Removal of Vignetting

[Abitconfused]

When vignetting is removed in post-production, I assume that some form of brightening is applied. However, isn't the underlying image's detail reduced however slightly by this process?

[Manfred M]

When vignetting is removed in post-production, I assume that some form of brightening is applied. However, isn't the underlying image's detail reduced however slightly by this process?

A vignette is usually an oval, neutral colour (gray) gradient applied to the image. As long as none of the channels are clipped, I don't see why there would be any loss of detail.

If I were asked to do this, I would probably convert the image to the LAB colour space and work on the L* channel using a soft, white brush. There should be no detail loss as the A* and B* channels containing the colour data would not be affected and I would just be lightening the darkened areas. It would not be "perfect" as this is a hand operation, but the final results should be quite acceptable.

[Abitconfused]

Sorry, I am thinking of vignetting from lenses as addressed in Camera Raw Lens Corrections, Lens Profile, Vignetting. Thank you!

[Manfred M]

Sorry, I am thinking of vignetting from lenses as addressed in Camera Raw Lens Corrections, Lens Profile, Vignetting. Thank you!

In that case the answer is even more clear; the vignetting we see from lenses, especially when we shoot fairly wide open is due to light being absorbed by the additional amount of glass it has to travel through. Light coming through the centre of the lens will pass through less glass and have lower losses. These light losses can be measured in the lab and can be compensated for during the raw conversion process.

Unless there is channel clipping during capture, these values can be calculated and corrected. I'm not sure why you would suspect there would be data loss.

When it comes to lens profile correction and CA correction, one could say there is some minor data loss. The green and purple fringes are due to a lens design decision where the colours do not converge on the sensor plane; certain wavelengths will be in front of it and others behind it, so the sensor is recording the wavelengths of light that fall there. The correction algorithms identify these areas and calculate the values that should be there without the lens defects that cause them. Is this a loss of data, sure, but one that is desirable as the corrected version works better. The same comment goes for the algorithms that correct for pincushion and barrel distortion; data processing gives us a more accurate image, but it is not exactly what the camera recorded.