Finessing focus stacking

I've been experimenting to see if I can improve the quality of microscope images I generate. The answer is - yes I can, but is it worth it?

DxO recently released their DxO PureRAW 6 image processing software. There are lots of reviews on YouTube if you'd like to know more. I downloaded a free trial copy and compared three different focus stacking variations. I'm not the first person to do this, just documenting it here in case it's useful to you:

1. Capture jpeg images in camera, focus stack with Helicon Focus, post-process with Affinity Photo. This is the standard method I have used until now. 
2. Capture raw (.ARW) images in camera, focus stack with Helicon Focus, post-process with Affinity Photo. Helicon will stack raw image files, so I tested this.
3. Capture raw (.ARW) images in camera, process with DxO PureRAW 6, focus stack with Helicon Focus, post-process with Affinity Photo. 

The results:

click for larger image

At this level, no real difference, and I need to pixel peep at 100% magnification to see the differences:

click for larger image

There is an improvement in quality from method 1 -> 2 -> 3. But what is the cost? 

• To make a focus stack using ~40 jpeg images typically takes me something like 15 minutes (sample processing, image capture, focus stacking, post processing). 
• Using raw images, Helicon stacking increases from 10 seconds to 240 seconds, the whole process taking ~20 minutes. 
• Processing with the sparkly new DxO DeepPRIME XD3 algorithm before stacking takes about 10 minutes, the whole process then taking ~30min. Although Helicon can output a raw (.DNG) image, DxO will only accept original images so it is necessary to batch process captured images before stacking rather than process a single stacked image. The more images, the longer it takes - I estimate making a stack with 100 images would take over an hour from start to finish. This isn't necessarily hands-on time, DxO will happily batch process in the background while I do something else, e.g. prepare the next specimen and capture the images. Sadly it's not possible to process the DNG file Helicon can put out - DxO processing really has to be the first stage of the workflow after image capture. 

There is an improvement in resolution dropping DxO into the workflow at the cost of much longer processing time. Leaving aside the cost of the DxO software, is it worth it? Not for routine images where I'm just documenting an ID. But on occasion, yes, maybe the extra quality is useful.