A display of tiles can have the pitch of the light emitting devices be greater than the threshold distance for a front observer but within the threshold distance for an observer at the middle or back position in the theatre. A blending region can compensate for a pitch that is more than a threshold distance to a front row observer such that the light emitting element spatial density near adjacent tile edges is greater than the light emitting element spatial density at other areas on the tiles. The pitch of the light emitters in the more spatially dense regions can be within the threshold distance for observers in a front row to help distribute slight misalignments between adjacent tiles over an area that is the blending area. The brightness of the light emitting elements in the blending region can be adjusted to compensate for spatial differences or variations in pitch between light emitter sources that are within a threshold distance in a blending region. An observer may be unable to directly resolve any varying gaps between adjacent tile edges or sudden angular transitions or translational transitions between rows or columns of light emitting elements of adjacent tiles of light emitters. Having greater spatial density of light emitters around the edges of tiles can smooth out angular misalignment or translational shifts between adjacent tiles by spreading the alignment differences over a larger area next to the tile edges between the adjacent tiles. By increasing the area of greater spatial light emitting density next to the edge of the tile, a larger blending region between tiles can be created, in effect, to reduce what would otherwise be a noticeable suddenness of differences of angular misalignment or translational transitions between tiles. Increased blending area can result in discontinuities of LED alignment between two tiles being less noticeable or unnoticeable.

Gaps between tile edges can have a blending region in a variety of configurations that can help reduce the visual artifacts associated with mismatches between adjacent tiles by spreading the mismatch over a larger area or over a longer distance in the area that is the blending region. Mismatches can be angular and translational spatial misalignments of rows and columns of LEDs in adjacent tiles. There can also be differences tile visual characteristics, such as in color, brightness, surface texture, angular projection of emitted light, or other visual mismatches between adjacent tiles.

One characteristic of the human vision system (HVS) is the ability to detect sudden spatial discontinuities even though the discontinuities may be low in magnitude. A sudden small change in color, light level, or alignment can have a high spatial frequency that the HVS can recognize. If a region between tiles extends the transition of the small visual difference between adjacent tiles over a larger area, then the spatial frequency of the discontinuity can be reduced and the ability of the HVS to detect the change can be reduced. Blending regions can reduce the spatial frequency of discontinuities. Increasing the blending region area with a greater spatial density is one approach, but alternate approaches are also possible.

One example of a situation in which a small visual difference can occur between adjacent tiles is when flat tiles that have a planar surface are positioned to create rectilinear curved screens that are curved either horizontally or vertically. A curved display can result in adjacent tiles having a slightly different angular position with respect to a viewer. Light emitted from a tile can vary based on the angle from which the tile is being viewed. If adjacent flat surface tiles of the same type of tile are positioned next to each other in which their respective surface is oriented on two different plane orientations, a viewer may view the same tile in two different positions and at two slightly different angles. This slight difference in angle can result in a viewer seeing a first level of light from the first tile and a second light level of light from the second tile where the first and second light levels are different. If the two adjacent tile edges are straight and do not include a blending region, the sudden spatial transition of light level at the adjacent tile edges can be easily perceived by a viewer. A blending region can spread the sudden transition in light level over a larger area to reduce or eliminate the perceived light level difference. If the display is a dome display surface, there can be both a horizontal curvature and a vertical curvature that use adjacent flat surface tiles positioned for causing angular viewing differences along adjacent vertical tile edges and along adjacent horizontal tile edges.

The blending techniques can be applied to a display that has multiple tiles, whether the display is a curved rectilinear surface or the display is a dome surface that has spherical curvature.

