An autostereoscopic screen that is considered to be of photographic quality

Abstract

The present invention relates to a screen (10) for the autostereoscopic display, considered to be of photographic quality, of autostereoscopic images with M x N viewpoints, where N is an odd integer and M is an integer selected from 1, 2, 3 and 6, said screen comprising a panel (20) of pixels arranged by rows and columns, each pixel (26) consisting of a plurality of P active sub-pixels of different colors, and a regular array of cylindrical lenses (30) inclined at an angle equal to arctan(1 / 12) with respect to the columns of the screen, said array comprising , with a pitch of N / 2 pixels wide so that even and odd lenses (30) of the array can be distinguished by the different distribution of the active zones (26) and black zones (24) of the pixels located under the lenses, and so that each of the M×N viewpoints can be decoded, distributed in N vertical blocks of 6 pixels high, consisting of M viewpoints by 12 rows when the panel's subpixels are horizontal, and in P×N vertical blocks of 6 / P subpixels high, consisting of M / P viewpoints by 12 / P rows when the panel's subpixels are vertical.

Description

[Technical Field]

[0001] 1.Technical Field The present invention relates to an autostereoscopic screen and a method for displaying autostereoscopic images on such an autostereoscopic screen. [Background technology]

[0002] 2.Technical background Autostereoscopy is a technique that allows images to be displayed in three dimensions without the need for the viewer to wear special glasses. This technique is known per se, in particular from patents WO 2006 / 024764, WO 2014 / 041504, WO 2013 / 140363, WO 2014 / 016768, WO 2019 / 063897, WO 2019 / 207235, WO 2022 / 175053 in the name of the applicant. An autostereoscopic image is composed of a plurality of alternating strips of elementary images, each corresponding to a view of the same object or scene from a different viewpoint. A selection device, typically consisting of an array of cylindrical lenticules or a parallax barrier, is placed in front of the display screen so as to project a pair of elementary images, corresponding to two different viewpoints of the scene, respectively, towards the observer's eyes, thereby enabling the illusion of depth to be created in the observer's brain.

[0003] The applicant has already proposed an autostereoscopic screen for displaying an image with N viewpoints, comprising a matrix of pixels arranged by rows and columns, each pixel being composed of a plurality of sub-pixels of different colors. Furthermore, the screen is covered with an array of identical cylindrical lenticules, each lenticule having a focal length configured to reflect light rays coming from the screen to infinity. The pitch of the array of cylindrical lenticules is precisely calculated so that an observer positioned at a predetermined distance from the screen can sequentially view the images due to the magnifying glass effect of the lenticule array.

[0004] This magnifying glass effect occurs because a lens positioned at the correct distance (i.e., focal length) magnifies sub-pixels that are on its optical axis and aligned with the pupil of the observer's eye. If magnified by a lens by a factor N, a sub-pixel viewed through the lens will appear N times larger than it actually is, thereby occluding the N-1 other sub-pixels that are not aligned in this way and becoming invisible to the eye receiving light through the lens.

[0005] Even the principle of lenticular arrays cannot provide a viewer with continuous parallax for a scene presented through a 180° viewing angle. Instead, the viewing angle is subdivided into strips called "lobes," and within each lobe, the parallax is discretized. Within each lobe, N viewpoints can be viewed sequentially. When one lobe is deviated from the other, the same sequence of N viewpoints reappears in the next lobe, and this phenomenon is repeated across the entire viewing area of ​​the screen. The angular width of the lobes is defined by both the pitch of the lenticular array and its focal length.

[0006] Autostereoscopic screens require the subdivision of lobes into discrete, highly visible subparts, and the width of the individual subparts within each lobe cannot exceed the mean interpupillary distance of 6.5 cm. Therefore, one of the keys to the performance of the autostereoscopic screen currently proposed by the applicant is the control of the resolving power of the optical components developed specifically for each pixel panel. In the case of printed autostereoscopic photographs (not autostereoscopic screens), it is possible to record the maximum number of printed points or dots that are adjacent to each other but still sufficiently distinguishable on a medium with good dimensional stability. With currently used printing technology, the applicant has succeeded in printing 100 dots per millimeter, and in some cases even more, which corresponds to the visible size of dots of 10 microns or less.

