An Approach to Post Processing.
I realize this is a long - and dense - article, especially for a blog post but please try to persevere through it. I am trying to unite the ideas of seeing what needs to be done with the importance of timing and the ability to retreat from dead ends. It is only when you can fuse all these issues into 'understanding' that you can get beyond the mechanical 'making things look better' into real expression.
Thanks for reading.
Much of the discussions of work-flow in post-processing (PPing) is aimed at the simpler images where most work is done globally and even bit-level corrections aren't ambiguous. However many images require extensive post-processing to get to the final point that the maker has in mind.
It is easy to talk about post-processing at a high level of generality and vagueness – what to do first and what to leave for last. It is also easy to talk about the mechanics of specific techniques in post-processing. But in the middle, how to look at the image, how to make decisions about what really to do and how to protect yourself from time-wasting dead-end avenues where the best path is often ambiguous.
Since I often do extensive post-processing and I hate having to redo intricate work like selections because I have taken a wrong avenue or if I just want to try different PPing techniques to see how to best finalize the image, I have adopted a work-flow that relies a great deal on making and using multiple layers to insure flexibility.
What I intend to do in this article is to talk very quickly about the generalities of my work flow and why I do things the way I do and then show two examples of how I approach any image.
Understanding and deciding what should be done to make the image looks best is the most important and the most difficult skill to acquire. For those new to this, I think it is best to take a very structured approach to diagnosing the PPing needs of any image; I have written about this before.
I always start with the few simple, basic changes - global adjustments of exposure, tint, contrast or even the slightly less obvious white balance. These adjustments require not too much knowledge or skill and even a novice photographer can tackle these because there are some external standards that can be used.
(Exposure shifts the values for the entire image up or down whereas brightness is essentially a mid-tone adjustment. The Levels adjustment is used to correct the tonal range and color balance of an image by adjusting intensity levels of image shadows, mid-tones, and highlights. )
These adjustments are often used to adjust the image back to a 'perfect' rendition of what our eye sees on a 'perfect' day.
But what if the scene isn't perfect, what if Mother Nature doesn't give the light and shadows that we want? What if there is no way to get a decent exposure of all the important parts in the camera?
What if we are not just editing to return the scene to the starting point we saw – or a bit better; what if our vision is more than that – to create an image that we have seen only in our mind's eye and for which the image as caught by the camera is only the starting point?
Then how do we proceed?
First, most important, you must know what you are making a picture of, what is important, what you want to show and what you want the viewer to see. I don't mean in a general sense like 'this is a pretty meadow' or 'here is a nice street scene' or 'there are flowers'.
You should/must know exactly what part of the scene is important because the whole intent of processing is to maximize the impact of the important parts of the scene and, of course, to minimize anything that is either not-important or even actually disruptive to the idea or impression you want to get across.
So you must be able to look very objectively at each picture, look with an intention of discarding pictures if they can't be edited to a good final image.
If you need to crop the image so that the important parts of the image are in better or more important places, is there sufficient room? (you need to have some understanding or feeling about composition. If you don't, go read this or some other source and then come back.)
Are the important areas in good enough focus and well enough exposed (no burnt out or blocked up – over/under exposed)so that they can stand up to being the centers of interest?
(let me step aside here to say two things:
First, if you never need to crop or post-process your images, you're much better than I am and don't need to be reading this.
Is he gone?
People who never crop or edit their images, or who say that, are more concerned with their own silly rules than they are with getting the best final image.
Second, if you don't shoot raw, you ought to have a really good reason because you are throwing away, not only information, but the chance to make the image much, much better with all the dynamic range and subtle tone detail that raw files have and jpegs do not.
Shoot jpegs if your need for post-processing is small, if you have a very well controlled lighting situation or if you need to get final images out immediately without time for editing. Otherwise, shoot raw.)
Now let's assume we have a good image with some real center of interest that is well-enough exposed, well-enough focused, which either is or can be cropped to a good final – can we happily go on?
