Ok to help Bernard not fall too deep into a FF hole, I created the attached PASeq. It inflates and image and makes it look like metal or plastic.
It has essentially 2 working steps after bringing the image to the canvas. The first is a smoothing step which repeated. For most image you can turn of one of the smoothing steps. The reason to smooth the image first is so that the "inflation" is not full of bumps.
the Second action (third step) is a MSG operation. This has also two "working" steps. the first 1GradientFloatSum is doing a gradient lighting operation. Leave the algo at 1 and test variations on the Dim, azimuth width and elevation. These all work in concert to create the look. The Azimuth essentially defines the directional source of the lighting. the Dim is a blur. The second working step of the MSG is the lighting. This step is controlling the specularity. The way I would tune a particular image is to get the gradient values close.. then pop down to the specularity get that to where you want it, then pop back up the the gradientfloatsum and fine tune it.
But you may have your own method.
cheers
Replies
Here's a quick edited variant of liveart's MSG preset that incorporates a blur option for smoothing directly in the MSG preset.
SrcSmoothLightGrad1.zip
Interesting, doesn't actually give the same results as posted, and when I went to tune the blur filter SA cashed..
OK, the crash is related to the edited blur parameter being out of range. Fixing it now. Don't set it to a negative value for now.
I must have accidentally scrolled to a neg number.
However, even when you fix it it is not going to generate quite the same quality of result as the PASeq. Is there a different blur operator in the MSG that might be used to better approximate the result of the Smooth ImOp using the preserve edge algo and triangle as kernal. Plus I was mixing at only 80% in the ImOp.. so that might be part of it.
Maybe a minmax followed by a rankarea...
The nature of the blur is going to be different. I wasn't really trying to exactly duplicate your paseq preset, just show another approach to smoothing it. If you add a nonlinear transfer curve (BiasGain processor) to the gradient processed luminance channel you can also shape the overall effect.
...
I don't think there's anything directly related to the Smooth Ip Op in MSG at present. FSA filters are cool because they are resolution independent. But they aren't really something that can be adaptively modulated like the Smooth filters are to preserve edge details.
and (crash) again.. to get results like my PASeq the blur dim% will have to set at a decimal value less than 1.0
Thanks Michael, I appreciate your sense of pity...
I had similar results :
https://studioartist.ning.com/photo/volume?context=user
The difference, is that you affine the process with the smooth imop operations in your PASeq preset.
Good idea.