peony with AI assistance

[DanK]

Last night I did a stacked shot of a bunch of peonies, testing out how well the focus bracketing of my R6 II would work for this type of work. Flawlessly is the answer. I created the stack in camera, culled it in LR, and stacked it in Zerene.

I was very tired when I finished, and the first edit below is overcooked. (Comments welcome, of course.) I'm posting it because of the somewhat irritating little peony bud in the bottom right. I shot this from a bunch of flowers cut from our garden and stuck in a vase without arranging them for the shoot, and I found a partially open bud with badly discolored petals where that bud is now. I should have removed it, of course. Instead, I selected it with the lasso tool and replaced it with Adobe's AI. First, I gave it no prompt, and the results weren't usable. I then added the text "use only a peony bud", and voila. My selection overlapped slightly with the adjacent pink petal, and that resulted in a faint blue (why?) line that I had to remove with the clone stamp, but otherwise, I didn't touch it.

[Round Tuit]

Dan,

The AI did a credible job of filling the lower right corner and fixing the slight overcooked look should not be a problem for you. The one thing that keeps nagging at me is the position of the flower. I keep wanting to move it toward the left. As it is, I find the composition very unbalanced. I tried flipping the photo horizontally which helps but does not completely fix the imbalance. A square crop fixes the imbalance but does not work well for me. I don't know what the solution could be.

[DanK]

Andre,

Thanks for the comment.

The shot was mostly a test of (for me) a new technique, an in-focus flower against a background of out of focus flowers, rather than the isolated flowers I usually photograph. I also wanted to test again the automatic focus bracketing on the R6 II, which had worked several times before but failed in a dark, low-contrast scene outdoors when I used my EF macro with an RF adapter. it worked flawlessly in this case, through several different series of shots.

If I had been doing this more carefully, I would have arranged things, for example, removing the bud and probably the leaf lower left. I also probably would have shot wider open (this was f/7.1) or moved the red flowers back to increase the focus difference.

However, I was deliberately not aiming for a centered framing. I think the problem with this is that it's centered in one direction but not the other, and the crop on the right is tight.

Unfortunately, these flowers don't last long when cut, and I don't know whether any more of these will be in usable condition after the rain we are having today.

Here's another photo of the same type of flower from a few years ago. I should have made the stem brighter.



Dan

[Cantab]

Dan, for the in camera focus bracketing, do you simply place the camera on a tripod, set some parameters and then the camera decides on the focus steps between each bracketed shot and then automatically takes the shots? (It sounds like cheating!)

[DanK]

Bruce,

This was done with a tripod, but at less close distances outdoors, like landscapes, I've done it handheld.

It starts wherever you initially focus, manually or automatically, and works back to greater distances. It has only two parameters: the number of images you want, and how far apart the focus should be from one shot to the next. In my limited experience, it will often but not always stop before the maximum you set if you get to a point where nothing is in focus. It didn't in this case.

The hard part is that the numerical scale is an arbitrary set of numbers, not a fixed distance. It has to be because the same scale has to work at close distances, where the difference in focus may be a millimeter or less, and outdoors, where it may be half a meter or more. I think the camera sets the distance between shots for any given number on the scale using the initial focusing distance and perhaps aperture. So, you need some trial and error to figure out what number works for a given type of photography, not leaving gaps. In this case, "5" on the scale worked fine.

I don't see it as cheating. There is nothing creative in what I used to do with my DSLRs: take a shot, manually change the focus a bit, hit the shutter with a remote release once to flip the mirror up, wait for the motion all of that caused in the flower to stop, and trip the shutter again. Some evenings I would do this perhaps 100 times. It's exacting but extremely tedious. I bought software from Helicon (the same company that makes one of the leading stacking packages) to automate this years ago but didn't get in the habit of using it.

This gets all of that out of the way so that I can focus on the creative parts: how I want the flowers arranged, how I want them lit, how I want to do the compositing in the stacking software, and how I want to edit the composite. Now the only hard part is figuring out where the nearest point is and making sure that the camera starts there or closer. I do this by holding a pencil or something in front of the nearest point while looking at it through the LCD at a high level of magnification. I usually then focus a little closer to be safe, since there is no harm in having a few out of focus. I then pause a bit to let any vibrations end, hit the shutter, and just wait. Wonderful.

After I upload the photos, I cull all of them that are out of focus at either end and decide which of the survivors I want to use (most often, all of them).

What I don't do is allow the camera to do the actual stacking, for two reasons. First, I think there are important decisions to be made in stacking, and I want control of them. Second, some cameras, including mine, create a JPEG or HEIF file when they make the composite, which I wouldn't want.

Dan

[Cantab]

Dan, thank you for your detailed comments. I'm going to print them off; I don't think I'm in the market for an R6 II but one day I'll need to upgrade from my 60D.

[DanK]

There are also less expensive models that have this feature, including the R7 and R10. Canon UK has a list online. You also need to make sure the lens you use supports this feature.


Sent from my iPad using Tapatalk

[Manfred M]

Excellent tutorial with description here.

I can see why you wanted to test the focus bracketing on the camera. Would the result have been different had you used your focusing rail?

[DanK]

Excellent tutorial with description here.

I can see why you wanted to test the focus bracketing on the camera. Would the result have been different had you used your focusing rail?

Good question. Possibly worse.

