Monday, July 31, 2017

For goodness sake Garry: KISS



As we know, KISS means Keep It Simple Stupid. Well, from feedback, it is clear my focus bar fails the KISS test :-(

As I’m introducing a ‘new’ way of (landscape) focusing Canon cameras with the focus bar, ie optimising focus between the hyperfocal distance and infinity, I clearly need to do a better job of explaining what is going on: that is in KISS language.

Having said that, if you are using a DSLR, then you are using a complex piece of hardware and software, ie you are not a point & shoot person. In addition, if you are using Magic Lantern you most probably are aware of depth of field and want to get more out of your camera.

As we know the defocus, that occurs as you move away from the point of focus, is a function of the focus distance, the focal length, the aperture and a blur criterion, usually called the circle of confusion (CoC).

Simply put, the circle of confusion is the defocus blur that ‘just’ looks sharp on the sensor, ie a larger blur would look out of focus.

Thus to get the best out of the focus bar it is essential to understand the basics of depth of field and the concept of blurs.

For a full frame DSLR producing images for digital display, ie on the web, a CoC or blur of 29 or 30 microns is considered (just about) OK. For more exacting work, ie a judged print, a CoC of about 15 microns might be a better choice. This is a simplification and in reality you should choose the blur to fit the viewing conditions, the size of the print and the distance that the viewer is away from the print. But let’s keep it simple for now.

For the full frame landscape photographer, simply set the Magic Lantern CoC to 29 or 30 microns. For a crop shooter reduce this by the crop factor. That’s it: don’t touch this again :-)

So let’s look at an example: a FF camera and a 16mm lens, at F/8 and focused short of the HFD, the defocus field (without diffraction) looks like this:


The left hand axis shows the blur, with a unity blur being the set CoC criterion, ie 30 microns. The plot clearly shows a zero blur at the point of focus and the focus field defocusing either side. By definition the depth of field is between the curve where it crosses the unity blur line. In this zone, our eyes think everything is tact sharp (relative to our criterion of 30 microns). Out side this zone the image gets progressively out of focus. Note that the near and far fields behave radically differently, ie the focus field is not symmetrical about the point of focus.

Let’s now move the focus to the hyperfocal distance. Now we see the classical HFD focus field, ie the near field depth of field (unity blur) is just short of the focus point, i.e. at HFD/2, and the far field (unity blur) is at infinity (although it is not shown here you can see the far field curve approaching unity blur, i.e. 30 microns in this illustration, where it will at infinity):


Finally, let’s now focus beyond the HFD and short of infinity, ie use the focus bar. This is where the focus bar ‘magic’ starts. As you can see the far field curve now asymptotes to a blur that is half of that at the HFD. Remember, in all the charts the unity (CoC) blur remains constant at 30 microns.


Finally, to emphasise what has happened, let’s show the depth of field curves for the HFD case and for the focus bar, together:


Here we see the advantage of using the focus bar to tell us when to stop focusing short of infinity. In other words, by focusing slightly beyond the HFD, ie at about 2m in this illustrative case, we have created an infinity defocus blur, ie no diffraction effect yet, of half of the ML set 30 microns; ie half of the HFD blur. 

Put another way, we have focused for a high quality print, with a near depth of field of about 1 meter at a blur of 0.5 of the CoC, ie about 15 microns, or a slightly larger depth of field if you accept the 30 microns blur for the near field.

Finally, for defocus blur you really can’t do much better than two sensor pixels; and on a 5D3, for example, that means a defocus blur at infinity of about 13 microns. Therefore the full frame guidance is to aim for a reported (defocus) focus bar blur at infinity of, say, 29 or 30 microns, for digital projection; and about 15 for high quality print work.

Note, the ML set CoC stays at 30 throughout. All we need to do is act on the reported focus bar blurs.

I truly hope this post has helped those that are struggling with using the focus bar; as I believe for landscape photography, the focus bar is about the best you will get from a focusing perspective.

4 comments:

  1. Hi, OK i understand the basics of images for stacking landscapes but sorry the way you put this forward (and only my opinion) is to complicated for your every day ameture who wants to improve there photography. All these graphs above make it look like something from Einsteins lab. I do have your Focus bar installed but and being honest I only use the distance scale read out near/far. Just a suggestion but a short video explaining how to use it would I'm sure be more benificial but then maybe I am in the minority on that one. Thanks. Russ

    ReplyDelete
    Replies
    1. Russ

      Sorry I couldn't help you get more out of the focus bar. I'm afraid to really understand its power, you need to understand how the different blurs, in both the lens optics and diffraction, degrade your image; and why hyperfocal focusing can be improved on. As shown by the curve above.

      Good luck with your photography.

      Cheers

      Garry

      Delete
  2. Hi Garry, Trying to use the latest version of the Focus Bar for the EOS M. I run the script and make sure that Display is set to ON in Shoot, Focus Bar. I am afraid that nothing is displayed on the rear screen, it just flicks between a blank screen and the top and bottom lines on the ML information screen flicker. In order to make it reproducible to you and to make sure it was not any settings I am using, I have formatted an appropriate SD card on my Mac using a low level formatter, downloaded and installed (https://builds.magiclantern.fm/jenkins/job/lua_fix/445/artifact/platform/EOSM.202/magiclantern-lua_fix.2018Mar31.EOSM202.zip) the experimental LUA build (I assume that this is the build I should use). Moved all the shipped scripts to a separate directory and copied the latest Focus Bar script for the EOS (2m.lua) to the scripts directory. I even did a camera Clear settings and then set things like RAW image quality up from scratch. All other ML settings were set to the default. Do you have a spare SD card that you could try and reproduce or have I missed something obvious? I have the 2.02 Firmware and the EF-M 18-55mm lens. I have read the Introduction in the scripts initial comment and could not identify any pre-reqs that had not been met. I look forward to your response. I have managed to install on my 6D and it appears to work fine. Although, dualISO does not work in the 6D LUA experimental build :)

    ReplyDelete
  3. Hi Garry

    I think I have found the problem... there appears to be a dependancy on having the ND Bracketing script installed. I did not expect that.

    ReplyDelete