Zerene: Difference between DMap & PMax with scaling OFF

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

Moderators: rjlittlefield, ChrisR, Chris S., Pau

Lou Jost
Posts: 5944
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Zerene: Difference between DMap & PMax with scaling OFF

Post by Lou Jost »

I was experimenting with telecentric Nikon MM 10x measuring objective pushed down to lower magnifications than its designed m. Stacking was in Zerene. I knew it would no longer be telecentric when pushed down to a wider FOV, but still I wanted to see what would happen if I turned off scaling in Zerene. Zerene's reported scaling factor near the end of these stacks was around 1.01 so definitely not telecentric. With scaling turned on, there was almost no difference between the resolutions of the images produced by the two stacking options, PMax and DMap, and both showed the fine ridges on the butterfly scales. When I turned scaling off, I still got quite good resolution of the ridges using PMax, but in DMap I lost most of them. This was a curious result which I do not understand.

You may wonder why anyone would turn off scaling when the lens is not telecentric. I would want to do this when I need to stitch several stacks together, to avoid artifacts. I am sure this produces its own artifacts, but these might be manageable.

Here are the two images with scaling turned off, blown up to 200%; on the left is PMax and on the right is DMap. Both are made from the same stack. Pardon the purple fringes, the MMs have a lot of that. You can see that the PMax image shows good ridges on the scales, but these are largely lost in the DMap image.

Image

rjlittlefield
Site Admin
Posts: 23561
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

This was a curious result which I do not understand.
Yeah, me either.

The characterization "difference...with scaling OFF" is one that I have not seen before. I strongly suspect that what you're seeing is due to some other cause that is just coincidentally correlated with scaling.

To briefly review, DMap works by selecting pixels from individual as-adjusted source frames, as indicated by the depth map, or forming weighted averages of two adjacent source frames where the depth map is transitioning between source frames.

When DMap misses detail that is present in some source image, it must be that the depth map did not select that image. Common causes include estimation radius and/or smoothing radius too big, and/or contrast selection slider too far to the right (too much "black in preview").

If this stack were sitting on my computer, the first thing I would do is to compare the as-adjusted source images against the DMap output to try identifying how the depth map had gone wrong.

It is best to do this in retouching mode. That's because when not retouching, the source image panel is likely to be showing a low quality "screen preview" image rather than the full quality source image. In retouching mode you also have the S key to flash between source and output to facilitate comparisons. To get faster transitions between source frames, it helps to select one of the advanced brushes such as Pixels.

From what I can see here, it looks like maybe the depth map is properly tracking the high contrast edges of the scales, but not the lower contrast bases and interior areas where the fine ridges are.

--Rik
Last edited by rjlittlefield on Fri Oct 06, 2017 2:19 pm, edited 1 time in total.

mawyatt
Posts: 2497
Joined: Thu Aug 22, 2013 6:54 pm
Location: Clearwater, Florida

Post by mawyatt »

Lou,

Very interesting, I would not have thought this would have so much effect on Dmap and not Pmap. Certainly a good reason to keep scaling ON in most cases!!

Best,

Mike

Lou Jost
Posts: 5944
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

I strongly suspect that what you're seeing is due to some other cause that is just coincidentally correlated with scaling.
But with the very same source images, there is no difference between DMap and PMax when scaling is left on. I'm not sure what I should look for in the adjusted images....I'll re-run the stacks now.

Lou Jost
Posts: 5944
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

Aha, I see that I took four shots at each camera position, and I may have changed the "use every n-th shot" setting between stacks. Will run everything again. Maybe that's the explanation.

Lou Jost
Posts: 5944
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

OK I've redone the four stacks, using every fourth frame.

The result is just as I presented above. When Scaling is checked, there is no difference in resolution between DMap and PMAx outputs. When Scaling is unchecked, there's a huge difference in resolution between the two outputs.

Also, comparing PMax outputs alone between "checked" and "unchecked", there very little resolution difference.

All these are done on the exact same set of source images.

rjlittlefield
Site Admin
Posts: 23561
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

Are you able to upload the stacks so that I can take a look at them?

--Rik

Lou Jost
Posts: 5944
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

Rik, do you mean the final results, or the whole set of source images? I can send you the final stacked images but the bandwidth here is not great enough to send you the source images.

rjlittlefield
Site Admin
Posts: 23561
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

I was thinking of the whole set of sources. I don't know how to investigate the relationship between sources and outputs unless I can access the sources.

Can you isolate the problem to a smaller set of images, or cropped to a smaller ROI?

--Rik

Lou Jost
Posts: 5944
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

I can see if the set of images can be reduced to the few which contribute to the scales I am looking at. But this may have to wait, I am about to travel to the US and have not finished my "real" work...or maybe I can copy them and send them when I am in the US. Where to send them? It is about 3 Gb.

rjlittlefield
Site Admin
Posts: 23561
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

Most people would upload to DropBox or send via wesendit.com, but I'm pretty sure that 3 GB at one time would exceed their limits.

If you want to send DVD, the best place would be the address shown at https://zerenesystems.com/cms/about_us .

For reducing the frame count, try first just restricting the stack to be the images contributing sharpness to the crop that you've shown.

--Rik

Post Reply Previous topicNext topic