I no longer know the details but could probably find them. A few years ago, I chanced upon a very long video by an expert macro photographer who compared various approaches for creating the stack. To his surprise, he found that with one stacking method (I don't recall whether it was depth map stacking, which is the most common), the results he got were a bit off when he used a rail. He wrote to the guy who wrote Zerene, Rik Littlefield. Rick wrote back with a technical explanation of why that happened. I vaguely recall that it had something to do with movement of the entrance pupil, but I may be remembering completely wrongly. However, Rik's rule of them was that if the subject is bigger than a raisin, don't use a rail. Instead, change focus with the lens.

The real benefit of a rail is for extreme macro. I do up to 2x and have never needed it. I use a rail to help get the camera positioned initially, but I use the lens after that.

[Manfred M]

Thanks Dan - that's good to know.

My current camera requires manual focus stacking, so I tend to get lazy and don't use that technique. The one that I have on order has that feature and want to try using it in some of my landscape and still life work. I have been wondering about how the shifting of the camera body compares with focus breathing issues.

[manfred9]

Great examples and very good and detailed description. When I was reading the post it got me started to check if my 90D has the focus stacking feature as well. And yes it does. Did some first test and did the stacking in PS. Then I started looking for alternative software so I downloaded trials from Helicon and Zerene.

At the moment I tend to like the results from Zerene a bit more. Definitely better than PS.

What have you used for the shot with the peonies?

Thanks for posting your experience.

[DanK]

I always stack with Zerene. Haven't used anything else for years. There are various postings comparing Helicon and Zerene, but I'm happy with Zerene, so I haven't pursued those in depth. Photoshop is in my view only adequate for the least demanding stacking, e.g., stacking a few images when doing landscapes. It has limited functionality and is slow.

Zerene is very good at its default settings, but to get the most out of it, you really have to look at the online tutorials about DMap settings and retouching.

[manfred9]

I always stack with Zerene. Haven't used anything else for years. There are various postings comparing Helicon and Zerene, but I'm happy with Zerene, so I haven't pursued those in depth. Photoshop is in my view only adequate for the least demanding stacking, e.g., stacking a few images when doing landscapes. It has limited functionality and is slow.

Zerene is very good at its default settings, but to get the most out of it, you really have to look at the online tutorials about DMap settings and retouching.

Thanks for your input. Yes, checking the tutorials is my next step.

[DanK]

A few more thoughts, if you want to dig into this seriously:

Zerene has two stacking methods. (I think Helicon may have three.) One is DMap, a depth-map method that is similar to PS's and many others. The second is PMax, a "pyramid" method. PMax is faster, better at preserving tiny details, and less prone to the halos that are the bane of macro photography, but it doesn't preserve colors well. DMap preserves colors very well, but it is more prone to halos and has more parameters to tweak.

My most common approach when I think I will have halos (for example, when there is a substantial front-to-back distance between an edge and whatever is behind it) is to tell Zerene to use both methods. Zerene has a superb retouching tool, and rather than retouch from a single image when I find a halo, I retouch from the PMax to the DMap, effectively using PMax only where there are halos. I blow the images up to higher magnification to do this.

DMap has two types of settings: a contrast adjustment it will always ask you to set, and two radius settings. This is what the tutorials explain. I use the contrast adjustment most of the time but--for what I do--rarely have to change the radius. The defaul settings work well for most of what I do.

Good luck!

[manfred9]

Thanks for sharing @DanK. I am working with both stacking methods at the moment. I realized the halos in the DMap but I didn't think of combining the two methods. Your tip about retouching the halos from the PMax sounds like a great idea. Will test it.

Thanks.

[manfred9]

@DanK May I ask for another advice. How do you deal with noise in PMax images. I had some test shots with high ISO. The PMax output is very noisy. So I exported the PMax to Lightroom and did a manual denoise. Imported the file into the ZS project and used it as a source file in the editing process. Worked well for me.

Just wondering if you have an alternative solution.

Thanks.

[DanK]

Manfred9,

This is something that's discussed in the voluminous FAQs on the Zerene site. They have a lot of detail you won't want but some essential information you will. This is from https://zerenesystems.com/cms/stacke...it_suppression

As mentioned in other FAQs, the PMax method tends to accumulate noise. That's because in general it is relentless about preserving the sharpest detail at all size scales, but at the very finest scale it has trouble distinguishing between focused detail and pixel noise. The “grit suppression” option, which is selected by default, makes a small sacrifice in saving fine detail, in exchange for the benefit of getting significantly less noise in the final image.

Even with grit suppression on, PMax can create a crunchy and noisy image. The more noise in the base image, the more extreme that will be. That and its inferior color preservation are why I don't often use it for the primary composite, instead using it to retouch a DMap image as needed. However, there are cases where it's fine, or where the problems with DMap are too extensive to fix.

I almost never stack high-ISO images. I do my flowers under controlled lighting and almost always shoot them at ISO 100. I usually shoot bugs with a diffused TTL flash and a fixed aperture and shutter speed, and in that case, I have to vary ISO to set the balance between ambient light and flash (for example, to avoid having the background too dark). However, even there, I rarely shoot about ISO 400, in part because I use an old, noisy camera (7D first generation) for bugs in the field. Given that fine detail is what most macro is all about, noise is a real problem, and noise reduction comes with loss of detail.

Dan