Two questions about Zerene

soldevilla · Post by **soldevilla** » Fri Mar 15, 2013 11:11 am

Hello.

Zerene ask me if I want download the latest upgrade every time I open the soft. But this upgrade is not useful for me. I my engeenering work we say: If it works, don´t touch it!

¿How can I skip this message??

And another question. Alfter my "discovery" for to see the stereo pairs, I´m atemting to do my first pair.

But if I turn the camera 90º because the image is vertical, the automatic generated pair no have the correct "distorsion" for wachting. Need I to rotate first all images in my photo editor or Zerene have a tool for automate this?

rjlittlefield · Post by **rjlittlefield** » Fri Mar 15, 2013 11:33 am

You can turn off the automatic update check at Options > Preferences > Updates & Stats. If you want to check for updates later, just visit http://zerenesystems.com/cms/stacker/softwaredownloads. The modification history is always posted along with new versions, so you can also follow that link on the downloads page to see if there is anything you care about.

For image rotation, see Options > Preferences > Preprocessing. Put a checkmark on "Image Pre-rotation", and select one of the 1/4-turn options.

--Rik

soldevilla · Post by **soldevilla** » Fri Mar 15, 2013 2:05 pm

Rik, Thanks and thanks!

I have thought that the noise pattern would be different with the horizontal and vertical stacking, so I stacked the two possibilities and then I have superimposed the two images in my photo editing software. Noise reduction is important, and I guess it will be much better averaging the four orientations of stacked every 90 º.

Here mi first stereo pair. A stibnite from Escarlatti mine, Tarna, Leon, Spain.

rjlittlefield · Post by **rjlittlefield** » Fri Mar 15, 2013 3:17 pm

soldevilla wrote:Noise reduction is important, and I guess it will be much better averaging the four orientations of stacked every 90 º.

I look forward to hearing how that works out. The 90 degree rotations just change pixel ordering before the images are processed. No doubt there is some interaction between pixel ordering and the exact operation of the stacking algorithms. But I would expect differences in the results to be small compared to noise in the input that essentially just gets copied through. If that's correct, then you'll be averaging four copies of almost the same noisy values and getting not much improvement. On the other hand, if you do see big differences in noise in the average, then I will be interested to hear more about exactly how you're doing the processing.

One thing to be cautious about: if noise is important, then be sure to stack the images at full size. Do not select Options > Preferences > Preprocessing > Image Pre-sizing with any fraction below 100%. If you want a smaller image in the end, then stack at full size and resize later with careful anti-aliasing.

The stereo image looks great, by the way. I really like to look at those!

--Rik

soldevilla · Post by **soldevilla** » Fri Mar 15, 2013 5:02 pm

Hello Rik. Here is a crop of 100% image, the two single images with the same previous overprocess. First is a single image and the second one is an average of two images stacked rotated 90º. Is easy so see the difference. Interesting is that the images are no exact, you can see it if superimpose them and do a blinking. As we talk in another thread, we have a uncertainty of one pixel...

When I find some free time, I will test the average with 4 images.

rjlittlefield · Post by **rjlittlefield** » Fri Mar 15, 2013 6:07 pm

Thanks for the test. I'll do some investigation with my own test stacks to see if I can get a better handle on what's going on.

--Rik

rjlittlefield · Post by **rjlittlefield** » Fri Mar 15, 2013 9:14 pm

As far as I can tell, you got lucky in producing an improved result.

What's happening under the covers is that the alignment process is sensitive to the image orientation and even has a bit of randomness in it. Over the depth of the stack, small errors can accumulate to be noticeable differences whenever the alignment is repeated, especially in a different orientation.

In the case you show here, the differences turned out to be enough to shift the noise pattern so that averaging helped, while not being so much that you got obvious blurring or echoing.

Just now I ran a test with the Zerene fruitfly and got very different results. Averaging the +-90º versions introduced obvious softening at most places in the image while having no visible effect at others. The 0 and 180º versions had a similar relationship to each other. But the 0º/180º versions and +-90º versions had clearly different geometries, with the subject appearing noticeably more "bug-eyed" in one set than the other. I would not have predicted this result, but in retrospect it makes sense because that stack is known to be very sensitive to alignment issues due to having only a little bit of the frame in focus most of the time.

It's probably worth noting that in regular use, Zerene simply re-uses whatever alignment it figured out the first time through the stack, so for example PMax, DMap, and any Stack Selected outputs all line up exactly with each other. Discrepancies arise only when the alignment has to be recomputed, as it does when the images are rotated.

--Rik

Craig Gerard · Post by **Craig Gerard** » Sun Mar 17, 2013 3:32 am

soldevilla wrote: Here mi first stereo pair. A stibnite from Escarlatti mine, Tarna, Leon, Spain.

Excellent stereo. 'Visually Immersive' would by no means be an understatement when describing the overall viewing experience.

Craig

soldevilla · Post by **soldevilla** » Sun Mar 17, 2013 4:07 am

Thanks. Now I have my table full of prism... I´m mounting a device for to see easy the stereo pairs. It is working, it only need the "aestethic touch"

soldevilla · Post by **soldevilla** » Sun Mar 17, 2013 8:00 am

A question to the members that are the luck of to see the stereos... This my first stereo pair is correct? after some test here with prisms, I have discovered that the images are inverted, the right picture must go to the left side...

rjlittlefield · Post by **rjlittlefield** » Sun Mar 17, 2013 8:41 am

The pair is correct for "crossed eye" viewing, which is the usual way of doing it. As you say, the right-eye picture must go on the left side in this format.

--Rik

soldevilla · Post by **soldevilla** » Sun Mar 17, 2013 9:29 am

I understand... My brico is a classic stereo viewer, right picture in right eye. The prisms are only to see bigger images, because the non optics viewer work only with pictures around 60mm wide, similar to the interpupilar distance.
In this forum is better to publish direct eye or cross eye stereo pairs?

rjlittlefield · Post by **rjlittlefield** » Sun Mar 17, 2013 9:42 am

It is better to publish cross eye. That is the only format that can be seen without a viewer.

If someone who has a viewer wants to swap the sides, that is easy to do using StereoPhoto Maker software (on Windows, anyway).

--Rik