I've been running into the limitations of Photoshop when processing focus-stacked images for my greenhouse blog. After some internet research I found five ways to do focus stacking with my particular camera, which is a Canon R5 Mark II.
- Photoshop (my current method)
- Canon in-camera Depth Compositing
- Canon Digital Photo Professional Depth Compositing
- Zerene Stacker program
- Helicon Focus program
To start the testing, I shot two new sequences of my plants that I thought would provide a good workout for each of the methods. All of the images were shot in AV mode at f/8, ISO 100, exposure time set by the camera. I didn't use any supplemental light in our kitchen, so exposures were several seconds and a tripod was mandatory. If anyone happens to read this and they are wondering "Why didn't you...?" then you can do your own testing. This is real world testing for me and I don't know or care what anyone else's needs are. Zerene and Helicon both have 31-day trial programs and DPP comes included with Canon cameras, go for it. The subjects are:
- An Echeveria Neon Breakers. I figured the overlapping leaves and crinkly edges would provide a real challenge. I shot 48 exposures with focus steps of 2.
- A little Powder Puff cactus with a flower. Photoshop chokes on the flower, I think because the background has higher contract. I shot 18 exposures with focus steps of 3.
The in-camera method is quickest because it produces an image immediately. But the output is only JPG. Also, there is no opportunity to select which images to use. All of the images in the stack are included. It did OK, not great, as there was extensive haloing around the Neon Breakers leaves. I usually use steps of 3 or 4, so I wasn't sure how many images to take using steps of only 2. It turns out that 48 was way too many. Fourteen was the right number, omitting the first image and the last 33.
DPP took 21 minutes to process the 48 images of Neon Breakers. It's a hands-off process and there are only a couple of parameters to fiddle with. I would call the result identical to the in-camera processing, which makes sense since they are both Canon. Reducing the number of images to the 14 which were actually needed cut the processing time to six minutes, and eliminated much of the haloing. There is none around the back leaves with a distant background, but still some around front leaves where they are close to other leaves. I'm actually impressed.
Photoshop required a few more mouse clicks, but only took four minutes to process the 14 source images. Alas, in my opinion the image was inferior to the DPP version. There was stronger haloing around the front leaves.
Next up was Zerene Stacker. It can't handle RAW CR3 files so I loaded the JPG files. It took three minutes to stack and save two images using different algorithms, PMax and DMap. According to the documentation, "PMAX excels in preserving fine detail and sharpness, while DMAP is ideal for smooth transitions and natural results." In my opinion, PMax produced a better image, but has weird artifacts scattered around the image in the form of short lines. Except for these artifacts, Zerene produced the best image. Condescending interweb twits claim these are dust spots that somehow become trails as the focus changes. That doesn't explain why these show up in Zerene and not in the other methods.
Finally, Helicon Focus took only three minutes to process and save the images using three different algorithms, Weighted Average, Depth Map and Pyramid. I loaded the CR3 images and the output was DNG, which would seem to give Helicon an advantage over Zerene. In my opinion, the Weighted Average was the best of the three, and in fact produced the best image of any of the methods. (And didn't have any streaks.)
I already had a winner in mind, but I wanted to see how the various programs handled the cactus flower. I determined that it was appropriate to use all 18 of the images that I shot so I just looked at the in-camera processing and did not use DPP. The flower is not in focus and that is all I will say about that. Same with Photoshop. Let's move on to the contenders.
Zerene got the flower into focus with the PMax algorithm, but the little streaks in the background remain. DMap was inferior for this subject.
In Helicon, all three algorithms got the flower into focus, mostly, with some haloing. Zerene PMax was better except for the streaks. Here is the Zerene image with the streaks cleaned up.
These are my conclusions:
- Zerene and Helicon produced the best results. DPP (surprisingly) was better than Photoshop.
- Helicon did the best on the Neon Breakers.
- Zerene did the best on the cactus flower, but those streaks are annoying.
- Both commercial programs have different levels of licensing. Zerene calls them Student, Personal, Prosumer and Professional for $39, $89, $189 and $289 for lifetime licenses (no yearly option). Helicon calls its levels Lite, Pro and Premium costing $30, $55 and $65 for one year; $115, $200 and $240 for lifetime. The cost is close enough that it is not a factor in which one I choose.
- Helicon can handle Canon CR3 files and can output DNG files. Zerene only deals with JPG and TIFF files. For the supposed benefit of using TIFF over JPG, that means an extra step for me and more files to clean up at the end.
- Helicon's interface is cleaner and appears to give more customization options which I did not fully explore. I just want something that works.
It's a difficult decision. Helicon nailed it on the Neon Breakers but didn't do great on the flower. There are a lot of settings in the program and maybe there is a combination that would handle it better. Zerene did the best job on the flower, BUT, those streaks. I could pick one, get both, or get neither. I have a month to decide. If I don't buy one, I will use DPP instead of Photoshop.
