Focus Stack Feedback

[Asha]

I finally had a chance to put the Affinity focus stacking tool to work today with RC1. I did a fairly modest 16-layer stack, and processed both from RAW files and from 16-bit TIFF files. Below are the results of this test, compared with output from an "other" dedicated focus stacking software.

 

For the most part, I find the processing time slightly faster for Affinity than the "other" software. In both softwares, processing from RAW was slower than from 16-bit TIFF, even though the TIFF files are several orders of magnitude larger than the RAW. Final resolution and quality of detail was roughly the same on both, however, Affinity had more artifacts. This is balanced by the fact that Affinity has high ease of editing from any one of the source images.

 

What I missed most in the Affinity tool is the capability to customize the stacking process. In the "other" software, I can select one of three algorithms, whereas Affinity has only one default algorithm. For this test, I chose depth map in the "other" software, because I think that is what Affinity is using, and I wanted a consistent comparison. In the "other" software, I can also adjust the algorithm parameters, which sometimes helps the artifacting--Affinity does not have this feature. I also did not see a way to select the RAW engine for focus stacking in Affinity, but that is something I would like to have since the different RAW engines may produce more pleasing results. Generally, the RAW processing seemed more streamlined in Affinity. The "other" software opened Adobe DNG converter for every single slice of my stack, which is somewhat laughable, however, the unretouched result was the best from RAW processing through the "other" software. One more thing that I have seen in dedicated software is the ability to process from top-down or bottom-up of the stack. This is a very helpful option since (again) it can reduce artifacting. This would be a nice improvement to the Affinity tool.

 

While my stack was rather benign, I've seen some crazy high-magnification macro stacks with several times the number of files I used. Those might not do as well with Affinity, due to the lack of adjustability. Having said that, Affinity is the most accessible and affordable software I've seen with this advanced capability, and I have nothing but the highest praise for the Serif team.

 

To summarize, here is my wish list for Affinity's new focus stacking capability based on this test:

--Ability to choose from different algorithms

--Ability to choose algorithm parameters

--Ability to choose the RAW engine that is invoked when using RAW source files

--Ability to choose the order of stack processing (top-bottom, bottom-top)

 

File legend:

Macro Roses from RAW--Affinity processed from Panasonic RAW format files

Macro Roses from RAW HF wDMap--"other" software processed from Panasonic RAW format files

Macro Roses from Tiff16--Affinity processed from 16-bit TIFF files

Macro Roses from Tiff16 HF wDMap--"other" software processed from 16-bit TIFF files

[TheLostVertex]

Some comments. First, I assume the "other software" is helicon focus. 

Second, as a personal rant, I do not understand the point of focus stacking raw files. While both softwares allow you to do it, it makes very little sense to take away the executive decision making that raw gives the user after the images are made, replacing it with what ever raw defaults the software will use. Maybe in the future AP will allow creation of raw developing presets you can apply when doing stacking/panoramas. Till then I think it is conceptually a bad idea. Best to convert your raw files the way you like then proceed, with any focus stacking or panorama software. You can chance the default raw engine used for AP in the preferences, but doing so requires relaunching the app i believe. 

 

I can select one of three algorithms, whereas Affinity has only one default algorithm. For this test, I chose depth map in the "other" software, because I think that is what Affinity is using, and I wanted a consistent comparison. 

That is not the case. What Affinity Photo does is more similar to Helicon Method C, or Zerene Pmax. The difference is Affinity Photo appears to also create what is either a depth map or a blend map(their video tutorials say it is a depth map, but I still feel a bit iffy about it, I might investigate further later on). The map generation is what I assume maps the images out of AP appear less noisy and with less contrast changes that what I would see in Pmax or Method C.

 

The artifacts that you see in the AP images, you will also see when using Pmax or Method C. They are very characteristic of pyramid style image stacking. You can see Helicons help page for more information about its stacking methods, http://www.heliconsoft.com/helicon-focus-main-parameters/Their help page also shows the different types of artifacts you might see with each method. 

 

For a pyramid style algorithm, i have found AP to be quite good so far. I certainly would like to see a depthmap based algorithm added as well though.

 

 

While my stack was rather benign, I've seen some crazy high-magnification macro stacks with several times the number of files I used. Those might not do as well with Affinity, due to the lack of adjustability. Having said that, Affinity is the most accessible and affordable software I've seen with this advanced capability, and I have nothing but the highest praise for the Serif team.

 

I have tested through out the beta with 20-350 image stacks. Currently the performance is very close to what I get with Zerene with "overlap image I/O" enabled. To see the history of my tests, look here https://forum.affinity.serif.com/index.php?/topic/24336-focus-merge-feedback/You can see through out the beta there has been a large performance improvement. 

 

For instance, from that thread, a 283 image stack Zerene Pmax: Completed in 0:28:00 and a much earlier beta, Affinity Photo Beta 5: Completed in 0:34:30. A small speed difference, but not anything huge. I have not had a chance to rerun that stack with the current release candidate, but I think you get the idea. 

 

Last, how did you change focus points? Did you move the camera, or focus with the lens? The main artifacts I see around the rose petals are most likely caused by a change in perspective or camera movement. 

 

Hope this helps. 

[aslusers]

I am barely following the conversation above, as I am new to the technology of focus stacking.      However, I hope my experience/image might still be helpful.

 

I stacked 12 raw shots using the new stacking feature and am very pleased with the resulting image.     The processing took about 4 minutes (just guessing - didn't time it).

 

I set my fuji xt-10 up with an old pentax 50 1.4 and about 8 mm of extension tubes.  Lens was set at f/8.  camera was on a tripod and I manually adjusted the focus between shots.

 

After the stack was finished, I brought the final image into the develop persona and applied my recipe for Fuji sharpening, adjusted a bit of exposure then back in the photo persona I punched up contrast a bit and unpainted out some dirt on the gummy bears.

 

For not paying a penny more, I couldn't be happier to have this additional function!

 

 

Question - TheLostVertex - do you process each of your raw images before stacking them?  (the new batch processing capability will surely help with that!)

 

 

EDIT - trying to remember how to attach an image.... working on it....

Edit - well, i have a nice image to share but can't figure out how to load it into this message.  If someone can explain how to upload a photo, I'll do it.  

[Asha]

TheLostVortex,

 

Thanks so much for linking to your thread. I wish I had seen that before I started a new one--perhaps the mods can merge the two threads. I also appreciate that you've been putting Affinity through some major exercises with respect to focus stacking. I only work at 1x, so having somebody testing with higher magnification and bigger stacks is super.

 

Thanks also for pointing out that Affinity uses the pyramid algorithm. Like you, I had watched the video and heard the narrator speak of "density maps" as the stack processed, so I assumed "Dmap". If I have time, I'll reprocess my stack with the pyramid algorithm in my software and post the results.

 

Regarding my stack, it is pretty old, probably one of the first I did. In that time frame, I was doing both focus-by-ring and focus-by-rail with a Panasonic m43 camera, and I don't really remember what I did with this particular setup. However, I think I see what you're getting at--if I had been smarter in building my stack, I could have minimized artifacts regardless of what algorithm is being used in the stack processing.

 

Regarding RAW--well, I just see it as an advantage to process in RAW for a RAW workflow, so it is a personal preference, especially if I want something quick. Plus, RAW files are much more compact than 16-bit TIFF. I hope some day to have a composition that is good enough to warrant committing to a batch RAW development and massive hard drive space to store the TIFFs, lol.

[TheLostVertex]

Question - TheLostVertex - do you process each of your raw images before stacking them?  (the new batch processing capability will surely help with that!)

I do. Currently my workflow is:

Take images

import into lightroom

mark off each group of images to be stacked with a color tag

edit one image with a central focus that is representative of the other images in the stack and then copy and paste those settings to the rest of the images in the stack

select and export each set of images as a sequence into its own folder

 

It may sound like a lot of extra hassle, but the only real time consuming part is marking off which images will be in a stack. Which I have to do anyways.

 

As far as I know, there is no way to adjust the develop settings when doing a batch job in affinity photo. So it is not really suitable for processing a lot of raw images at a time currently. 

 

Regarding RAW--well, I just see it as an advantage to process in RAW for a RAW workflow, so it is a personal preference, especially if I want something quick. Plus, RAW files are much more compact than 16-bit TIFF. I hope some day to have a composition that is good enough to warrant committing to a batch RAW development and massive hard drive space to store the TIFFs, lol.

 

Thats understandable. Hopefully in a future update affinity photo allows applying some raw develop presets when processing raw in operations like focus merge, panoramas, and batch processing. 

[ruprecht]

I also have a problem with the algorithm, perhaps an option to edit settings or mask parts of an image as "keep this" would solve it. I attach the frontmost layer with some text in focus (blurred in subsequent layers focused further away) and the result from RC4. I would expect that since one layer has text in focus (first picture), that will be kept instead of blurred versions, not combined with them into text (slightly deformed and not smooth) with halo (second picture).

 

https://1drv.ms/i/s!Al7e-lTZyLzbgeYcLtI3CiCxYVfxlw

 

https://1drv.ms/i/s!Al7e-lTZyLzbgeYdQ7g369EiMbzzQA