Hugin Stitcher tab
The rest of hugin is all about setting up the project and aligning images, the Stitcher tab is where the final output file is created.
Here you can set the output Projection of your project, there are lots to choose from, each with different advantages and disadvantages:
- Rectilinear, this is the same projection as a photo taken with a 'normal' camera and lens. Use this if you are just stitching a handful of photographs together with a narrow Field of View or correcting perspective in a single shot.
- Panorama, actually a simple Cylindrical Projection as used by traditional rotating panoramic cameras. A good projection for printing a 360 degree panorama, though you may prefer Mercator Projection.
- Equirectangular, the all purpose format for representing an entire spherical scene. It covers 360 degrees horizontally as well as the zenith and nadir.
- Fisheye, the same projection as a photo taken with a fisheye lens. Better for representing a wide Field of View than rectilinear, but in many cases Stereographic Projection gives less distortion than simple fisheye.
- Stereographic, a conformal fisheye image. Objects in a stereographic image keep the same shape and show less distortion than simple fisheye.
- Mercator, a conformal cylindrical image. A good projection for printing a 360 degree panorama.
- Trans Mercator, a mercator image rotated 90 degrees, suitable for displaying tall or overhead objects.
- Sinusoidal, an equal area projection of an entire spherical scene.
- Lambert Equal Area Conical
- Lambert Equal Area Azimuthal
- Albers Equal Area Conic
- Miller Cylindrical
Field of View
This is the horizontal and vertical angle of view of the output image, clicking Calculate Field of View will shrink or enlarge the field of view of the output to fit the arrangement of the input images - The Fit button in the Hugin Preview window does the same thing.
Note that some Projections have a limited field of view, notably:
- Rectilinear has to be less than 180 degrees both vertically and horizontally.
- Panoramic (cylindrical) has to be less than 180 degrees vertically.
- Stereographic has to be less than 360 degrees both vertically and horizontally.
- Mercator has to be less than 180 degrees vertically.
- Transverse Mercator has to be less than 180 degrees horizontally.
Panorama Canvas Size
Set the width and height of your output panorama in pixels. Calculate Optimal Size will estimate a size that has about the same resolution as your input images.
Some examples: a three megapixel image has pixel dimensions of 2048 x 1536, an A4 print at 300 pixels per inch will have a pixel size of 3500 x 2480, a full screen spherical Equirectangular Projection image will have pixel dimensions of 6000 x 3000 or greater and a gigapixel image has a pixel size of 32768 x 32768.
Note that the interpolation used by hugin doesn't handle downsampling very well, so output images smaller than about half the size of the Optimal Size will show aliasing artefacts. If you want to create high quality small images, it is better to create an Optimal Size image in hugin and downsize it later in an image editor such as the Gimp.
The crop settings allow just a portion of the panorama to be stitched, there are various reasons to do this:
- When correcting perspective large areas of the panorama output will be empty anyway.
- Large 'gigapixel' style panoramas can be stitched in sections then blended later.
The cropped-out areas are shown darkened in the hugin Preview window.
Hugin can output 'normal' stitched images, exposure blended images or high dynamic range (HDR) images. The following options determine which kind of image is created, and allow keeping the intermediate images created during the process.
If Blended panorama is enabled then enblend is used for blending. In the final stitching process nona reprojects and distorts images to fit, enblend takes these images as individual TIFF files and merges them using sophisticated seam positioning and blending. Further enblend settings can be found in the hugin Preferences.
Enable remapped images if you want to keep the intermediate images that enblend uses as input - For example modifying the alpha channel of these images and then blending manually is one technique for including and excluding people or objects that move between shots.
If Blended panorama (enfuse) is enabled then hugin will group the input images into exposure stacks by comparing positions and EV exposure values. Each of these bracketed exposure stacks will be exposure blended with enfuse and the results stitched together into a panorama with enblend.
Note that for this to work, the scene has to be photographed multiple times using exposure bracketing and the EV exposure values set either manually in the hugin Camera and Lens tab, automatically from EXIF data or by optimising exposure in the hugin Assistant tab or hugin Exposure tab.
Note also that unlike Normal and Merge to HDR options where images are exposure corrected as part of the remapping process, enfuse requires that each exposure layer is supplied uncorrected - Hugin takes care of this automatically and will not apply correction in this case.
Enable Blended exposure layers to additionally stitch each exposure layer into a separate panorama - Useful for manual exposure blending.
Enable Remapped images to keep the intermediate images supplied to enblend.
Merge to HDR
Note that like the Exposure blending option above, this generally only makes sense if the scene has been photographed multiple times using exposure bracketing, and the EV exposure values optimised in the hugin Exposure tab.
Enable Stacked HDR images to keep copies of the remapped HDR images as supplied to enblend.
Enable Individual non merged images to keep copies of each image remapped in linear colour space before deghosting and merging to HDR.
Click Save project and stitch to generate output panoramas immediately (or Save project and send to batch to create a Makefile for later batch stitching - TODO doesn't work yet).
Set the Interpolator (i) to change the sampling interpolation. You probably won't notice much difference between the various options except that Nearest Neighbour is fast but with very low quality. The default of Poly3 (bicubic) is generally good for most purposes.
Normal Output can be in one several formats:
- TIFF, various compression options. 16bit and 8bit depth supported.
- JPG, lossy compression suitable for web/email.
- PNG, lossless compression. 16bit and 8bit depth supported.
HDR Output can be either:
- floating-point TIFF, various compression options.
- EXR, this is a high dynamic range format which is more compact than a high dynamic range TIFF.
TODO where is this set? Cropped TIFF files are smaller and more efficient because unused parts of the image are not stored in the file. You should always save cropped layers unless you need to open them in an image editor without Cropped TIFF support.