Miniatures
Picture Time!
In this project, we create fake miniatures by simulating the effects of selective focus cameras, also known as Tilt Shift. We do this by first setting a focus line, so that we can specify our depth of focus, which will return the same amount of focus as the original image. Next, we increasingly blur the image around the focus plane to simulate a depth of field effect. We do this by applying Gaussian filters multiple times to sections of the image that are above or below the depth of focus. In our case, we segment the rest of the images into 5 sections and progressively add another Gaussian blend to each section further away from the depth of focus. In this way, we will imitate photos of miniature models. In this project, we use two images taken on campus and two images from Flickr.
Outside Doe Library
Trees on Campus
Greece (by Bilwander)
Seoul (by Bilwander)
Blurring
One challenge faced while trying to increasingly blur these images was to find the perfect sigma value so that the blur would not show the lines where the blur was increased and to create a more natural blur that looks more authentic. We eventually find that a sigma value of 1 is the most natural while still achieving the blurring effect we want. Although it might not be as apparent, upon closer examination blurs with sigma values above 1 have lines that show when the blur intensifies.
Sigma=5
Sigma=2
Sigma=1
Sigma=5
Sigma=2
Sigma=1
Failures
Failures were encountered when the resulting image was not normalized or if the image was in CMYK instead of RGB. These images were also by Bilwander.
Results!
Finally, all the cases that looked great!
Outside Doe Library
Trees on Campus
Greece (by Bilwander)
Seoul (by Bilwander)
- © Untitled
- Design: HTML5 UP
In this project, we create fake miniatures by simulating the effects of selective focus cameras, also known as Tilt Shift. We do this by first setting a focus line, so that we can specify our depth of focus, which will return the same amount of focus as the original image. Next, we increasingly blur the image around the focus plane to simulate a depth of field effect. We do this by applying Gaussian filters multiple times to sections of the image that are above or below the depth of focus. In our case, we segment the rest of the images into 5 sections and progressively add another Gaussian blend to each section further away from the depth of focus. In this way, we will imitate photos of miniature models. In this project, we use two images taken on campus and two images from Flickr.
Blurring
One challenge faced while trying to increasingly blur these images was to find the perfect sigma value so that the blur would not show the lines where the blur was increased and to create a more natural blur that looks more authentic. We eventually find that a sigma value of 1 is the most natural while still achieving the blurring effect we want. Although it might not be as apparent, upon closer examination blurs with sigma values above 1 have lines that show when the blur intensifies.
Sigma=5
Sigma=2
Sigma=1
Sigma=5
Sigma=2
Sigma=1
Failures
Failures were encountered when the resulting image was not normalized or if the image was in CMYK instead of RGB. These images were also by Bilwander.
Results!
Finally, all the cases that looked great!
Outside Doe Library
Trees on Campus
Greece (by Bilwander)
Seoul (by Bilwander)
- © Untitled
- Design: HTML5 UP
One challenge faced while trying to increasingly blur these images was to find the perfect sigma value so that the blur would not show the lines where the blur was increased and to create a more natural blur that looks more authentic. We eventually find that a sigma value of 1 is the most natural while still achieving the blurring effect we want. Although it might not be as apparent, upon closer examination blurs with sigma values above 1 have lines that show when the blur intensifies.
Sigma=5
Sigma=2
Sigma=1
Sigma=5
Sigma=2
Sigma=1
Failures
Failures were encountered when the resulting image was not normalized or if the image was in CMYK instead of RGB. These images were also by Bilwander.
Results!
Finally, all the cases that looked great!
Outside Doe Library
Trees on Campus
Greece (by Bilwander)
Seoul (by Bilwander)
- © Untitled
- Design: HTML5 UP
One challenge faced while trying to increasingly blur these images was to find the perfect sigma value so that the blur would not show the lines where the blur was increased and to create a more natural blur that looks more authentic. We eventually find that a sigma value of 1 is the most natural while still achieving the blurring effect we want. Although it might not be as apparent, upon closer examination blurs with sigma values above 1 have lines that show when the blur intensifies.
Failures
Failures were encountered when the resulting image was not normalized or if the image was in CMYK instead of RGB. These images were also by Bilwander.
Results!
Finally, all the cases that looked great!
Outside Doe Library
Trees on Campus
Greece (by Bilwander)
Seoul (by Bilwander)
- © Untitled
- Design: HTML5 UP
Failures were encountered when the resulting image was not normalized or if the image was in CMYK instead of RGB. These images were also by Bilwander.
Results!
Finally, all the cases that looked great!
Outside Doe Library
Trees on Campus
Greece (by Bilwander)
Seoul (by Bilwander)
- © Untitled
- Design: HTML5 UP
Failures were encountered when the resulting image was not normalized or if the image was in CMYK instead of RGB. These images were also by Bilwander.
Results!
Finally, all the cases that looked great!
Outside Doe Library
Trees on Campus
Greece (by Bilwander)
Seoul (by Bilwander)
- © Untitled
- Design: HTML5 UP
Finally, all the cases that looked great!
Outside Doe Library
Trees on Campus
Greece (by Bilwander)
Seoul (by Bilwander)
- © Untitled
- Design: HTML5 UP
Finally, all the cases that looked great!
- © Untitled
- Design: HTML5 UP