I say, 'No.'
Often when people ask in what order to edit in, others will refer them to the Light Room Develop panel. “Start at the top and go down,” they say. Generally, this is good advice – except in this circumstance.
When people look at an image, their eyes are draw to bright areas, colorful areas, in-focus areas but also they are distracted by objects that are expected to be 'standard' in some way that aren't.
By 'standard' I refer to things like horizons – which are level – as are roofs and standing water. Verticals, like walls and telephone poles, should be, as much as possible, vertical. Corners, that we know intuitively to be right angles if they are flat to the viewer, should be right angles. That is because, unless there is an artistic reason, things that the viewer expects to be standard, should be so that the insignificant issues don't attract undesired attention.
This is the last criterion that I think is important and that must be correctable before I know an image is 'editable.' (In LR, correcting this perspective distortion is done in the Lens Correction panel down at the bottom and is, unfortunately, not as flexible as the transformation options in Photoshop.)
What is the sense of going through lots of effort to edit an image only to find out that it can't be transformed to remove the unfortunately distracting perspective distortion.
So to reiterate: center of interest, adequate focus, adequate exposure, correctable brightness issues, correctable perspective distortion?
All good. Now we can look at the picture and actually consider the task of editing.
I do a tiny bit of noise reduction and very little sharpening of the photo in the raw state. Because much of my photography is of people and I am often fairly aggressive in post-processing to isolate the centers of interest I often want to subdue the impact of the backgrounds, I see little reason to sharpen aggressively first only to reduce the sharpness later.
I work with three 'prime directives' in mind:
The latter two I'll work on in reverse order.
I generally do correction of the white balance and overall global changes in exposure in Lightroom. It's easy, sensible and generally not an exquisitely difficult issue. Although it is possible to do much of the correction of simple perspective distortions in Lightroom, if more complicated transformation (unilateral torquing) seems necessary, I will export the image to PhotoShop and do both the correction and the initial crop there.
I often try different post processing avenues with more complex editing and I don't like the working-on-a-canvas process in Lightroom. Like painting in watercolors, editing in LR isn't very flexible and relatively difficult to be both exacting and able to retreat step by step.
So, this is typically the point in my workflow when I export a copy of the original from LR to Photoshop where I can make selections in a variety of ways, save these selections, do intricate masks, use blending techniques and, above all, stack changes in layers.
If needed, my first step in PS is to correct the distortion and crop, if necessary.
Since, it is rare that the centers of interest and the non-important areas are to be treated the same so my next step is invariably to separate the centers of interest from the background onto one or more new layers. I do this by selecting them and then moving the selection to a new layer.
When selecting, it is sometimes advisable to 'feather' the edge so that the transition from the selected portion that has been edited to the unedited portion is not too abrupt. That will minimize the cut and pasted look. How much to feather any selection is one of the many ambiguous areas that are left to the artist's eye. It is important to avoid the 'cut-out' look so a refine every selection with a small feather edge.
De-focusing the background has gotten a bad reputation because, if not done carefully and well, the topmost un-blurred layer looks like a cutout laid on the lower layers. Feathering the selection helps with that.
When making complex selections I almost always save the selections as an alpha channel* for two reasons:1) I may want to try this again later and there is no reason to redo a perhaps difficult selection and 2) selections can be difficult and, since selections can be added to a previous alpha channel selection, a complex selection can be made up of incrementally added pieces. Alpha channels are 'cheap' in terms of bytes so the moment taken to save a selection, particularly a difficult one, is well worth it.
*Selections and masks are the largest and most powerful part of Photoshop. If you don't understand layer masks, channel masks, 'alpha channels' – this is the time to read about them. The best book I have found on these topics is 'Photoshop Masking and Compositing' by Katrin Eismann. My preference is the 2004 edition.
Now I have at least two layers, the top one(s) with my centers of interest and the bottom one(s) with the background.
