Lightfield Camera
For this project I used Stanford’s Light Field Archive to create change and aperture after taking the pictures.
Depth Refocusing
To change the focus point, I had to realign the images before all stacking them. Without chaging the images at all, I get a focus of the back row of the board:
Versus displacing them by the difference to the center image’s displacement gives an image that is aligned to the front row:
Here’s an animation of refocusing accross multiple depths:
Aperture changing
To change the aperture, I displayed a subset of the images. To emulate a small aperture, choosing only the center image gives the following result:
Whereas displaying all the images gives a more blurry image:
.
To choose a subset of the images, I used a circular weight function. Here are the weights for a radius of 5:
I also tried to mess around with a gaussian distribution of weights in attempts to get smoother images:
Here are the different results:
Circle |
Gaussian |
|
|
You can see that the gaussian distribution results in some more smooth transitions through different radiuses.
What I learned
Lightfields are pretty cool! The ability to change depth/focus, aperture, and even the center of focus is pretty neat. It was also simpler than I thought it would be!
Link to Image Quilting
Link to AR
Lightfield Camera
For this project I used Stanford’s Light Field Archive to create change and aperture after taking the pictures.
Depth Refocusing
To change the focus point, I had to realign the images before all stacking them. Without chaging the images at all, I get a focus of the back row of the board:
Versus displacing them by the difference to the center image’s displacement gives an image that is aligned to the front row:
Here’s an animation of refocusing accross multiple depths:
Aperture changing
To change the aperture, I displayed a subset of the images. To emulate a small aperture, choosing only the center image gives the following result:
Whereas displaying all the images gives a more blurry image:
.
To choose a subset of the images, I used a circular weight function. Here are the weights for a radius of 5:
I also tried to mess around with a gaussian distribution of weights in attempts to get smoother images:
Here are the different results:
You can see that the gaussian distribution results in some more smooth transitions through different radiuses.
What I learned
Lightfields are pretty cool! The ability to change depth/focus, aperture, and even the center of focus is pretty neat. It was also simpler than I thought it would be!
Link to Image Quilting
Link to AR