Image-stitching and virtual tour solutions My account Updates
It is currently Wed Aug 20, 2014 1:46 pm

All times are UTC + 1 hour




Post new topic Reply to topic  [ 11 posts ] 
Author Message
PostPosted: Wed Nov 03, 2010 10:18 pm 
Offline
Member

Joined: Thu May 03, 2007 9:43 pm
Posts: 198
Location: the Netherlands
Hi, I've been testing APG v2.5b2 extensively since it became available. I really like the speed, especially of the new rendering engine. However, I have a feeling something is wrong wrong with the multiblend+remove ghosts option of APG.

I'm getting strange coloration and loss of contrast in part of the images where previous versions rendered perfectly. One example of this is when I render my panorama using linear blending colors are good but when using multiblend+remove-ghosts the contrast decreases in some areas and colors change extensively (no color-correction in use), in this example of the nadir-patch.

Also, when I change the pitch -90º (looking at the nadir) and render again without changing any other setting, the color is rendered correctly. It's like the color and brightness of the water below heavily influences the color of the nadir patch witch overlaps the water but that part is masked.

Examples:
first: linear blending, good colors but hard transitions.
second: multiblend+remove-ghosts, loss of contrast, odd color on nadir.
third: pitch -90, looking at nadir, color seems to be better, no changes except pitch -90º.
fourth: normal render, pitch -90º render example, check out the color and contrast differences...









_________________
Ronald

Fujifilm S5 Pro, Nikon D300, NodalNinja 3mk2, Sigma 10mm f/2.8 fisheye HSM,
Sigma 30mm f/1.4 EX HSM, Sigma 24-70 f/2.8 EX HSM, Sigma 70-200mm f/2.8 EX HSM


Last edited by Ronald Tichelaar on Wed Nov 03, 2010 10:39 pm, edited 1 time in total.

Top
 Profile  
 
 Post subject:
PostPosted: Wed Nov 03, 2010 10:38 pm 
Offline
Member
User avatar

Joined: Fri May 05, 2006 8:16 am
Posts: 1228
Location: Bulgaria
It may be due to the late hour or the amount of mulled wine, but I fail to see anything fundamentally different (let alone - wrong) with those images.

_________________
Some of my panoramas, posted in the Autopano Pro flickr group.


Top
 Profile  
 
 Post subject:
PostPosted: Wed Nov 03, 2010 10:41 pm 
Offline
Member

Joined: Thu May 03, 2007 9:43 pm
Posts: 198
Location: the Netherlands
Hi Bo, I've added another example illustrating the difference better. By only changed the pitch to -90º the complete nadir rendering changed (for the better)

_________________
Ronald

Fujifilm S5 Pro, Nikon D300, NodalNinja 3mk2, Sigma 10mm f/2.8 fisheye HSM,
Sigma 30mm f/1.4 EX HSM, Sigma 24-70 f/2.8 EX HSM, Sigma 70-200mm f/2.8 EX HSM


Last edited by Ronald Tichelaar on Wed Nov 03, 2010 10:42 pm, edited 1 time in total.

Top
 Profile  
 
 Post subject:
PostPosted: Thu Nov 04, 2010 8:47 am 
Offline
Administrator
User avatar

Joined: Mon Nov 14, 2005 4:56 pm
Posts: 5892
Location: Francin, France
Let's open an issue on that one : 406.


Top
 Profile  
 
 Post subject:
PostPosted: Thu Nov 04, 2010 2:44 pm 
Offline
Administrator
User avatar

Joined: Mon Nov 14, 2005 4:56 pm
Posts: 5892
Location: Francin, France
Seems really closely related the multiband pyramid level.
Have a try with -4 instead of -2 and it should be okay. See the explaination here : http://www.autopano.net/forum/p72046-today-14-43-12#p72046
( We'll change the default value for RC )


Top
 Profile  
 
 Post subject:
PostPosted: Mon Nov 08, 2010 9:57 am 
Offline
Member

Joined: Thu May 03, 2007 9:43 pm
Posts: 198
Location: the Netherlands
Setting the multiblend level to -4 has improved the image but there is still a colorcast in the center of the image. Setting the level to -6 removes the colorcast but creates a defect somewhere else in the pano...

The images below are rendered 'normal' orientation and have the pitch changed afterwards in PS + flexify to improve visibility of the defects.



_________________
Ronald

Fujifilm S5 Pro, Nikon D300, NodalNinja 3mk2, Sigma 10mm f/2.8 fisheye HSM,
Sigma 30mm f/1.4 EX HSM, Sigma 24-70 f/2.8 EX HSM, Sigma 70-200mm f/2.8 EX HSM


Last edited by Ronald Tichelaar on Mon Nov 08, 2010 10:33 am, edited 1 time in total.

Top
 Profile  
 
 Post subject:
PostPosted: Mon Nov 08, 2010 10:49 am 
Offline
Member

Joined: Thu May 03, 2007 9:43 pm
Posts: 198
Location: the Netherlands
Okey, I've also changes the pitch in AGP and rendered the same sequence again, note that defects in the images compared to the previous sequence have changed... only by changing the pitch in APG.

