Sunday, July 10, 2016

Bringing it all together

It’s Wimbledon week and Men’s Final day, so it’s raining! Of course. But that gives me an opportunity to write a few things that I hope will bring into focus (sorry!), what I’ve been trying to do over the last few months: at least until the final starts, albeit undercover.

I started with the idea of using the power of Magic Lantern Lua scripting to create an in-camera tool to carry out auto bracketing sequences for exposure and focus.

The script is now close to what I envisioned: it works and, on my 5D3, it’s stable and fast.

The script allows me to capture exposure stacks using two basic workflows. The first is to set the base, shadow end exposure and the script will set the highlight end exposure and grab the required brackets (with 1Ev or 2Ev separation). The second workflow is fully auto, with the script setting both shadow and highlight and grabbing the required number of brackets (once again with a user selected 1Ev or 2Ev separation).

For focus stacking the script assumes the user has set a base/starting focus point. The user then has three focus stacking workflow options: to focus stack to infinity stop (FP2INF); to focus stack to the macro stop (FP2MAC); to focus stack to the hyperfocal distance (FP2HFD), as long as the FP is less than the HFD.

In addition, if in FP2HFD mode, the script provides realtime feedback to the user on the estimated number of focus brackets that will be taken. This feature is important if you are using ‘long’ lenses; as the number of focus brackets to cover the full depth of field will potentially be large. The script also allows the user to change aperture to see the impact of the number of focus brackets. For optimum results, ML depth of field feedback should be set to diffraction aware.

The final bracketing options in the script cover long exposure use cases. The first allows the user to simulate a long exposure by taking many shorter exposures, and joining these up in post. The second case is related to the first LE one, but rather than setting the number of exposures based on time, this option lets you specify the number of images to capture. This mode is useful for super resolution photography or eliminating people from a scene, for example.

The latest version of the script may be found on the right.

However, getting the ‘data’ is only part of the story. Until you bring it all together in post, all you have is a ‘load of images’.

I have played around with several post processing workflows and the following works for me. As an example, let’s assume we took a focus and exposure stack (which I just did sitting in our front room, to illustrate things in this post):

  • Ingest all images into Lightroom. 
  • Decide on your exposure blending strategy: fusing or tone mapping. 
  • Process your exposure stacks. I tend to use either the inbuilt ‘HDR’ tool in LR (select images in a given exposure set and press Ctrl+Shift+H and repeat for the other exposure sets) or LR/Enfuse (put each exposure set into a LR stack, select all the stacks and do a round trip to LR/Enfuse in Enfuse batch mode). 
  • Once I have the exposures back into LR I will correct one for Lens Correction and tonal balance. 
  • I then do a round trip to Helicon Focus and, according to the scene, select an appropriate focus stacking method: 
    • Method A computes the weight for each pixel based on its contrast and then forms the weighted average of all pixels from all source images. This method works better for short stacks and preserves contrast and color. 
    • Method B selects the source image containing the sharpest pixel and uses this information to form the "depth map". This method imposes strict requirements on the order of images - it should always be consecutive. Perfectly renders textures on smooth surfaces. 
    • Method C uses pyramid approach to image processing dividing image signals into high and low frequencies. Gives good results in complex cases (intersecting objects, deep stacks), though increases contrast and glare.
  • Once I have the exposures back into LR I will correct one for Lens Correction and tonal balance. 
  • I then finish off in Lightroom and/or Photoshop as required.

Here are a few screen shoots and the ‘final’ image showing things at various states in the above workflow. The use case was with a 5D3, a 24-105 F4/L at 24mm, F/7.1 and ISO-100, with a macro to HFD focus request and a 2Ev full auto exposure request. The scene, as calculated in-camera by the script, required 4 focus stacks of 4 exposures, ie 16 images in total.

Bottom line: At first, many may be put off by the amount of work required to undertaken bracketing. However, until we have a light-field camera that is able to take a full depth of field in one shot, and cover a 20Ev, say, dynamic range; photographers need to learn how to bracket. I hope, at least for the Canon Magic Lantern shooters, my script and my workflow allows you to achieve your vision.

No comments:

Post a Comment