Using a DSLR or mirrorless camera to shoot and stitch a 360 photo

Learn how to use a regular camera and PTGui to create 360 photos that are higher resolution, have a better dynamic range and are more precise and accurate than a 360 camera can produce.

Hardware:

DSLR or mirrorless camera, fisheye lens, panorama head, tripod

Software:

RAW processing software (Lightroom, DxO PhotoLab, or other)

PTGui (Pro version required for masking and Viewpoint Correction)

360 cameras can shoot a complete 360 scene in one go, but while the quality these can produce can be very good if handled carefully, it doesn't match what’s possible using a regular photographic camera (DSLR or mirrorless body) and a fisheye lens. Shooting and stitching multiple photos into a final composite 360 equirectangular image requires some care and attention, but the results can be head and shoulders above what a dedicated 360 camera can produce.

Fisheye lenses are normally used because they capture a wider angle of view than more traditional rectilinear lenses. The curved line distortions these produce are corrected automatically during the stitch process. For the following steps we suggest either an 8-11mm fisheye with an APS-C (crop) sensor size or a 15-16mm fisheye with a full-frame sensor. Both options will require taking six shots around plus one directly up and another directly down in order to create an ‘equirectangular’ image that contains the complete 360x180 degree scene in a single file. Non-fisheye lenses can be used for 360 photography, these just require more shots to cover a complete 360-by-180-degrees scene. That makes them more challenging to work with, and the final panorama pixel sizes will be much larger than is required for use in Meta Quest headsets.

A panoramic head is important to use. This holds the camera in the right position so that it turns around the optical center of the lens, known as the ‘no parallax point’ or NPP for short, rather than the traditional rotation point of ordinary tripod mounts. This ensures that where one shot overlaps with another for stitching, there will be no problem with things in the foreground appearing to shift relative to things in the background.

It is possible to shoot hand held, using a weighted string hanging from the lens to help turn around one point in space. This tripodless method even has a name, Philopod, and it is invaluable where tripods aren’t allowed. But, just as with using non-fisheye lenses, this is best attempted after gaining experience with the normal process.

Once an equirectangular image has been stitched it can be used in tour-building software such as Pano2VR, 3DVista Virtual Tour Pro or KRPano, in online virtual tour services such as Teliportme, Kuula, 360Cities and many others, and of course in Facebook as a 360 image post. For use in a Meta Quest 2 headset the optimal pixel size is 8192x4096. The maximum size advised for use in a Facebook post is 14000x7000 pixels, but the smaller Meta Quest-optimized 8K size is fine unless opened in large display mode.

1. Camera settings

The shots you take will be assembled into one composite image. This means all the settings for your exposure should be locked down; they shouldn’t change from one shot to the next, or PTGui will have to work hard to blend the differences away. Best practice for this work is to have every setting set to manual, and to shoot in RAW rather than JPEG mode. (Shooting in RAW also means you can adjust white balance later, although every shot should be treated the same.) Expose for the most important part of the scene, and consider shooting bracketed exposures if the dynamic range between shadows and highlights is very wide.

Your panorama head should have click-stops to help you rotate the ideal amount between each shot. For the lens and camera body we suggest here, six stops around is ideal, plus a shot directly upwards (the zenith shot) and down (the nadir). In general there should be around 25-30% overlap from one image to the next. Avoid overlapping by 50% or more as this can cause problems with stitching. Turn the panorama head gently, being careful not to jog the tripod out of position. Also watch out for your shadow as you go. A wireless remote trigger can help if you need to step away from the camera.

When processing your RAW files make sure you apply the same settings to every shot, for the same reason that you should use the same exposure settings throughout when shooting. Feel free to adjust the white balance of your shots, but avoid setting different values on different shots; that’s a technique that can be difficult to manage. Save the images as 16-bit TIFF files for the highest image quality when you stitch; working with JPEGs from the start saves disk space but the format’s lossy image-damaging compression and limited dynamic range is not ideal for intermediate steps.

Drag and drop your images into PTGui, in the Project Assistant tab. If the images don’t have the metadata identifying the lens and sensor size you’ll need to supply this, otherwise you can move straight to the Aligning stage. In our experience PTGui almost always selects the right values based on the image metadata. If you want to experiment with the values in the Lens Setting tab, first read the information at https://ptgui.com/examples, as this can require quite technical levels of detail.

You can simply click the Align Images button in the Project Assistant tab, and PTGui will look for matching areas and set control points to link and align image pairs. It can help, however, to align things roughly first. Choose Tools > Panorama Editor, then in this window click the Edit Individual Images and the Show Seams buttons. Click on each image thumbnail, drag the image to the approximate location in the main area, then once this is done for all images switch back to the main window and click the Align Images button.

At this point it’s often worth choosing Control Points > Delete Worst Control Points. If you want to refine control points switch to the Control Points tab, then pick adjacent image pairs and select a control point. You can drag these or simply nudge them with the arrow keys on your keyboard. To add a new point, find a detail that’s in both images and click on a specific part in one image. PTGui may generate a point to match the one you just created. Make sure it’s in the right spot, otherwise click in the correct place in the second image to complete the point pair yourself. Choose Project > Optimize to have PTGui recalculate and adjust the fit of the images.

