Need some help with focus stacking impasse please

[irlien]

Hi everybody, thanks for reading!

I'm trying out my new focus rail, a Phottix. I'm trying to do some pack shots for a client and decided to give focus stacking a go. I'm not shooting macro, just using a 50mm at f11. My dof is 3.25cm according to the spec sheet for my lens & camera. I have made several attempts, making more and less stacks (5; 10 and 20). I used both Photoshop CS6 and Zerene to compile the layers to compare results.
The results are disastrous: the photos are either jagged or appear to be melting. Does anybody have an idea what's going wrong, the procedure is quite straightforward?
Have a look at my results please:
Sharpened version.jpg
Untitled_Panorama2.jpg

TIA, irlien

[Andrew76]

Hey Irlien, I'm not an expert, but I've used Zerene fairly extensively since another member here (DanK), recommended it to me. I've done a ton of macro work with it, and I have NEVER seen anything like that!

I'm sorry - I'm sure there are other members here who have more experience than I do. It almost appears as though the images are not aligning properly before the stacking software takes over.

Best of luck!

[irlien]

Thanks Andrew. I think you're on to something there about alignment.

I might do a series of melting pictures in the future!

[James G]

Irlien,
I don't use a rail, I now use Helicon remote mounted on a Nexus tablet to control my Canon. The advantage is that I do not need to touch the tripod or camera once set up.

You need to ensure that if you are using a rail that you allow the system to settle before you take each shot, after you advance the focus.

I have usedZerene to do stacks and got reasonable results, but just recently I have been using cs6 only.

Try the following with the set you captured: -

1) in CS6 use Load files into a stack Files>Scripts>Load Files into Stacks. ( Make sure you tick the attempt to align box).
2) once the images are into CS6 and aligned, then use CS6 to Auto blend .....Edit>Auto-Blend-Layers (as stack)

James

[chauncey]

Referring to CS6...assuming that you are "loading into layers" from LR and "selecting all" in CS6.
There are numerous options available in the "auto align" dialogue box, it is not a "one method fits all" scenario...try different methods and, be aware that using the same method anew may/may not cough
out different results. It's often a crapshoot.
Fortunately, the "auto blend" option has but one viable option..."stack images".

Your glasses problem...are you positive that your not altering the perspective by going off to the side as you move the camera in/out on the rail?

I don't use a rail...merely lock down camera position and manually move focus point in and out.

[Glenn NK]

I've used Zerene quite a bit, and any time I've had a problem, it's because something moved - either the subject (flower), or the camera (tripod settled).

NOTHING must move from frame to frame, and vibrations can also be a problem. On another thread (now locked), I posted the following (edited a bit):


Balance a small dish of water on the hot shoe (carefully so as not to tip over and rain on the camera), and using a cord type remote release and with MLU set, I carefully started the procedure:

1. Any vibration (movement) will result in ripples in the water.

2. The mirror-up slap was very obvious - ripples immediately.

3. When the ripples stopped (this took just over one second), I pressed the remote button again (this opens the shutter).

4. Surprisingly from the shutter opening, there were still ripples, although noticeably less than from the mirror slap.

5. At the end of the exposure, the shutter closed resulting in more ripples - this of course has no effect on the image.

6. What I did notice was that I had to keep the remote release cord perfectly still or the very small force from the moving cable would cause ripples in the water. Note what DanK said (Northern Focus) in his post: "most times when using a remote, I take the strap off the camera. Straps flapping in the breeze are also an innocuous source of camera shake when working on a tripod".

7. After the mirror went up, it took more than a second for the ripples to stop, so using MLU is effective but one doesn't have to wait very long for the vibrations to stop before releasing the shutter.

8. Added 26 October: I had to be very careful to not move my feet, or even shift my balance as this would induce ripples in the water.


Glenn

[Glenn NK]

Your glasses problem...are you positive that your not altering the perspective by going off to the side as you move the camera in/out on the rail?

