Long Walls: Which Panoramic Format for Corridors, Bands and Back Walls
Long walls are a case of their own in image planning. A corridor running fifteen metres, a band across three walls, a back wall that continues around a corner — there is no standard format for surfaces like these, and the obvious solutions lead nowhere: a normal image does not fit, four small ones look fragmented, and a squeezed motif is noticed immediately. The question is therefore not which picture to choose, but which projection and which crop. Behind this article stands Gigapixel GmbH — the specialized large-format stock portal: real gigapixel captures from 100 MP, no AI upscaling.
Fifteen metres of corridor, and no picture fits
The problem with long surfaces is their aspect ratio, not their size. A wall of 15 × 2.4 m has a ratio of a little over 6:1; a motif straight out of a camera sits at 3:2 or 4:3. Between those numbers lie three ways out, two of which are bad. Squeezing the motif visibly distorts every face, every tree trunk and every building edge — and it does so even when nobody can name the reason. Splitting the surface into four separate pictures creates rhythm where breadth was wanted: the gaze catches on every frame edge instead of following the wall.
The third route is the right one, and it requires a motif that was captured for the purpose. A panorama is not a cropped ordinary picture but a capture with a different geometry — and that geometry decides how the surface behaves at its edges and around a corner.
Why breadth carries a long surface
On a long wall, breadth is not merely a fitting format but the right class of subject. Browning, Ryan and Clancy describe the pattern prospect in the 14 Patterns of Biophilic Design (2014): the unimpeded view across distance, associated with lower stress and less boredom and irritation. The term goes back to Appleton (1975, The Experience of Landscape), who described the need for an overview as a basic pattern of landscape perception. A corridor is its exact opposite: narrow, without depth, the next wall an arm’s length away.
That images of wide landscapes are actively sought in confined surroundings was described by Clearwater and Coss in 1991: in isolated and confined settings, photographs of expansive landscapes are preferred in order to substitute for missing environmental cues. Franěk and colleagues compared eye movements while viewing images of differing restorativeness in the Journal of Environmental Psychology in 2018 and found that nature scenes are processed with fewer fixations and shorter scan paths than visually busy surfaces. For a corridor that many people walk through every day, that is the practical point: the surface should guide the gaze, not occupy it.
The projection question decides the edges
How a panorama behaves at its edges depends on the projection — and that is chosen during stitching, not at the printer.
| Projection | Behaviour | What it is for |
|---|---|---|
| Rectilinear | Straight lines stay straight; beyond roughly 80 to 120° of field the edges stretch severely | Architecture and interiors at moderate angles, where edges absolutely must stay straight |
| Cylindrical | Vertical lines stay straight, the horizon stays stable, edge stretching is damped | The normal case for long wall images — wide landscapes and cityscapes beyond 120° |
| Equirectangular | Fixed 2:1 geometry across the full sphere, with pronounced distortion at the poles | The source format for 360° captures; usable on a wall only after reprojecting the section required |
| Mercator / Panini | Compromise projections between straight lines and damped edge stretching | Very wide wall images where neither rectilinear nor cylindrical alone convinces |
The values and the behaviour are described in the PTGui documentation (as of 2026), which serves as a reference in panorama production. In practice this means that for a band along a side wall the cylindrical projection is almost always the right choice, because it holds the horizon and does not pull the edges apart. Rectilinear is worth it only while the field of view stays moderate — beyond that the edge regions grow so much that a tree at the edge of the frame looks twice as wide as the same tree in the middle.
That the projection need not be a fixed property of the file was shown by Kopf, Uyttendaele, Deussen and Cohen as early as 2007 in ACM Transactions on Graphics: when zooming into a gigapixel image it is possible to interpolate continuously between a cylindrical overview and a perspective detail view, so that both ways of looking work within the same image. For print, by contrast, a decision has to be made — a wall can show only one projection.