PTGui uses the control points to understand how the different images should be aligned. This means it is critically important to only place control points on details that haven’t shifted position between the two images; beware points on leaves or clothing as those things are likely to have moved about. Delete points that are on the wrong things, then choose Project > Optimize again to recalculate things. You’ll be told the results of the optimizing step; aim to have the average control point distance in the low single figures and the maximum below 10; any unexpectedly large values should be checked by examining the points directly.

Vertical control points help PTGui know how to level the scene. As well as fine-tuning a normal scene’s leveling this can fix a wildly tilted scene, for example shot at an angle over a balcony.

In the Control Points tab, pick the same image in both sides; this switches the control point type to Vertical Line. Look for something that has a straight vertical edge. Click near one end of that edge in one image, then at the other end in the second image. Do this for a couple more vertical lines and ideally across at least one other image, then choose Project > Optimize again and see how it snaps to a more perfectly level position. You can switch the control point type from Vertical Line to Horizontal Line, but this is only really useful if you can see a definite, real horizon in your scene.

To do a first check of the image there are two options. A standard equirectangular view can be seen in Panorama Editor window with the ‘Blended’ icon active, but to see the scene interactively go to the Preview tab and click the Preview button. This will create a low-resolution stitch of your images and open it in the PTGui Viewer tool. Drag the image to look around, paying careful attention to areas where images overlapped and looking for breaks in lines. If you see alignment problems try adjusting some control points then optimize again.

If you see unwanted elements from one of the images, for example someone walking into a shot or parts of the tripod appearing, you can mask these out in PTGui. Switch to the Mask tab, then paint red on image areas you want to remove and green on areas you want to have override other overlapping areas. Shift-clicking will paint lines between click points, and command-clicking (Mac) or control-clicking (Windows) will fill an area.

Switch to the Create Panorama tab to get to the final stitching tools. Click the 100% button to see the maximum pixel width and height your images can produce. If you want something smaller, for example 8192x4096 pixels for the Meta Quest headset, set the width or height accordingly – the other value will match. For the output format pick TIFF (a non-lossy format) if you’re going to process it in dedicated tour building software or JPEG if you plan to view this directly in a headset. (JPEG can be used instead of TIFF but best practice advice is to avoid using lossy compression formats until generating final deliverables.) Finally, choose the output color space: sRGB is best for anything that’s going straight to a player or the web. Other color spaces can sometimes be useful for further work in Photoshop or other editors and when going to print, but the image should always be converted to sRGB before final processing in tour-building software or services.

Tripod removal

Capturing a spherical 360 image in this way delivers very high quality results, but the tripod and pano head will be visible when the viewer looks down. This can be dealt with in a number of different ways. The simplest is to use a ‘tripod cap’ graphic. This can be added to the stitched equirectangular image by converting it into cube faces (in PTGui choose Tools > Convert to cube faces), then opening the bottom (nadir) image in your preferred image editor and adding your graphic there. For a more natural look you can use your image editor’s clone stamp and smart fill features to cover the tripod with what looks like uninterrupted ground. Once done, drag the cube face images into a new PTGui window and publish your reassembled panorama from there.

If these methods aren’t suitable, the pro version of PTGui includes a feature called Viewpoint Correction. This uses a photo of the floor shot from a few feet to one side and adjusts its scale and perspective so it fits into the rest of the scene. This relies on the area being patched to be relatively flat; any significant depth changes or three-dimensional objects in this area can look obviously distorted.

After you’ve taken all the shots for your complete 360 image, move your tripod to one side so you can see the original spot and it is free from tripod shadows. You can tilt the camera, the panorama head and even the tripod itself to make sure the important area of the ground is relatively near the middle of the frame. Some panorama heads also have an ‘offset nadir’ option, as shown here.

In the Panorama Editor window, drag the offset image roughly into position. You’ll need to have the Edit Individual Images button and the Show Seams button clicked in the toolbar of this window, then select the correct image thumbnail, then finally drag that image around in the main part of this window. This won’t fit perfectly as it was shot from a different position; start by dragging it straight down, then shift it around to get as close as you can to the right position at this stage.

In PTGui Pro’s Optimizer tab, in the Viewpoint column, set this image’s value to Optimize. The other images should be left set to Reset. This tells PTGui which image can be warped to match the perspective of the others. It also ensures the other images aren’t moved to fit to this one; that would mess up the work you put into refining the main set of shots, so it is very important that you don’t forget to set this image to Optimize.

In PTGui Pro’s Control Points panel, select your offset image on the left side, then go through other images on the right-hand side and assign control points manually. It is vital that you only have control points placed in the same flat plane; any control points placed on other parts of the image (for example on a wall or furniture as well as a floor) will cause incorrect and rather weird distortions to the composite image.

Once done choose Project > Optimize once more. The maximum control point distance will jump up dramatically, but only for the offset image. Check in the Panorama Editor window and by previewing to make sure this new image fits correctly. Move or delete any control points that seem suspect and optimize the panorama again, repeating until you’re happy.

Once the image is in the right place, use PTGui’s Mask panel to hide any part of this shot that isn’t on the one flat plane or that is so far away from the area with the tripod that it’s not needed. PTGui will probably create the panorama without unwanted parts of this offset image, but masking out areas means you’re not leaving that to chance.