I start my work on the background of the image, removing obnoxious highlights, or at least toning them down, perhaps lowering the overall contrast, correcting color casts – all this to affect anything that would distract the viewers' eyes. As much as possible, I make the changes on adjustment layers to insure flexibility.
By flexibility I mean the ability to manage the effect of the adjustment by changing the opacity of the adjustment layer, change the layer effect with blending modes properties (read about layer blending modes) or remove the adjustment by turning off the layer.
Adjustment layers and masks allow a precision and flexibility that just isn't available in Lightroom.
If I want to clone anything, to remove wires for example, I do cloning on a transparent layer above the layer I'm working on but below adjustment layers, with the clone stamp tool.
Dodging and burning is also done on new layer filled with 50% grey and the layer property set to overlay. (read on burn/dodge new layer)
If I want go as far as actually defocusing the background with lens blur, I will make a copy of the background layer and try the changes there tentatively so if it doesn't work I can delete the defocused layer or tone down its opacity.
Do I end up with lots of layers? Yes, well not exactly because when I get the final image and I think it's just perfect, I'll merge or delete as many of the layers as I can to make the file smaller and more easy to handle.
Now I start corrections on the center(s) of interest on their layer(s), making the changes also on adjustment layers but always with the option of 'clipping mask' on. Thus the effects of adjustment layers are confined to the specific layer.
What kind of 'corrections' do I make to the layer with center(s) of interest (COI)?
Well, I make the same kind of technical changes to the centers of interest except that my intent to to make this layer the one that attracts the eye but, at the same time, looks coherent with the changes on the underlying layers.
Well, as a general rule, the COI should be the area that is the lightest, the best in terms of contrast contrast, most in-focus part(s) of the image – thus embodying the characteristics that attract the viewers' attention. If the COI is a person, I often bringing down the color or tone articles of clothing that attract attention from the face.
Sharpening the center of interest in general or the face in particular is often my last step. Sharpening is an art and taste is unteachable but there are numerous references to be found on the web.
If I am planning to convert to B/W, I will often insert a temporary b/w adjustment layer to be certain that my tonal adjustments will look good in the final. Then I turn off the layer or delete it until I am finally done.
If I am ready to finalize the image, I will merge all the usable layers to a final layer on top. Then I might use a plug-in such as Color Efex Pro, Silver Efex Pro or DxO Filmpack to give a final unifying punch to the entire frame. There is nothing in these plug-ins that can't be done using PS (I think) but the advantage of having presets for the huge range of effects is too much of a time savings to ignore.
If things don't look 'perfect' or if I want to try some more alternatives, I can go back and pick out the steps/layers that don't look good and try new alternatives. Much of the time, I may be on the fence about where I want to go, so I'll toss the layers I don't need, merge any that work into smaller grouped layers and save the file with layers. If things look great – and I am satisfied, I will just collapse everything and save it as a full resolution PSD.
If I have worked very hard and get to the end of a session with lots of work accomplished, I don't leave things to chance. I run a backup of my photo directories to two external drives right then. Thus I can sleep soundly not worrying about data loss.
First example: I will start with the image shown taken of a corridor inside the Forbidden City in China by a colleague, Steve Levitas of Maryland. (This is the image above seen above.)
While the almost total blackness of the wall is a problem, my eye is much more drawn by the strange distortion of the walls. They are at angles with the floor – and the angles don't seem to be so much of settling by time as of strange lens distortion.
The unusual and asymmetric tilt of the walls is obvious and unsettling but most important of all, it is distracting. It is clear that this distortion is not within my ability to correct in LR with sliders. I need to push and tug using free transform in Photoshop. So my first chore is to expand the margins and do some free transformation to torque the various angles close to what is normal to the eye.
Once the picture looks fairly square I crop it and go on to some cosmetic changes, making the red redder and the dark wall less of a complete black.
