GH2 Best way to sharpen footage in Premiere CS6?

[L1N3ARX]

I'm doing a comparison between the GH2 and the 1DX. On the 1DX, the general recommendation by most everyone is to take sharpness down to 0 and then add it back, in post. I've never had to do this before even once with the GH2 for obvious reasons... so how do you guys add sharpness generally?

The only thing I've tried so far was to add the "sharpen" filter and then proceed to adjust the slider incrementally from 1 on upwards. For some reason it doesn't feel right, the sharpness is not as natural as the stuff philip bloom has done with the 1DX / 5d III by adding sharpness in post. What filter or effect does sharpening the best in Premiere CS6?

 

[L1N3ARX]

Nevermind, Unsharp mask works perfect for this, came out just like I wanted!

 

[downsouthstudio]

I just installed CS6...can you explain the "unsharp mask" and how it works, and why its better to sharpen in post rather than shoot sharp.
Thanks,
Jeff

 

[j1clark@ucsd.edu]

I just installed CS6...can you explain the "unsharp mask" and how it works, and why its better to sharpen in post rather than shoot sharp.
Thanks,
Jeff

"Unsharp Mask" is a filter which increases 'edge contrast' in images that gives the human observer the impression of 'sharper'. When filters are used on data, and the filter operation is limited to a 'neighborhood', such as a small 'square' of pixels(*), the small area is called a 'mask'. Using the word 'unsharp', means that the mask is not 'sharp'... but smoothly goes to zero at the edges.

Example for a 1 dimensional 'unsharp mask'... One wants to average 5 pixels in a row. A 'sharp' mask would be (1,1,1,1,1), that is all pixels get equal weight in the average. An 'unsharp mask', would be something like, (.25,.5.1,.5,.25) so the pixels are not given equal weight in the average. The 'central pixel' is given 'full value', but the pixels at the edge of the mask are only given 1/4 value in the average.

Performing this operation in 'post' is better because the amount of 'sharpening' used can be tailored to the output device. An 'in camera' sharpener is 'tuned' to a specific 'assumed' output device, which may or may not fit actual device used.

*The mask can be arbitrary, and need not be a 'square'. That shape is 'convenient' for rasterized images...

Historically the 'method' used a full sized 'blurred' image and the 'normal' image to achieve the apparent 'sharping' effect on the human vision system. When one 'blurs' a digital image one uses a 'gaussian' (frequently) which has the 'averaging' shape mentioned above.

 

[Chris Adler]

I've been finding pretty good results with tracked mattes that have Unsharp Mask applied to them. I had a few shots where the wall behind the actors was in focus and the face wasn't so much. Sharpening the whole frame wouldn't have bought much since the ratio between the foreground and background doesn't change if you apply the effect to the whole frame, not just elements of the face.

 

[SteB]

I haven't got much experience sharpening video footage, and I've had to put aside my plans to work on this until I get some other things sorted out. However, I have had a lot of experience trying out different sharpening regimes on stills. I take lots of greater than life-size macro photographs with apertures well into the diffraction limited zone. So I've worked on ways to restore the apparent sharpness, whilst also avoiding ugly looking oversharpening.

What I discovered was that the pixel radius was key to the appearance. What's more, the pixel radius effect depends on the resolution of your image. So a pixel radius of 1 is fine on full resolution still images. However, if you reduce the resolution to make a web image similar in size to video format resolutions, it will look oversharpened and produce halos. I find a pixel radius of about 0.3 much better. I haven't mentioned specific sharpening types because I've found the same approach works with all of them. The pixel radius also alters the effect a lot. If you use a very wide pixel radius of several pixels at a lower strength it will greatly increase the micro-contrast of the image, which can also greatly increase apparent sharpeness. Care is needed though or it looks coarse. Whereas a pixel radius of less than 1 works more on fine detail like hair. In fact I often use several different applications of sharpening at different pixel radii to change different parts of the image. Although it needs to be done carefully and subtley to avoid looking over-sharpened.

The other thing I noticed is that halos, and their avoidance are key to natural looking sharpening. Even when you can't clearly see the halo, it seems as if the eye is still aware of it. Also it is much easier to apply sharpening to a flat image without a lot of clipping in it. Clipped whites especially tend to produce ugly artifacts. So generally I don't set the black or white points until after sharpening. The sharpening itself can produce regional clipping which can exaggerate the over-sharpened look.

 

[goodgoings]

SteB, thanks for explaining. Very Nice!