Monday, January 2, 2017

Further experiments in Super Resolution Photography

In a previous post I wrote about how to use a Magic Lantern Lua script (on the right) to create auto super resolution brackets. In this post, I offer some insight into why SR photography may be of interest to you, and a Photoshop action to automate PS-CC post processing: well at least for a four layer SR stack.

As usual my test image is a view in our house.

I first used the SR Lua script to create a four bracket SR set, using the jiggle and move option.

After ingesting into LR, I simply exported as layers into Photoshop.

I had previously created a script that took the four layers, scalled up by 200%, aligned the layers, reduced the opacity according the usual recipe for SR processing (100%, 50%, 33%, 25%), flattened the layers, and reduced the image scale back down, ie by 50%. For those interested here are the script commands:

Action: 4 Layer SR

Select back layer

Without Make Visible

3

Select front layer

Modification: Add Continuous

Without Make Visible

1, 3, 4, 5

Image Size

Width: 200%

With Scale Styles

With Constrain Proportions

Interpolation: nearest neighbor

Align current layer

Using: content

Apply: automatic

Without Vignetting Removal

Without Geometric Distortion Correction

Select back layer

Without Make Visible

6

Select forward layer

Without Make Visible

3

Set current layer

To: layer

Opacity: 50%

Select forward layer

Without Make Visible

4

Set current layer

To: layer

Opacity: 33%

Select forward layer

Without Make Visible

5

Set current layer

To: layer

Opacity: 25%

Flatten Image

Image Size

Width: 50%

With Scale Styles

With Constrain Proportions

Interpolation: nearest neighbor


As for a comparison, I made good use of the new LR-CC compare mode (in the Library module).

First, here is the base (RAW) image:


Here is the LR comparison after processing both the a single RAW .CR2 and the PS-CC processed SR image with the same settings, ie lens correction and detail sharpening.


I hope you agree, that if you are interested in getting the maximum image quality and detail out of an image, a simple 4-layer SR approach appears to have real benefits (accepting the subject shouldn't be moving). 

The SR (on the left) clearly exhibits detail and the single RAW is softer and lacking the detail in the SR version.

As usual I welcome any feedback on this approach to SR photography, ie using an ML script and a PS-CC script (at least for a four image bracket set).




1 comment: