Image comparison of ZS and PS CS5 on deep high mag stacks

A forum to ask questions, post setups, and generally discuss anything having to do with photomacrography and photomicroscopy.

Moderators: rjlittlefield, ChrisR, Chris S., Pau

ChrisR
Site Admin
Posts: 8668
Joined: Sat Mar 14, 2009 3:58 am
Location: Near London, UK

Post by ChrisR »

Errm, it being an experiment, it needs a "control".

A person has a set of images, whatever program he/she's used to , though not all equally, and the same brain.
Several people could share the same images, but even with the same program, abilities and opinions would vary.
The opinion may change with a different type of image. Folk on this forum may well tend to have a similar type of subject.
I was reading today of a pro photographer, who indicated that to get everything between four feet and infinity in focus, he routinely stacks six images. He's a pro stacker, but not a big macro man. (He uses Helicon in preference to Photoshop). He may not have even heard of Zerene.

We may find that a particular individual's computer soon falls over with photoshop. He might have little experince of CombineZm, and has never used Hellicon - with no desire to pay for it because people whose work he follows didn't get on as well with it.

If you listen to a person with a particular type of photo about what he's tried and what he finds best, that has to be a positive vote for the winner, but what mark do you award the one he never tried? That's a notorious statistical problem.

"7 out of 10 cats said they preferred Kittyfish."

Great,
but

It killed the other three
another ten wouldn't try it because of the smell
you've only tested talking cats.

So I think we're back to a "scientific theory" paradigm.
The maths and the current understanding say it should be right, though not many understand the maths really well.
The not-so-good fails, in ways predictable according to that maths.
Fair chance the theory is right, then, at least basically.

A rather crucial matter has to be addressed when someone claims to have produced "scientific" evidence which erodes the edges of the theory - is there another factor involved, or was it just bad science?
So if someone claims to have seen a neutrino going faster than light, but he can't tell you how many types of neutrino there are (3), you'd tend not to shake your faith. If he doesn't know about probablility density functions, then he can't measure it. It's only when you have to start analysing the application of something relatively deep, like that, that there's even a question to ask of the original theory.

Zerene seems to be best for the stuff I play with, which isn't all macro by any means. Photoshop works but has serious flaws. Helicon isn't rated as better (quite) by anyone credible. CombineZm seems to be less clever. One or other might show advantages in some areas though I fear it's likely that collateral nasties would overwhelm them. I have not, in any case, seen any such advantages demonstated, by me or anyone else anyone who's heard of a neutrino.

Pau
Site Admin
Posts: 6053
Joined: Wed Jan 20, 2010 8:57 am
Location: Valencia, Spain

Post by Pau »

ChrisR wrote: ...you've only tested talking cats.
Brillant!
Pau

rjlittlefield
Site Admin
Posts: 23564
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

Another member recently suggested that I re-run the first test, but "try allowing geometric distortion". That seemed like an interesting test, so I let my machine work on it while I slept last night.

Here are the results:

Image

This is very promising in one respect -- CS5 did manage to get all the images pretty well aligned in this case. I did not expect that allowing geometric distortion would make any difference, since these images don't have any geometric distortion, or perhaps more formally, whatever distortion they do have is equal for all images. Nonetheless, giving Photoshop the extra freedom to torque around the geometry did let it get the alignment pretty good.

There are, however, some downsides. The computation time increased from 95 minutes to over 4 hours, and while it's not immediately apparent with this subject, CS5 actually introduced quite a bit of barrel distortion in the images. This is more apparent in the following animation, which flashes between the ZS and CS5 results, aligned as best I could.

Image

The final observation is that while CS5 did an admirable job of seamlessly blending between images, it did that at the cost of leaving out quite a bit of detail. This is typical behavior, seen over a wide range of stack types.

--Rik

Rylee Isitt
Posts: 476
Joined: Fri Apr 13, 2012 3:54 pm
Location: Canada
Contact:

Post by Rylee Isitt »

Wow... over 4 hours of processing and CS5 still couldn't figure things out quite that right. This is one area that their engineers definitely need to work on.

There really aren't too many settings to fine-tune in CS5, so there's not much else you do at this point aside from manually tweak positions of images and paint into the masks. Which is not worth it if ZS is giving you good output right out of the box.

I haven't yet needed to do a stack with more than 20 images, as my voyage into macrophotography beyond 1:1 is very recent development and I'm still working at 3x for the most part (where I generally like the shallow DOF).

I'll try ZS when I find myself in this boat, though. It looks like it's doing a spectacular job.

Post Reply Previous topicNext topic