever you can right in camera. Motion is going to be an issue at 1 sec. and at 8 sec. So you're not really avoiding an issue there - just expose right, and have much better data to start with.</p>

Link to comment
Share on other sites

<p>Leszek: We're not even on the same page.<br /> Mike: I was really hoping I wouldn't have to explain myself. That's why I was trying to keep things simple and to the point. But to satisfy your curiosity and maybe open up the eyes of other people I will describe a few a scenarios where the application of this technique would be very useful. Not "would be" but rather "is useful" as I personally have used multiple exposures in so many scenarios when I was still shooting on film.<br /> 1. Imagine that you need to photograph a cityscape at night. The main point of interest is a road or a bridge. You're interested in portraying the street as 'busy' by capturing many light trails created by vehicles' head and tail-lights. As you arrive at the scene, you realize that there are hardly any cars at all. In a 30-second exposure you would get 4, maybe 5 cars. One solution would be to set your camera to 6 exposures, each lasting 5 seconds. You would then open the shutter only when there were 3 or 4 cars in the frame. Thus in the merged image you would have light trails from about 20 cars, versus 4 or 5. This approach is more useful in film, as with digital, you can get lots of light trails without underexposing or even shooting in the multiple-exposure mode.<br /> <br />2. You are photographing a distant landscape with a 300mm lens. The light is getting low and you realize that with the lens stopped down to f-11 and the ISO already being at 200, you would still need to expose at 1/15 of a second. From your previous tests you know that even with the mirror locked up, the shutter alone introduces enough vibrations that will impart noticeable blur to your photo. To negate this effect, one could simply raise the shutter speed to 1/125 and make 8 separate exposures, each of which would be grossly underexposed but when merged together, they would provide ONE properly exposed image, free of any motion blur.<br /> 3. In a studio: This is what I call "flash banking". Imagine you're photographing a piece of jewelry with a macro lens. The magnification is close to 1:1, thus the compensation for tube extension is enormous. The DOF dictates the aperture of f-16. The strobes are right over your subject but they still can't put out enough light. Instead of bumping up the ISO, one could do the following. Set the camera to 4 exposures and take the same picture 4 times to compensate for the lack of 2 stops of light (that your strobes can't put out). Out of those 4 exposures the camera would create one good picture... Not really good, because it would be blurry. The reason is that each time the shutter is released it introduces enough vibration, which in term offsets the image plane. The flash 'freezes' the picture but each of those 'pictures' is taken from a slightly different angle because the lens is moved in each of those instances. Everything is as stable as it can be and with the mirror locked up there is hardly any vibration at all. However, at this kind of high magnification, the offset of the image plane is very, very noticeable in the picture and it manifests itself as lack of image sharpness . Fortunately, there is a way around this, which I've used with film. Make sure the studio is completely (or almost completely) dark. Open the shutter in bulb mode and fire the strobes manually as many times as you need. When done, close the shutter. This way the camera is not moved in any way between the "exposures" that in this case are the flashes of light. With film you can keep the shutter open for a week. As long as it's completely dark, it won't matter. With digital you have a fairly narrow window to fire your strobes before bad things start to happen to your picture. Of course, there's the Long-exposure NR but it introduces its own artifacts, that get progressively worse with the exposure time.</p>

<p>I hope this was enough.<br /> If merging exposures in-camera can be done with such high-quality results, why can't it be done by a proper computer with even better results?</p>

Link to comment
Share on other sites

<p>Hi<br /> You could have added pixel values (after aligning images) if you could do it on the RAW files themselves (before gamma encoding). The gamma encoding (which is reversed by your screen hardware before you see the image on your screen) makes the relationship between the RAW pixel value and the final image pixel value (JPG, TIFF or whatever) nonlinear. A very simple example to show what I mean: If you have two values, say 2 and 5, and add them together you get 7. But if you take 2^2 = 4 (2 squared) and add to 5^2=25, you don't get 7^2=49. Instead you get 4+25=29. So adding after squaring makes a much smaller value (29) than adding before squaring (49). Gamma encoding works just like the squaring, but it makes it the opposite way since it uses power of 1/2.2=0.45 instead of power of 2. You will get a very overexposed image as you have experienced (but not only overexposed, the colors and tones will be completely messed up)</p>

<p><br /> Emulating multi exposure:<br />To emulate an 8 picture multiple exposure with layers in photoshop you might use the opacity control on each layer. If we count layers from the bottom, i.e. the bottom layer is layer 1, the one above is layer 2 etc, then:<br /> <br />Layer 1 has 100% opacity<br />Layer 2 has 100% / 2 = 50% opacity<br />Layer 3 has 100% / 3 = 33.333% opacity (rounded to 33%)<br />Layer 4 has 100% / 4 = 25% opacity<br />Layer 5 has 100% / 5 = 20% opacity<br />Layer 6 has 100% / 6 = 16.667% opacity (=17%)<br />Layer 7 has 100% / 7 = 14.2857% opacity (=14%)<br />Layer 8 has 100% / 8 = 12.5% opacity (= 13%)<br /> <br />All layer blend modes are set to "Normal".<br /> Now, this is if each image was correctly exposed, i.e. each image exposed for 8 seconds in your example. What you get is an average of each image, i.e. the result will contain equally much from each image (about 12.5%).<br /> Since you had each image exposed for only 1 second, you will have to adjust the exposure 3 stops up, either on each image or on the final resulting image. Each image will then have the same quality as a 3 stop underexposed image (since that is exactly what they are), but when averaged you will get the same quality as a correctly exposed image (8 seconds in your example).</p>

<p><br />The value of multiple exposures is grossly underestimated. It has a multitude of uses. :-)<br /> <br />Cheers,<br /> <br />Hope this was of any help. If not, print it out, curl it up and throw it in the dust bin.</p>

Link to comment
Share on other sites

<blockquote>

<p>or on the final resulting image</p>

</blockquote>

<p>Beware that this will also not be correct, but much better than just adding the layers. The correct result will be achieved by adjusting each image 3 stops up <em>before</em> you blend them together (usually in the RAW converter). This will be indistinguishable from the single 8 seconds exposure.</p>

Link to comment
Share on other sites

<p>A while ago I've actually stumbled upon an online demonstration where the guy was using different opacity values for each of the layers. I didn't dwell on it because I figured there must be a more straightforward approach. It would have been nice to have a dedicated software for this or at least a Photoshop or Lightroom plugin that would do all the figuring out and what not. Given that there are 4, 8, 16, 32 or whichever number of exposures of the same thing, proper noise-removal algorithms could really work wonders... Did I say "super-resolution"? Oh, well...<br>

Anyway, by now I know that there's no such thing and I'll have to adapt my technique to the tools available in Photoshop. Wouldn't be the first time.<br>

Thank you, Frode! I will try what you suggested as soon as I'm free.</p>

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...