What has to happen on the capture side
A band fifteen metres wide is not made in one exposure but in a multi-row grid. The camera is rotated about the entrance pupil of the lens and captures row by row; so that the individual frames can be matched later, they typically overlap by 30 to 40 %. Less overlap leaves the software too little common structure to work from; more costs capture time without gain. In uniform areas of the image — sky, water, mist — it is better to sit at the upper end of that range, because there is little structure available there in the first place.
Two things then decide the quality of the result, and both are treated in detail in the article on stitching quality: seam placement, which should not run down the middle of the overlap but along structural edges and through the quieter of the two areas, and exposure control across the entire capture period. The second point matters especially for long panoramas: if the sun moves during the capture, brightness changes from row to row. On a 15-metre band that does not show up as a seam but as a slow brightness gradient across the wall — and it is noticed, because the eye can compare across the length.
Wall planning: measure the aspect ratio first, then choose the crop
The surface dictates the format, not the motif. The order is therefore: measure the length and height of the surface, work out the aspect ratio, and only then look for the crop within the motif that can carry that ratio. What must not happen is squeezing — a motif whose proportions are wrong is not rescued by size. A wide crop from a high-resolution capture is the right route, and it costs resolution only where there is plenty of it.
Bands that continue around a corner need a single continuous file rather than two separate prints. The reason is the corner itself: two separately computed images meet there with different perspectives and often different brightness, and the break is visible from every angle. If the surface is planned as one file and only divided during production, the motif runs across the edge unbroken.
That leaves the arithmetic. In a corridor the viewing distance is usually one to two metres — according to Ashraf, Chapiro and Mantiuk (2025, Nature Communications) the eye resolves 137 and 68 ppi respectively at those distances. Large-format print practice works at 70 to 120 ppi, in special cases at 50 to 150, and no process takes in more than around 240 ppi. Pixel width then follows from wall width in inches times ppi: fifteen metres is roughly 591 inches, giving about 41 300 pixels at 70 ppi, around 59 100 at 100 ppi and around 70 900 at 120 ppi. Anyone who wants to serve 137 ppi for a viewer standing one metre away arrives at roughly 80 900 pixels of width. How to set up this calculation for an individual surface is shown in the article on image resolution for 10 m². And anyone who wants to see in advance how a band works across three walls can walk the surface in the VR gallery before anything is printed.
One long picture or several?
It depends on whether the wall is meant to read as one surface or as a sequence — and on long walls the single picture is usually right. A continuous file carries the gaze along the length and creates exactly the breadth that the pattern prospect stands for (Browning, Ryan & Clancy 2014). Several separate pictures create rhythm instead: at every frame edge the viewing starts again, which in a corridor reads more like a row of framed pictures than like a view. There are situations where that is precisely what is wanted, for instance when the wall is interrupted by doors, light switches or projections anyway — in that case it is more honest to make the interruption part of the design than to let a continuous motif fail on it. Two practical points argue additionally for the continuous file. First, a band only continues cleanly around a corner if it comes from one file; two separately computed images meet at the edge with different perspectives. Second, panels can still be divided during production, whereas separately planned pictures can no longer be joined after the fact.
Why do the edges of very wide motifs look stretched?
Because a rectilinear projection pulls the edge regions further apart as the field of view grows. That projection keeps straight lines straight, and it can only do so by giving the surface more room at the edges than in the middle. Up to roughly 80 to 120° of field the effect stays inconspicuous; beyond that it becomes obvious, and a tree or a face at the edge of the frame appears noticeably wider than the same subject in the centre. On a long wall this stands out particularly, because the viewer walks along the surface and sees the regions one after another from close up. The solution is the cylindrical projection: it keeps vertical lines straight and the horizon stable and damps the edge stretching, at the cost of slightly curving horizontal lines away from the horizon — practically invisible in a landscape, but visible in strict architecture. For special cases, the Mercator and Panini projections offer compromises between the two properties. The overview for this is in the PTGui documentation (as of 2026). What matters is that this choice is made during stitching and not at the printer — on the wall it can no longer be changed.
