Hot answers tagged

16

Using Hugin Yes, since Google PhotoSphere panos are stored as equirectangular projections you can use Hugin to remap to other projections. Go into the Interface → Advanced (or Expert) mode. Click the Add Images... button to load the stitched panorama. Set the Lens type to Equirectangular and the HFOV to 360. This will load your 360x180 as a 360x180. Go ...


15

Increase the number of control points. The standard is eight (-c 8). Raise this number to 20, 50, 100 or even 500. You will not get a worse result because of that - align_image_stack will just use (much) more time. A good time for a coffee break! Play around with the required correlation between control points. The standard value is --corr=0.9, so if ...


11

Many people use Hugin to do not just panostitching, but also exposure stacking (for high-dyanmic range (HDR) or exposure fusion) at the same time. Stitching 360x180 panoramas shot outside almost inevitably means you have the sun in your shot and are covering a very large dynamic range, so pano shooters were among the first users to do HDR processing. So, ...


9

I think you should, in fact, use the slightly-smaller value. That's not because I've measured, but because I can resolve the apparent contradiction from exiftool: it's showing you a rounded value. Try giving it the -n flag, to disable what exiftool calls "print conversion": $ exiftool -n -ScaleFactor35efl sample.jpg Scale Factor To 35 mm Equivalent: 0....


6

No, you need to enter either the field of view (FOV), or the lens and crop factor information. Luckily for you, for the Apollo mission images, the camera and lens information is well documented. The still photos taken by the lunar module team were taken exclusively by Hasselblad cameras modified to accept 70 mm film backs (56mm frame size, 1:1 image ratio), ...


5

After some more studying the program and testing around I finally got it to work! Here is every step I took: Launch Hugin (version 2019.0.0.a369cbe55179, on Manjaro Linux); Interface menu > Expert; In the Photos tab, add images (In my case they are a 3x4 grid, all are 1920x1080, but I have also tried a set of images that don't have a clear shape after ...


4

What you're missing is a lot of scene coverage. To get a 360ºx180º full spherical view, you have to cover the entire sphere. Without a fisheye lens, this is likely to mean multiple rows as well as a zenith and nadir shot, with precise coverage. In hardware terms, this probably means you need a panohead on a tripod--or at least a plumbline--to make sure ...


4

Broken-Line stitching errors There may be no way to fix these in Hugin itself, as it's very probable you have a parallax issue when shooting. If you do not rotate the camera/lens combination precisely around the no-parallax point (NPP), particularly in smaller enclosed spaces, then you are creating issues that no stitcher can fix with any amount of warping,...


4

The pto files produced by Hugin are basically just configuration files. Some programs, such as Hugin itself, may be able to open the files and allow you to preview something that resembles what the final panorama may eventually look like, but because it is not the finished panorama, the preview will only roughly approximate the final output. Generally, pto ...


4

With the images you supplied, Hugin (version 2020.0) had no problem automatically finding control points. As you can see, the distance was very good (zero is desired outcome) with average control point distance = 1.36, standard deviation = 0.82, and maximum distance = 3.39. From the Photos tab: I used Hugin's CPFind to generate control points. Optimized ...


3

You don't have sufficient scene coverage for an equirectangular. 360x180 coverage means you took enough photos and stitched to cover the entire sphere. If you didn't point your camera straight up and straight down at any point to take a member image, you have no coverage of the zenith (top) and nadir (bottom) of the sphere, which is why those areas are black ...


3

Using align_image_stack, there is no need to load all the images at once. The whole point is to remap the images to one target image, so if image nr. 1 is your anchor image you can just load that image and another one, get that one aligned to the anchor and then you repeat this to process for your next image. This can be automated using a suitable script ...


3

My two cents. 1) Take a look at blender to stabilize video: https://www.youtube.com/watch?v=nU8zqn091rM 2) Do shorter pices of video at a time. Try for example doing sequences of 30-45 frames. 3) Use one frame as a reference to stabilize the rest. So you have a frame 1 on all the aditional sequences. You can remove this repeating frame after. 4) You now ...


3

It's hard to tell, but it looks like in the stitched image your camera might have been aimed slightly above the horizon in at least 2 of the unstitched shots. By aiming up a little bit, the horizon at the widest part of your image "curls upwards" somewhat. This effect is similar to barrel distortion, and is more prominent with wider fields of view (shorter ...


