Post-processing (PP) techniques? (w/pics)

[Blazer88]

I was hoping to start a discussion about post processing and how to get the most out of our images. People always talk about how to setup the camera correctly and techniques to get good pics, but I don't notice much stuff about using photoshop and other programs to help along the way. I understand that PS is in no way a "crutch" to turn a bad pic into a good one but it is very helpful.

Here is one picture that I took the other day. The first picture is a jpeg straight from the camera, nothing other than a crop to try and match the finished product. The second is a picture that has about 10 mintues or so with a few various programs.

My workflow is something like this. I process my RAW's using a program called DxO Optics. It's an amazing program that processes the RAW using camera (and lens) specific information to do color/tone correcting, fix lens distortion, lighting, and other cool things. I then take the jpeg that DxO generates and use Neat Image (depending on the ISO I shot at) to remove as much noise as possible. Then the photoshoping starts. I feel nothing is more important than color correction. I set a Curves adjustment layer and set the light/dark/gray points with the help of a Threshold layer. That makes the largest difference IMO. I then (sometimes) use a multiply adjustment layer to darken the background a bit to help the subject stand out more. It's easy to take out any floating debris with the spot heal brush and I'll also clone-stamp out any annoyances that distract from the picture. Maybe a tab of minor sharpening and I'm pretty much done. Maybe I'll have the time to do some step-by-step instructions on what I do with pictures if people are interested. Feel free to post up your own ideas on how to improve our pictures.

JPEG from Camera:

10 minutes later:

 

[jwedehase]

I'd love to know more about the curves adjustment layer and threshold layer. Could you explain a little more specifically what you're doing here? I've used PS heavily since v3.0, but I've never used layers in this fashion. I still find myself correcting a flat image, and using a lot of masks.

 

[einsteins]

Blazer88 Nice work on your fish photo...
Nice mini tutorial there...Might want to look at Nikon Capture NX it is an amazing tool....just out recently and a major improvement over the previous versions. The tools and technology they use is not like
anything else and its not complicated at all. 30 day trial is free....give it a try.
http://www.nikondigital.com/main.html?page=soft_capture_nx
go to that link and watch the tutorials for a demo of the capabilities...way cool!
I have to look into Neat Image I could use a good noise reduction tool

jwedehase
Threshold allows you to find the darkest and lightest spots in your picture...by identifying those it allows PS the make adjustments to the levels to get better color correction.

You need to work with adjustment layers because they are non-destructive to the original. Its a must for good post work in PSCS2

 

[Blazer88]

I was bored this morning so I thought I would make a little better tutorial on what I did. Here the picture straight from my DxO Optics program. This image was shot at a fairly accurate color to begin with so the difference wasn't as dramatic as some of my other pictures in the end, but this process still helps. This should answer your questions jwedehase.


This picture is pretty far from perfect, mainly the soft focus and slightly under exposed subject. I'm not going to worry about that just yet as I'm just going to color correct first.

Here is how to add an adjustmet layer. This is a non-destructive layer that doesn't change the actual image. I generally ONLY use these and never modify the original background. So the first thing to do is add a Curves layer and just click OK when the box pops up.

The next step is to add a Threshold layer. This turns the image to strictly black and white. It is very easy to identify the white and black points by chaning the slider.

The next stop is to start fixing the dark and light reference points. This picture shows the slider to the left which is showing the absolute darkest points on the picture to be used for the shadow reference. Click OK on the box and open up the Curves box again buy clicking on the black/white symbol next to the layer.

After I set the black point, I set the white one below (with droppers). Here is the lightest point to be set a white. Click the white dropper and sample the area.

So here is the picture when the black and white points set. This took away the blue color cash color cast from the picture and looks more how it does in real life.

But the problem is the middle-tone, or gray, point hasn't been set. This is probably the most confusing part on how to find the true gray point on the picture. If you open up the Curves box again and select the middle dropper, click on a few points on the picture and watch how much the picture can change. Guessing where the gray spot is is tough to do so I found a cool trick.

Here is what to do. Go to Layer, New Layer, and select OK. This will make a new layer for you. From there, go to Edit, Fill, and from the contents drop down select 50% Gray. This will turn the whole image gray, but don't worry. From the drop down box as pictured, select Difference which makes the picture look strange.

Now turn the Threshold layer back on (click on the eye next to the layer) and you should see a box similar to this. Now open up the Curves box again and select the blackest point with the middle dropper. Now you have an easy way to accurately find the gray or middle tone point in the picture.

Now turn off the Threshold and 50% Gray layer to see what you got. Now, assuming you have a somewhat calibrated monitor, you should see a picture that looks better without a blue color cast.

Keep in mind there are tons of different way to fix color. I've just found this one to be the most accurate so far. Next I'll see if I can fix the rest of the photo.

 

[scothew]

Great write up Blazer, Im defintely going to bookmark this to try these techniques next time I shoot

 

[jwedehase]

Yes, thank you for the extensive detail. I'll have to try this myself to "get it" fully. I'm excited to try something new, though. Thanks again!