[0007] As an observer moves in front of such a photograph, each element in the scene appears to move continuously relative to one another, and in perfect accordance with the original physical three-dimensional structure that would be perceived if the object were viewed in real life from the same distance. Because the two eyes simultaneously see two different perspectives of the scene, the brain is able to reconstruct a perfectly natural sense of three-dimensionality. As long as these screens consist of a matrix of pixels of varying resolution, arranged in orthogonal rows and columns, separated from one another by opaque black surrounds of varying thickness depending on the panel, forming a grid-like pattern around the active zones of the pixels, it is in principle impossible to achieve autostereoscopic photographic quality on the screens. The surface area ratio between the black surrounds and the colored active sub-pixels not only affects the brightness and contrast of the screen, but also its autostereoscopic resolving power.

[0008] Throughout this specification, an autostereoscopic screen is considered to be of photographic quality if it is able to eliminate the discrete jumps that occur when switching from one stereoscopic pair to another in image regions where the disparity is greatest. In other words, "photographic quality" means a continuity of three-dimensionality within the lobe, which is thought to be difficult to achieve in principle with an autostereoscopic screen that does not have a tracking system.

[0009] In addition to resolution problems, problems related to moiré also arise. In effect, the lenticular array magnifies both the active zones of the pixel panel and the black surround that surrounds them. As a result, some sub-parts within the viewing lobe may appear darker or brighter than others. These brightness variations are typically perceived at the transitions between adjacent viewpoints. The moiré effect on light intensity varies depending on the specific structure of the panel and the orientation of the screen. Summary of the Invention [Problem to be solved by the invention]

[0010] The inventors have attempted to overcome the technical preconception held by those skilled in the art that it is in principle impossible to obtain photographic quality autostereoscopic images (characterized by the continuity of three-dimensionality within the lobes) on a screen with a pixelated panel.

[0011] 3. Purpose of the invention The present invention aims to provide an autostereoscopic screen that is considered to be of photographic quality. In particular, the invention aims to provide an autostereoscopic screen that makes it possible to obtain a continuum of disparities without a perceptible discretization of the viewpoint. Accordingly, the present invention, in at least one embodiment, aims to provide a screen that provides a continuous parallax sensation within each lobe. Another object of the present invention is to provide an autostereoscopic screen that does not exhibit noticeable moire or ghost images. Furthermore, the present invention aims to provide a screen that is considered to be of photographic quality, regardless of the structure of the sub-pixels (vertical or horizontal sub-pixels) in the pixel panel of the screen. Finally, the present invention aims to provide a method for displaying an autostereoscopic image on a screen according to the present invention. [Means for solving the problem]

[0012] 4. Description of the invention To achieve this, the present invention relates to an autostereoscopic display screen of considered photographic quality for autostereoscopic images having M×N viewpoints ordered from 1 to M×N, where N is an odd integer strictly greater than 1 and M is an integer selected from 1, 2, 3 and 6. A screen according to the invention comprises a panel of pixels arranged in rows and columns, each pixel consisting of a number P of active sub-pixels of different colours and opaque inactive zones forming part of the black inter-row and / or inter-column areas of the panel.

[0013] The screen according to the invention further comprises a regular array of cylindrical lenses mounted on the pixel panel, each lens being inclined at an angle equal to arctan(1 / 12) with respect to the columns of the screen. Furthermore, the array has a pitch of N / 2 pixels wide, making it possible to distinguish between even and odd lenses in the regular array by distinguishing the distribution of the active zones of the pixels located under each even and odd lens and the black zones of the panel. Furthermore, this array makes it possible to decode each of the M×N viewpoints, which are distributed in N vertical blocks of 6 pixels high, consisting of M viewpoints (6 / M pixels per viewpoint) if the panel's subpixels are horizontal, with each block repeating periodically every 12 rows on the lens axis, and in P×N vertical blocks of 6 / P subpixels high, consisting of M / P viewpoints (6 / M subpixels per viewpoint) if the panel's subpixels are vertical, with each block repeating periodically every 12 / P rows on the lens axis.