3

From http://hugin.sourceforge.net/docs/manual/Hugin_Control_Points_table.html: Distance, the distance in pixels between a perfect alignment and the actual alignment achieved by the optimiser. Otherwise, after selecting Fine-tune all Points from the Edit menu, this column shows the correlation between the points (0.0 indicates no correlation and 1.00 ...


3

If you use 6.03 for the Focal length multiplier field you should be good to go. The Sony PowerShot DSC T-100 lists the 5.8-29mm focal length as 35 equivalent to 35-175mm. This figures out to 6.035, but the 1/2.5" sensor of your camera is in a 4:3 ratio instead of the standard 35mm 3:2 ratio. The diagonal of a standard 35mm film frame divided by the 7.18mm ...


3

Your question is not clear because we do not really know what you have, but if your image is misaligned because you did not use a panorama head, you can not correct it. Take a look at this: Do I "need" a panoramic head to shoot 360 panoramas?


2

The answer to your question is this. It is possible to stitch an equirectangular panorama or a stereographic projection from handheld offset images but you are likely to end up with problems like the one in your images, especially when you are that close to things you are photographing. You see the problem with the ceiling may not necessarily be connected ...


2

Try using Darktable instead. The 'lens correction' module shown above should help remove the fisheye effect. If your lens is in the database it can automatically make the correction, or if not you may adjust manually.


2

I suspect you haven't "Aligned" the images. In the latest version (2013.0.0, released a few weeks after your question was posted) it's a bit simpler / more obvious what to do in the 'Simple' interface mode (the default). For a lot of simple cases it's a matter of just clicking the 3 buttons in order: Load images... Align... Create Panorama...


2

Straight line features apply to a single image. You set your view to have the same image on the left and right and set the two end points. If you have three images, you'll do this once for each image.


2

My guess is that you're not taking care to keep your camera level/consistent in pitch. When you change the orientation of the camera between member shots, you can get errors like this. I'd highly recommend shooting with a two or three-axis spirit level on the hotshoe. Most cameras will only show you if you're level in roll, not in roll and pitch.


2

This FFMpeg command takes as input an equirectangular image and provides as output an image as if it was shot by a camera poiting in the direction specified by yaw and pitch parameters: ffmpeg -i input.png -vf v360=e:flat:yaw=-30:pitch=-20 -y flat.png -vf - video filter v360 - v360 filter e:flat - reproject from equirectangular to flat (see other ...


2

As suggested by @HarryHarrison, batch export the images in substantially smaller size like 600x400 and in 8 bit per channel. At least as a test. This will require 50-100x less memory. Repeat your program tests on that size - it should run way faster and should finish. If you have plenty of memory, the small images will process possibly 50 x faster, but I ...


2

The issue may be the size of the images, you're doing images which are 4000x3000, which is considerably larger than 4k resolution, batch scaling your images to a more manageable size before loading into the software may make it finish faster (using less resources). If you plan to turn this into a video, then think practically about where it will be shared, ...


2

If memory serves me you can force Hugin to make certain points form a horizontal or vertical line in the final image. You might look into that. Note however that forcing one part of the image to adopt a particular geometry will typically result in other parts of the image being distorted. Manually selecting matching point and lines can also help, but ...


2

If you're trying to remap six cube faces back to an equirectangular, the easiest way to do it is with the commercial software package, Pano2VR. Pano2VR lets you feed it six cube face images individually, or in a t-cross format, and will transform it into an equirectangular and vice versa. If you're looking for an open source command-line way to do this, ...


2

What you are missing is the type of projection used to create a 360 degree tour. The software you are using seems to be building a cylindrical projection and then unrolling it. In other words, it is blending the images together to remove seams along the sides, but it maps them on to a flat circular wall around the viewer. This results in a very natural, ...


2

Hugin used to use the lensfun database. But the 2015 version of Hugin's release notes state: Lens database Hugin now has its own camera and lens database, and hugin_lensdb is the maintenance tool. Lens data is automatically added to the database when stitching a project file from hugin. Data can also be manually added to the database from hugin and from ...


2

I don't think there's any computational way of doing this. The main problem is that whatever computational software you have isn't going to know your true HFoV (how far you rotated), and you can't use the aspect ratio, since it's not a true equirectangular mapping, and you're missing VFoV (i.e., you haven't covered the entire ceiling or the entire floor) for ...


Only top voted, non community-wiki answers of a minimum length are eligible