I luckily fixed the line when I masked the angled images:
Is there an easy way I could fix parallax line errors in PTGui Pro before blending? In the following photo:
the stairs appears in 4 images (2 at 0 degree and 2 at -45 degree). I tried to add control points on each line of the stairs between the 4 images but there were already enough control points and so it was automatically assigned. I also tried to mask out the stairs in the two shot at -45 but encountered other small line problems on the chair and floor (see below picture) but it was better result. I had to align the images every time I use mask.
I’m using D7000 with 16-85mm at 16mm on 303SPH head & 190XPROB tripod. I use tripod legs to level the head so levelling is not 100% right but I think levelling doesn’t affect parallax point! am I right?
I take around 32-38 images for one panorama:
In the Show Seams mode, the line of the stair looks just fine unlike the blending mode:
The way I use to find the nodal point on the 303SPH is to point the camera downward @-90 degree and turn on the viewfinder grid. move both the base plate and the camera plate to the centre of the head. Then, point the camera at 0 degree and move the 3rd base plate back and forth. I guess my head setting is right so why I'm always getting line errors!? shall I just use fisheye lens to make life easier!
Thanks for your reply. Yes I tried my best to make the lens rotation in the centre (I use two doors edges to find the Non Parallax Point). But even when I don't do it 100% right, is there a way in PTGui to fix these lines? I'm not fully aware of all the features the software offers.
on my panohead, I found that the vertical arm rotate a little despite the screw is secured on the base plate. As you can see in the picture of my camera grid, the line is not straight. Is that a problem?
Am 05.11.2014 um 21:52 schrieb Moafaq [hidden email] [PanoToolsNG]:
> But even when
> I don't do it 100% right, is there a way in PTGui to fix these lines?
You can set (more) control points in the affected area and you can use
masks to move the seam line. In general, as Willi and John wrote, there
should be no deviation if anything is set up properly.
However, there are other possibilities that cause that misalignment: You
could have accidentally moved the tripod, camera is mounted skewed, you
have misplaced control points in other images, your lens correction
parameters are off (due to wrong defaults, insufficient freedom for the
optimizer or bad settings) etc.
You also can try smartblend, an alternative blender used as a plugin in
PTGui which tries to place the seam lines such, that no errors are visible.
It's difficult to say whether slight rotation of the vertical arm might significantly affect the stitching errors. I'm sure we would be able to get a better understanding of the source of your problems if you could make available the complete set of images (half size jpegs would be fine, preferably grouped together in a zip or rar file). If you could upload them to some convenient web space or the free file sharing site http://www.ge.tthttp://www.ge.tt (no need to register), that would be great. Post a link here.
Having now checked the images, there is a small amount of both horizontal and vertical parallax visible, This needs to be checked with something really close to the camera set against a background really far away. Using two objects indoors doesn't work as well since both objects are then relatively near to the camera and both are affected by parallax. My tutorial that Willy referred you to suggests a strip of sticky tape on a window looking out onto a distant view, but don't focus on the near object - use a small aperture.
Your nadir shot is largely redundant as it offers very little extra useful coverage. Nevertheless, for test purposes it is helpful to take two nadir shots, rotating the head 180 degrees in between. Stitch them to PSD layers and check for parallax like this: http://ge.tt/6nqUsi32/v/0?chttp://ge.tt/6nqUsi32/v/0?c . This is better than simply centering the centre screw in the camera's viewfinder.
Wow, that's impressive result John. How did you manage to solve the stairs problem? I always work on the simple mode and not the advanced where there are extra tabs. Did you manually set values in Lens Settings, Image Parameters or Optimizer? I’ll try finding the NPP again using your techniques, the tape on a window and the two nadir shots.
I didn't take any particular action over the stairs. I did my usual workflow when presented with "these images won't stitch - I've tried everything". In brief:
I load in the images and input all the information I have about the project on the Lens Details and Image parameters tab. I set the image parameters (y,p,r) to their nominal known positions - either manually (using fill yaw etc) or by applying a template or Align to Grid. I then generate control points from the Control Points menu only between horizontally and vertically adjacent pairs of images. Optimize. Then depending on the result I do whatever seems necessary to get rid of bad points, assign additional points to fill areas where there aren't any (making use of the Generate control points here feature, now that the images are reasonably aligned).
If things initially go awry, I will optimize first the horizontal row (by checking only those images on the Use control points of list on the Create Panorama tab). When those are optimized ok, further images or groups of images can be successively included until all are used.
When I checked for NPP using stick on a window, it wasn’t right hence the stitching issue.
Here it shows before (as GIF):
And after (as GIF):
From your workflow:
“only between horizontally and vertically adjacent pairs of images”
How to generate control points to selected images? Is it not good to just generate control points to all the images? I only add control points when it’s asked for after pressing Align Images. Like the zenith shot in my example, I had to add points.
“I set the image parameters (y,p,r) to their nominal known positions”
the values of y,p,r in my project is different than yours! Did you set the image parameters manually i.e. entering values for each image!?
In the lens detail, will it make a different if I change the values? In my project it’s set to 16mm and 1.5 but yours it’s slightly different :s
Finally, I found that the option that solve the stairs issue (if I did everything automatically) is when I see you clicked on Horizontal/Vertical shift in Optimizer tab. Not sure what does it actually do! by default it’s not clicked.
Q1: To generate control points between a pair of images, select them for display in the Control Points tab and then go to the Control Points menu and select Generate control points for image x and image y. You can also select a number of images on the Source Images tab or Image Parameters tab and use the Control Points menu option Generate control points for selected images. Generate extra points in specific areas by shift+drag a rectangular selection on one of the images and then right click in the rectangle to select Generate control points here from the sub menu (but note that this can only be used when the images have been reasonably aligned).
Q2.To quickly assign the nominal image parameters: on the Image Parameters tab select a set of 12 boxes for one row in the yaw column and click the Fill Yaw button at the bottom of the page. Enter the number of shots per row and the starting angle (0) and let it generate the appropriate yaw values. Select the corresponding 12 boxes in the pitch column and enter the pitch angle for that row (eg 30 degrees). Roll is 0 by default.
Q3. The lens settings can be set to 16mm with a crop factor of 1.5. These are simply used to calculate an initial value for the horizontal fov parameter. These values (and the y,p,r values) will all be changed a little by the optimizer to align the control points correctly. You could set the lens to 24mm with a crop factor of 1 or 12mm with a crop factor of 2 and they would have exactly the same effect, as they all give a 35mm equivalent focal length of 24mm.
Q4. The horizontal and vertical shift parameters compensate for the optical axis of the lens not being exactly centered on the centre of the camera's image sensor. The parameter values are in units of pixels. It''s usual to include these in the optimization (but not the shear parameters, which may be needed when processing scanned images).
Finally - don't overlook the Help for this screen option, which will usually display much useful information about the current tab/window.