Hi, I was going to do some testting with the combo in the subject line and the NN3 MKII
If anyone has also tried this combo, pleaes share your experiences...
I will post back with my findings
using the suggesting setting for the upper and lower rail based on the NN List.
(i.e. 58mm and 93mm)
and was also going to take shots according to the suggestions on vrwave.com (listed below)
Min. shots: N, 6 images every 60¬? at -30¬? pitch, 6 images every 60¬? at +30¬?, Z. Output resolution: 11600 x 5800 pixels - Canon 20D (8mpx) Recommended: N, 4 images every 90¬? at -60¬? pitch, 8 images every 45¬? at 0¬? pitch, 4 images every 90¬? at +60¬? pitch, no zenith image required Alternative: N, 8 images every 45¬? at -30¬? pitch, 8 images every 45¬? at +30¬? pitch, Z
I took shot as directed by the bold Recommended Settings in the 1st post.
I used PTGui mostly automatic, although had to pick control points for 1 image.
I would have to say Im pretty happy with the stitched result, although there is definately some areas of the ceiling which are not correctly aligned...
Would this be due to parallax? Do I need to choose better control points than the auto ? Note this was made without taking a dedicated zenith or nadir photo.
If I take those would it help??
Your shots do not seem to be affected by the parallax error at all. Even the nadir circle is just perfect.
Misaligned parts are where the control point genarator usually fails, so I would suggest to add some more CP's in those areas and reoptimize the project.
If you could share the images and the PTGui project, I would be more than happy to have a look at them!
Do as Mauro says and check/add control points in the areas where you see stitching errors. Then I would suggest that you:
1. Switch into Advanced mode via the button on the right of the Project Assistant tab.
2. Select the Optimizer tab.
3. In the Minimize Lens Distortion option, select Heavy + Lens Shift.
4. Then click on the Run Optimizer button.
5. On the top bar, select Control points->elete Worst Control Points and optimize once again.
Display the Control Points table and check the placement of the points with the largest distances (just double click a table entry to display the point in the Control Points tab). Correct as necessary and reoptimize. Proceed in like manner until you get a good optimization result, with an average of around 1 pixel and a maximum of 2 or 3 (or as good as you can manage!). Then generate the panorama. It should be very much improved.
You've done really well to get such a good result. You can find a detailed comparison of features for the standard and pro versions in PTGui's support section at http://www.ptgui.com/features.html.
I've got a similar setup, 40D with 10-22mm. I've posted my rail settings on the bottom of this thread http://nodalninja.com/forum/index.php?topic=124.0 but using the T20 adaptor as I also use the 17-55mm lens. I had a similar question to you about ideal sequence: http://nodalninja.com/forum/index.php?topic=589.0 and I've been using the alternative, i.e. zenith, row at +30, row at -30 and nadirs. This works pretty well except you have a join around 0 degrees when you stitch and sometimes it's hard to align the zenith as there can be few control points on the zenith. All the same I've had some good results, see http://nodalninja.com/forum/index.php?topic=604.0 though I'm still very much in leaning mode and can probably take advantage of more experience with PTGui etc. I think the recommended flow with extra zenith coverage from the +60 shots would help with the stitching and I'll try this next time I shoot some panos.
I had a look at your first attached pano and I see there are some stitching glitches at the top, but your second pano looks perfect. I'll have to try those tips myself. Sometimes I add extra control points to PTGui, especially if it added its own points on clouds that were moving rather than on buildings etc.
RE shadows and Nadir, I setup the camera into Manual exposure and once everything is set I Register the setting to the C2 position on the dial. Then I change the drive to 10 second timer and register it to C3. Then I take the sequence of photos and switch between C2 and C3 when I need to put it on self timer so that I can move myself out of frame, and out of shadow casting. Hope this helps.
... just took a second look at your pano. I see you mean the Tripod shadow is the probem. I'm still working on this, I've not taken many indoor shots, but for most of the panos I've done I've used a handheld nadir (shutter speed was fast enough) and spent a lot of time with the clone brush to copy it in on top of the nadir shots with the tripod. The for the tripod shadow which may be outside of this frame I've used more clone brushing from nearby objects, e.g. floor boards in your case. It tends to be time consuming and painful, sometimes I need to use some kind of perspective correct to distort my handheld pano.
You should also look at the viewpoint correction in ptgui which allows you to take a photo from a distance away from your original tripod location: http://www.ptgui.com/examples/vptutorial.html but I've not managed to get this to work yet (though I've only tried one example).
I don't use HDR for my panoramas, but I do make use of features that support HDR such as linking images and generating blend planes (stitches of all the images shot at a particular exposure) for blending manually. I very much like the Viewpoint feature in the Pro version, and the vignetting and colour correction options are useful too as an aid to blending.
neil...nice job with all the people in your panos - must have been some time consuming masking to get that the way it is...
Thanks for looking at my panos, and yes it did take a lot of time to get it to look right, especially with people that are on the overlaps between pictures, or who move between pictures (walking around). One advantage of the +30 row and -30 row is that people in the distance are captured by both the +30 and the -30 shots so you can chose between them. People closer in the foreground, e.g. this one in Punta Arenas are captured mostly by the -30 row, so normally there is a single photo that includes the whole body. Be careful though of people moving horizontally as you rotate the camera, sometimes you need to take extra shots in one place to make sure you have something you can stitch later.
One tip I would offer is that PTGui has parameters that allow you to chose the blending priority from 0 to 100. The default is 100, so I set my +30 upper row to 90. This means the lower row will have priority, and it will try and blend in the majority of the lower picture. The advantage here is that the seam between pictures will be moved, e.g. for people on the horizon line you will get a whole person rather than the default 1/2 a person (just a top or bottom half due to the seam). Hope this makes sense (it's a little difficult to explain).