Focus stacking: Oly E-M1 in camera vs Photoshop

I was interested in comparing the Oly ooc merged focus stack against the same merged stack in Photoshop.  I often use the 40-150 Pro at f2.8 when out and about in pretty gardens, so set up the Lens Align target to more or less fill the frame (landscape orientation of the camera) and snapped away.

The first shot is a crop of the "base" snap, 150 mm focal length at f2.8.  The target was around 114 cm away (Lens Align "0" to E-M1 sensor plane), focus was on the "0", and the target was angled at 20 degrees, meaning each of the 32 "up" and 32 "down" target steps had around 3 mm depth of field.  We can see that the base snap depth of field is from around 1 up to maybe 2 down, so approximately 12 mm.

Base 150 mm f2.8 snap

I set the camera to take a set of 8 focus stack snaps, using "focus differentials" of 2, 4, 6, 8, and 10 for each set, resulting in 5 stack merges, one per set.  The next shots are of the cropped in-camera Oly merge followed by the cropped Photoshop merge for the "2" differential.

Oly merged stack in camera, differential = 2

Photoshop merged stack, differential = 2

From the same focus stack set of 8 snaps, the Oly in camera merge compares well with the Photoshop merge, although it has extracted a slightly narrower depth of field from around 7 up to 3 down (maybe 33 mm) than the PS merge from around 8 up to 4 down (perhaps 40 mm).

The next pair of merges are for a differential of 4:

Oly merge in camera, differential = 4

Photoshop merge, differential = 4

The Oly merge gives a depth of field from around 9 up to 6 down, 48 mm, while Photoshop manages around 14 up to 7 down, perhaps 66 mm, from the same set of 8 snaps.  IQ seems similar, though we can note some Photoshop artefacts at the edges of the merge (here and in the other images) which are not present in the Oly merge.  Part of the reason for this is that the Oly merge crops the result in camera, while Photoshop leaves it to the user to crop a merged stack as needed.  Another part of the reason could be the PS image alignment algorithm (it was set as "auto") isn't quite as good as the Oly algorithm, and I unchecked "smooth transitions" at the PS blend step.

At differential = 6, things begin to change for the Oly in camera merge:

Oly merge in camera, differential = 6

Photoshop merge, differential = 6

The Oly in camera merge has acquired some artefacts in the image, being fringe ringing or ghosting that we can see on the upper left "12" and the large upper "6".  And the depth of field in focus hasn't increased much, though the slightly out of focus field has.  The PS merge is clearly superior.  An interesting point we can pick up on is that the E-M1 focus stack of 8 snaps seems to bias the stack so that one third of the depth of field is in front of the nominal focus point, and two thirds of the depth of field is behind.

The final comparison is at differential = 8:

Oly merge in camera, differential = 8

Photoshop merge, differential = 8

We can see focus banding creep in, quite marked with the Oly stack between 13 and 16 up, more subtle with the PS merge which only seems to miss a beat on 13 up...

obsolescence wrote:

I wonder why 150mm was chosen

I often use the 40-150 Pro at f2.8 when out and about in pretty gardens...

Wouldn't the Olympus automatic focus stacking work better at, say, 60mm?

The Oly 60 will be my next post in a few days.

obsolescence wrote:

I wonder why 150mm was chosen, since I would rarely if ever use such a long focal length for macro work or for studio still life work. Wouldn't the Olympus automatic focus stacking work better at, say, 60mm?

You might not (use 150mm) for your own macro work, but then you apparently do not shoot macros of live insects...especially the flying sort. For that, a 105-150mm FL is often the one of choice, and Sigma's 150mm macro (at one time available in 4/3 mount) is a highly prized lens.

Phil

Thanks for all your work on this test! Very helpful!

I do have to ask....  1/10,000/sec shutter speed??  I thought my Mark ii only went up to 1/8000/sec.