Cheers,
Remco
- 51 stack
- Canon 50D
- 100mm macro lens
- f2.8 - 1/40s
- WeMacro rail with 100um steps via Helicon Remote
- led lights
Blurring around complex shapes
Blurring around complex shapes
Hi all! Recently started playing with macro a bit more using my old Canon 50D and a wemacro rail. One of my hobbies are carnivorous plants which is my current subject to take pics off at the moment. When taking stacks of sundew plants, you have the lovely dew drops. Around these complex shapes there's a blur most of the time which I can't seem to get rid of whatever setting I pick. Any suggestions?
Cheers,
Remco
Cheers,
Remco
Re: Blurring around complex shapes
Hi Remcon,
I've spent a lot of time on this issue. But even so, I welcome others who are more knowledgeable about optics to correct or improve my answer.
The normal set up of a camera has the sensor fixed in place, and the lens moves when you refocus.
Distance to the image (Di) divided by distance to the object (Do) = Magnification
When you move the lens closer to the sensor to focus on the the farther object, Di decreases and Do increases. The magnification gets smaller.Apparently, HeliconFocus deals with this by changing images sizes to match the magnification. That's good, but it leaves a big out of focus foreground object next to the in-focus background in the faraway images. You end up with halos around near objects since you your faraway images have a big "shadow" of the nearby object blocking the in-focus far away object, and the nearby images have no in-focus pixels of the faraway objects.
I spend a lot of time and cloning background data onto the halos if they are annoying. I use some combination of the stamping tool and the spot healing tools in Photoshop.
See the images I am posting later today of the orchids for an example. Look at high contrast original and final, and compare the halos around the nearby leaves. It's not perfect, but it's a useful improvement.
Phil
I've spent a lot of time on this issue. But even so, I welcome others who are more knowledgeable about optics to correct or improve my answer.
The normal set up of a camera has the sensor fixed in place, and the lens moves when you refocus.
Distance to the image (Di) divided by distance to the object (Do) = Magnification
When you move the lens closer to the sensor to focus on the the farther object, Di decreases and Do increases. The magnification gets smaller.Apparently, HeliconFocus deals with this by changing images sizes to match the magnification. That's good, but it leaves a big out of focus foreground object next to the in-focus background in the faraway images. You end up with halos around near objects since you your faraway images have a big "shadow" of the nearby object blocking the in-focus far away object, and the nearby images have no in-focus pixels of the faraway objects.
I spend a lot of time and cloning background data onto the halos if they are annoying. I use some combination of the stamping tool and the spot healing tools in Photoshop.
See the images I am posting later today of the orchids for an example. Look at high contrast original and final, and compare the halos around the nearby leaves. It's not perfect, but it's a useful improvement.
Phil
Re: Blurring around complex shapes
Hello Remco/Phil,
This is my first post. I agree completely with Phil on the Halos. It is quite frustrating. I was hoping to hear that Phil's Phase One would have solved the problem. What lens are you using, Phil? How are you advancing the setup? Is it the NovoFlex? I am working a number of different algorithms with no success yet. It has to be an Adaptive Interpolation. The only group that I have heard may be able to help after we use our Helicon is "Smart Edge", but I can't find out who owns it. It's not on the market yet. I know it is not Adobe.
I only photograph flora/grass. Mother Nature's middle name is Chaos. I have learned to smile, and I realize she will win most of the time, because I stack at least 200+ images into her final beauty. I adore the challenge.
Greg
This is my first post. I agree completely with Phil on the Halos. It is quite frustrating. I was hoping to hear that Phil's Phase One would have solved the problem. What lens are you using, Phil? How are you advancing the setup? Is it the NovoFlex? I am working a number of different algorithms with no success yet. It has to be an Adaptive Interpolation. The only group that I have heard may be able to help after we use our Helicon is "Smart Edge", but I can't find out who owns it. It's not on the market yet. I know it is not Adobe.
I only photograph flora/grass. Mother Nature's middle name is Chaos. I have learned to smile, and I realize she will win most of the time, because I stack at least 200+ images into her final beauty. I adore the challenge.
Greg
Re: Blurring around complex shapes
Hello once again Remco and Phil,
The concept of the halos is full of trade-offs. Could it be Blurring or maybe Alliasing? This seem to be similar to the complex nature of enlarging a photograph. Algorithms don't treat all pixels equally.
I decided to step away from the math for some experiments.
