Sharpening

[Chris Bell]

I'd be curious to hear about favored methods of sharpening. Where do you set the OLPF compensation? Low, Medium, or High detail? No sharpening in RedCine...do you sharpen in post with the tool in Final Cut?

Advice welcome....

Chris

 

[michael zaletel]

Hi Chris:

Sharpening is tricky enough for still images. There are so many variables to consider. My preference is to leave sharpening until the very last step after grading even. Reason being that when you make any changes to color balance, gamma, contrast, tint to match shots, you modify the edges of images. If you do the sharpening prior to grading, you run the risk of emphasizing or downplaying your sharpening in certain parts of the image. Also, just like with still images for print, you want to apply different sharpening settings (or unsharp mask or smart sharpen settings) for different output resolutions. For example, you would use different settings for the HD version than you would for SD version or web resolution versions. Also, compression for web can do crazy things to pre-applied sharpening. Therefore, my advice is to avoid sharpening, unsharp masking or smart sharpening until you've finished the master, then apply sharpening based upon how you are planning to output or display the footage and remember to use different settings for different output sizes, resolutions, compressions.

EDIT: Also, many times it is best to sharpen locally within a feather-edged mask or use same method to apply varying levels of sharpening.

-shooter

 

[Chris Bell]

What software are you using to perform sharpening?

 

[Costelloe Michael]

Hi Guys,

This is a topic that I'd quite like to hear more about and therefore post this reply as a way of getting it back onto the top of most recent posts.

Where are your favoured methods of sharpening aplied and what do you find gives best results. (I find my workflow is based on whatever the client is most used to and a lot of the time I have little control over paying jobs.)

So come on guys get with the replies!

Mike C

 

[Shawn Booth]

What software are you using to perform sharpening?

I follow the "sharpening" whenever it is brought up. I believe most employ the After Effects>Unsharp method. At the end of all work. Final step before output.

 

[Chris Bell]

Shawn,

What settings do you use in Red Cine? Detail setting? OLPF?

 

[Ramesh Jai]

I follow the "sharpening" whenever it is brought up. I believe most employ the After Effects>Unsharp method. At the end of all work. Final step before output.

What settings have you found to be good in AE's unsharp mask? Thanks

 

[Dan Hudgins]

Sharpen at final size

Sharpen at final size

I would add that you should only sharpen up at the final output size, if you sharpen a 4K image then down res to DVD size you can get worse aliasing.

In fact before you down size you may need to blur the image to avoid aliasing in the smaller version.

If you sharpen early on too much you can get the video look and worse noise in the images.

Sharpen/unsharp mask can help with using fast lenses like f/1.2 since they have a halo around the image details.

 

[Brad Stoddard]

When using Red Alert to output 2k ProRes HQ for pre color grading, what settings should be used for Chroma Denoise/De-Moire and PLPF Compensation if I want to maintain a pristine workflow and not introduce artifacts?

The default setting are maximum and high respectively. Debayer Detail is also default at high.

Thanks

Brad

 

[Shawn Booth]

Shawn,

What settings do you use in Red Cine? Detail setting? OLPF?

RC -

Sharpen - leave on default 50.00
detail: Medium
NR: Enabled

I'm the first to admit though, I'm still learning.
I also believe though, every shot requires its own settings. "This shot gets HIGH Detail" while "That one gets 0 Sharpen"....

Honestly, I don't know...

 

[Dan Hudgins]

"best" setting...

"best" setting...

every shot requires its own settings.

Although everyone will have their own idea of what the "best" setting is for any value that changes the image, it is true that you cannot use the same settings for shots done on location under various lighting conditions, lenses, and f/ stops.

When you open up the lens the "plate effect" of the OLPF and cover glass on the sensor may make the image softer, so you may need higher settings of the unsharp mask for shots done at f/1.2 than at f/6.3.

Also the radius of the unsharp mask to use depends on the size the image will be viewed at, for IMAX you would use a smaller radius than you would use for an ART HOUSE size (viewing angle) screen, but for an Ipod you would need a large radius in relation to the screen width, so it can be hard to tell what will work, its best to under sharpen than to over sharpen.

Since LCD TVs may have built in sharpen if you make the source frames too sharp you may get the result over sharpened on the end viewers TV.

If you sharpen too much in the 4K files, when you down res to DVD size you can get more aliasing artifacts and noise.

Sharpen was not a problem with 35mm film, it just got softer with each printing.

 

[Chris Bell]

Sorry to sound like a noob, but how does unsharp mask sharpen an image? Is this tool in Final Cut or is there a good plug-in available?

Also, I find the sharpening in my broadcast monitor can greatly effect the decision making process. Where do you guys set the sharpening on your monitors?

 

[Dan Hudgins]

image minus un-sharp = sharp

image minus un-sharp = sharp

Sorry to sound like a noob, but how does unsharp mask sharpen an image? Is this tool in Final Cut or is there a good plug-in available?

Also, I find the sharpening in my broadcast monitor can greatly effect the decision making process. Where do you guys set the sharpening on your monitors?

In the old days of optical printing you would shoot a pan low con positive off the color negative, then put that positive in the back projector of a two head optical printer a little out of focus, then make a color master positive off the combo of the out of focus pan positive and the color negative, that gave a lower contrast sharper master positive.