[0014] In other words, the screen has the feature that, on a panel extending below the lenses, each of the M×N viewpoints can be coded by 6 / M different vertical pixels within N vertical blocks of 6 pixels high that are repeated every 12 rows under each lens if the panel's subpixels are horizontal, or by 6 / M different vertical subpixels within P×N vertical blocks of 6 / P subpixels high that are repeated every 12 / P rows under each lens if the panel's subpixels are vertical. The array is configured to decode, by a magnifying glass effect, each of the MxN viewpoints coded according to the principles described above. A screen according to the present invention also comprises an array of cylindrical lenticules, each lenticule being inclined at an angle of arctan(1 / 12), or approximately 4.76°, relative to the columns of the screen. Here, the "arctan" function refers to arctangent. The arctangent of a real number is the value of the oriented angle whose tangent is equal to the real number.

[0015] The screen is also characterized by having a lenticular array configured to distribute the viewpoint over 12 consecutive rows under each lens if the subpixels are horizontal, or over 12 / P consecutive rows if the subpixels are vertical. In the following, for the sake of simplicity, the "horizontal mode" or "horizontal version" of a screen refers to a screen composed of subpixels arranged horizontally, and the "vertical mode" or "vertical version" refers to a screen composed of subpixels arranged vertically. It should be understood that these terms do not imply a portrait or landscape orientation of the screen. The horizontal and vertical modes refer only to the arrangement of the subpixels, not to the vertical or horizontal orientation of the entire screen.

[0016] In the following, for the sake of simplicity, unless otherwise specified, "pixel" refers to the content-addressable active zone of a pixel, and "black space" refers to the opaque inactive zone of a pixel that forms the inter-row and / or inter-column areas of the screen. In horizontal mode, with a tilt angle of 4.76°, each lens of the lenticular array passes through a group of basic vertical blocks of 6 pixels, then through black space before arriving at a new group of basic blocks of 6 pixels. Note that all pixel matrices have black opaque zones between each row of active pixels, but these zones are part of the pixel's composition and, depending on their proportion within the pixel, can provide good resolution.

[0017] Since N is an odd integer, the regular array of screens has a non-integer pitch of N / 2, which at least partially offsets the distribution of light zones (active pixel zones) and black zones (opaque pixel zones) on the screen, making it possible to distinguish between even-numbered and odd-numbered lenses. In other words, if the lenses in the regular array are numbered sequentially from the left edge of the screen to the right edge, pixel panels located under even-numbered lenses share a common distribution of active pixel zones and black zones on the screen, and pixel panels located under odd-numbered lenses likewise share a common distribution of active pixel zones and black zones on the screen, but the distributions corresponding to the even-numbered lenses are different from the distributions corresponding to the odd-numbered lenses. That is, when a portion of a pixel panel arranged under an even-numbered lens and a portion of a pixel panel arranged under an odd-numbered lens are overlapped, it is confirmed that at least a portion of the black zone in the panel under the even-numbered lens faces at least a portion of the active zone in the panel under the odd-numbered lens, and that at least a portion of the active zone in the panel under the even-numbered lens faces at least a portion of the black zone in the panel under the odd-numbered lens.

[0018] For an array of 11xM viewpoints (N equals 11), the lenses have a pitch of 5.5 pixels wide (in the unsimplified definition of a pixel). For an array of 7xM viewpoints (N equals 7), the lenses have a pitch of 3.5 pixels wide. The array of screens according to the invention makes it possible to use a magnifying glass effect to decode each of the M x N viewpoints, distributed in N vertical blocks of 6 pixels high, consisting of M viewpoints (6 / M pixels per viewpoint), with each block repeating periodically on the lens axis every 12 rows, if the panel sub-pixels are horizontal, or in P x N vertical blocks of 6 / P sub-pixels high, consisting of M / P viewpoints (6 / M sub-pixels per viewpoint), with each block repeating periodically on the lens axis every 12 / P rows, if the panel sub-pixels are vertical.

