Olympus In-camera Focus Stacking

[Nigel]

Ever since Olympus updated the firmware for in-camera focus stacking to allow selection of the number of images (EM1-2 ver 3.0) in a stack I have had the feeling the final composite image has not been as good as previously. I certainly get much better composites using the individual images acquired in Helicon Focus.

So I have investigated this using a test target photographed at an angle - see the sequential images posted here. This was set as an 8 image in-camera focus stack and going through the images one by one you can see the initial focus at 0 on the scale then 2 images closer and 5 images farther away as expected. The image marked 9 is the composite. You can see that the in-camera composite does not cover the depth of focus that should be possible from the individual images. The individual images have sharp focus regions ranging from -2 to + 5 on the scale but the composite is sharp only from -1 to +3.

I have repeated this test with 15 images selected and to my surprise the composite showed no more depth of focus (i.e. it was still -1 to +3) even though the individual images had sharp focus regions ranging from -5 to +10.

Do other Olympus owners find the same?

If anyone still is using firmware prior to the focus-stacking update (Ver 3.0) I would be very interested to know if you also find this effect.

 

[Bushboy]

It looks to me like inadequate aperture has been used for those shots? Could you do this test again at different apertures? And post the results? Nice target you’ve got. I shot a focus stack yesterday, and the first stack was way off. I had to adjust the aperture and the diff to get the proper result. Lens used, subject size, aperture, distance to target, colour of your underwear, and differential, all need to be in alliance I find. But saying that, I always get a great result in camera quite easily.

 

[Nigel]

It looks to me like inadequate aperture has been used for those shots? Could you do this test again at different apertures? And post the results? Nice target you’ve got. I shot a focus stack yesterday, and the first stack was way off. I had to adjust the aperture and the diff to get the proper result. Lens used, subject size, aperture, distance to target, colour of your underwear, and differential, all need to be in alliance I find. But saying that, I always get a great result in camera quite easily.

I don’t hint this is the problem as I used a diff of 3 which is narrow and you can see that each individual image overlaps - I believe the step size is varied according to the aperture used in any case. Also, there was no difference in the composite between 8 shots and 15. I am tempted to do a factory reset before trying again.

 

[Nigel]

I don’t hint this is the problem as I used a diff of 3 which is narrow and you can see that each individual image overlaps - I believe the step size is varied according to the aperture used in any case. Also, there was no difference in the composite between 8 shots and 15. I am tempted to do a factory reset before trying again.

Also Helicon Focus generates smooth stacks from the same images

 

[Michael Meissner]

If you wanted to test it more thoroughly, you could back out the firmware release in question via lightsnowdev.com to see if it was a particular release.

 

[ColorBlind]

Interesting. I haven't noticed anything like this on mine (EM1.2, f/w version 3.2). I just took this quick stack outside (12 frames, diff = 2). I've also included the final pre stacked frame #12 - and it is clear that it has also been included in the composite stacked version.

Stacked version

Subscribe to see EXIF info for this image (if available)

Frame #12

Subscribe to see EXIF info for this image (if available)

I have no doubt that 3-party s/w can likely improve on the in-camera results, if the frames are taken using a tripod. However almost all my stacks are taken handheld, and in my experience, stacking software (Workspace, Affinity) can't produce good stacks if the image isn't perfectly stable. This is where the in-camera stacking excels IMO, as it creates surprisingly good images even with some camera shake or subject movement.

 

[Growltiger]

I see a significant problem with your test. Look at image 8 for example. There are fairly sharp areas in the far right of the image at positions 1 and 2. The camera is taking these into account too, not just the squares you are looking at. Repeat the tests without the objects on the right of the image.

When I did some tests I had the camera near the floor above a patterned carpet, there was nothing else in the images. I used a wide aperture and the results were exactly what they should be with 15 images (or 8).

 

[Nigel]

I see a significant problem with your test. Look at image 8 for example. There are fairly sharp areas in the far right of the image at positions 1 and 2. The camera is taking these into account too, not just the squares you are looking at. Repeat the tests without the objects on the right of the image.