Chancey - very good point:

Moving the camera WILL alter the perspective (parallax). It can be minimized by using a longer lens - the longer the lens and the farther the camera is from the object, the less parallax shift there will be.

Also with a wider object, parallax shift will be more noticeable. This is the same phenomenon that applies to portraits - shooting up close makes a person's face look bloated - moving back improves the perspective. The problem is that we have two lenses that form a single image in our brains - thus compensating for parallax error, but the camera has only one lens (and no brain).

Glenn

[irlien]

Thanks guys,

@Chauncey I tried a few alignment options coming from Capture One and Bridge alright, they all give some kind of error. Thanks for the tip on perspective, I have a feeling my camera is not exactly straight-on. I'm definitely not shifting between shots though. I thought perspective might shift vertically due to the camera weight at the extremity of the rail.

Keep 'em coming folks!

[irlien]

I've just read your aquabatics post Glenn and realised I'm not using MLU...

I'm going to reshoot now.

Can anyone explain the odd background in the first photo, it's pure 255 in each shot? I'm using a mix of Profoto and Elinchrom...

[Glenn NK]

I've just read your aquabatics post Glenn and realised I'm not using MLU...

I'm going to reshoot now.

Can anyone explain the odd background in the first photo, it's pure 255 in each shot? I'm using a mix of Profoto and Elinchrom...

I'm suspecting that parallax shift is a large part of your problem. This little sketch shows the effects of changing camera position.



Glenn

[chauncey]

In addition to the potential parallax shift problem...consider your choice of glass. The 50mm lens has a lot of distortion at the edges of the frame, magnified when you use f/11. Focus stacking distorted images might not work very well. Using a longer lens, with less distortion might serve you better.

[irlien]

I've locked the mirror now and switched to my 85mm. Merging in CS6 using different modes seems to be helping too: perspective mode improved results significantly.
I'll continue with the different merge modes for now and try a longer lens later. Thank you all for the tips.

[Stagecoach]

Hi Irlien,

Another area to consider is that you are looking at quite a long travel on a rail and it's build quality will affect your results.

I have a lower end unit of which I always ensure that at the middle of the planned travel the centre of gravity of camera/lens is in line with the centre of the tripod. The play in the slide if set too free can allow additional vertical movement of the camera the farther its weight is off centre to the tripod.

Grahame

[DanK]

I think the problem has to be either parallax or motion. I'm guessing the latter. My suggestion is that you try this without the rail, which is most likely source of the motion. MLU is always a good idea, but there is no way that mirror slap will move your camera enough to cause the ghosting your image shows.

Here is how I would test this:
--set the camera on a very solid tripod
--use MLU and a remote release.
--focus on the nearest edge and take the first image.
--now change focus slightly to the rear. How slight depends on DOF, but the in-focus areas of the successive images must overlap. This is the same principle as with a rail.
--let the camera fully settle, and take the second image.
--keep going until you have an image with the farthest part in focus
--choose align and stack with PMax in Zerene. PMax is faster and halos a bit less from parallax than DMap.

If this is clean, you have your answer. In that case, just use this technique routinely and put away the rail. You don't need one just to focus stack. I do a lot of macro work with stacking and don't even own a rail.

If you still have halos, then they are from parallax. Then there is one more simple fix to try:

--repeat the above, but using DMap
--when DMap finishes its first pass, which is approximate, it will give you a contrast slider. Slide that to the right and see if it eliminates most of the problem. If it does, accept and let the program finish.

You can also touch this up using Zerene's retouching tool, but unless you get most of it out first, it can be a tedious process.

[chauncey]

While I totally agree with everything that has been said, sometimes ya just get lucky...like when I broke every rule on focus stacking. Handheld 180 macro mounted on a Canon 1Ds3.



BTW, this guy is the accepted source...http://photography-on-the.net/forum/...d.php?t=807056

[irlien]

Thank you everybody for your input!