[0019] In the horizontal version, the advantage of having vertical blocks of 6 pixels (also called basic blocks) is that each of these pixels can be assigned the same viewpoint or different viewpoints. If the displayed scene has a shallow depth, there is little problem in displaying the same viewpoint, called the main viewpoint, for six consecutive pixels in a basic block. On the other hand, large disparities degrade the perceived image quality. When the array is precisely focused, content with large depths or large forward projections clearly exhibits a discretization of the content, which detracts from the desired photographic visual experience (continuous parallax). When high-contrast details in the image move horizontally by more than two or three lens widths with a change in viewpoint, the discretization becomes excessive, making it necessary to increase the number of viewpoints. While this increase in viewpoints sacrifices the initial resolution of the displayed primary viewpoint, it clearly improves the perceived image quality.

[0020] By assigning slightly different viewpoints to each of the six pixels in each basic block, discretization is eliminated even in scenes with high parallax. The screen according to the present invention therefore allows the display of so-called intermediate viewpoints between the N main viewpoints of the image to be displayed. These intermediate viewpoints are distributed within each basic block, and are therefore slightly offset from each other between the main viewpoints. In this way, the continuity between successive main viewpoints creates a photographic effect of creating a sense of depth in the distant background and in the foreground, without causing any noticeable discretization. In other words, the screen according to the invention is remarkable in that it uses groups of six pixels high per main pixel (which forms part of the group) to encode slightly different viewpoints, filling a discrete series of main viewpoints as a continuum, making it possible to achieve a display approaching the quality of an autostereoscopic photographic image.

[0021] This new approach, referred to throughout this specification as the "tiling effect," achieves the goal of reproducing on screen the parallax continuity found in printed photographic images. In particular, in 3D images, the parallax between viewpoints becomes more pronounced as one moves away from the plane of the screen (also called the plane of collimation). The overlap between common viewpoints, called "crosstalk," exists uniformly across the entire screen and affects the entire displayed image, but is only evident in areas of the image where the parallax between viewpoints is large. In particular, the inventors have discovered that blurring in photography not only potentially contains information related to three-dimensionality, but is also a phenomenon that visually represents depth of field. Therefore, contrary to conventional wisdom in the art, the inventors have attempted to realize depth or prominence in an image not through discretization between viewpoints, but through gradual fading (blurring). The overall impression aimed for is an image with depth and continuity, where the fading between viewpoints becomes more pronounced as objects move away from the collimation plane (considered the focal plane). Therefore, inter-viewpoint discretization becomes apparent when distinguishable details in the image are repeated multiple times. When the disparity in a scene is large and the number of viewpoints is insufficient to distribute it smoothly among the viewpoints, high-contrast objects far from the collimation plane are often perceived as ghost images, degrading image quality.

[0022] Meanwhile, depth details across multiple viewpoints overlap simultaneously, with diminishing intensity and contrast as they move away from the central viewpoint, allowing them to be visually reconstructed as a single object, even if the image is somewhat blurred. The screen according to the invention allows the number of displayed viewpoints to be adjusted (by changing the value of M) depending on the parallax in the image, both in depth and in projection, with ghosting of details at the boundaries of the viewing volume being a determining factor. The novel screen provides the observer with continuous parallax within the lobes, without a clear viewpoint discretization, but rather with a gradual fading of details as the limits of the viewing volume are approached. This approach contrasts with previous methods proposed, which believed that aiming for complete discretization of viewpoints was the way to achieve high-quality screens.