A viewer can easily perceive sudden spatial transitions along adjacent tile edges between areas of an orderly arrangement of LEDs, or sudden transitions in visual characteristics of LEDs that differ between areas of LEDs. Gaps between tile edges that are a continuous line can be more recognizable than gaps that are not in a straight line (e.g., zig zag line, sinusoidal line, interleaving profile, etc.) or broken up and distributed over a larger area such as gaps being redistributed over a blending region or area. Breaking gaps up into smaller pieces that are distributed over an area can allow for alternate compensation methods to be applied more effectively. Examples of alternate compensation methods include placing additional LEDs about the sections of gaps and using brightness uniformity algorithms to define the “brightness weighting” of light emitters over the area or blending region about the gap to ensure light uniformity about the gap and in the blending region appear consistent with light uniformity over other sections of the tile. These are different techniques in which the spatial frequency of the discontinuity or disturbance can be reduced by spreading out the discontinuity. Resampling of image content onto the varied spatial light emitting placement can further ensure continuity (i.e., reducing discontinuity) of images over the areas with seams to make them less perceptible. A blending region may not eliminate these differences, but it can provide a more gradual transition between the mismatches and can significantly reduce the perceptual impact of a viewer noticing the image artifacts associated with edge gaps and blending regions between tiles, which may be referred to as seams.

FIG. 2 shows a configuration of one blending example based on the spatial density of light emitting elements being higher at the edges of the tiles and thus at the area about the gap between tiles. Shown are two tiles 202, 204 positioned proximate to each other in which the tiles 202, 204 are aligned vertically so the rows of the LEDs (indicated by dots) on a first tile 202 are aligned with rows of LEDs (indicated by crosses) in a second tile 204. Both tiles 202, 204 have regions 210, 212 with greater spatial density of LEDs in the vertical direction about the vertical tile edges 206, 208 than the vertical spatial density of LEDs in the area away from the vertical tile edges 206, 208. The blending region 214 is formed by the region 210 of greater spatial density of LEDs in the first tile 202 and the region 212 of greater spatial density of LEDs in the second tile 204. In FIG. 3, two tiles 302, 304 of the same size are depicted, but the tiles 302, 304 do not have the vertically increased LED density near the vertical edges 306, 308 of the two tiles 302, 304, for comparison purposes.

In FIGS. 4 and 5, first tiles 402, 502 and second tiles 404, 504 have a translational vertical shift between them with a displacement in an amount d1, as shown in FIG. 4. Thus, a straight horizontal line to be shown by the LEDs in the first tiles 402, 502 and the second tiles 404, 504 has a sudden discontinuity between the respective tiles of each FIG. To smooth out the vertical discontinuity transition of a horizontal line of LEDs from the first tile 402 to the second tile 404 in FIG. 4, the appropriate additional LEDs in the higher density regions 406, 408 can be allowed to emit light to depict a straight-line image as shown in FIG. 6. In FIG. 6, the LEDs that best represent the straight horizontal line have two transitions (e.g., transition “a” 602 and transition “b” 604) that are only ½ d1 vertical tile displacement. If the tiles 402, 404 have a greater density of LEDs about the tile edges, then the displacement between the LEDs can be further divided from d1 than the ½ d1 displacement used in the example. For example, if the vertical density of LEDs is doubled then there can be four equal transitions of a ¼ d1 magnitude spread out in some manner, such as evenly as possible, through the high density LED area (i.e., the blending region) to make the transition even more gradual. FIG. 7 shows a larger single sudden transition (i.e. transition “c” 702) that can occur at the tile edge area when not using increased LED density in the blending area.

The increase of LED density can also be applied to the top edge and the bottom edge of two tiles that are placed one above the other. Instead of the LED spatial density being increased vertically to produce more rows of LEDs by the tile edge for the vertical tile edge of horizontally adjacent tiles, more columns of LEDs can be positioned along the top edge and the bottom edge of stacked tiles, where one tile is positioned above the other tile to form the blending area suitable for compensating horizontal displacements between the vertically stacked tiles.

A calibration process can identify which LEDs in the blending region for each of the tile edges can be activated to provide the best way to have a uniform transition between adjacent tile areas to display an image.

A blending region between adjacent tiles can also have LEDs placed as close as possible to the tile edge in the blending region.

