Image Sharpening

Image Sharpening and Raw Conversion.

A lot of people imagine that there is some sort of ‘magic bullet’ method for sharpening images.

Well, here’s the bad news – there isn’t !

Even if you shoot the same camera and lens combo at the same settings all the time, your images will exhibit an array of various properties.

And those properties, and the ratio/mix thereof, can, and will, effect the efficacy of various sharpening methods and techniques.

And, those properties will rarely be the same from shoot to shoot.

Add interchangeable lenses, varied lighting conditions, and assorted scene brightness and contrast ranges to the mix – now the range of image properties has increased exponentially.

What are the properties of an image that can determine your approach to sharpening?

I’m not even going to attempt to list them all here, because that would be truly frightening for you.

But sharpening is all about pixels, edges and contrast.  And our first ‘port of call’ with regard to all three of those items is ‘demosaicing’ and raw file conversion.

“But Andy, surely the first item should be the lens” I here you say.

No, it isn’t.

And if that were the case, then we would go one step further than that, and say that it’s the operators ability to focus the lens!

So we will take it as a given, that the lens is sharp, and the operator isn’t quite so daft as they look!

Now we have a raw file, taken with a sharp lens and focused to perfection.

Let’s hand that file to two raw converters, Lightroom and Raw Therapee:

Image Sharpening

I am Lightroom – Click me!

Image Sharpening

I am Raw Therapee – Click me!

In both raw converters there is ZERO SHARPENING being applied. (and yes, I know the horizon is ‘wonky’!).

Now check out the 800% magnification shots:

Image Sharpening

Lightroom at 800% – Click me!

Image Sharpening

Raw Therapee at 800% – Click me!

What do we see on the Lightroom shot at 800%?

A sharpening halo, but hang on, there is NO sharpening being applied.

But in Raw Therapee there is NO halo.

The halo in Lightroom is not a sharpening halo, but a demosaicing artifact that LOOKS like a sharpening halo.

It is a direct result of the demosaicing algorithm that Lightroom uses.

Raw Therapee on the other hand, has a selection of demosaicing algorithms to choose from.  In this instance, it’s using its default AMaZE (Alias Minimization & Zipper Elimination) algorithm.  All told, there are 10 different demosaic options in RT, though some of them are a bit ‘old hat’ now.

There is no way of altering the base demosaic in Lightroom – it is something of a fixed quantity.  And while it works in an acceptable manner for the majority of shots from an ever burgeoning mass of digital camera sensors, there will ALWAYS be exceptions.

Let’s call a spade a bloody shovel and be honest – Lightrooms demosaicing algorithm is in need of an overhaul.  And why something we have to pay for uses a methodology worse than something we get for free, God only knows.

It’s a common problem in Lightroom, and it’s the single biggest reason why, for example, landscape exposure blends using luminosity masks fail to work quite as smoothly as you see demonstrated on the old Tube of You.

If truth be told – and this is only my opinion – Lightroom is by no means the best raw file processor in existence today.

I say that with a degree of reservation though, because:

  1. It’s very user friendly
  2. It’s an excellent DAM (digital asset management) tool, possibly the best.
  3. On the surface, it only shows its problems with very high contrast edges.

As a side note, my Top 4 raw converters/processors are:

  1. Iridient Developer
  2. Raw Therapee
  3. Capture One Pro
  4. Lightroom

Iridient is expensive and complex – but if you shoot Fuji X-Trans you are crazy if you don’t use it.

Raw Therapee is very complex (and slightly ‘clunky’ on Mac OSX) but it is very good once you know your way around it. And it’s FREEEEEEEEE!!!!!!!

Iridient and RT have zero DAM capability that’s worth talking about.

Capture One Pro is a better raw converter on the whole than Lightroom, but it’s more complex, and its DAM structure looks like it was created by crack-smoking monkeys when you compare it to the effective simplicity of Lightroom.

If we look at Lightroom as a raw processor (as opposed to raw converter) it encourages the user to employ ‘recovery’ in shadow and highlight areas.

Using BOTH can cause halos along high contrast edges, and edges where high frequency detail sits next to very low frequency detail of a contrasting colour – birds in flight against a blue sky spring to mind.

Why do I keep ‘banging on’ about edges?

Because edges are critical – and most of you guys ‘n gals hardly ever look at them close up.

All images contain areas of high and low frequency detail, and these areas require different process treatments, if you want to obtain the very best results AND want to preserve the ability to print.

Cleanly defined edges between these areas allow us to use layer masks to separate these areas in an image, and obtain the selective control.

Clean inter-tonal boundaries also allow us to separate shadows, various mid tone ranges, and highlights for yet more finite control.

Working on 16 bit images (well, 15 bit plus 1 level if truth be told) means we can control our adjustments in Photoshop within a range of 32,768 tones.  And there is no way in hell that localised adjustments in Lightroom can be carried out to that degree of accuracy – fact.

I’ll let you in to a secret here!  You all watch the wrong stuff on YouTube!  You sit and watch a video by God knows what idiot, and then wonder why what you’ve just seen them do does NOT work for you.

That’s because you’ve not noticed one small detail – 95% of the time they are working on jpegs!  And jpegs only have a tonal range of 256.  It’s really easy to make luminosity selections etc on such a small tonal range work flawlessly.  You try the same settings on a 16 bit image and they don’t work.