[0023] Advantageously, according to the invention, each pixel of the pixel panel is made up of a number P of horizontal sub-pixels of different colours arranged next to each other in a column direction, such that each of the M×N viewpoints can be coded using 6 / M different vertical pixels repeated every 12 rows under each lens, where M is equal to 1, 2, 3 or 6. In other words, the screen according to the invention makes it possible, in a horizontal version, to display, as needed: - N main viewpoints (where M is equal to 1) each coded with 6 different consecutive vertical pixels, - 2xN viewpoints (where M is equal to 2), each coded with 3 different vertical pixels, - 3xN viewpoints (where M is equal to 3) each coded with two different vertical pixels, - 6xN viewpoints (where M is equal to 6) each coded with a single pixel. Advantageously, according to the invention, each pixel of said pixel panel is made up of a number P of vertical sub-pixels of different colors juxtaposed to one another in the row direction, making it possible to code each of M × N viewpoints using 6 / M different vertical sub-pixels repeated every 12 / P rows under each lens, M being equal to 3 or 6. In other words, a screen according to the invention makes it possible to display, in vertical version, as needed: - 3xN viewpoints (where M is equal to 3), each coded with two different vertical sub-pixels, - 6xN viewpoints (where M is equal to 6) each coded with a single subpixel. Furthermore, the screen according to the present invention has narrower lobes than screens previously proposed by the applicant. To achieve such lobes, the optical components have a relatively long focal length, narrowing the lobes as much as possible without compromising the convenience of observer positioning. This improves resolution and continuity in the viewing angle. The narrower lobes increase the transition zone for the viewpoint, but also divide the viewing space in front of the screen into thinner strips, allowing multiple observers to be positioned simultaneously and very comfortably at different distances.

[0024] Therefore, the present invention contradicts the preconception of those skilled in the art that wide lobe widths are essential for the quality of an autostereoscopic screen. Contrary to assumptions, the perceived image quality when all six pixels in a basic block belong to the same high-resolution viewpoint is comparable to that achieved when lower-resolution intermediate viewpoints are "tiled" within the basic block, rather than distributing the main viewpoint. In other words, a display with a theoretically very high resolution is not necessarily superior to a display in which intermediate viewpoints partially overlap due to the "tiling" effect. Meanwhile, depth details that would otherwise be discretized when resolution is optimized are spread as a blur with a circular appearance, similar to the depth-of-field effect. Because the intermediate viewpoints are spread simultaneously in both the horizontal and vertical directions, a nearly isotropic improvement in image quality is possible in the resulting image. Humans are accustomed to depth-of-field blur, and perceiving blur in the farthest planes of a 3D image is a physiologically unavoidable phenomenon: the brain naturally accepts this blur as a result of depth distance, without adversely affecting the perception of overall image sharpness.

[0025] According to one variant of the invention, the screen comprises a pixel panel which, when arranged in horizontal mode (or vertical mode), has a number of columns (or rows) separated from one another by black inter-column (or inter-row) areas, the width (or height) of which is close to the width (or height) of the active zones of the pixels of the panel. These black areas form a black surround, the area of ​​which is close to the area of ​​the active display surface of the screen, allowing for very good contrast levels to be obtained.

[0026] Furthermore, according to this variant, when horizontal subpixels are used, the active zones and dark zones in the panel portions located under the even-numbered lenses and odd-numbered lenses are arranged in reverse, i.e., the active zones under the even-numbered lenses correspond to the dark zones under the odd-numbered lenses, and the dark zones under the even-numbered lenses correspond to the active zones under the odd-numbered lenses. Advantageously, according to the invention, N is equal to 7 or 11. The inventors have found that a finer lenticular array, which allows for the display of only 7 or 11 main viewpoints, can make the structural effect even less noticeable, although the number of viewpoints used is not limited to this and there is of course nothing that prevents the use of other numbers of viewpoints.