When I did some tests I had the camera near the floor above a patterned carpet, there was nothing else in the images. I used a wide aperture and the results were exactly what they should be with 15 images (or 8).

I have added the numbers to the images in the far right to make sure they uploaded in the right order - they are not on the originals.

 

[Nigel]

OK. Problem solved. I have done some more tests. Firstly, I repeated the tests with my EM1-2 and the in-camera focus stacking worked fine. I factory reset my EM1-3 and tried again but the in-camera focus stacking was still not working correctly. So I updated my firmware (I had not updated as from V1.1 to V1.2 as the improvements were mainly for video and I don't have the 150-400 lens) and now the in-camera focus stacking is working!!

 

[daleeight]

So you now have E-M1mk3 with v1.2 on it?

 

[hernan]

I don't understand why om5 mkiii with focus stacking saves 8 jpg + ninth as final result.
for me it's just confusing to find the right shot.
if you want to do a more serious job I will use focus bracketing, but for a jpg ready immediately why save 9 shots?
Is my thinking right or wrong?

 

[Bushboy]

It is easy to find the stacked shot.
While your in playback/ review mode, press the info button until slightly more info becomes available on your screen. The layer icon will display when your looking at the stacked shot.
I agree, keeping the original jpegs is a pain in the butt. I delete them immediately after they’re taken. My delete button has the paint worn off... ?

 

[Thai-Mike]

Keeping the shots makes sense to me cos if you are not satisfied with the end-result your camera-in-stacking does you can use an external software to stack your photos.

 

[Bushboy]

I like your thinking, but I can see no difference between in cam stack and the Affinity Photo that I use. Haloing, hot pixels, it’s all there.
I watched this guy on ytube, explaining the relationship between aperture and differential amount . It was very interesting. One of the things I got from it was, upping the aperture really brings out the ugh distracting background. Even with closeups.
YouTube, what a cool internet thing.

 

[Thai-Mike]

When I use my Helicon Focus 6, I can just to some pp in LR than export to Helicon Focus 6 and it looks much better. I speak of my personal experience when I was using my EM5ii.

BTW, I love focus stacking all by myself. I choose the focus points I want to be. I know it depends on the subject.

Honestly, my best in camera focus stacking was using the Olympus tough TG6

 

[Bushboy]

Wow. That’s interesting!
TG6! I want to try one now...?

 

[Ross the fiddler]

I see a significant problem with your test. Look at image 8 for example. There are fairly sharp areas in the far right of the image at positions 1 and 2. The camera is taking these into account too, not just the squares you are looking at. Repeat the tests without the objects on the right of the image.

When I did some tests I had the camera near the floor above a patterned carpet, there was nothing else in the images. I used a wide aperture and the results were exactly what they should be with 15 images (or 8).

I also think 'real world' subjects are going to look better than a 'test chart' too.

 

[retiredfromlife]

I like your thinking, but I can see no difference between in cam stack and the Affinity Photo that I use. Haloing, hot pixels, it’s all there.
I watched this guy on ytube, explaining the relationship between aperture and differential amount . It was very interesting. One of the things I got from it was, upping the aperture really brings out the ugh distracting background. Even with closeups.
YouTube, what a cool internet thing.

Do you remember which video it was? sounds like a good one to watch.

When you say upping the aperture do you mean smaller or larger "F" number?

 

[Bushboy]

I will try n post a link tomorrow when I’m in town.
Stopping down from f4 to f16 made a dramatic difference in the pics he was taking. f16 was really making the background much clearer. Not something I like, I prefer backgrounds blurred.
He experimented with the differential numbers to get all of his subject in focus. To higher a number, i.e, say 10, and you could see bands of in focus and out of focus areas. That’s something I’ve noticed before, it ruins the pic.
I usually start at f4-5.6, differential no.5 with my closeups. The dudes vid was very interesting and well done. I hope I can find it again.

 

[Ross the fiddler]

Do you remember which video it was? sounds like a good one to watch.

When you say upping the aperture do you mean smaller or larger "F" number?

Go up to f5.6 or f8 etc to give plenty of overlap between layers (steps).