The profile of the tile edge can be different than having one long straight edge between adjacent tiles for the full length of the tile. For example, FIG. 8 has a first tile 802 and an adjacent second tile 804 in a configuration 800. Each of the tiles 802, 804 can have an LED spatial density as described above where there is a greater LED spatial density at the edges of the tiles 802, 804 in the blending region 806 as compared to the LED spatial density away from the tile edge blending region 806. The LEDs are not depicted in FIG. 8, but the configurations described in FIG. 2 can be applied to the various combinations of edge profiles. The blending region 806 in FIG. 8 can be the areas of the tiles 802, 804 that include a tile edge profile. If there is an area near the edge that has higher spatial density of LEDs, this area may also form part of the blending region 806. For example, the blending region 806 can be the space between the dotted lines. In FIG. 8, the two tiles 802, 804 are aligned in that the two tiles 802, 804 butt up next to each other with no gap between the tiles 802, 804. In this situation, the LEDs can have no spatial misalignment. In FIG. 9, two tiles 902, 904 are horizontally displaced by a distance of d2, as shown. The gaps or spaces such as “space a” 908 and “space b” 910 in FIG. 9 that form as a result of the displacement become more distributed over the blending region 906 instead of being the one long gap when the edge was just one straight edge.

FIG. 10 shows another tile edge profile in a blend ingregion 1006 that distributes the gap between two adjacent tiles 1002, 1004. FIG. 11 shows a displacement of d2 in the blending region 1106 between tiles 1102, 1104, but the spaces (i.e., space “a” 1108, space “b” 1110, and space “c” 1112) formed when the tiles are separated by a distance d2 are distributed over a larger blending region 1106. This distribution of a gap between tiles 1102, 1104 that is broken up into segments and distributed spatially throughout the blending region 1106 can reduce the perceived visual artifact associated with tile seams.

LEDs from two adjacent tiles can be spatially interleaved. The greater the distance of the finger length dp shown in FIG. 8, the greater the area of the spatial interleaving between the two adjacent tiles 802, 804 in the blending region 806 of LEDs. If there are differences, such as color balance differences or light level differences caused by different viewing angles to occur between tiles 802, 804, which may be in a curved display, these differences can become spatially interleaved or spatially intermixed by the tile edge profile in the blending region 806. Tile edges profiled that are interlocked as shown in FIG. 8 can create a blending region 806 with light emitters on the interleaving protrusions that are spatially positioned to create an area that has emitted light that alternates over the area between light coming from a first tile 802 with light coming from a second tile 804. Increasing the number of interleaving portions or making the interleaving portions longer can spread out visual differences between adjacent tiles and appear less noticeable to a viewer. Other differences can include slight differences in tile texture or tile surface look, or the light dispersion angle of the LEDs themselves on the two tiles 802, 804 being slightly different. Color balance of light emitters between tiles 802, 804 can become different when a tile is replaced that is next to a tile that has remained. The color differences can be related to differences in light emitters used on the tile being different or differences related to length of time the light emitters have been powered up causing the light emitters to age and in turn develop a shift in their spectral composition.

Increasing the number of interleaving features along a tile edge can result in an increased number of smaller gap areas that become distributed over the blending region. The increased number of distributed gaps in the blending region can further reduce a viewer's perception of visual artifacts associated with the space between adjacent tiles, in particular if all dimensions of the gap are within the threshold distance dimension.

From the above example, the spatial frequency of the discontinuity can be reduced by the shape of the edge profile, the number of interleaving protrusions, and by the length of the interleaving protrusions.

In some examples, the space width between finger protrusions can be larger than the width of the protruding finger for both adjacent tile edges, as shown in FIG. 12, where Wf is the finger width and Wg is the width of the gap in the blend region 1206 between a first tile 1202 and a second tile 1204. This can allow for easier fitting of tiles 1202, 1204 with less spatial precision when interconnecting many tiles to create a large display area. The increased gap width between fingers can be targeted to be within a desired threshold distance. For example, a tile edge that has finger widths and gap widths between fingers of the same dimension the widths can be altered within a threshold distance to reduce the precision with which two tiles 1202, 1204 need to be mounted next to each other. If a threshold distance is 2 mm, than the width of the gap between fingers can be increased by an amount that is the threshold distance or within the threshold distance. When adjacent tiles 1202, 1204 are mounted, the gap between the finger edges can vary from tile to tile in the range of 0 mm to 2 mm. Gap variations within the threshold distance may not be perceptible by a viewer positioned at the minimum distance for the threshold distance determined. Light emitting elements can be aligned within a threshold distance along one axis, such as the vertical axis, and can be aligned with a greater displacement in another axis, such as the horizontal axis. For example, a displacement that is 1 pitch in distance, where the pitch is significantly greater than the threshold distance, a greater spatial freedom to align tiles along a horizontal axis can be achieved, while keeping closer gap spacing along the vertical axis or greater spatial freedom along the vertical axis and closer gap spacing along the horizontal axis.