Light! I now only use macro flashes (2-4). Each can be calibrated to whatever the individual flower or grass tells me it needs. I then have found that the lower the better. Raising in Post. This improved the edge halo about 25%, if it is edge halo.
That led to 2 sleepless nights. Passive white reflection (small, 12"). The more the better. I need each of the 12" reflectors to have 100 tiny reflectors. You can probably understand where I'm going with this. I happens more with grasses or when there is a lighter object on top of a darker object. No body makes this type of reflector. I don't have time or the inclination to invent one at my age. It has solved another 25% of the problem.
I'm still not sleeping. So, decided to go crazy and really pump up the stacks. It has definitely helped.
I am still experimenting with all of these and other variables. Six more methods planned.
Enjoy,
Greg
The concept of the halos is full of trade-offs. Could it be Blurring or maybe Alliasing? This seem to be similar to the complex nature of enlarging a photograph. Algorithms don't treat all pixels equally.
I decided to step away from the math for some experiments.
Light! I now only use macro flashes (2-4). Each can be calibrated to whatever the individual flower or grass tells me it needs. I then have found that the lower the better. Raising in Post. This improved the edge halo about 25%, if it is edge halo.
That led to 2 sleepless nights. Passive white reflection (small, 12"). The more the better. I need each of the 12" reflectors to have 100 tiny reflectors. You can probably understand where I'm going with this. I happens more with grasses or when there is a lighter object on top of a darker object. No body makes this type of reflector. I don't have time or the inclination to invent one at my age. It has solved another 25% of the problem.
I'm still not sleeping. So, decided to go crazy and really pump up the stacks. It has definitely helped.
I am still experimenting with all of these and other variables. Six more methods planned.
Enjoy,
Greg
-
shutterbug
- Posts: 59
- Joined: 17.12.2022 15:14
Re: Blurring around complex shapes
I know this is a year old thread but still:
If you have a lot of complex structures in the foreground and background you may want to use Method C (pyramid) instead of Method A or B.
Method A and B are great for flat surfaces without too many changes in dimensionality or rather geometrical figures. With complex overlapping, dimensional stuff Method C often gets better results, though the contrast sometimes suffers.
Personally I would suggest combining methods for the cleanest results and/or render the foreground and background objects separately and then combine them.You could even combine these rendered images again in Helicon Focus without resorting to Photoshop.
EDIT: as for the original image I would probably change your shooting parameters as well.
With a 100mm macro lens f 2.8 at 100um steps isn't quite necessary or beneficial unless you're shooting at 3x or 4x magnification and you're diffraction limited. At f2.8 you will have a very shallow DOF and that will cause a very sharp separation of foreground/background elements - which will make if more difficult to then stack them properly.
Instead you might want to try f4.0 or maybe even f5.6 - as a benefit you will need fewer images and your step size doesn't have to be quite as small.
If you have a lot of complex structures in the foreground and background you may want to use Method C (pyramid) instead of Method A or B.
Method A and B are great for flat surfaces without too many changes in dimensionality or rather geometrical figures. With complex overlapping, dimensional stuff Method C often gets better results, though the contrast sometimes suffers.
Personally I would suggest combining methods for the cleanest results and/or render the foreground and background objects separately and then combine them.You could even combine these rendered images again in Helicon Focus without resorting to Photoshop.
EDIT: as for the original image I would probably change your shooting parameters as well.
With a 100mm macro lens f 2.8 at 100um steps isn't quite necessary or beneficial unless you're shooting at 3x or 4x magnification and you're diffraction limited. At f2.8 you will have a very shallow DOF and that will cause a very sharp separation of foreground/background elements - which will make if more difficult to then stack them properly.
Instead you might want to try f4.0 or maybe even f5.6 - as a benefit you will need fewer images and your step size doesn't have to be quite as small.
Re: Blurring around complex shapes
Hello Gregory and shutterbug,
Continuing this antique thread. Sorry for not getting back to you soon.
I'll try method C, but I predict it won't help much because of the magnification issues I posted earlier.
Paraphrasing what I wrote in 2022:
As you move the lens closer to the object -- as when you focus on a near point -- the magnification increases. As you move the lens away from the object towards the sensor -- as when you focus on a far point -- the magnification decreases.
Consequence: Halos happen around nearby objects because the far-away in-focus pixels near the edge of the nearby objects are blocked by the in-focus nearby and larger objects.
The good news is I've found a new way to fix a lot of halos. With the combination of careful use of the lasso tool and "Content-Aware Fill", you can fake a lot of the halo into looking like the background.
