Storm and weather pictures - how to sharpen?

  • Thread starter Thread starter Christopher E. Kincaid
  • Start date Start date

Christopher E. Kincaid

More advice needed...

My "cure" for SDS is to play with my RAW files from chases past. I am trying to learn, but I figured I would ask you guys because there are many experts here and I would rather not re-invent the wheel. Lately I have been playing with shapening so I want to ask how do you guys sharpen your photos? I have PS CS3 and I just found these scripts:

http://www.thelightsrightstudio.com/TLRProfessionalSharpeningToolkit.htm

I have tried to use them but I have no idea what I am doing and have had less than stellar results. I found out that photos shot in RAW can be soft and with my less than pro Sigma 17-70 lens I am sure they are real soft.

Again thanks in advance...
 
I don't do much sharpening at all, actually. Sharpening tends to bring out the noise in the image more than it has a positive effect on discerning details. There is also no real way to correct a photo that is fuzzy or blurred to begin with, as you can't add more data into the image (like all the bogus forensics shows and movies do). I'd say the most important thing is trying to get the image sharp when you shoot it. Filters that use an edge preserving smooth might work better as they preserve sharp lines and cancel out adjacent noise which makes the lines more pronounced. Local contrast enhance filters also make features more discernible from adjacent areas. Maybe somebody else has a technique that works, but I tend to avoid sharpening.
 
A single pass of the Photoshop Smart Sharpen filter is all you need to use. If that doesn't get the job done, you're pretty much up the creek. There are plenty of scripts out there that have lots of steps, heavy CPU usage, and claim to do it better. They're basically the photo editing equivalent of snake oil.
 
As Skip said, it's best to get it right when you shoot it. But that's easier said than done and sometimes not possible.

I use unsharp mask. If you can't fix the picture using that tool, then your picture is most likely toast. Keep in mind when shooting in RAW, you're photo is going to need a little sharpening anyway even it looks sharp.

There are some ways to make your pictures sharp as possible. Set your focus to infinity. Shoot on a tripod (with a shutter release, if you want) as much as possible, even during daytime. I've found this more annoying that helpful, but I guess it's worth noting, turn on the mirror lock function (if your camera has it).
 
I'm not big on these plugins and toolkits for sharpening.

In CS3, try using Smart Sharpen with the following settings, after flattening all layers and resizing down for web:

Amount = 50%
Radius = 0.3
Remove = Lens Blur

That should be a good starting point. Tweak the amount/radius to your liking for individual shots.

Remember, sharpening should be saved until the tail end of the workflow. Make sure the sharpening parameters in Adobe Camera RAW are turned down pretty low when you convert your RAW files. Sharpening the original, full-size image too much can make additional sharpening after downsizing (which is more effective) look bad.

As a side note, I also own the Sigma 17-70 mm. While I don't particularly like it due to color/contrast issues, it's always seemed reasonably sharp to me. Make sure to use a tripod and get the focus right, like Andrew said above. If you open a shot in the RAW converter and find it looks really soft, chances are that attempts to sharpen it in Photoshop will be less effective than if it was shot properly.
 
In my experience every image needs to be sharpened.
The question is how much.
It is different for every image.
It's even different for various areas within the same image.

I use the following settings for Unsharp mask in Photoshop CS3.

Amount 200 (always 200, I never change amount)
radius 2.0 (radius is always adjusted up or down as needed. (.25, .5, 1.0, 1.5, 3.0 etc)
Threshold 0 (always 0, I never change threshold)

I've created actions for for a half dozen sharpen levels.
Actions allow you to experiment with the image and find the correct level quickly.

Many factors contribute to sharpening side effects.
Here are just a few that come to mind;
Soft lenses need more sharpening.
Sharp lenses need less sharpening.
Underexposed images tolerate less sharpening.
Low resolution images tolerate less sharpening esp. if they've been cropped
even smaller.

Better to error on the side of softness as over sharp images just have that look of having been over-sharpend.

After sharpening examine the image at various magnifications on your screen to be sure it does not appear coarse.

EDIT; Sharpening should be the very last thing you do to an image. I always keep an unsharp version of my image in case I want
to further modify color, levels, etc. Sharpening is the last thing I do right before publishing/printing.

EDIT; EDIT; Turn sharpen and noise reduction OFF when opening the raw file.
 
Last edited by a moderator:
Another thing to consider is using a higher F-stop. There's a lot of temptation to shoot wide open, but if you have a tripod and reasonable daylight, stopping the camera down a little helps sharpness quite a bit. The usefulness diminishes around sunset with storm pictures, however, as fast cloud motion starts to become a problem with longer exposures.
 
Raw Therapee, a freeware RAW processor has a Richardson-Lucy deconvolution sharpening module that is really worth trying.
http://www.rawtherapee.com/forum/viewtopic.php?t=933&sid=f2c751c15ec10bf3123870a3dfec178a

Note: You can apply RL sharpening to ANY image, RAW, TIFF, JPG, etc.

The procedure is highly CPU intensive, and there is no quick preview function. I suggest you make small test tiles from the interesting areas on your (large) image, and use these to fine tune the numerous RL parameters. Once happy, load the full sized pic, enter the magic parameters, hit 'Go,' and retire for a minute or three to enjoy a cup of coffee.


In PS, I've found myself 'cheating' and using USM to increase local contrast. Select a huge radius, hundreds of pixels, and very low 'amount.' (maybe 10 or so.) This is much easier than layer fiddling and produces subtle yet noticeable 'pop' without any overt side effects. You can do this at any stage of processing. I usually apply this after any curve tweaking, when various subjects aren't separating from one another quite as well as I'd like.

