First comparison Tufuse vs CZP

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

Moderators: ChrisR, Chris S., Pau, rjlittlefield

Post Reply
lauriek
Posts: 2402
Joined: Sun Nov 25, 2007 6:57 am
Location: South East UK
Contact:

First comparison Tufuse vs CZP

Post by lauriek »

I ran my first comparison stack between Tufuse and CZP last night, with the same input images (well I ran the alignment/stack in CZP then exported the aligned images and re-ran the stack on those in Tufuse)!

Nikon 10x objective on OM bellows at minimum extension on E-330. Macro twinflash and one sheet of diffuser foam.

Results follow:-

Full image resized for web, tufuse first:-

Image

And the CZP full image:-

Image

Both these full images have had minor sharpening, no other PP after the stacking...

Now a couple of crops, this first one shows that Tufuse definitely has the edge when it comes to overlapping detail, first the Tufuse crop:-
(although the contrast increase from Tufuse is unwelcome!)

Image

And the first CZP crop:-

Image

But this second crop shows that Tufuse's tendancy to increase contrast has some serious downsides, the front of the eye is much clearer in CZP. First Tufuse:-

Image

and the CZP second crop:-

Image

So I think it's still swings and roundabouts really - each handles some bits better than the other. I will try to combine the best bits from each output into one final decent image!! :)

ETA Further comments after perusing these images a little more:

The background is a lot noisier in Tufuse, but the output suffers really weird streaking in CZP, I think the noise although it looks bad will be easier to sort out than what CZP has done. I don't know how obvious this will appear to readers of this thread, probably depends on your monitor setup. I doubt the noise will be too obvious resized for web... CZP also made a bit of a mess of one of the antenna. Tufuse didn't do this perfectly but it did a much better job...

augusthouse
Posts: 1195
Joined: Sat Sep 16, 2006 1:39 am
Location: New South Wales Australia

Post by augusthouse »

Thanks for sharing that Laurie! Reading your post with great interest.

I downloaded Combine CZP when Rik first brought it to our attention but I haven't done much with it yet and I was wondering if anyone else had taken it for a ride.

Craig
To use a classic quote from 'Antz' - "I almost know exactly what I'm doing!"

rjlittlefield
Site Admin
Posts: 20770
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

Laurie,

When you ran CZP, did you use one of the new pyramid stacking commands, or did you run the old Do Stack macro or something like it?

I suspect the latter, since the defects in the CZP image look just like the ones produced by the depth-map algorithm used by Do Stack and friends.

To get the new pyramid algorithm, you have to use one of the new commands too.

--Rik

lauriek
Posts: 2402
Joined: Sun Nov 25, 2007 6:57 am
Location: South East UK
Contact:

Post by lauriek »

Nope I definitely ran a pyramid stack - I just re-ran it to be absolutely sure, the resulting image is almost identical (differences are probably due to me stacking a slightly different number of images, or alignment - see next paragraph!).

I /may/ not have run an alignment on these images, as I previously thought that align was run as part of the stacking macro (In fact I'm positive it was before, whether by default or by me modifying the macros!). I just realised in the last couple of days that the "pyramid do stack" macro at least, does no alignment.

Does this suggest that the software is not really doing a pyramid stack due to some bug, or that one of the other macro components (like fill gaps) is creating the mushy areas?

rjlittlefield
Site Admin
Posts: 20770
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

Laurie,

Oh, I think I understand now. I'm not in a place where I can re-test right now, but as I recall there are two different pyramid algorithms. The one named something like "maximum" is very similar to what TuFuse does. That's the one you want.

There is another algorithm with a name that might be something like "pyramid do stack" that uses a pyramid algorithm to calculate a depth map, complete with filling gaps, then uses that depth map in the old-fashioned manner, to select actual pixel values from individual frames. I have not seen any cases where the pyramid depth map is significantly better than the old do-stack.

My recollection is that the pyramid macros as released do not run alignment by themselves. I have no idea why Alan made that choice; it seems less than optimal to me.

--Rik

lauriek
Posts: 2402
Joined: Sun Nov 25, 2007 6:57 am
Location: South East UK
Contact:

Post by lauriek »

That's correct, there are actually three 'pyramid' macros, the one I've been using is just "Do pyramid stack" - there is also "Weighted average pyramid" and "Pyramid max contrast". I will try the max-contrast one post-haste!!

lauriek
Posts: 2402
Joined: Sun Nov 25, 2007 6:57 am
Location: South East UK
Contact:

Post by lauriek »

Okay here are the same shots stacked with the "Max contrast pyramid" in CZP...

Full shot, minor sharpening same as above:-

Image

First crop, tried to do the same bit, might be slightly mis-aligned from the ones above!

Image

Second crop:

Image

These do look much more like the Tufuse output! I need to peruse the images a little more to make my mind up!!

lauriek
Posts: 2402
Joined: Sun Nov 25, 2007 6:57 am
Location: South East UK
Contact:

Post by lauriek »

Hmm pyramid weighted average looks interesting, some detail has definitely been lost in some areas with this one but it's done an impressive job with that tricky overlapping hairy eye area...

Image

Note this is unsharpened, I did use some sharpening on the first set of images but didn't note exactly how much! This is all getting a bit out of hand, I think I might do another test from scratch now I know what's going on a bit better!!

Post Reply