Because the pan positive was un-sharp it subtracted from the sharper negative and made the result sharper, that is you subtract un-sharpness to get sharpness.

The same idea works in software to get more sharpness. In my programs the radius of the un-sharpness is adjustable as is the edge detection, using both of those adjustments and using different values for the R,G, and B helps reduce noise and make edges sharper. The radius needed changes with the size of the image and the angle of view since if it is too small it will be less than what the lens in the movie projector can resolve, or your eyes see as sharpness. Other programs have other adjustments but the idea is the same.

In a TV sharpness is sometimes called "peaking" for the electronic circuit that makes the edges sharper, it boosts high frequencies.

There may not be an adjustment in all TVs and monitors, but most monitors and TVs have peaking built in.

You should load a known image that looks good on everything, such as test pattern, and adjust the sharpness for flat MTF and then do not tuch the monitor controls again. If you see a white or light outline around a light gray square on a dark gray background you have the sharpness too high.

 

[Shawn Booth]

Posted this in another thread - more appropriate here though.

"We know sharpening is not added unless we add it. Most say it's up to you as to how much to add - fine. But the question is, Should we add sharpening? Should it be added in RC/RA? Why not at this point? I understand waiting until the final step, but it makes logical sense that if it's added when 1 lighting, the image stays sharp and can withstand the riggers of post production all the way 'til the end. I don't see how creating a Quicktime that has been sharpened will adversely affect the footage."

1st Response (from Branden Kramer)= "Personally, I usually add sharpness back in to some degree in Redcine when I am generating my conformed transcodes. Depending on how much scaling I am doing I might add more or less sharpness. I also adjust the detail and if i need to I add noise reduction. I would much rather add sharpness while reading the raw files than affecting the transcodes later on. Sometimes I would choose to hold of on noise reduction till Color, but I rarely need this with RED footage anyway."

Generally speaking - Would you use more 'Sharpening' if delivering to DVD/HD source or less? Theatre distribution (digital/film)?

Better question - how do you tell what's too much versus not enough?

 

[Charles Angus]

If you can tell its been sharpened, you've used considerably too much.

You would probably sharpen your 4k files little if at all for DVD, the sharpness would come through as aliasing.

For HD, again, little if at all.

I would wait until the footage was resized and projected/viewed as it is intended to be viewed to make decisions about sharpening.

Doing sharpening wrong is the easiest way to make your footage look like crappy daytime TV soaps.

 

[Chris Bell]

I am find it interesting that this is not the number one topic discussed here. I am so bewildered and confounded by this, I am beginning to think something is wrong with my camera. My lenses are in top shape (Angenieux HR and Zeiss Primes), the backfocus is properly set on the camera, but I am still not convinced I get a sharp image. I find the proxy files total unusable as they are dark and out-of-focus. I have to all but max out the detail and sharpness in Red Cine to get what I consider proper sharpness. I tried Red Rushes, and the results were unacceptable. I would never show a client unsharpened footage.

What the hell am I doing wrong?

 

[Nicholas Shields]

Chris, what shutter speed/angle are you using?

 

[Shawn Booth]

180

I always have to do something...

 

[Dan Hudgins]

sharp?

sharp?

but I am still not convinced I get a sharp image. What the hell am I doing wrong?

In part you are not doing anything "wrong", the OLPF in the camera divides the light over four pixel groups, so unless you sharpen the edges will look soft.

The whole point of the OLPF is to soften the edges so that aliasing and chroma moire are reduced. Also there are software filters to low pass the images to reduce chroma moire in part caused by the compression used and wavelet artifacts.

Without all the low pass filters the images would be full of colored splotches, which can still be seen if you bost the chroma matirx in post.

35mm Film does not look sharp at 4K without some sharpen as well, RED ONE probably looks lower noise and sharper for equal noise if exposed well.

Another issue is that the OLPF may add more softness when fast lenses are used, if you want sharper results try shooting at f/6.3 to f/8, not smaller or larger.

Sharpen is part of the de-mosaic (de-Bayer) process, and it needs to be adjusted for:

1) The lens used.

2) The f/ stop used.

3) The subject contrast and detail.

4) The lighting ratio and angle, i.e. raking light etc.

5) The sensor area used.

6) The thickness of the OLPF.

7) Any "noise" filters selected.

8) The "quality" settings and output size.

9) The size of the monitor and type viewed on.

10) The way the filmout will be done and projected.

11) The way the images will be downsized to avoid aliasing and noise.

12) The compressed formats for distribution to be used.

13) If 16bit 48bpp TIF, lower bpp, or compressed frames are to be made.

14) The distance of the subject from the camera and its size in the frame.

15) The K value of the light and if an IR/UV filter is used.

16) Do not focus by scale the OLPF and sensor cover glass may make small shifts when the lens is at various distance and f/ stops.

 

[Anson Fogel]

What tools?

What tools?

Lots of great info here, but almost nothing on the tools themselves - what specific pieces of software/plugins are people using? The tools in FCP and color off the shelf seem less than sophisticated. Is there an adobe smart sharpen equivalent plugin for FCP, Color for those of us not using Adobe workflows? How do they compare to redcine?