I've noticed recently that many of my focus stacked pics using Combinezm have unnaturally sharp DOF boundaries. I initially thought this was because with a a very wide DOF the eye/brain expects to see a much more gradual drop off in DOF. However some focus stacks I've played with recently have made me realise this is not the whole reason. CZM seems to have a problem with slightly OOF or low contrast detail and just does not include it in the stacks. Funnily enough the older combinez5 is not quite as bad at this. I've started a discussion with the software author re this problem and sent him some sample pics and stacks.
The pics below illustrate the problem
#1 is just the first photo used in the stack and shows the extent of the stack but also the expected DOF drop off at the boundaries.
#2 is the result of the CZM stack with just the borders cropped off. You can clearly see the extreme DOF boundaries esp at the back.
#3 is the same stack but with some of the lost detail layered back on from the original shots.
That is pretty extreme -and I've seen it in other stacks (not just yours).
FWIW I actually like the first image better -and the only thing I'd do to improve it is to clone out the critter in the upper left corner. With the stacked photo my eye doesn't know where to go (no clearly defined subject) and the areas that are OOF are distracting. What would be really cool is if you could take enough frames to make the entire image sharp...
I think you are on to something. #3 is quite a bit better than #2, #2 looks as if the back louse if being cut by the DOF line. Interesting phenomenon - do other stacking programs correct for this issue?
Dalantech wrote:
That is pretty extreme -and I've seen it in other stacks (not just yours).
FWIW I actually like the first image better -and the only thing I'd do to improve it is to clone out the critter in the upper left corner. With the stacked photo my eye doesn't know where to go (no clearly defined subject) and the areas that are OOF are distracting. What would be really cool is if you could take enough frames to make the entire image sharp...
Thanks John
Think I might agree with you on this one - I liked the non-stacked shot best. I actually did this set as a test to see if the phenomenon I'd seen in other stacks of mine and other peoples was real or not - not really a fair comparison of the actual shots taken - just a test set.
MichAg92 wrote:
I think you are on to something. #3 is quite a bit better than #2, #2 looks as if the back louse if being cut by the DOF line. Interesting phenomenon - do other stacking programs correct for this issue?
I'm really not sure if helicon focus suffers from similar problems - I only tried it a few times a couple of years ago and it seemed to make more of a mess than combinez5 at the time. Mind you i was very used to using cz5 and not used to helicon - so not a very fair comparison. I often stack pictures by hand now- a lot easier with a little bit of intelligence behind the process
I'm pretty certain most of these stacking programmes started life as a tool in microscopy for building up a sort of 3-D model. They may therefore not like the gradual drop off in DOF you can get when shooting at an angle rather than straight onto a subject.
[EDIT] - just tried a stack of shots causing the problem with czm eith an oldish version of helicon I had around. Whilst it made a bit of a mess of alignment it did not seem to suffer from the same DOF boundary problem
LordV wrote:
Thanks John
Think I might agree with you on this one - I liked the non-stacked shot best. I actually did this set as a test to see if the phenomenon I'd seen in other stacks of mine and other peoples was real or not - not really a fair comparison of the actual shots taken - just a test set.
Brian V.
It is an excellent test series -really shows the problem well!
well pointed out.I have noticed this effect in some of my shots mostly with the really small stuff as I often shoot down onto the bug and whatever background it is on,I must have a go at stacking by hand
michael kilner wrote:
well pointed out.I have noticed this effect in some of my shots mostly with the really small stuff as I often shoot down onto the bug and whatever background it is on,I must have a go at stacking by hand
michael - it's quite easy to repair some of these problems manually just using the healing brush tool in PS from one of the originals back onto the stacked photo although you may have to resize the original photo first.
Glad to say the Author has already cured the problem - found a bug in the code after trying some sample stacks I sent him
He should be posting the modded exe file shortly. How's that for speed of service- thanks again to Alan Hadley the author of this wonderful programme.
same stack as above run through the new executable with no hand interference.
Wow. Big difference. Many kudos to you for identifying the error and to the author for addressing it so quickly! - if only the rest of us could get that kind of response from Bill Gates
Yes, Helicon does have some of this same problem--'though I've never seen it that bad!
I'm glad to see that CZM gets updated and fixed so fast--since my Helicon trial ran out about the same time as my wealth, I think I'll have to learn the ins and out of CZM for a while.