Multiband level 4 gives is the best results but it's not perfect yet.

I've never had these problems using smartblend, at least not to this extent. Would it by posible to include smartblend as a rendering option in APG2.5? Smartblending seems to be something else as removing ghosts. Ghosts are often detailed objects in a picture and are perfectly removed, but blending the same objects with different lighting conditions (changing shadows, lensflares, moving clouds) are better blended with smartblend until now it seems...



_________________
Ronald

Fujifilm S5 Pro, Nikon D300, NodalNinja 3mk2, Sigma 10mm f/2.8 fisheye HSM,
Sigma 30mm f/1.4 EX HSM, Sigma 24-70 f/2.8 EX HSM, Sigma 70-200mm f/2.8 EX HSM


Top
 Profile  
 
 Post subject:
PostPosted: Mon Nov 08, 2010 11:04 am 
Offline
Member

Joined: Thu May 03, 2007 9:43 pm
Posts: 198
Location: the Netherlands
One last question Alexandre...

I understand that APG2.5 is still beta, I'm only posting my experiences stitching 360º x 180º panos. I'm still relying an APG2.09 for my final rendering, ...unfortunately it doesn't load APG2.5 files.

Does the multiband anti-ghost procedure see the image as a 3D sphere or as a flat image? Since the zenith and nadir are stretched at the top and bottom of the flat rendering, details are also stretched and anti-ghost probably blends these different. This would explain some of the differences in the sequences I've posted above.

again some sequences to illustrate the difference at varying levels and pitches...

first, level -2, PS pitch -90 <> APG pitch -90
second, level -4, PS pitch -90 <> APG pitch -90
third, level -6, PS pitch -90 <> APG pitch -90







_________________
Ronald

Fujifilm S5 Pro, Nikon D300, NodalNinja 3mk2, Sigma 10mm f/2.8 fisheye HSM,
Sigma 30mm f/1.4 EX HSM, Sigma 24-70 f/2.8 EX HSM, Sigma 70-200mm f/2.8 EX HSM


Last edited by Ronald Tichelaar on Mon Nov 08, 2010 11:19 am, edited 1 time in total.

Top
 Profile  
 
 Post subject:
PostPosted: Mon Nov 08, 2010 12:22 pm 
Offline
Member
User avatar

Joined: Fri May 05, 2006 8:16 am
Posts: 1228
Location: Bulgaria
The original Smartblend was rewritten from scratch and as far as I understand the current logic has little to do with the old one. So I don't think there will be a way to use the "vingage" Smartblend in 2.5.

As a side note, testing 2.5 I haven't encountered such problems with blending. Probably because don't create 180- or 360-degree panoramas, but mainly "flat" landscapes. It's altogether different matter that I really have to strain to see the difference in the demo images in this thread and that probably means I'm a lot less demanding in that area.

_________________
Some of my panoramas, posted in the Autopano Pro flickr group.


Top
 Profile  
 
 Post subject:
PostPosted: Mon Nov 08, 2010 1:04 pm 
Offline
Member

Joined: Thu May 03, 2007 9:43 pm
Posts: 198
Location: the Netherlands
Hi bo,

I agree with you that stitching 'flat' panos is less demanding then stitching a spherical. The problem with these sphericals is that they often are viewed at 100% on screen (unless they're gigapixel spheres) so everything has to be perfect. shooting with a fisheye or extreme wide-angle lenses often causes images not to be perfectly aligned at every location in the pano. Blending these small errors out puts extra strain onto the blender. Also, shooting 360º you always have to shoot straight into the sun at least ones, causing lens flares en loss of contrast, resulting in images with different brightness in the skies and shadows. Smartblend handled these differences very good, multiband antighost has less margin for these variations so it seems.

_________________
Ronald

Fujifilm S5 Pro, Nikon D300, NodalNinja 3mk2, Sigma 10mm f/2.8 fisheye HSM,
Sigma 30mm f/1.4 EX HSM, Sigma 24-70 f/2.8 EX HSM, Sigma 70-200mm f/2.8 EX HSM


Top
 Profile  
 
 Post subject:
PostPosted: Mon Nov 08, 2010 1:18 pm 
Offline
Member

Joined: Thu May 03, 2007 9:43 pm
Posts: 198
Location: the Netherlands
[bo wrote:
]It's altogether different matter that I really have to strain to see the difference in the demo images in this thread and that probably means I'm a lot less demanding in that area.

I can imagine it's hard to see the defect, clicking on the images opens a larger higher quality image, maybe that helps. I agree, they're little defects... and yes, I'm a perfectionist :-)

Still, I'm very happy with APG 2.5 and where it's going.

_________________
Ronald

Fujifilm S5 Pro, Nikon D300, NodalNinja 3mk2, Sigma 10mm f/2.8 fisheye HSM,
Sigma 30mm f/1.4 EX HSM, Sigma 24-70 f/2.8 EX HSM, Sigma 70-200mm f/2.8 EX HSM


Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 11 posts ] 

All times are UTC + 1 hour


Who is online

Users browsing this forum: No registered users and 1 guest


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB® Forum Software © phpBB Group