Depth Refocusing
How?
Since all the images were taken from a single axis, we know that the front parts of the image change from angle to angle, yet the back part of the image stays roughly the same. Therefore, if we were to take the mean of all the images the front parts will tend to be blurry, while the back will remain clear. We take this a step further by shifting the images so that different parts of the image stay in the approximate same place, while the other parts vary, creating a blurring effect. In our case, we shifted it towards the middle of the image and changed the specific shift using an "alpha" scalar value. In our case we used a range of alpha values [-4.0, 1.0] with 0.25 steps.
Results!
Details
Alpha = 1.0
Alpha = 0
Alpha = -1.0
Alpha = -2.0
Alpha = -3.0
Alpha = -4.0
- © Untitled
- Design: HTML5 UP
Since all the images were taken from a single axis, we know that the front parts of the image change from angle to angle, yet the back part of the image stays roughly the same. Therefore, if we were to take the mean of all the images the front parts will tend to be blurry, while the back will remain clear. We take this a step further by shifting the images so that different parts of the image stay in the approximate same place, while the other parts vary, creating a blurring effect. In our case, we shifted it towards the middle of the image and changed the specific shift using an "alpha" scalar value. In our case we used a range of alpha values [-4.0, 1.0] with 0.25 steps.
Results!
Details
Alpha = 1.0
Alpha = 0
Alpha = -1.0
Alpha = -2.0
Alpha = -3.0
Alpha = -4.0
- © Untitled
- Design: HTML5 UP
Alpha = 1.0
Alpha = 0
Alpha = -1.0
Alpha = -2.0
Alpha = -3.0
Alpha = -4.0
- © Untitled
- Design: HTML5 UP
- © Untitled
- Design: HTML5 UP