Although this is a jpeg original, which means a narrower dynamic range, the histogram shows that the area of the wall I want to brighten isn't clipped thus does have some detail, so all is well there. I don't want much out of those shadows, just a hint of detail.
The order of doing this is, in this case, irrelevant but the straight side of the dark wall make for an easy selection, so I start with that.
I make a copy of the background layer. I made a selection of the dark wall and then create a layer mask that reveals the dark wall . (On a very simple obvious edit like this I don't bother to save the selection or label the layers because the layers palette reveals what is going on.) Then I brighten the wall to reveal a tiny bit of detail. OK, done there.
I actually brightened the wall a bit a bit too much so I lowered the opacity until it looked good.
Now I used Select>Reselect to and then inverted that selection (Ctrl, Shft 'I') to grab the remainder of the image so I could edit that. Then Ctrl 'J” to make a new layer with only the selection. Now I can work on just that part of the scene.
I made it brighter, sharpened it a bit, then realized I went too far and turned down the opacity and there it was. As good as I expected it to be.
Took about as much time to do as to read and much less than to type.
Second example is a totally excellent grab shot taken in Myanmar in February of 2013 by Saul Pleeter of NIHCameraClub.
At first glance, this pictures looks pretty good, good focus, decent framing (although more of the little boy would have been perfect) but it does look rather bland. Rather than the figures jumping out of the background and presenting themselves, everything is this sort of overall flat color. So I decided to try some edits to give it some impact.
While still in LR I lowered the darks a bit; there weren't any real shadows and there really weren't any areas that needed to be black but this did help the contrast some. I noticed that both the little boy's and the woman's face looked pretty contrasty already and I didn't want to increase that any more with any global changes because lightening them would have looked wrong and they are contrasty enough as it is.
The higher tones were fine and it was clear that I needed to get a little less global and more specific; as I said before, I don't like the one-way, no-going back of Lightroom, so I moved this picture into Photoshop.
So my entire PPing needed to be centered around leaving the faces as they were and changing everything else so that the faces were more the real centers of interest. My intent was to darken the background, darken the woman's clothing, adding more saturation and keep away from any real changes of tone or contrast on their faces. To do that I needed to separate the figures from the background and separate the woman's clothing from the figures.
I chose to darken her clothing first. That way I would know how much change I needed to make in the luminosity of the wall.
First I made a quick selection of both of the figures, as seen in Quick Mask, saved that selection and used it to lift the figures to another layer.
Then I edited that selection to include only the clothing of the woman, that which needed darkening and removed the clothing alone to yet another layer.
As you can see from the layer panel, one doesn't need to work from the bottom up. I darkened the clothes with a brightness layer with a layer masked that was linked as a clipping mask, thus affecting only the layer directly below.
Then I went back and darkened the background wall so that the clothing and faces stood out well against it.
When I was happy with the overall look, I merged everything to a new top layer and warmed that a tiny bit and then sharpened everything a bit.
So one can see by this approach that there are two Rules in any Post-Processing effort: First, have some idea of where you want to go with the final edit and second, create your processing files so that, if your desired end result changes, you can make changes without redoing too much of the delicate work.
I use layer masks, clipping masks, channel masks, layer opacity - anything I need to create a final image that makes me happy.
This result was a picture that, at first glance was the same as the original, but had more depth, more punch and was, to me, infinitely more satisfying.
(For some odd reason the option to subscribe by email is given above and the option to subscribe using an RSS reader is below the following text. you figure that out.)
Recent PostsComposition and Critique - understanding a photograph Customizing the Lightroom Metadata Panel An Approach to Post Processing. The Salt of the Earth – a film about Sebastiao Salgado - review The Rules of Composition, The Rules of Art. Is post processing cheating? Combat Photography Using Fluorouracil for Actinic Keratosis - case study Review: Focal Point: Fine Art and Creative Photography - MD Fed Art Review- Naturevisions Traveling Exhibit at the Oakland Mills Interfaith Center