Is there anything I can do at either the stage of taking the initial shots or additional tweaks I could do in Photoshop to improve my focus stacking results?

I am finding that the 'rear' of the stack is forming quite well; the correct areas are being chosen to use the sharpest point. However, the front-most layer seems to not be receiving sufficient attention.

This is a very small crop from a photograph of a flower I took at 4 focus points [though only 3 are relevant to this segment] The crop centres on one 'petal' from the very front of the flower.

This is Photoshop's stacked result...
Right click any image & 'open in new tab' for full sizes

It is made from these three layers...

As you can see, photoshop seems to have favoured the middle & rear layers rather than bringing this front petal in as sharp as possible from the available material.

Is there anything I can do to influence Photoshop's decision as to which areas to favour at any given depth? I'm aware I could go back through the masks by hand & change the end result; but that seems a lot of work for a process that in auto-mode takes mere seconds.
I've tried changing the order of the layers before stacking, front to back/back to front, but this seems to have little effect.

Alternatively, is there anything I can do at the photography stage to let photoshop favour the 'front' more - perhaps one more layer with the focus pulled even closer? I haven't had chance to investigate that possibility yet.

Equipment used:-
Nikon D5500 Nikkor AF-S 24-120mm 3.5-5.6 @120mm ISO 400 1/125s f5.6
Full manual mode. AF & VR off.

Full-size, VLQ jpg of the whole stacked frame, marked with the approximate crop

enter image description here


2 Answers 2


I don't know that much about focus stacking algorithms, but judging by your final image it looks like the problem has to do with contrast. The algorithm obviously has to pick out areas that are sharply focussed somehow, and we can see from the image that it's most successful where you have good contrast between the in-focus areas and the out-of-focus areas. For example, the petals against the blue background are relatively sharp except in the spots where they overlap -- the red-on-red areas are much blurrier even though they're obviously at the same distance as the rest of the petal:

flower detail

I think the area you're asking about suffers the same problem -- there's just not enough contrast between the red petal and the red background in your first (leftmost) image, or between the sharp background and the blurry petal in your third (rightmost) image, for the algorithm to be able to discern which parts are in or out of focus.

  • \$\begingroup\$ You may be onto something - however, where it crosses the paler, yellowish part in the bottom left, it's really made a mess of that section too. Though it has made a much better guess at the top right of that 'petal', where it's backed by a darker brown. Maybe I should try sharpening first - that stack was pretty much straight from the camera. \$\endgroup\$
    – Tetsujin
    Commented Feb 2, 2017 at 14:51
  • \$\begingroup\$ A few more weeks of experimentation would say you're right, it's a contrast/edge issue. It has few problems if there's high contract & sharp edges. \$\endgroup\$
    – Tetsujin
    Commented Mar 13, 2017 at 9:12

I don't use photoshop, but I'd expect you to be able to edit the result once done. In the gimp it would be a case of adjusting the layer mask for the top layer. That might be a bit of a pain but at least the image would be recoverable, within even starting at the beginning of post-processing. You may find remaking the layer mask using the contrast in a different channel (I assume the default is composite) works in a few clicks. This is mainly based on the crops but this should help in the other areas as well, as the tips look like a slightly different colour.

  • \$\begingroup\$ This is actually what I've started to do over the past few weeks, if the resulting stack feels worth the effort - it isn't always. Sometimes I just change my lighting & start over. It's definitely contrast-based. \$\endgroup\$
    – Tetsujin
    Commented Mar 13, 2017 at 9:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.