Home | News
Lesson Materials
Course Information
Resources

Assignment, Week 4: Image Stabilization Technology to Reduce Camera Shake

Last week we discussed various techniques photographers can use to stabilize their cameras and reduce blur. While photographers have used those techniques throughout the history of photography, in modern times we've also developed technology that can help to stabilize images and reduce blur.

In general, technology that compensates for camera shake is called 'Image Stabilization' (IS). There are many other marketing names, of course - "Vibration Reduction", "Super SteadyShot", "Mega O.I.S.", but all refer to the same concept - shifting some sort of camera element (usually the sensor or the lens) to compensate for camera movement.

For this week's assignment, we're going to test how effective these technologies are. Like last week's assignment, you'll take a series of photos with the image stabilization function off, and take another series of photos with the IS on, measuring how many shots were acceptably sharp (not blurry), under your own subjective judgment.

For students without IS on their cameras, we are going to simulate image stabilization using just a plain old ordinary camera. For a camera with IS, the stabilization works by shifting a lens element or the sensor to 'realign' the light onto the correct part of the sensor. While we can't do this in real-time like the camera can, we can easily achieve the same effect by processing the pictures afterward.

To simulate this, we're going to divide up our image up into multiple slices, and then rearrange the pieces to align with each other. This is similar to what a camera's IS is doing - taking the light or sensor and shifting it to line up in a consistent position. Take a photo with a long exposure time (maybe 1s). Then take a series of photos at a fraction of the time (keeping all other settings like aperture and ISO the same) - say 8 photos at 1/8s, or 10 photos at 1/10s. Load all of these photos on top of one another and add the light information - the result should be an image with the same exposure as the full exposure, and also with a similar amount of blur. The key is that since we have multiple image slices, we can shift these around - by realigning each of the image slices on top of one another, we can give the photos the same blur properties as a 1/8s photo.

For the software stabilization assignment, you will need some sort of image editor. Adobe Photoshop and Photoshop Elements will work fine. Alternatively, there are a couple of free image editors that will do the same trick. You can download either GIMPShop or Paint.NET from the links on the resources page. The exact process is detailed out below for Photoshop (Elements should be similar) and Paint.NET. You can also try GIMPShop and/or GIMP, although I have not yet figured out the user interface and probably won't be able to help you. If you do not have access to a Windows PC, Paint.NET will not work, so let me know if you don't have access to Photoshop or a Windows computer.

Samples
Full exposure, 1s shot - blurry
1/8 exposure, 1/8s shot - dark
Full exposure, eight 1/8s shot combined
1/8 exposure, 1/8s shot - noisy
Completed image stack of 8 1/8s exposures - Photoshop .psd format, 32MB
Completed image stack of 8 1/8s exposures - Photo.NET .pdn format, 29MB

Adobe Photoshop (Photoshop Elements should be similar)

  1. Take a picture at full shutter speed (1s).
  2. Set your shutter speed to some fraction of the full time - say 1/4s. Make sure you leave all other settings - aperture, ISO - the same. If your camera can't manually adjust these parameters, there should be an "Exposure Compensation" (+/-) option on your camera. Dial it down to -2 or -3EC.
  3. Take a series of photos at the reduced shutter speed. The number of photos should correspond to the time fraction - if the full expsore is 1s, and you took 1/4s shots, you need to take 4 pictures to get the same exposure.
  4. Load all the images into Photoshop.
  5. Stack an image on top of another other by selecting all (Ctrl+A) of the 1st image, and copying and pasting on top of a 2nd image. Under Layer->Layer Style->Blending Options, select the "Linear Dodge" blend mode - linear dodge does a mathematical addition of the two images
  6. Shift the top image around until it aligns with the bottom image. (try looking at a bright feature like lights or text, and use the up/down/left/right arrows to fine tune).
  7. Repeat step #5, adding each of the images onto the stack
GIMPShop
  1. Download GIMPShop
  2. After trying this for an hour, I've realized that I'm completely clueless as to navigating the user interface of GIMPShop - despite its claims it's not anything like Photoshop whatsoever. If you'd still like to give GIMPShop a try, the process is similar to the Photoshop steps above - set the blending mode to "Additive", and align the images on top of each other.
Paint.NET
  1. Download Paint.NET
  2. Take a picture at full shutter speed (1s).
  3. Set your shutter speed to some fraction of the full time - say 1/4s. Make sure you leave all other settings - aperture, ISO - the same. If your camera can't manually adjust these parameters, there should be an "Exposure Compensation" (+/-) option on your camera. Dial it down to -2 or -3EC.
  4. Take a series of photos at the reduced shutter speed. The number of photos should correspond to the time fraction - if the full expsore is 1s, and you took 1/4s shots, you need to take 4 pictures to get the same exposure.
  5. Load the first image into Paint.NET (drag and drop the file in)
  6. Load another image in, but this time choose "Add into Layers". This will automatically stack it on top of the first photo.
  7. Under Layers->Layer Property->Blending->Mode, select the "Additive" blend mode - addidtive does a mathematical addition of the two images
  8. Shift the top image around until it aligns with the bottom image. (try looking at a bright feature like lights or text, and use the up/down/left/right arrows to fine tune).
  9. Repeat step #6, adding each of the images onto the stack

Assignment Summary: