Halo artifact from NIK SEP2

[Snarkbyte]

The B&W conversion for this shot was done in Nik Silver Efex Pro 2, and there is a slight halo visible around the bell and along the upper arch on the inside. I used a lot of small control points along the edge of the bell, trying to eliminate the effect, but I'm not having much luck getting rid of it completely. Does anyone know of a reasonable way to avoid/eliminate this? (Without using a hundred tiny control points, that is.) I noticed in another thread that rpcrowe was getting a similar effect using Topaz. Any help appreciated, surely there must be some way of making local adjustments in SEP2 with control points that does not leave a halo.

[MiniChris]

I generally find most halos come from over-sharpening.

[Snarkbyte]

I generally find most halos come from over-sharpening.

Now why didn't I think of that? Thanks, Chris, you're probably right. SEP2 has 2 separate sharpening controls, called "Structure" and "Fine Structure". The problem is, these controls are simple sliders, so you can only adjust the amount of sharpening, but not the radius or detail, and there's no masking. I suppose the simplest solution would be to ignore the structure controls in SEP2 and perform sharpening in LR or Photoshop, where I have more control, after B&W conversion in SEP2. I'll give it a shot and see what I can do. Thanks!

[Bobobird]

I generally find most halos come from over-sharpening.

Pushing up the clarity does that too.

[Rob Douglas]

I'll second the sharpening Chris mentioned and also add the possibility of the sky being darkened a bit too much in conjunction will show a halo also.

[Snarkbyte]

Here's another experimental version. There is still some halo visible inside the arch, but I've learned that the halo only appears when I use control points in SEP2 for local adjustments. Global adjustments in SEP2 do not show a halo. I'll keep experimenting with this to see if I can find a solution, but it may be that the area adjusted by the control points just have to be small to prevent the halo, which means a control point will need to be replicated many times to cover an odd-shaped area.

[revi]

Another source of wide halos could be local contrast enhancement (which, at least in one form, is unsharp masking with large radius and low intensity, like 100 pixels @ 0.1). Some tone-mapping routines also seem to cause halos (judging from some images in here).
Not knowing SEP2, I can't tell if that plays here, though.

[MiniChris]

Almost anything you do to an original image is capable of causing a halo. To a large degree, they can be controlled but I am not convinced they can be totally eliminated if there is more than two or three layer adjustments. It is the nature of the pixel to bend, distort, recolor and change tonal values, but always at a price.