FIGS. 13-15 show another tile edge profile that takes further advantage of distributing the gap area between adjacent tile edges resulting from tile displacement between two adjacent tiles that can accommodate large horizontal or vertical tile displacements. In this example, any discontinuity between adjacent tiles can have an angular component with respect to the vertical and horizontal alignment grid of the light emitting elements in the two adjacent tiles. When used with higher spatial density of LEDs at the edges and calibration techniques, maximizing suppression of gap artifacts or other artifacts associated with the blending region that would otherwise be more noticeably visible can be suppressed.

Complementary first and second serrated sawtooth tile edge portions 1302, 1304 in FIG. 13 are fitted together with no gap. The blending region 1306 is indicated by the space between the two dashed lines and can be an area with increased LED spatial density. In FIG. 14, the first tile 1402 and the second tile 1404 are horizontally displaced by an amount d3. As a result of the displacement in the amount of d3, there is also a vertical gap created in the amount of d4 in the blending region 1406. Depending on the length of the triangular tooth of the serrated edge, the vertical gap of d4 can be dimensionally very small compared to the horizontal displacement of d3. Effectively, the horizontal gap is distributed over a long but narrow gap. The edges of the adjacent tiles can have high spatial density of LEDs, which can be used to aid with creating light uniformity around the gap area to compensate for LEDs that would have otherwise been positioned in the gap area. For smaller displacements of d3, the gap d4 may have a dimension that is less than a targeted visual acuity limit, such as a threshold distance. For example, if a targeted visual acuity dimension is 2 mm, the dimension d4 can be less than 2 mm for a horizontal tile displacement d3, where d3 may be as much as or more than an average pitch between two LEDs in the area outside of the high spatial density LED area. The resulting d3 displacement of this amount can result in the image being warped to match the newly shifted light emitting positions over the area of the shifted tiles. Resampling or warping the input image data can be necessary to prevent image distortion caused by display light emitters that no longer correspond with each other between shifted tiles.

Edges that are serrated or contoured can facilitate alignment between tiles by forming a natural interlocking feature. This can facilitate making alignment easier during the assembling of larger display areas with multiple tiles.

The serrated edges are not limited in shape to what has been disclosed. For example, the edges can have a sinusoidal profile, a trapezoidal profile, random shaped edge or other shape that is more complex in profile than discussed or a combination of different edge profiles.

Another variation related to the blending region relates to the arrangement of the LEDs in the high spatial density region. For example, the edge profile may benefit from LEDs positioned in a line profile that is at right angles to the edge profile that is a different orientation than the LEDs positioned in the area outside the blending region within the tile. Alternatively, the LEDs can be positioned in a random manner in the blending region and in an orderly manner in the area that is within the tile but outside of the blending region. LEDs can also be positioned in the blending region as close as possible to the tile edge for a portion of the tile edge, as shown in FIG. 15. The light emitters on the first tile 1502 are shown as dots and the light emitters on the second tile 1504 are shown as crosses. At the serrated edge, there are a number of light emitters positioned along the edge as close as possible to the edge. Other areas of the blending region 1506 show the light emitters spaced out more evenly. Outside of the blending region 1506, the light emitters are spaced out at least with twice the pitch of the light emitters in the blending region 1506.

