Rodney Campbell's Blog

One of the “issues” raging on the Internet today concerns the use of lens designs that “require” the use of software corrections (in camera and/or in post) to “correct” the lens output

Lens corrections in software aren’t new. They’ve been around in software post processing packages to correct things like chromatic aberration, vignetting and distortion for ages. Typically however whether you apply these corrections or not, or to what degree, is entirely at the users discretion

What is new is that modern manufacturers of cameras and lenses are “baking in” and “relying” on these software corrections as part of the design process. Ultimately leading to them having to do less work in trying to do the corrections in the optical lens design itself. Whilst expecting that part of the work is done by the software correction that follows. This presumably allows for smaller, lighter, cheaper lens designs that don’t need as large and as complex systems inside the lens to deal with the issues

This has mostly been brought on by the new Mirrorless (D-SLR, etc, replacement) systems with their goal of smaller and lighter

So why does this matter …

Well in two ways …

The “purists” say this should all be done in the lens and they want to decide what post processing software based activities occur. When they read lens reviews comparing two or more lenses they “only” want to compare the underlying raw optical data

Secondly…

We have on one side, lens review sites that are reviewing the “newer breed” of lenses. But doing so in the same way as the traditional lens designs. i.e. photographing test charts, but ignoring the manufacturer intended software corrections when analysing the data and scoring. Possibly resulting in some level of poor performance (especially for wider lenses where things are trickier)

On the other side we have the owners and users of said lenses complaining that the test charts and “scores” are not indicative of what they are seeing in real world use. Their in the field results appear to be much “higher” than what this would indicate. Including against other lenses which in theory “score” much higher

A controversial view … Lets try this sequence of events …

In the good old days lenses were designed “proper”. You attached them to a camera (film or digital) and shot with them and WYSIWYG (What You See Is What You Got)

What I mean by this is…

A lens projects a circular imaging area at the back of the lens. Typically if designed properly this circle is larger than the sensor (film or digital sensor) it’s designed to project over. The sensor “crops” out a rectangular area of this projected view. Lens designers design for this – they don’t expect the area outside of the sensor borders to be used – even though the lens does of course project there

So now first question – when “you” “score/rate” lenses do you score the circular projection area or what the lens designer designed for – the rectangular sensor crop area?

On a “traditional” camera (film or D-SLR say (or even digital P&S)). If you had a 100% coverage optical viewfinder or use liveview on the back LCD screen of digital to compose an image you’d expect the resultant image to represent exactly what you framed and composed. If you placed an object at the exact corner of the frame, or multiple objects at exact positions in the frame – thats exactly where they were in the resultant image. If you composed for something to be excluded from the frame – it would be. i.e. it was WYSIWYG

No matter what lens you attached – the view you saw through the viewfinder or liveview is what you got in the final image. Nothing cropped off or nothing outside of the frame. In fact if you had a less than 100% optical viewfinder you the photographer had to factor this in and include less in your frame to cater for the fact that the final image was going to include more. Or you cropped the final image in post

Note: This only applies if you did NOT apply any post/software lens corrections. In this case if you apply “aftermarket” lens corrections you may find that distortion correction controls move parts of your intended composition outside the frame (i.e. cropped out). Which is I suspect why many people don’t want to apply lens correction profiles in some instances and mess up their carefully crafted WYSIWYG composition

So again I ask – do you score/rate this lens based on the projected imaging circle or what both the lens designer and “you” the photographer actually exactly intended/composed for in the viewfinder or on liveview?

Today we come to new lens designs with designed in software corrections. In this case the designer expects these corrections to take place. In fact has an intended lens design that typically goes outside the expected final image. Much like the projected imaging circle above. This is because they expect their software corrections to counteract lens distortion and thus crop or move parts of the image (especially at the extremes/borders/corners of the frame). So the optical lens design might include some extra area around the “advertised” FOV (e.g. a 14mm lens might actually be more like 13mm in reality). Fully intending that when software corrections are applied it gets cropped down to the desired frame/FOV

Let’s take a system like my Nikon Z mirrorless camera with a Nikkor Z lens attached. When you look through the viewfinder or compose using the rear LCD you are seeing the view WITH these anticipated/designed lens manufacturer software corrections being applied. i.e. it is designed to be WYSIWYG. You can compose your scene as you desire with things exactly along the edges or right at a corner and fully expect it will be there in the final image. Likewise compose to exclude things from outside the frame and they won’t be there in the final image as intended

That is… if you purposely DON’T apply the manufacturer designed software corrections, then your carefully composed scene MAY NOT match the image you get. If however you DO apply the software corrections, your composed scene WILL match the image you get in post. In the case of the Z system the “baked” in software corrections are automatically applied by a number of popular image software applications (e.g. Adobe)

You can of course load the raw image files into software which does not do these intended software corrections. This includes imaging/lens testbed systems. However what you may get is data around the sides of the image that the lens designer, and in fact you the photographer, never intended to be seen or used

So finally I ask – do you score/rate this lens based on the projected area of the lens or what both the lens designer and “you” the photographer actually exactly intended/composed for in the viewfinder or on liveview?