Depth Blurring

 

WHY?

In reality a camera has a focus effect which means that elements will be in focus (sharper) while others will be blurred. 

Usually you can add depth fading  and render your scene with it, other times you may have rendered the images without depth fading and wish to add it afterwards such as this case.

This procedure gives a hands-on approach, since you can see the level of grays and can always fine tune results. This effect also helps in reducing aliasing at long distances where things get really small and lack anti-aliasing.

You may wish to add Depth Of Field within the software and avoid this process , but this often means that once the images are rendered you can't restore to the original images that don't contain DOF .

 

 

PROCEDURE

 

It consists of :

1 : the actual images themselves :

wpe2.jpg (20024 bytes)

2 : creating a new layer with the same scene which will contain depth information by using depth-fading :

wpe3.jpg (5862 bytes)

To create this image

-all lights must be removed

-all materials and textures must be removed and transferred to constant white to simplify this task you may want to use the speadsheet .

The distance must be calculated between the camera and the furthest point in your scene (where you want the elements to be the blurred the most)

calculate the distance by using Info-->distance .

Apply depth-fading to the scene : use black for color, 1 for starting (or more depending on when you want it to start fading)

and the distance value (obtained by using info-->distance) for ending distance.

In this case I put 80 and 800 which resulted to this :

wpe4.jpg (5862 bytes)

 

You need not use anti-aliasing when rendering the grayscale images.

 

COMPOSITING

Combining the 2 images the (depth-faded image and the original image ) using one of the r g or b values of the depth-faded image as the alpha to the blurred (original image) and then by compositing the results on top of the original , here are the basic steps:

wpe5.jpg (19261 bytes)

1-blur the original rendered scene image.

2-replace it's alpha-channel with the r g or b channel of the depth-faded image.

3-composite the result in alpha on top of the original (you may have to invert the alpha).

RESULTS :

 

before

wpe2.jpg (20024 bytes)

after

wpe6.jpg (20819 bytes)