Digital photo stitching application is the workhorse in the panorama-making process, and will range from providing a completely automatic one-click stitching, to some more hours-consuming manual process. This is part 2 of the tutorial, which assumes all individual photos are actually properly captured (stage 1 below is finished); for stage 1 and a review of the complete stitching process kindly visit part 1 of this tutorial on digital panoramas.
In order to begin processing our series of photos, we must select a suitable software application. The most significant difference between options is at the way they choose to address the tradeoff between automation and suppleness. Generally speaking, fully customized stitching software will always achieve better quality than automated packages, but it can possibly cause being overly technical or time consuming.
This tutorial aims to boost knowledge of most software stitching concepts by keeping the discussion as generic as you possibly can, however actual software features may talk about a course called PTAssembler or PTGui (front-end for PanoTools or PTMender). PTAssembler incorporates a fully-automated one-click stitching option, together with providing for pretty much all possible custom stitching available options in other programs.
During the time of this short article, other notable programs include those that come packaged with all the camera, including Canon PhotoStitch, or popular commercial packages like Autostitch, 360 photo software, Arc Soft Panorama Maker, Panorama Factory and PanaVue, among others.
STAGE 2: CONTROL POINTS & PHOTO ALIGNMENT
Panorama stitching software uses pairs of control points to specify parts of two camera photos that refer to a similar point in space. Pairs of control points could be manually selected by visual inspection, or these might be generated automatically using sophisticated matching algorithms (like Autopano for PTAssembler). With a lot of photographs, greatest results are only able to be achieved with manual control point selection (which is usually the most time-consuming stage from the software stitching process).
The example above shows a selection of four pairs of control points, for two photos in a panorama. The most effective control points are the type that are based on highly rigid objects with sharp edges or fine detail, and are spaced evenly and broadly across each overlap region (with 3-5 points for each overlap). This means that basing control points on tree limbs, clouds or water is ill-advised except when absolutely necessary. It really is for that reason recommended to always capture some land (or some other rigid objects) in the overlap region between all pairs of photographs, otherwise control point selection may prove difficult and inaccurate (such as for panoramas containing all sky or water).
The example below demonstrates a predicament where the only detailed, rigid portion of each image is in the silhouette of land at the very bottom-thereby making it tough to space the control points evenly across each photo’s overlap region. During these situations automated control point selection may prove more accurate.
PTAssembler includes a feature called “automatically micro-position control points,” which works by with your selection as being an initial guess, then seeking to all adjacent pixels within a specified distance (for example 5 pixels) to determine if these are typically an improved match. When stitching difficult cloud scenes including that shown above, this effectively combines the advantages of manual control point selection with the ones from automated algorithms.
Another consideration is the way far awayfrom the digital camera each control point is physically located. For panoramas taken with out a panoramic head, parallax error may become large in foreground objects, therefore better results may be accomplished by only basing these on distant objects. Any parallax error from the near foreground may not be visible if each one of these foreground elements are certainly not contained within the overlap between photos.
STAGE 3: VANISHING POINT PERSPECTIVE
Most photo stitching software gives the opportunity to specify the location where the reference or vanishing point of perspective is found, in addition to the sort of image projection.
Careful range of this vanishing point will help avoid converging vertical lines (which could otherwise run parallel), or possibly a curved horizon. The vanishing point is usually where one would be directly facing if they were standing within the panoramic scene. For architectural stitches, such as the example below (120° crop in the rectilinear projection), this time is additionally clearly apparent by following lines in to the distance which are parallel to one’s type of site.
Incorrect placement from the vanishing point causes lines laying within the planes perpendicular towards the viewer’s collection of site to converge (even though these would otherwise appear for being parallel). This effect can even be observed using a wide angle lens within an architectural photo and pointing your camera significantly below or above the horizon- thereby giving the sense of buildings that are leaning.
The vanishing point is likewise critical in very wide angle, cylindrical projection panoramas (for example the 360 degree image shown below). It could exhibit different looking distortion if misplaced, causing a curved horizon.
In case the vanishing point were placed way too high, the horizon curvature would be from the opposite direction. Sometimes it can be difficult to locate the particular horizon, because of the presence of hills, mountains, trees or some other obstructions. For such difficult scenarios the positioning of the horizon could then be inferred by placing it with a height which minimizes any curvature.
Panorama stitching software also often allows the option to tilt the imaginary horizon. This is very useful once the photo containing the vanishing point had not been taken perfectly level. For this particular scenario, even when the vanishing point is put in the correct height, the horizon may be rendered as owning an S-curve in case the imaginary horizon is not going to align with all the actual horizon (from the individual photo).
When the panorama itself were taken level, then your straightest horizon is the one that yields a stitched image whose vertical dimension is the shortest (and it is an approach sometimes utilized by stitching software).
STAGE 4: OPTIMIZING PHOTO POSITIONS
When the control points, vanishing point perspective and image projection have got all been chosen, the photo stitching software may then begin to distort and align each image to make the final stitched photograph. This might be probably the most computationally intensive step in the process. It works by systematically searching through combinations of yaw, pitch and roll to be able to minimize the aggregate error between all pairs of control points. This method can also adjust lens distortion parameters, if unknown.
Keep in mind that the aforementioned photos are slightly distorted; this really is to emphasize that anytime the stitching software positions each image it adjusts for perspective, and that the quantity of perspective distortion is determined by that image’s location in accordance with the vanishing point.
The key quality metric to be aware of is definitely the average distance between control points. If it distance is large relative to the print size, then seams may be visible no matter how well these are typically blended. One thing to check is whether any control points were mistakenly placed, and that they adhere to the other guidelines listed in stage 2. If the average distance is still too large then this may be due to improperly captured images, including parallax error from camera movement or perhaps not by using a panoramic head.
STAGE 5: MANUALLY REDIRECTING & BLENDING SEAMS
Ideally one may wish to position the photo seams along unimportant or natural break points in the scene. In case the stitching software supports layered output anybody can perform this manually by using a mask in photoshop:
Without Blend Manual Blend Mask from Manual Blend
Note the way the above manual blend evens the skies and avoids visible jumps along geometrically prominent architectural lines, such as the crescent of pillars, foreground row of statues and distant white building.
Ensure that you blend the mask over large distances for smooth textures, like the sky region above. For fine detail, blending over large distances can blur the photo if you find any misalignment between photos. It really is therefore wise to blend fine details over short distances using seams which avoid any easily noticeable discontinuities (see the “mask from manual blend” above to view how the sky and buildings were blended).
On the other hand, manually blending seams could become extremely cumbersome. Fortunately iphoto software has an automated feature which may perform this simultaneously, as described over the following section.
STAGE 5: AUTOMATICALLY REDIRECTING & BLENDING SEAMS
Among the best approaches to blend seams within a stitched photograph is to apply an approach called “multi-resolution splines”, which could often rectify even poorly captured panoramas or mosaics. It functions by breaking each image up into several components, just like how an RGB photo might be separated into individual red, green and blue channels, except that in this instance each component represents some other scale of image texture. Small-scale features (including foliage or fine grass) have a high spatial resolution, whereas larger scale features (such as a clear sky gradient) are said to possess low spatial resolutions.