[0027] Advantageously, according to the invention, P is equal to 3. According to this variant, when the screen is oriented in horizontal mode, each pixel of the pixel matrix consists of three horizontal sub-pixels of different colors arranged adjacent to one another in a column, the columns being separated from one another by black inter-column spaces. Of course, when the screen is oriented in vertical mode, each pixel of the pixel matrix consists of three vertical sub-pixels of different colors arranged adjacent to one another in a row, the rows being separated from one another by black inter-row spaces. However, this does not prevent the use of pixel panels where each pixel consists of four subpixels, in particular panels that have an additional white subpixel in addition to the usual three colors red, green and blue.

[0028] Advantageously, according to the invention, said pixel matrix is ​​constituted by an 8K panel. Such pixel panels allow for the highest resolutions currently available, such as a 31.5-inch panel with a resolution of 7680 x 4320 pixels, oriented in portrait mode, with a pixel matrix where each pixel is made up of horizontal RGB stripes. In fact, higher screen resolutions allow for more simultaneous viewpoints to be displayed, and these additional viewpoints can be used either to shorten the focal length and enlarge the lobes while keeping the pitch constant, or to dedicate a portion of the narrower lobes to each viewpoint, reducing the width each viewpoint occupies and creating a more photographic effect.

[0029] The present invention also relates to a method for photographic quality autostereoscopic display of autostereoscopic images having M×N viewpoints ordered from 1 to M×N, wherein N is an odd integer strictly greater than 1 and M is an integer selected from 1, 2, 3 and 6, characterized in that said method comprises: Selection of an autostereoscopic display screen according to the present invention; A distribution of MxN viewpoints under each lens, where the MxN viewpoints are distributed in N vertical blocks of 6 pixels high, consisting of M viewpoints (6 / M pixels per viewpoint), with each block periodically repeated every 12 rows on the lens axis, if the panel's subpixels are horizontal, or in PxN vertical blocks of 6 / P subpixels high, consisting of M / P viewpoints (6 / M subpixels per viewpoint), with each block periodically repeated every 12 / P rows on the lens axis, if the panel's subpixels are vertical.

[0030] In other words, the method for an autostereoscopic display according to the invention makes it possible to distribute, under each lens, the M×N viewpoints of the image into N elementary blocks that are periodically repeated every 12 rows if the pixels of the screen are composed of horizontal sub-pixels, or into N×P elementary blocks that are periodically repeated every 12 / P rows if the pixels of the screen are composed of vertical sub-pixels. The advantages and technical effects of the screen according to the invention apply equally to the display method according to the invention, mutatis mutandis. Advantageously, according to the invention, each of the MxN viewpoints of the image is coded using 6 / M different vertical pixels within N basic blocks of 6 pixels in height, where M is equal to 1, 2, 3 or 6, when the panel is constructed from horizontal sub-pixels.

[0031] Advantageously, according to the invention, each of the M×N viewpoints of the image is coded using 6 / M different vertical sub-pixels within P×N basic blocks of height 6 / P sub-pixels, where M is equal to 3 or 6, when the panel is constructed of vertical sub-pixels. When a screen according to the invention is oriented in vertical subpixel mode, the pixels are arranged in a configuration in which the subpixels extend vertically and have a height P times their width. If each pixel consists of three RGB subpixels (in this case P is equal to 3), then each square pixel consists of three vertical subpixels arranged side by side in the horizontal direction. In this case, content is addressed sub-pixel by sub-pixel, rather than pixel by pixel as in a horizontal sub-pixel screen. Therefore, the distribution of viewpoints (also called "mix") can be done in groups of four consecutive rows (12 divided by 3) instead of 12 rows. For the same number of viewpoints, the viewpoints are distributed horizontally at three times the rate of the mix in the horizontal version. Each viewpoint is coded over at least two rows, which makes it even possible to use fewer high-resolution viewpoints, or to "tile" intermediate viewpoints, depending on the depth of stereoscopic depth in the scene.