How much pixel width does a 15 m corridor need?
Between roughly 41 300 and 80 900 pixels, depending on how close people get. The calculation is simple: wall width in inches times the desired ppi figure. Fifteen metres is about 591 inches. At 70 ppi that gives roughly 41 300 pixels, at 100 ppi around 59 100 and at 120 ppi around 70 900. Anyone wanting to meet what the eye resolves at one metre — 137 ppi according to Ashraf, Chapiro and Mantiuk (2025) — ends up at roughly 80 900 pixels of width. Which figure is right is decided by the viewing distance in the finished room: in a corridor people usually pass at one to two metres, and for two metres the same study puts the requirement at 68 ppi. Two notes from practice. First, the pixel count is only half the answer: an upscaled file meets the number but holds no measured detail — and on a surface people walk along and stop at, that shows. Second, 16-bit material is worth it, because a sky running fifteen metres can form visible banding in 8 bit where 16 bit gives a smooth gradient.
Can an ordinary motif be cropped to a panoramic format?
Yes — by cropping, never by distorting. Cropping to a wider aspect ratio is a legitimate and often the best approach, as long as the motif brings enough resolution and the crop holds up in terms of content. This is exactly where a genuine gigapixel capture pays off: from an image with a very high pixel count a 6:1 strip can be taken that still has the width required. Three things need checking. First, the content: a crop removes foreground and sky, and some motifs lose their sense of depth as a result while others only gain it. Second, the resolution after cropping — what counts is the pixel width of the remaining strip alone, not that of the original. Third, the projection: a very wide strip cut out of a rectilinear image carries that image’s edge stretching with it, whereas a cylindrically computed capture stays even across the full width. What never works is squeezing or stretching to the target ratio: the intervention is immediately visible on faces, trunks and building edges, and across several metres of width it is noticed by everyone who walks past.
Conclusion
On long walls it is not the size of the motif that decides the outcome but its geometry. An aspect ratio of 6:1 tolerates neither a squeezed ordinary picture nor four framed panels; it needs a panorama captured for that shape. The projection determines how the edges behave: rectilinear keeps straight lines straight and stretches the edges beyond roughly 80 to 120°, cylindrical holds the horizon and the verticals with damped edge stretching, equirectangular is the 2:1 source format for 360°, and Mercator or Panini are the compromises for very wide surfaces. In terms of content, breadth carries such surfaces because it supplies what a corridor structurally denies — an unimpeded view across distance, processed with fewer fixations than a visually busy surface. On the capture side this calls for 30 to 40 % overlap in a multi-row grid together with clean seam placement and exposure control; on the planning side the rule is: measure the aspect ratio first, then choose the crop, use one continuous file around corners, and calculate pixel width as wall width times ppi — fifteen metres needs around 59 100 pixels at 100 ppi. Gigapixel GmbH — the specialized large-format stock portal: real gigapixel captures from 100 MP, no AI upscaling.
Sources
- Browning, W., Ryan, C. & Clancy, J. (2014): 14 Patterns of Biophilic Design. Terrapin Bright Green.
- Appleton, J. (1975): The Experience of Landscape.
- Clearwater, Y. A. & Coss, R. G. (1991): Functional esthetics to enhance well-being in isolated and confined settings.
- Franěk, M. et al. (2018): Differences in eye movements while viewing images with various levels of restorativeness. Journal of Environmental Psychology.
- Kopf, J., Uyttendaele, M., Deussen, O. & Cohen, M. F. (2007): Capturing and viewing gigapixel images. ACM Transactions on Graphics.
- Chang, H. & Cohen, M. F. (2017): Panning and zooming high-resolution panoramas in virtual reality devices. UIST '17.
- PTGui documentation (as of 2026).
- Ashraf, M., Chapiro, A. & Mantiuk, R. K. (2025): Resolution limit of the eye — how many pixels can we see? Nature Communications.