Phil
Continuing this antique thread. Sorry for not getting back to you soon.
I'll try method C, but I predict it won't help much because of the magnification issues I posted earlier.
Paraphrasing what I wrote in 2022:
As you move the lens closer to the object -- as when you focus on a near point -- the magnification increases. As you move the lens away from the object towards the sensor -- as when you focus on a far point -- the magnification decreases.
Consequence: Halos happen around nearby objects because the far-away in-focus pixels near the edge of the nearby objects are blocked by the in-focus nearby and larger objects.
The good news is I've found a new way to fix a lot of halos. With the combination of careful use of the lasso tool and "Content-Aware Fill", you can fake a lot of the halo into looking like the background.
Phil
Using Content-Aware Fill on halos
Example of Using Content-Aware Fill on halos
original draft image
https://www.dropbox.com/s/wxuy73g1igs4p ... y.jpg?dl=0
after two minutes of using Content-Aware Fill on the left and upper left halos:
https://www.dropbox.com/s/2z4zd3n0w57sl ... s.jpg?dl=0
A decent effect is sufficient for me in this image since I will darken the background to almost black and low contrast in the final image.
original draft image
https://www.dropbox.com/s/wxuy73g1igs4p ... y.jpg?dl=0
after two minutes of using Content-Aware Fill on the left and upper left halos:
https://www.dropbox.com/s/2z4zd3n0w57sl ... s.jpg?dl=0
A decent effect is sufficient for me in this image since I will darken the background to almost black and low contrast in the final image.
Method B vs. Method C for fighting halos
Method B, as I posted earlier today
/https://www.dropbox.com/s/wxuy73g1igs4p ... y.jpg?dl=0
Method C, as suggested by shutterbug
https://www.dropbox.com/s/c7gbdiiw1u42k ... y.jpg?dl=0
I was wrong. Method C is definitely better. Look at the left edge. And more interesting -- look at the horizontal curved line at the bottom of the top center large area. C made both of these better. Not perfect, but better in both cases.
I was wrong. shutterbug was right. C works better at decreasing halos! Thank you, shutterbug. I am switching to C for images like these!
Phil
/https://www.dropbox.com/s/wxuy73g1igs4p ... y.jpg?dl=0
Method C, as suggested by shutterbug
https://www.dropbox.com/s/c7gbdiiw1u42k ... y.jpg?dl=0
I was wrong. Method C is definitely better. Look at the left edge. And more interesting -- look at the horizontal curved line at the bottom of the top center large area. C made both of these better. Not perfect, but better in both cases.
I was wrong. shutterbug was right. C works better at decreasing halos! Thank you, shutterbug. I am switching to C for images like these!
Phil
-
shutterbug
- Posts: 59
- Joined: 17.12.2022 15:14
Re: Blurring around complex shapes
Hmmm well Phil you could also use Method B and increase the radius gradually and reduce smoothing until you're satisfied with the result.
That's a subject where I think both methods could work.
That too should prevent that sort of glow from occurring.
That's a subject where I think both methods could work.
That too should prevent that sort of glow from occurring.
Re: Blurring around complex shapes
Thanks, shutterbug, I'll try that if I have the patience for the multiple experiments. As I was writing this note, my camera was making 76 images; each will be 907MB. Helicon Focus experiments take a long time. I am exploring getting a faster machine.
Phil
Phil
Re: Blurring around complex shapes
I've got a new analysis/understanding.
I did the math about the magnification change, checked it at least three times, and had another mathematician check my work. If anyone is interested, I'll post the equations. They require no more than high school geometry and algebra.
Conclusion 1.
The nearer the focus, the greater the magnification. Therefore, if you have a tight composition, compose it while focused on the nearest object. Otherwise you will lose parts of the image around the edges.
Conclusion 2. The near object shrinks more than the far object when you focus on the far object. This is the opposite of what I thought was happening. Therefore, the blurring is due to the out-of-focus spreading of the near object when we're focused on the far object.
I shoot my focus stacking images at f8. My next experiment will be to do a second set of focus-stacked images at f32 or something like that to see if I can use a closer-focused image of the far object with a less-blurry near object image. I'll give up the lack of diffraction at f8 if I can get a more focused image where the halo usually resides.
If anyone has other ideas or tries this, please post your ideas and results.
Thanks,
Phil
I did the math about the magnification change, checked it at least three times, and had another mathematician check my work. If anyone is interested, I'll post the equations. They require no more than high school geometry and algebra.
Conclusion 1.