[0032] In the vertical subpixel version, the subpixels of the same view placed under adjacent lenses are located four rows down and on the next color component. If N is equal to 11, two coding schemes are obtained: 33 views, each coded with two different vertical subpixels (in this case M is equal to 3), and 66 views, each coded with one subpixel (in this case M is equal to 6). Regardless of the screen configuration (horizontal or vertical sub-pixels), the method according to the invention allows adaptively adjusting the number of displayed viewpoints depending on the parallax, depth and prominence in the image, with ghost images of details at the edges of the viewing volume being the determining factors. Regardless of the screen version (horizontal or vertical), the method according to the invention makes it possible to adjust the number of displayed viewpoints depending on the parallax (depth and projection) in the image, the occurrence of ghost images of details at the boundaries of the viewing volume being a determining factor in this adjustment. Also, depending on the stereoscopic depth in the scene, it is possible to either use fewer viewpoints with higher resolution, or to "tile" intermediate viewpoints. Furthermore, the invention also relates to a display screen and a display method characterized in that it combines all or some of the features mentioned above or below. [Brief explanation of the drawings]

[0033] 5. List of drawings Other objects, features and advantages of the present invention will become apparent on reading the following description, given by way of non-limiting example only, and made with reference to the accompanying drawings, in which: [Figure 1] FIG. 1 is a schematic diagram of a display screen according to one embodiment of the present invention. [Figure 2] FIG. 2 is a schematic diagram of a display screen according to one embodiment of the present invention, having eleven main viewpoints. [Figure 3] FIG. 3 is a schematic diagram of a display screen according to another embodiment of the invention, having seven main viewpoints and oriented in portrait mode. [Figure 4] FIG. 4 is a schematic diagram of a display screen according to another embodiment of the invention, having seven primary viewpoints and oriented in landscape mode. [Figure 5] FIG. 5 is a schematic diagram of a display screen according to another embodiment of the invention having 42 viewpoints and oriented in portrait mode. DETAILED DESCRIPTION OF THE INVENTION

[0034] 6. Detailed Description of the Embodiments of the Invention In the drawings, for clarity of illustration and explanation, scale and proportion are not strictly adhered to. 1 shows a schematic representation of a screen 10 for displaying autostereoscopic images with 6×N viewpoints. The screen 10 comprises a matrix 20 of pixels arranged by rows and columns. In other words, in this embodiment, M is equal to 6. Above the pixel matrix is ​​an array of cylindrical lenses 30 (also referred to as a "lenticular array"), each tilted at an angle of arctan (1 / 12), or 4.76°, to the column direction. In Figure 1, the tilt shown is exaggerated for illustrative purposes only. The array of cylindrical lenses 30 has a widthwise pitch of N / 2 pixels, with each lens in the array spanning one vertical block of six active zones (hereafter referred to as a "basic block") of pixels in one column of the screen, the next black inter-column space, and a new basic block of six active zones in the next column of pixels to the bottom of the screen.

[0035] As mentioned above, each pixel consists of three active sub-pixels (P equals 3) of different colors and opaque inactive zones that form part of the inter-row and / or inter-column areas of the panel. In this configuration (for simplicity), the term "pixel" will be used herein to refer to the content-addressable active zone of the pixel, and the term "black space" will be used to refer to the opaque inactive zone of the pixel that forms the inter-row and / or inter-column areas of the screen. Reference numeral 26 denotes a pixel in the unsimplified version of the invention, ie a group consisting of an active zone of a pixel and a black zone surrounding this active pixel zone. The lenticular array is, for example, an array manufactured by the method described in International Application WO2020 / 178506 in the name of the present applicant. 2 shows a schematic diagram of a portion of a pixel panel of a screen according to an embodiment of the present invention with horizontal sub-pixels on which lenticular array lenses 30 are shown, the screen being designed to display an image with 11 main viewpoints (i.e. N equals 11).