 

[argo]

Blazer, thanks for your PS tip. It took me a few (well a lot) of tries to "get" it. But the the end result was worth it IMHO. It seemed to help remove that "behind-the-glass" haze. Can you recommend a good PS book for newbies? The difference between both images is more dramatic in PS. Thanks.

This image was converted from NEF file and resized:

This image was produced per your write-up:

 

[Blazer88]

Hey Art,
I'm glad someone found my write-up useful. The best book I've gone through so far are Adobe Photoshop Restoration and Retouching by Katrina Eismann and The Photoshop CS2 Book for Digital Photographer by Scott Kelby. Both are great reads and cover some different techniques. Nice looking pic BTW!

 

[argo]

Blazer, thanks for your reply. I'll check out the books you're recommending.

 

[astrogazer]

Here's a quick primer on unsharp masking I posted in an earlier thread ...


I learned how to apply unsharp masking years ago and it can make and/or break an image. For those interested I'll try to explain how it works in Photoshop.

In order to be able to really use, and get the benifits of unsharp masking you have to start with a high quality, well exposed and close to full frame size image (not a cropped postage stamp sized portion of the image). The image should not appear grainy (low ISO setting), as depending on your settings you will highly amplify that grain.

An image is usually perceived as sharp if the edges in the image are well defined, reasonably crisp and have a good edge contrast. A soft image would have a smooth transition between edges and does not have a clear line of distinction.

In Photoshop, applying unsharp mask requires adjusting 3 different parameters. Too little USM may not add much to the sharpness of the image. Too much USM can add lot of grain to the image and makes the edges brighter than rest of the image making it look unreal.

The USM filter detects the edges in the image, and then creates halos around the edges. The Radius field is used to specify the size of the halo. The larger the specified radius, larger will be the halo produced at the edges. Keeping this in mind, it is good to use a small radius for an image with lot of details - such as feathers of a bird. A larger radius could be useful for images that do not have sharp edges and need a smooth transition. Images such as foggy mornings or sunsets do not have and do not require sharp edges or fine details and these are better off with slightly higher radius.

The threshold field indicates what part of the image should be considered an edge for applying USM. A threshold of 0 indicates every edge to be considered a candidate for USM. A higher threshold means that the edge contrast needs to be higher to consider that edge for applying the filter.

The Amount field indicats the amount of contrast to be applied on the halos created. A higher value amount means dark edge will be made darker and lighter edge is made to look lighter.

When shooting my marine images I start with an Amount of 150-200, a Radius of 1.2-2, and a Threshold of between 10 and 4, make sure the preview box is checked and you can observe the changes in real-time.

 

[jwedehase]

Wow, interesting, Astro. I never take the radius beyond 0.3, and use 0.2 most often. Then I'm able to bump the amount all the way to 500 if I want.

I've recently swapped over to Smart Sharpen, and like it a lot better. My typical setting with Smart Sharpen is a radius of 0.2 and an amount of 400-500. I've found it to produce a little less halo than USM.

 

[Blazer88]

astrogazer, great writeup on USM. I never really knew what some of those parameters meant and wondered why I would get halos. I generally use smart shapen now and have been very happy with the results.

 

[einsteins]

Smart sharpen will work EVEN better if you change the dropdown from gaussian blur to lens blur.

September version of Popular Photography has an awsome tutorial on how to use smart sharpen.

 

["Umm fish?"]

I can also _really_ recommend _Landscape and Nature Photography with Photoshop CS2_ by Rob Sheppard. Instead of being program oriented, the whole book is based on a photo-oriented approach. That is, he takes specific photos with specific problems and takes you through the steps in Photoshop to fix the problems. In the process, he shos you how to use most of the program. It really starts to make you look at your photos, and that's the big part of the battle.

 

[argo]

Blazer, not to hijack your thread... I tried USM procedure posted by Astro - it gave my pic an artificial kinda look; maybe I'm doing something wrong. In the meantime found a tutorial to try the smart sharpening - http://www.gimp.org/tutorials/Smart_Sharpening/

The more time I spend with PS the deeper it pulls me in.

Thanks for all the advice.

 

[Blazer88]

Argo, can you post a picture? I found a very cool way to just apply the sharpening to the areas that you want, which may help the problem of USM looking bad since you really just want it on the edges. Maybe that could help.

 

[Lpabsolute]

argo - Do you have a Full Tank Shot? Great looking pic....thatnks for sharing....

 

[Blue Deuce]

I don't know if anyone here has used TLR sharpening but it blows away smart sharpen and USM. You make a pass before you convert your Raw image to tiff (you are getting the most out of your camera and shooting raw aren't you) then once afterwards when editing and finally at the end of postprocessing.

 

[argo]

Sorry for a delayed response guys...

Blazer, below is the same pic as #2 above with Astrogazer's USM technique applied (mine has 80_1.5_8 applied):

Thanks.

 

[astrogazer]

I generally apply heavier USM to the black layer of a CMYK file, then do minimal sharpening (USM) to the entire image