Noticeability of seams between tiles can be further reduced by having areas in the tile make gaps or blending regions (e.g., seams) that are regular in profile look less regular in profile. For example, a seam pattern on a display that has a rectangular grid appearance can be made to appear less like a rectangular grid by adding angular seams to form triangular seams, which can be considered in making the adjacent tile seam pattern stand out less. For example, FIG. 21 shows four tiles 2102, 2104, 2106, 2108 positioned together. Each of the tiles 2102, 2104, 2106, 2108 has an area near the edge shown with single crosshatched edges 2112, 2114, 2116, 2118. The edges 2112, 2114, 2116, 2118 can create a blending region 2120 that has a higher density of light emitting elements than the area outside of the blending region 2120, along with complementary edge profiles with other tiles 2102, 2104, 2106, 2108. An angular seam pattern 2122, 2124, 2126, 2128 is included within each of the tiles 2102, 2104, 2106, 2108. The angular seam patterns 2122, 2124, 2126, 2128 can have the same density and pattern of light emitting elements as the blending region 2120 between the tiles 2102, 2104, 2106, 2108 but without the complementary edge profile of the joint between the tiles 2102, 2104, 2106, 2108 and a different density of light emitting elements as other areas of the tiles 2102, 2104, 2106, 2108. The angular seam patterns 2122, 2124, 2126, 2128 can, in effect, form additional blending regions that emulate the blended regions 2120 between the tiles 2102, 2104, 2106, 2108 to create a camouflage effect with the blending region 2120. To emulate the blended region 2120, a characteristic of the blending region 2120 can be emulated. For example, the light emitting density can be one characteristic in which the light emitting density of the additional blending has a next effect of being equivalent as the light emitting density of the blended region 2120 between tiles 2102, 2104, 2106, 2108. Another example can include creating a spacing between light emitting elements in the additional blended region that characterizes a physical gap of the space between tiles 2102, 2104, 2106, 2108 in the blended region. Although the angular seam patterns 2122, 2124, 2126, 2128 are shown as angular in FIG. 21, additional seam patterns according to other examples can be a different shape or pattern. For example a rectangular section at the center of the tile with greater density of light emitters or an additional pattern that has horizontal or vertical seams that are parallel to the existing seams or blending regions between tiles. This solution can be more effective in combination with other suppression techniques associated with gaps between tile edges and blending regions to further improve display quality.

Dark or black image scenes, such as outer space scenes, can allow differences in adjacent tiles to be viewable. For example, if light emitters in different tiles have a different minimum light level or the reflective surface in the blending region that is different than the reflective surface of the rest of the tile area, visible artifacts may be viewable between adjacent tiles. Having increased spatial density of light emitters in the blending region can allow for increased accuracy of spatial brightness distribution control in the blended region. The increased accuracy of spatial brightness distribution in the blending region can allow for smoother transition between different minimum light levels of adjacent tiles.

The space between light emitters can be a black surface with a diffuse surface that may reduce specular reflections so that the transition between tile surface textures is minimized, such as transitions originating from tiles being mounted with angular misalignments with respect to the plane of the display surface and hence emit narrow angle specular reflections in different directions. The tile surface can also have different profiles of black or low gain to camouflage other discontinuities, such as the differences in the reflective surface of areas, with greater density of light emitter elements (blending regions) compared to other areas that have a lower density of light emitters (i.e., areas that are not blending regions). The reflective surface area of the tiles, between light emitting elements, can be tailored such that there is reflective uniformity across the surfaces of adjacent tiles. A blending region can be between adjacent tiles to minimize visual artifacts perceived by the HVS using one or a combination of different methods (some of which are listed below) where the visual artifacts that can be caused by discontinuities of small differences in light emitting characteristics (e.g., light level, color, tile surface texture, and tile surface reflectance) between adjacent tiles.

By modifying light emitter positions with respect to the tile edge:

- Adjusting light emissions of adjacent tile light emitters that have a pitch that is less than a minimum threshold distance.
- Increasing the spatial density of light emitters along adjacent tile edges.
- Increasing the spatial density of light emitters in the blending region that is an area next to the tile edge.
- Improving brightness uniformity across adjacent tile edges and blending regions by calibrating adjacent tiles with brightness uniformity algorithms.