Final sharpening, after resizing, etc. is USM with radius of 0.3 ~ 0.4 and a strength/amount of anywhere from 50 to 150 or so, depending on the image. I want just enough to highlight any fine textures in the image without creating the dreaded 'overcooked USM' halos.


All these attempts are downright primitive compared to what can be achieved with layers, magic wands, and newt tails. My version of PS (6) doesn't have Smart Sharpen, but I've tried a number of drawn-out actions that probably emulate it. All this seems overkill and is generally more than I care to routinely undertake.
 
Last edited:
In PS, I've found myself 'cheating' and using USM to increase local contrast. Select a huge radius, hundreds of pixels, and very low 'amount.' (maybe 10 or so.) This is much easier than layer fiddling and produces subtle yet noticeable 'pop' without any overt side effects. You can do this at any stage of processing. I usually apply this after any curve tweaking, when various subjects aren't separating from one another quite as well as I'd like.

.

Greg I just tried that. Pretty cool. I'm always looking for ways to pop local contrast without killing the entire image. Not sure how this will be different than other contrast techniques but I'm adding it to my tool box. Thanks!
 
Greg I just tried that. Pretty cool. I'm always looking for ways to pop local contrast without killing the entire image. Not sure how this will be different than other contrast techniques but I'm adding it to my tool box. Thanks!

:)

Do experiment with the radius value. At smaller R values, tens of pixels, the haloing effect can actually be a good thing, if applied in measured doses. (It can look awful if overcooked, be careful!) Since most software gives you a fast preview, it's easy to fiddle and tune.

You might try two large radius USM passes; one at ~100+ R, then a second at smaller R to highlight fine features.
 
In PS, I've found myself 'cheating' and using USM to increase local contrast. Select a huge radius, hundreds of pixels, and very low 'amount.' (maybe 10 or so.) This is much easier than layer fiddling and produces subtle yet noticeable 'pop' without any overt side effects. You can do this at any stage of processing. I usually apply this after any curve tweaking, when various subjects aren't separating from one another quite as well as I'd like.
Yeah, I do the same thing, and have found it to be almost a necessity for storm images. Oftentimes, even cranking global contrast as far is it will go doesn't achieve good separation between clouds and sky, sky and ground, etc.

In fact, it seems like using any radius value above ~0.4-0.5 ends up affecting local contrast more than "sharpness" (granted, the line between the two is very subjective). That's why it puzzles me that Adobe sets the default USM radius to 1.0. That's way too high for the purposes of gracefully sharpening an image, in my opinion.
 
Yeah it seems to me low dynamic range scenes are worse than high ones on digital. I get so annoyed shooting a snow or fog scene with low contrast, because the camera file is always way too low contrast. Do about anything to bring it out and wala, green-red casting to snow or fog....as in both that don't remove with white balance adjusting. I'd much rather deal with too high than too low for some reason.

As for sharpening and USM contrast, I always do them on new layers as you can sharpen/or USM contrast the majority of the image more than you can the horizon line or other sharp contrast edges. So sharpen for the rest of the image then mask down those parts that get oversharpened in the process.

Some use the high pass filter method. Layer/duplicate/filter/other/high pass. Then set it pretty low, just bring it up so the detail starts to show. Click ok. Go to layer palette and change it to overlay. That one I think is to mostly sharpen without sharpening noise. I never use it, but may down the road I guess. You can mask out noise/sky sharpening the regular way with the layer mask too.

Layer masks are the answer to about everything in photoshop lol. Not sure lightroom has gotten very good with those yet, or not. So yeah.....photoshop isn't pointless yet. Last lightroom I saw, version 2, masking was similar to what CS4 has in raw conversion. Ok and useable, but not great like in photoshop.

And if you don't know how to layer mask, with any new adjustment layer it sticks a layer mask/white box next to the layer in the layer palette. There is an icon to add one in the layers palette if you duplicated teh layer or something and don't have one showing. Or in layers/add layer mask. Back conceals white reveals....depending on how you view it I guess lol. But just do one for contrast, layer/new adjustment layer/contrast then nuke it for an experiment. Now grab the eraser set to black or a paintbrush set to black. Click on the white box/layer mask so that is what you'd be painting on. Then go back over to the image and draw around on it and that back will conceal the changes you just made while revealing it as it was below(you can click a degree of grey instead of black for some masking instead of 100%). You can do that with selection tools like select/color range and really get precise with what you are masking. Then you can blur that after the fact so it blends in. There's just a ton of control one can have over an image if they want it, all in an attempt to deal with dynamic range issues that come with any image.
 
Last edited by a moderator:
You may also only want to sharpen certain areas of an image (right click the 'blur tool' to get the 'sharpen tool') after doing a gentle Unsharp Mask in Photoshop - my settings are generally 40%, 0.5pixels radius and 0 threshold...tend to find this is pretty well all I use.
 
I use smart sharpen these days - there is a great plugin called "intellisharpen II" that I used to use, but that plugin may soon become incompatible. If you shoot RAW, some sharpening is required; if you shoot JPG, the camera will do the sharpening for you. I generally leave the final product JPGs unsharpened and then sharpen things based on whatever output the file will be. For example, if you downsize for the web, you use one sharpen setting, if you upsize for a massive print, you use another.

Re contrast and color hues: the way to get around this is to change the curves layer you've made to "luminosity" blending mode or to switch to LAB mode from RGB. The former is probably the better option of the two, as switching to LAB and then back again may change colors ever so slightly.
 
Last edited by a moderator:
Back
Top