PTA 6.2 Tufuse blending thru Masked Areas

Discussion forum for Tawbaware's PTAssembler software, Helmut Dersch's Panorama Tools software and any other photography related software

Moderator: spamkiller

Post Reply
seigell
Posts: 90
Joined: Tue Mar 21, 2006 10:26 pm

PTA 6.2 Tufuse blending thru Masked Areas

Post by seigell » Fri Oct 18, 2013 7:07 pm

I found some odd behavior for Tufuse in PTA 6.2 x64 related to Masks.

I have been trying to "rescue" some Fall Color Panos from problems due to strong motion in the foreground bushes. There were 20-30mph winds with stronger gusts and spits of near-freezing rain with an early Winter Storm flying over the mountain peaks - nothing was going to stop that motion. But the impending snow made the area unapproachable for the next two days, and the clay-ish roads difficult for even longer. So, motion is what I got...

Given the difficult light due to the heavy but rapidly changing clouds, I was relying on Exposure Bracketing to capture the sceen. I was expecting Tufuse to perform Exposure Blending (as I've come to do for a number of Pano Sets). But I believe this is likely the first time that I've used Exposure Blending with Masks... I masked the foreground bush with eggregious motion-blur in the two longer exposures, intending to "live with" the darker/longer-exposure version of the bush coming through from the unmasked 3rd image. However, I was disappointed to find that Tufuse seems to be applying some Exposure Blending even in those Masked areas - resulting in a "pastel-like washed lighter" image where the masked darker bush should be.
I changed Blenders from Tufuse to Enfuse v4.1, and retried... I got results more in keeping with my expectation, but not as sharply imaged as I've come to expect via Tufuse.
I checked that the Exposure/Whitebalance/Vignette settings were all 0% for all images.

I've posted the pertinent portion of my Pano Efforts into a Dropbox, and shared it here: PTAssembler
(the Tufuse version images have the "a" in the filename, while the Enfuse version have an "e" instead.)

Please review and advise where I've gone wrong.

maxlyons
Posts: 3649
Joined: Fri Jun 20, 2003 8:55 pm
Location: USA
Contact:

Re: PTA 6.2 Tufuse blending thru Masked Areas

Post by maxlyons » Sun Oct 20, 2013 5:30 pm

This is a bit of torture test for any blender...I think Enblend is doing a better job than Tufuse here, but there really isn't enough data in the masked area of the image to produce a result where the illumination looks balanced with the rest of the scene. I don't think either program will produce a convincing result in this scenario, I'm afraid. You may have to resort to some manual patching after the result has been created in this case.

Max

seigell
Posts: 90
Joined: Tue Mar 21, 2006 10:26 pm

Re: PTA 6.2 Tufuse blending thru Masked Areas

Post by seigell » Mon Oct 21, 2013 11:45 am

Hmmm... Maybe my interpretation of the "Blender" actions is off.
Doesn't the Exposure Blending function take the Luminance values at each Pixel Location and statistically calculate a resulting value for that same Pixel ?? My perception is that if it is involving "surrounding pixels" in that calculation, that the ultimate outcome would be a rather uniform Grey for the entire image rather than the "best" Exposure Value from the source pixels. If the Masked-out Pixels are truly ignored for purposes of this Exposure Blending process, it would mean instead of all 3 contributing source values only those of Unmasked-at-the-current-Pixel Images would be included. No ?? Or is there a built-in Feathering function that I've failed to consider ??

Ultimately, I can resort to Manually Blending in a "Patch" from the intended Unmasked Image(s) using CS3. I'd just rather thought I had a sufficient way to do so within PTA...

Thanks

maxlyons
Posts: 3649
Joined: Fri Jun 20, 2003 8:55 pm
Location: USA
Contact:

Re: PTA 6.2 Tufuse blending thru Masked Areas

Post by maxlyons » Mon Oct 21, 2013 10:05 pm

seigell wrote:Hmmm... Maybe my interpretation of the "Blender" actions is off.
Doesn't the Exposure Blending function take the Luminance values at each Pixel Location and statistically calculate a resulting value for that same Pixel ??
No.
My perception is that if it is involving "surrounding pixels" in that calculation, that the ultimate outcome would be a rather uniform Grey for the entire image
No. The value of each pixel in the image is a mathematical combination of many other pixels in the input images (the more blending levels, and the more images, the more pixels are included). With the size of images that come out of modern cameras, the value of any single pixel in the blended image is calculated using the values of hundreds or thousands of other pixels in the input images.

If you want to wade through the math, then this is the paper to read.

Max

Post Reply

Who is online

Users browsing this forum: No registered users and 2 guests