I finally got around to trying a longer lens, a 135mm and doing a few less stacks since my DOF changed to 7cm. I used my Gitzo 1545GT, locked up the mirror and plugged in my release.

Still using the rail (I believed it to be a better technique, upon advice), I tried different editing suites. CS5 was better than CS6... Zerene is the best however, I'm going to buy it.

I'll try without the rail after lunch and post the result to compare with this photo stacked with Zerene

[DanK]

Glad it worked out.

Can you please add your real name to your profile? If you look at other postings here, you will see that we use real names.

Re Zerene: I couldn't agree more. Once I learned my way around with it, I lost all interest in any alternatives. Check out Rik's (the developer's) videos online that show how to use the retouching tool, which is an invaluable way to deal with halos and other such. Rik is also remarkably responsive to user problems. Last week, for the first time, I tried stacking images taken with my 5D Mark III. All of my others have been with my 50D, which created files less than 2/3 the size of those from the 5DMkIII. Zerene balked at the 16-bit TIFFs from the MkIII, requiring that I downsize the images. I sent an e-mail asking for help, figuring that I would hear several days later, but I had a detailed reply by the next morning.

Re a rail--just for the sake of others who are getting into this: unless you are going to do very high levels of magnification, you really don't need one for image stacking, and a low-quality rail can hurt more than help. if you look at my flower macros, the large majority are stacked, and I don't even own a rail. In fact, it is not even on my wish list. If you are going to work at magnifications higher than 1.5:1 or 2:1, the changes in focus have to be minute, and many people find that in those cases, a high-quality rail helps. I virtually never go higher than 2:1, and I have never felt the need. I will post an image below that is fairly deep, as flowers go, and that I think is a stack of around 20 images, using Zerene's DMap. I'm not writing this to pick a fight with those who like rails, but just to clarify to people starting out that the lack of a rail is not an impediment.

For small subjects, not like these eyeglasses, one piece of equipment that does matter is a good tripod head, because minor movement--like the inevitable sag after you cinch down a ball head--can completely ruin the framing. I eventually moved to a Manfrotto geared head for tripod-based macro work, but that is a story for another day.

Dan

[irlien]

Hey,
Neil again. Here is the non-rail version, processed with Zerene. Photoshop made a mess of it again. Grahame makes a valid point about the extremities of the rail. I suspect that had an influence to some degree. So it's bye-bye to the rail, if you see it on eBay don't buy it! I'm joking, it slides horizontally too so I will use it to stitch panoramas.
Thanks again everybody, particularly DanK for posting diagrams and a lot of useful links. Please excuse the dust on the glasses if you can see it, it's just a trial. Even the lighting needs to be worked on. Here it is:

2013-10-31-14.22.48-ZS-PMax.jpg

[Glenn NK]

Neil:

Glad it all worked out for you.

I have a focusing rail also - and it moves on both axes. It is also not very rigid - fortunately I bought it on consignment and didn't cost much because it's not worth much. Actually because it's worthless, I paid too much for it.

A rail is of no use in pano stitching without a TS lens - with a TS lens, the amount of lateral shift used in the rail should be the same amount that the lens is shifted (except in the opposite direction). This will avoid parallax errors.

Without a TS lens, the camera has to be rotated around the nodal point of the lens, and a rail would be of no use. In this situation, parallax shift isn't a problem, but edge distortion of the lens can be.

http://www.youtube.com/watch?v=k0HaRZi-FWs

Glenn

[Stagecoach]

Neil,

I would fully support what Dan has said from my own experience and seeing the results of others.

The rail is used with my macro work where I am going up to 1:1 and the main reason I find it at advantage is that physically it is easier for me to turn a small knurled knob whilst watching a clearly marked scale to count divisions.

Some time ago I devised a 'scale template' to wrap around the focus barrel of the 105 macro that gave equal linear divisions of focus distance for controlling your slices from what is a logarithmic distance rotation.

Grahame