So you end up thinking it’s your fault – your image isn’t as ‘perfect’ as theirs – wrong!

It’s a tale I hear hundreds of times every year when I have folk on workshops and 1to1 tuition days.  And without fail, they all wish they’d paid for the training instead of trying to follow the free stuff.

You NEVER see me on a video working with anything but raw files and full resolution 16 bit images.

My only problem is that I don’t ‘fit into’ today’s modern ‘cult of personality’!

Most adjustments in Lightroom have a global effect.  Yes, we have range masks and eraser brushes.  But they are very poor relations of the pixel-precise control you can have in Photoshop.

Lightroom is – in my opinion of course – becoming polluted by the ‘one stop shop, instant gratification ideology’ that seems to pervade photography today.

Someone said to me the other day that I had not done a YouTube video on the new range masking option in Lightroom.  And they are quite correct.

Why?

Because it’s a gimmick – and real crappy one at that, when compared to what you can do in Photoshop.

Photoshop is the KING of image manipulation and processing.  And that is a hard core, irrefutable fact.  It has NO equal.

But Photoshop is a raster image editor, which means it needs to be fed a diet of real pixels.  Raw converters like Lightroom use ‘virtual pixels’ – in a manner of speaking.

And of course, Lightroom and the CameraRaw plug in for Photoshop amount to the same thing.  So folk who use either Lightroom or Photoshop EXCLUSIVELY are both suffering from the same problems – if they can be bothered to look for them.

It Depends on the Shot

sharpening

The landscape image is by virtue, a low ISO, high resolution shot with huge depth of field, and bags of high frequency inter-tonal detail that needs sharpening correctly to its very maximum.  We don’t want to sharpen the sky, as it’s sharp enough through depth of field, as is the water, and we require ZERO sharpening artifacts, and no noise amplification.

If we utilise the same sharpening workflow on the center image, then we’ll all get our heads kicked in!  No woman likes to see their skin texture sharpened – in point of fact we have to make it even more unsharp, smooth and diffuse in order to avoid a trip to our local A&E department.

The cheeky Red Squirrel requires a different approach again.  For starters, it’s been taken on a conventional ‘wildlife camera’ – a Nikon D4.  This camera sensor has a much lower resolution than either of the camera sensors used for the previous two shots.

It is also shot from a greater distance than the foreground subjects in either of the preceding images.  And most importantly, it’s at a far higher ISO value, so it has more noise in it.

All three images require SELECTIVE sharpening.  But most photographers think that global sharpening is a good idea, or at least something they can ‘get away with’.

If you are a photographer who wants to do nothing else but post to Facebook and Flickr then you might as well stop reading this post.  Good luck to you and enjoy your photography,  but everything you read in this post, or anywhere on this blog, is not for you.

But if you want to maximize the potential of your thousands of pounds worth of camera gear, and print or sell your images, then I hate to tell you, but you are going to have to LEARN STUFF.

Photoshop is where the magic happens.

As I said earlier, Photoshop is a raster image processor.  As such, it needs to be fed an original image that is of THE UTMOST QUALITY.  By this I mean a starting raw file that has been demosaiced and normalized to:

  1. Contain ZERO demosaic artifacts of any kind.
  2. Have the correct white and black points – in other words ZERO blown highlights or blocked shadows.  In other words, getting contrast under control.
  3. Maximize the midtones to tease out the highest amount of those inter-tonal details, because this is where your sharpening is going to take place.
  4. Contain no more sharpening than you can get away with, and certainly NOT the amount of sharpening you require in the finished image.

With points 1 thru 3 the benefits should be fairly obvious to you, but if you think about it for a second, the image described is rather ‘flattish – looking’.

But point 4 is somewhat ambiguous.  What Adobe-philes like to call capture or input sharpening is very dependent on three variables:

  1. Sensor megapixels
  2. Demosaic effeciency
  3. Sharpening method – namely Unsharp Mask or Deconvolution

The three are inextricably intertwined – so basically it’s a balancing act.

To learn this requires practice!

And to that end I’m embarking on the production of a set of videos that will help you get to grips with the variety of sharpening techniques that I use, and why I use them.

I’ll give you fair warning now – when finished it will be neither CHEAP nor SHORT, but it will be very instructive!

I want to get it to you as soon as possible, but you wouldn’t believe how long tuition videos take to produce.  So right now I’m going to say it should be ready at the end of February or early March.

UPDATE:  The new course is ready and on sale now, over on my digital download site.

sharpening

The link to the course page is HERE.

Hopefully I’ve given you a few things to think about in this post.

Don’t forget, I provide 1to1 and group tuition days in this and all things photography related.

And just in case you’ve missed it, here’s a demo of how useful Photoshop Smart Sharpen can be:

Become a patron from as little as $1 per month, and help me produce more free content.

Patrons gain access to a variety of FREE rewards, discounts and bonuses.

2 thoughts on “Image Sharpening

    • Hi Hugh

      No, Hard Light is too aggressive a blend mode imo, I much prefer Soft Light as a rule over the standard de facto Overlay.
      But the way I use high pass, and the slightly subtler frequency separation is all explained in detail in my new course.

Add your comments and feedback

This site uses Akismet to reduce spam. Learn how your comment data is processed.