I don’t shoot a lot of panos but it’s something I want to start focusing on.
I’m using Version: 2019.2.0.b690aa0334b5.
I’ve got a good tripod head with a built in level and captured what look to be pretty level images.
I use the “Advanced” interface and add my 8 jpgs, I find control points (123 in this case) and optimize for geometric and photometrics. I move to the Control Points tab and everything looks good. When I click on the OpenGL preview I get this:
When move to the Stitch tab I click on field of view, optimal size (over 25K wide) and crop. The output is either what I see in the preview (expected) or sometimes I get an output of just image 0.
What am I doing wrong?
Here’s a link the a Google Drive folder with the jpgs: https://drive.google.com/drive/folders/10wNxIcSEWyQoPrkT4XuZRsHGRVXOwya_?usp=sharing
Make sure you don’t have control point in the clouds and add a few manual point to the buildings.
Are you using the level built into the camera? If you’re using a level built into the tripod legs, it usually isn’t level when you rotate the head. Turn the level in your camera on and pan your camera around before actually shooting the photos.
You can also get a leveling pano base that sits between your tripod legs and head: Sirui LE-60 Leveling Base LE60 B&H Photo Video
Sometimes when consecutive images are largely overlapping, Hugin adds control points to other images as well (for example between image 0 and image 2 or 3) so make sure you check those control points too.
Did you try to set the Anchor to the middle picture of your sequence like this?
I did remove the points from the clouds and that helped.
I have a level in the legs and in the head and I used the one in the head. My camera does have a built in level but I always thought the bubble level would be more accurate.
I didn’t but I’ll give that a try.
I’ll check that out, thank you.
@David_LaCivita Have a look at this pano I stitched from a lot of shots.
Because I was lazy, the horizon is curvated, but that could be corrected by adding horizontal line control points
How does the Horizontal Line mode work?
You select two images (maybe only one as in my following simplified screenshot example and add two control points (one on the left, one on the right) ) and select mode:
Be careful. The horizontal control point work only correctly when they are on the horizon or the output projection is rectilinear. A more general approach is to use vertical control points. (The assistant uses this approach.)
PS: You can also create horizontal and vertical control points when you select the same image in the control point tab and drag a line with pressed alt button.
So you’re suggesting I use vertical lines on the sides of the buildings instead of trying to identify a horizon? I like this approach because the “horizon” moves away from me in this image.
Also,for anyone looking, I’m on Linux and I need to hold ctrl+alt to draw a line.
Yes. Horizontal control points are (in general) only useful when re-projecting rectilinear images. Otherwise stick on vertical control points.