The nearer the focus, the greater the magnification. Therefore, if you have a tight composition, compose it while focused on the nearest object. Otherwise you will lose parts of the image around the edges.
Conclusion 2. The near object shrinks more than the far object when you focus on the far object. This is the opposite of what I thought was happening. Therefore, the blurring is due to the out-of-focus spreading of the near object when we're focused on the far object.
I shoot my focus stacking images at f8. My next experiment will be to do a second set of focus-stacked images at f32 or something like that to see if I can use a closer-focused image of the far object with a less-blurry near object image. I'll give up the lack of diffraction at f8 if I can get a more focused image where the halo usually resides.
If anyone has other ideas or tries this, please post your ideas and results.
Thanks,
Phil
Re: Blurring around complex shapes
Hi Phil,
Thanks for sharing your findings and tips!
I believe you are correct and the narrow aperture (higher F number) will have exactly the effect you have described.
Thanks for sharing your findings and tips!
I believe you are correct and the narrow aperture (higher F number) will have exactly the effect you have described.
Re: Blurring around complex shapes
Hello Phil,
I might miss a a point but I think this is a wrong assumption:
"Paraphrasing what I wrote in 2022:
As you move the lens closer to the object -- as when you focus on a near point -- the magnification increases. As you move the lens away from the object towards the sensor -- as when you focus on a far point -- the magnification decreases."
Why could this be wrong? When we do focus stacking, indeed we are moving the object towards the lens (or vice versa).
However -we do this to change the focus on the next area of sharpness! That means the distance between the lens and the sharp area does NOT change.
As an example: I use a 10x lens, the lens has around 7micrometer depth-of_view (DOV). While stacking I move the lens lets say 3-5micrometer towards the object. This results in the same distance between lens and sharp areas for each single image of a stack.
To my understanding Helicon does not scale the images during the stacking process, it can compensate small amounts of vibrations or rotations by scaling the images, but this is an "add-on" and not directly related to the stacking process itself.
"Consequence: Halos happen around nearby objects because the far-away in-focus pixels near the edge of the nearby objects are blocked by the in-focus nearby and larger objects."
Here I fully agree - because the blurry parts are the part of the image that moves towards the lens and hence they change their size because of a change in magnification.
as said - I might miss a big point here and I am happy if you proof me wrong ...
best regards
Marc
I might miss a a point but I think this is a wrong assumption:
"Paraphrasing what I wrote in 2022:
As you move the lens closer to the object -- as when you focus on a near point -- the magnification increases. As you move the lens away from the object towards the sensor -- as when you focus on a far point -- the magnification decreases."
Why could this be wrong? When we do focus stacking, indeed we are moving the object towards the lens (or vice versa).
However -we do this to change the focus on the next area of sharpness! That means the distance between the lens and the sharp area does NOT change.
As an example: I use a 10x lens, the lens has around 7micrometer depth-of_view (DOV). While stacking I move the lens lets say 3-5micrometer towards the object. This results in the same distance between lens and sharp areas for each single image of a stack.
To my understanding Helicon does not scale the images during the stacking process, it can compensate small amounts of vibrations or rotations by scaling the images, but this is an "add-on" and not directly related to the stacking process itself.
"Consequence: Halos happen around nearby objects because the far-away in-focus pixels near the edge of the nearby objects are blocked by the in-focus nearby and larger objects."
Here I fully agree - because the blurry parts are the part of the image that moves towards the lens and hence they change their size because of a change in magnification.
as said - I might miss a big point here and I am happy if you proof me wrong ...
best regards
Marc
Re: Blurring around complex shapes
Hello Marc,
Helicon Focus does scale images during alignment, if it didn't - it wouldn't have been able to produce acceptable output, as the magnification of the subject changes as you shoot at different focal planes. Like you said, the distance between the lens and the area of sharpness is constant (unless you change any optical parameters of your setup), but the area of sharpness moves along the subject, which means that the subject moves relative to the camera. And closer objects take a larger area on the camera's sensor, i. e. they are magnified (effectively). It's an inherent phenomenon of focus stacking.
Helicon Focus does scale images during alignment, if it didn't - it wouldn't have been able to produce acceptable output, as the magnification of the subject changes as you shoot at different focal planes. Like you said, the distance between the lens and the area of sharpness is constant (unless you change any optical parameters of your setup), but the area of sharpness moves along the subject, which means that the subject moves relative to the camera. And closer objects take a larger area on the camera's sensor, i. e. they are magnified (effectively). It's an inherent phenomenon of focus stacking.