The goal for this part of the project was to create the fake miniature effect by artifically manipulating the field of view of an image to simulate a tilt shift camera.
1 . In order to generate these images, the user selects a point on the image. This point dictates where the field of view is centered on. From there, the image becomes progressively blurry the farther away you get from that point vertically (across one row in an image, there is no difference in "blurriness"). This is achieved by creating several bands across the image that are taken from different levels of a gaussian stack of the original image. The lower and upper bounds for the field of view is determined by the number of these bands that are used. The final steps are to increase the saturation of the image and also to do one last convolution with a gaussian filter. The saturation gives it a more "fake" color to make it look more like a miniature. The final convolution is useful because it hides the high frequency parts of the image. This makes it look more like an actual miniature because in miniatures it is difficult to replicate all the details present in the real thing which substantialy less space. For bells and whistles I created a fake miniature video.
Here are my results. The last two images are taken by me.
| Original | Miniatures |
|---|---|
| SF Skyline | |
| Source | |
|
|
|
| Swiss Valley | |
| Source | |
|
|
|
| Ellsworth and Durant | |
|
|
|
| View from the Metropolitan | |
|
|
|
| Durant and Ellsworth video | |
|
|
|
The goal for this part of the project was to duplicate the Dolly zoom. This was done by taking many pictures of one object, each at a different distance away. Each of the images however, would be zoomed in such that the object took up the same amount of the image. This created an interesting effect with the background. I used a friend's camera for this project. One limitation is that the number of pictures that could be taken was determined by the amount that the camera could zoom. With that limitation, we had a hard time creating gifs with a lot of frames. Bells and Whistles: I have made gifs of the vertigo sequences and are shown below.
| Result 1: Jack | |||||
|
|
|
|
|
|
|
|||||
| Result 2: Sony Theakanath | |||||
|
|
|
|
|
|
|
|||||