By modifying adjacent tile edges:

- Configuring the adjacent tile edges to have complementary protrusions with a particular shape (e.g., sawtooth profile, sinusoidal profile, interleaving block profile, etc.).
- Increasing the width of the blending area by increasing the length a tile edge protrusion goes into an adjacent tile area.
- Increasing the density or frequency of complementary protrusions over a given length of adjacent tile edge.
- Orienting the gaps between adjacent tiles so that the tile edge lines form an angle that is not parallel or 90 degrees with respect to any of the rows and columns of light emitting elements on the tile.
- Creating additional blending regions within the tile area but outside of the blending regions at the tile edges where the additional blending regions create a camouflage effect with the tile edge blending regions. For example, the camouflaging regions can have a spatial density of light emitting elements that is different than the spatial density of light emitting elements in other areas of the tile.
- By aligning image content intended for a spatially defined array of light emitting elements with light emitting elements where a portion is not a spatially defined array of light emitting elements.
- Decreasing the discontinuity of image content between adjacent tiles and the blending region by resampling received image content to align to the actual position of the light emitting elements on adjacent tiles and the blending region.

A display system with tiles can be calibrated to ensure visual uniformity (e.g., with respect to brightness and color) of the light emitters across the tiles with blending regions and to ensure the image content intended for uniform light emitting grid arrays corresponds to light emitting elements in a display system that have regions and tiles that do not correspond in alignment with a uniform grid array of light emitters.

A calibration process providing data for pre-processing images may be performed to reduce or eliminate seam visibility. The calibration process can be performed, for example, on installation of the LED display, replacement or maintenance of one or more tiles, or after the LED display is physically influenced, such as by objects contacting the LED display. FIGS. 16 and 17 are flowcharts of examples of processes for calibrating for an LED display. The examples are described with respect to a monochrome display for simplicity, but the processes can also be applied to displaying color on an LED display.

In block 1602 in FIG. 16, tile positions are recorded. For example, angular misalignment orientations of tiles of an LED display, as mounted, can be captured via a camera that is viewing the display tiles or by other means obtained via manually measuring distances via angular measurement tools. The position of the tiles may be center positions of tiles defined in a display area. When using a camera, at least one recorded image of the display's surface can be captured while an alignment image is being displayed on the LED display. The alignment image may include a tracking pattern and a tile identifier encoded as a pixel pattern for each tile in the display. A pixel being one light emitting element (LED) or a cluster of light emitters that represent one light emitting element. The tracking pattern may include at least one feature designed for detecting the center and the orientation of a tile. An example of an identifier is a bar code, such as a QR (Quick Response) code. In another example, recorded images that each includes an alignment image with a tracking pattern and a tile identifier are captured for a subset of the tiles in the LED display.

In block 1604, the tile pixel layout is read by a processing device. For example, the processing device may receive the tile pixel layout from data storage. The pixel layout coordinates may be identical as another tile or all tiles for the tiles. The pixel layout data may include X and Y coordinate positions of each pixel on a tile relative to a center position of the tile.

In block 1606, the processor device calculates a table “P” in which table “P” represents the position of all pixels in an area that is the LED display or a portion of the display. The pixel positions in the table may be calculated from the data of recorded positions and angular orientations of tiles in the LED display obtained in block 1602 and from the tile pixel layout data obtained in block 1604. The calculation of positions of pixels in the display may include performing a translation of X and Y positions of pixels on a tile relative to a center position of the tile in the display area. It may further include rotating the X and Y positions around a center position of a tile in the display area.

In block 1608, the processor device calculates a table “D” in which table “D” represents the relative pixel densities at positions of pixels in the display. Calculating table D can include using a point density function (i.e., pixels may be assumed to be point light sources). The point density function may simulate the visual acuity limit (e.g., threshold distance) of a viewer located at a distance from the display (Dobs) when a viewer can no longer resolve spatial differences between display pixels that is a minimum distance of Dobsmin that the display is designed for. An example of the point density function is a Gaussian averaging function. The Gaussian averaging function may have a standard deviation equal to tan(1/60)×Dobsmin. An example of Dobsmin is 7 m, in which case the standard deviation is tan(1/60)×7 m=2.0 mm.

