Hi there
this is a tutorial showing how we can use a rather sophisticated combination of composite nodes for replicating the so-called “Unsharp masking” image (or video ) sharpening technique,known and used already from the '30s on analog photographic plates.
we’re going to be presenting both the “math node” (values-based) and the “color mix/L.U.T nodes” (image color data-based) variations of this technique
Image used as texture for sky background is courtesy of and downloaded from cgtextures.com and used for non-commercial purposes
thank you for doing this tutorial. I was able to follow the thought process behind it, nicely explained, I appreciate your work and sharing the knowledge with us. This process of sharpening I used it plenty on sharpening photos on gimp and I am happy to see its possible to do it on blender as well. In further simplifying your way of work, and makig a bit more controllable as far as the multiplying factors of the image goes I have gone a step further (hoping you dont mind) and made it possible to have it controlled by numbers than graphs which are not 100% accurate if you wanna adjust it in smaller amounts. I attach a picture of the node setup I used based on yours of course. All I have done is creating the vectors required based on a single input which is the “sharpening factor”. Then I multiply that with the blurred picture and the remaining… ( 1/(1-fac)) I multiply it with the subtracted image result thus restoring the brightness to the original levels. It would be nice if Blender had “vector math” node so that we could bypass such things… it wouldnt be hard to implement anyway
Hope you like this method too… and cant wait to see more of your tuts.
Mike
P.S. I havent trying making the factor 1 as it would cause a divison by 0 lol but you are welcome to try and tell me what happens!
excellent variation,indeed it offers better control and feels simple yet powerful,I will try to working further on this,thank you for the contribution,Peter