Damn, ain't those isopods primitive lookin' critters?
Yes, Helicon does have some of this same problem--'though I've never seen it that bad!
I'm glad to see that CZM gets updated and fixed so fast--since my Helicon trial ran out about the same time as my wealth, I think I'll have to learn the ins and out of CZM for a while.
Damn, ain't those isopods primitive lookin' critters?
Thanks Timm
great response time on the bug fix- Have to admit I thought maybe the author had been optimising the prog for speed and hadn't noticed a gradual problem appearing with the dof boundary detail but it turned out a genuine programming bug had crept in unnoticed.
Have to say I've always found czm better at aligning my stacks than helicon in the few runs I've had with helicon but suspect it may be due to my inexperience using helicon (ie I was missing out a separate alignment stage somewhere). One advantage Helicon does have is the file types it will handle and output - CZM does not handle 16 bit TIFFs as an output- not that I use them anyway.
The revised code does produce a much more pleasing effect. In the first example, it looked almost as if there was a clearly defined micro fog-bank from which the rear critter was emerging.
But I also believe that your original observation that the eye-brain expects to see a gradient of blur that is proportional to the depth of the sharp area is correct too. So even the "fixed" code does produce an effect that is a bit different than what (at least my) eye/brain expects. But it's FAR better, and is a great tool that produces a very handy and enjoyable effect.
It's probably as good as it can get unless you had a bunch more shots that had good detail farther away and closer in to use in the stack - and then at that point, rather than use them to achieve a "normal-looking" blur-gradient, you'd be able to do what John was talking about and just have everything in the frame be dead-sharp.
I think the author should provide a new control for the program that lets us have an adjustable application of the "bug" in the code so that if one wanted to get that odd effect of the very sharp DOF cutoff, you could just dial it in with a slider
This is all quite interesting. I really need to try some of this myself. It's amazing!
JimH. wrote:
Thanks for that whole post and the fix info too.
The revised code does produce a much more pleasing effect. In the first example, it looked almost as if there was a clearly defined micro fog-bank from which the rear critter was emerging.
But I also believe that your original observation that the eye-brain expects to see a gradient of blur that is proportional to the depth of the sharp area is correct too. So even the "fixed" code does produce an effect that is a bit different than what (at least my) eye/brain expects. But it's FAR better, and is a great tool that produces a very handy and enjoyable effect.
It's probably as good as it can get unless you had a bunch more shots that had good detail farther away and closer in to use in the stack - and then at that point, rather than use them to achieve a "normal-looking" blur-gradient, you'd be able to do what John was talking about and just have everything in the frame be dead-sharp.
I think the author should provide a new control for the program that lets us have an adjustable application of the "bug" in the code so that if one wanted to get that odd effect of the very sharp DOF cutoff, you could just dial it in with a slider
This is all quite interesting. I really need to try some of this myself. It's amazing!
There are all sorts of ways you can vary the DOF effects by mixing machine made focus stacks with hand focus stacks - In several of my stacked photos I'm purposely using Differential DOF where I'm stacking the main subject but not the BG but you have to do that by hand. Other ways of playing are to shoot a couple of pics at the start and end of the stack that have much smaller aperture to give greater DOF. The author on the yahoo group site acutally gave details of the programme error so you could always put it back in. As it is I've still got the old and new progs so could always swap them
out of curiosity I tried Helicon and it does seem to suffer from the same effects to some extent.One thing I did notice though is that helicons stacked image retains a much more accurate colour rendition of the the original images than CZM.Unless I am doing something wrong my CZM stacks look a little washed out,any thoughts?
michael kilner wrote:
out of curiosity I tried Helicon and it does seem to suffer from the same effects to some extent.One thing I did notice though is that helicons stacked image retains a much more accurate colour rendition of the the original images than CZM.Unless I am doing something wrong my CZM stacks look a little washed out,any thoughts?
michael- I don't tend to have much problem with czm colour rendition but it does have to do colour matching and I think it does contrast changes during the output. Are you using sRGB all the way through as your colour space ?
See what you mean but nothing drastic that a little bit of levels correction would not cure. Suspect that is because of the colour matching it has to do- you may find one of the shots was slightly lighter.