Additionally or alternatively, the standard deviation may be adjustable and a technician may observe a uniformly illuminated image displayed after the calibration procedure is completed, adjust the standard deviation based on how uniform the image looks, perform another calibration procedure, re-adjust the standard, observe the result and repeat this in an iterative process until the image viewed is satisfactory. Calculating table D may further include identifying the pixel in the display for which the pixel density is smallest and dividing the pixels in the display with the smallest pixel density, thereby normalizing the table D by making the pixel density values in table D relative to the lowest density on the display so that the lowest density value in the Table D is 1 and all other density values are higher. Identifying the pixel with the smallest value may include selecting any pixel not near the edge of a tile. Alternatively, calculating table D may include displaying an image where pixels have essentially uniform brightness and performing a manual adjustment of pixels along at least one tile edge of at least one tile while observing the edge from a distance equal to or greater than Dobsmin until the tile boundary appears less visible. Manual adjustment may include adjusting all pixels along at least one edge at the same time and may further include adjusting a brightness distribution that may be a gradient function. The gradient function may affect pixels at an edge with a gradient direction along the edge.

FIG. 17 is a flow chart of a process for modifying an image based on calibration information according to one example. In block 1702, a digital image is received from, for example, a storage medium. The data of the digital image may be pre-configured for display on an LED display with LEDs that are uniformly distributed to be equidistant from each other.

In block 1704, the digital image is resampled using the positions of pixels in the display stored in Table P calculated in FIG. 16. The resampled image may be calculated using a bicubic, bilinear, Lanczos or Mitchell resampling.

In block 1706, an output image is calculated by dividing the resampled image by the pixel density values in Table D calculated in FIG. 16. Each pixel in the resampled image may be divided by the corresponding pixel density value in Table D, thereby providing an output image that, when displayed by the tiles of the LED display, can result in less visibility of seams between the tiles than if the provided digital image data intended for displays with uniform distribution of the LEDs (pixels) over a display area was displayed without the pre-processing 1700.

FIG. 18 schematically shows four adjacent tiles 1802, 1804, 1806, 1808 according to one example. The tiles 1802-1808 may be configured with a pixel pattern supporting 3D with passive eyewear, such as polarized glasses or spectral filter glasses. Each tile includes a first set of LEDs 1810 (represented by circles) having a first set of light characteristics, and a second set of LEDs 1812 (represented by crosses) having a second set of light characteristics. An observer may wear a pair of glasses that allows light from the first set of LEDs 1810 to be transmitted to a first eye of the observer and light from the second set of LEDs 1812 to be essentially blocked from reaching the first eye of the observer. And the glasses can allow light from the second set of LEDs 1812 to be transmitted to a second eye of the observer and light from the first set of LEDs 1810 to be essentially blocked from reaching the second eye of the observer. A first perspective image may be observed by the first eye of the observer and a second perspective image may be observed by the second eye of the observer such that the observer perceives the images stereoscopically.

Each LED may have any dimension. In one example, the dimensions of each LED are 1 mm×1 mm, the horizontal spacing between adjacent edges of the LEDs on the same tile may be 1 mm, and the vertical spacing between adjacent edges of the LEDs on the same tile may be essentially 0 mm. The horizontal distances between the LED centers may be 2 mm and the vertical distances between the LED centers can be 1 mm (e.g., the LEDs are arranged in a matrix pattern having a horizontal pitch of 2 mm and a vertical pitch of 1 mm). An LED located in a row of the matrix pattern that is closest to an edge of a tile and an LED located in a column of the matrix pattern that closest to an edge of a tile may be referred to as an “edge LED” (1814, 1816). The tiles 1802, 1804 may be mounted with target mounting positions so that the distance between a first edge LED 1814 on a first edge and a second edge LED 1816 on a second edge facing the first edge is 0.5 mm, and the mounting tolerance may be 0.25 mm. The maximum distance across a tile boundary between two adjacent edge LEDs can be 0.5 mm+2×0.25 mm=1 mm.

In FIG. 18, the tiles 1802-1808 are shown as mounted so that the spacing between tiles 1802-1808 is the maximum distance such that the spacing is equal to the spacing between pixels of the same tile, and to the width and the length of a pixel.

In the vertical direction, LEDs may be arranged in a pattern, alternating between the first set of LEDs 1810 and the second set of LEDs 1812. The first set of LEDs 1810 can be arranged in a first-eye matrix pattern with a horizontal and vertical pitch of, for example, 2 mm, and the first eye of the observer can view the first-eye matrix pattern. The second set of LEDs