Light Falloff

[Marpel]

Latest version.

Experimenting, I took a number (8) of photos of a boat masts in a foggy harbour (you can only see the masts, against a, foggy, white background). In each photo a bird is seen flying across, from right to left. I processed the images out of my raw editor, with no changes. Normally, I would use the vignette control in raw, but did not in this case. I Composited each image, masking in each position of the bird, so the final image has the bird in eight positions across the frame. not a wall hanger by any means, but it was an experiment.

Once the final Composite was done, I then decided to do tone and other changes, including vignette control, as the perimeter of the image is noticeably darker. I could not use the raw editor's control as it apparently would not work on a TIFF, so went to PWP.

I found Light Falloff, as it suggested it could be used for vignette control.

However, I read the Help File and, although it provides some explanation, it does not describe how to actually go about using the Transform, so a couple questions:

A sentence mentions a default curve illustrated above, but this curve (only one above) appears to be a user-made curve. Is the default curve just a straight line across the top of the box?

By the way, before I forget, the Help File has a link to Cosine Falloff, which has a further link to Test Patterns (where I found a vignette mask). The first sentence directs the user to find Test Patterns through Transformations>Special Effects. It should be Transformations>Test Patterns.

The Help File has two examples of generated Curves with a great number (33?) of points, but there is no explanation as to how to accomplish these points. I initially thought one could Shift-Click in the image to place a point (like the Curves Dialogue), but this does nothing. Is one supposed to randomly place 33 points along the Curve line, and if so, how does one establish where these points should be placed and then what shape the Curve should be? And why so many points?

What is the practical purpose of "Compute Falloff Curve"? I tried this but got the image with a number of white circles, presumably because the image was not a plain white image with darkened perimeter? If it is only meant for a white image, where would the average user ever use this?

I finally threw in the towel and decided to use the Test Pattern as a reversed mask (to lighten the perimeter) in conjunction with Compositing the image on itself with the Subtractive Filter Operation. I had to do this a couple times until I reached the best, visual, result.

Marv

[jsachs]

Light Falloff is designed to let you calibrate out vignetting from a lens by first photographing a uniformly illuminated field such as a white computer screen or a white piece of paper that is evenly lit. Since you are not starting with this as an input image, it will not be as useful in your situation. If you wanted to use Light Falloff without a calibration image, you would start with a curve that starts out in the upper left corner and drops down toward the upper right corner, corresponding to the amount of vignetting you want to remove, something like this:

falloff v1.jpg

By playing with the curve you can vary the amount of corner lightening, using the output image as a guide.

For what you are trying to do, the Transformation/Gradient/Oval transformation might work better. By tweaking the gray levels on the Color Line control (you need to start by reversing the order so it runs from white to gray) and then adding more control points if necessary, you should be able to compensate for the composite vignetting. To add vignetting, I usually use Brightness Curve with a soft mask as it lets you shape the vignette to match the subject. You could also use this to remove a vignette, but it is harder to control precisely than the Oval Gradient.

Whatever method you use, without a calibration image to start with, there will be guesswork involved.

[Marpel]

Thanks Jonathan for the reply.

I may try the calibration image approach, as you describe, as it would be nice to have one handy for future use. However, as I have no real knowledge of this method, can you advise:

Would a single image suffice for each lens? I would imagine the f-stop would have some bearing (so grab an image for each anticipated f-stop and keep all in a file, perhaps), but for a zoom (in this case 70-200), does it matter much if the zoom factor is different, as long as the f-stop matches?

And, does the focus point have any influence? For the image(s) I was referring to initially, I used the 200mm end of the 70-200 and was focusing on subjects probably 75-ish feet away. Of course, a similar distance would not work for the calibration image of a piece of paper or computer screen.

Marv

[jsachs]

f stop and focal length both matter. I don't think the focus point matters significantly. The raw develop transformation can do the calibration if you create a set of images at different f stops and focal lengths, but you can only use it on raw files. The best place to remove vignetting is in the raw converter.

[Marpel]

Thanks again Jonathan,

I routinely use the raw editor vignette control, except those few times when I do as described in my initial post, as sometimes the raw editor vignette control will give slightly different results between images and it then becomes a bit of a pain to blend parts of one image into another (slight difference in the surrounding background). If I just turn the images into TIFFs without any raw editing, it is easier to blend the objects in the TIFFs. Then I just have a single image to process, including vignetting. Which is what led me to my post.

Marv