The main issue is the deliberate parallax for stereoscopic imaging. You will only be able to get good alignment if your control points are on objects as far away from the camera position. In principle, you will only need 2 pairs of control points to align 2 panoramas with y, p, r adjustment. Initially, try to find 2 pairs of control points that are 90 degrees apart from each other and on object as far away as you can find. You can then add more pairs of control points to see if you converge on a good solution.
If you want to fix verticals, then you can place a pair of vertical control points on each image. Yaw the images so that the vertical control point pair on the reference pano is at 0 deg yaw and then allow the roll of the reference image to also be optimised. The other image vertical control point pair should also try to be optimised to be vertical. See attached PTO file. You can see the very significant parallax (that you want) but this diminishes for parts of the scene further away from the camera. On Saturday, 26 November 2022 at 21:37:42 UTC DerekS wrote: > Hi dkloi > > Thank you again for your further assistance. I'm 99% of the way there > now. My control points were in different places than yours, and so I got a > different yaw, pitch, and roll than you. I tried both in a 3D viewer I > created and neither result are ideal. The reason that uprights post of the > signpost are not parallel to each other. Rather a Yaw of around 0.4, Pitch > of 1.3, and Roll of around -1.1 seem to be a better fit. Adding lots more > control points helps a little, but it seems to be trial and error rather > that a precise science. I'm going to have to give this some more thought. > > Best wishes > > > Derek > > On Friday, 25 November 2022 at 11:02:11 UTC dkloi wrote: > >> >> What you can do is custom optimise and choose what parameters you want to >> keep fixed on the optimise tab that comes up. Make sure the parameters for >> the 1st panorama are fixed (unticked) and that the lens parameters are also >> unticked. Then just optimise the y, p, r parameters of the 2nd image. >> >> To export just the 2nd pano with the adjusted parameters, in the preview >> window you can select only the 2nd panorama, hide the other one. Then when >> you export, it will only use the selected image. >> >> Since you have deliberate parallax, you should try to select control >> points on objects as far away from the camera as possible. >> On Thursday, 24 November 2022 at 09:18:05 UTC DerekS wrote: >> >>> Hi dkloi >>> >>> Thank you for taking this on. >>> >>> I should explain that the parallax is deliberate, as these two panoramas >>> are taken from a 360 camera rotated at a distance of approximately 32mm >>> from a centre point, by 90 degrees. As mentioned above, there are actually >>> four panoramas, the 1st taken at 0 degrees rotation, 2nd 90 degrees, 3rd >>> 180 degrees, 4th 270 degrees. A 3D panorama can then be viewed by showing >>> the left and right eye the appropriate section of each panorama, it works >>> well in a google cardboard style headset. Example of earlier work here at >>> https://www.virtualmountains.co.uk/GorskiKotarV50/TourControlV50_8_7.html >>> . >>> >>> Unfortunately some inaccuracy in the camera gyro and likely some >>> flexibility in my equipment means that what may be vertical in one panorama >>> is leaning a little in another, and the horizon is higher or lower from >>> panorama to panorama. >>> >>> So what I'm trying to achieve is to find an easy way to correct the >>> pitch and roll so that each panorama is better aligned. >>> >>> The pitch and roll of 0.4 and -0.6 found in your pto are typical of the >>> corrections I have to apply. >>> >>> When I added the two panorama to Hugin, a yaw of -135 is applied to the >>> first image, and 135 to the second. What are you doing to prevent this? >>> >>> After adding the control points, are you just selecting optimize to get >>> the best pitch and roll values, or am I missing a stage? >>> >>> Is there an simple way to export the 2nd panorama alone, with the pitch >>> and roll of 0.4 and -0.6 applied, or do you recommend manually applying >>> these correction values in the "Move/Drag" tab? >>> >>> All the best >>> >>> Derek >>> >>> >>> >>> On Wednesday, 23 November 2022 at 19:16:29 UTC dkloi wrote: >>> >>>> Here's the Hugin project file. I've tried to align it as best as I can >>>> but you have some noticeable parallax which prevents the two panoramas >>>> lining up perfectly. >>>> >>>> On Tuesday, 22 November 2022 at 20:49:30 UTC DerekS wrote: >>>> >>>>> Sure. Here are two examples. >>>>> >>>>> I would be looking for an easy process to align image 2 with image 1. >>>>> In reality there would also be images 3 and 4 that would also need to be >>>>> aligned with image 1. >>>>> >>>>> Many thanks and kind regards >>>>> >>>>> Derek >>>>> >>>>> -- A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ --- You received this message because you are subscribed to the Google Groups "hugin and other free panoramic software" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/hugin-ptx/54a19f1c-41a5-453f-b6ac-29bdd495531cn%40googlegroups.com.
Aligned Verticals.pto
Description: application/ptoptimizer-script