[0036] To achieve this, the pixel panel (or pixel matrix) comprises a number of pixels 26 arranged in rows and columns, each of which consists of three horizontal sub-pixels of different colours, corresponding to the R, G and B components respectively. Each column of pixels is separated from the others by a black inter-column space 24 having a width approximately equal to the width of the active zone of the screen pixels 26. Note that in Figure 2, for clarity, these black columns are shown as white. Each lens 30 constituting the array of cylindrical lenses is inclined at an angle equal to arctan(1 / 12), or 4.76°, with respect to the column direction of the screen. The array is designed with a pitch of 5.5 pixels (1 1 / 2) wide. Each lens of array 30 extends across a vertical block 22 of six pixels (called a basic block 22) in one column of the screen, followed by a group of equal-height black inter-column spaces, then a new basic block of six pixels in the next column, and so on down to the bottom of the screen.

[0037] Such a screen is capable of displaying 11×M viewpoints, where M is an integer that can take the values ​​1, 2, 3 or 6. The 11 main viewpoints are indicated by italic Roman numerals in Figure 2. Thus, for the basic block indicated by reference number 22, the main viewpoint for this block is viewpoint 1. The next basic block in the same row encodes viewpoint 11. Note that under the illustrated lens 30, the first row of basic blocks encodes odd-numbered viewpoints (1, 3, 5, 7, 9, and 11), and the second row of basic blocks encodes even-numbered viewpoints (2, 4, 6, 8, and 10). Thus, the 11 main viewpoints are interlaced under the lens across 12 consecutive rows in the horizontal version. If the screen were designed in vertical mode, the 11 viewpoints would be interlaced across 4 consecutive rows (as long as they could be addressed sub-pixels instead of whole pixels).

[0038] As can be seen from Figure 2, to recreate the same viewpoint under the illustrated lens, a shift of one column to the right and 12 rows down is required. Thus, if an observer is looking at viewpoint 1 with one eye and moves slightly to the left (the observer's movement is opposite to the horizontal shift of the viewpoint delivered to the eye by the lens), they will then see viewpoint 2, then viewpoint 3, and so on up to viewpoint 11. These 11 viewpoints are interlaced across the 12 rows of the pixel panel. In each basic block, it is also possible to assign intermediate viewpoints located between the main viewpoint of that block and the main viewpoint of the next block to other pixels within that basic block (in other words, an image with 11×M viewpoints can be displayed, where M is selected from 1, 2, 3 and 6).

[0039] The optical axis of the lenticular array 30 intersects the pixels it passes through at an angle of 4.76°, which allows each of the six pixels it passes through in one basic block to be aligned in turn with a subpart of the solid angle occupied by each of the main viewpoints. Since one pixel is made up of three sub-pixels, each basic block 22 of six pixels is made up of 18 stacked horizontal sub-pixels, so going through a basic block of six pixels actually means going through these 18 sub-pixels in order. If each of the six pixels of basic block 22 is assigned a different viewpoint, then 18 different overlaps actually occur. Therefore, the subpixel that is passed must coincide with a theoretical point located at the center of gravity of the surface of interest. When the six pixels of a basic block are assigned to six different viewpoints, overlap is evident and viewpoint mixing is inevitable, but the perceived image quality is better than that obtained with the higher resolution primary viewpoint.

[0040] Thus, a screen according to the embodiment of FIG. 2 may allow content to be viewed from 11, 22, 33 or 66 viewpoints, as required: - 11 main viewpoints (M equals 1), each viewpoint coded with 6 consecutive different vertical pixels; - 22 views (M equals 2), where each view is coded with 3 different vertical pixels, - 33 views (M equals 3), where each view is coded with two different vertical pixels, - 66 viewpoints (M equals 6), where each viewpoint is coded by a single pixel. FIG. 3 shows schematically a part of a pixel panel of a screen according to another embodiment of the invention, oriented in horizontal sub-pixel mode and intended to display seven main viewpoints (N equals 7). The difference with the screen of Figure 2 is that the lenticular array has a pitch of 3.5 pixels instead of 5.5 pixels. Each lens in the array 30 of cylindrical lenses is also tilted relative to the columns of the screen, the tilt angle being equal to arctan(1 / 12), or 4.76°.

[0041] The seven principal viewpoints are distributed over 12 consecutive rows under each lens, following the same principle as described in connection with FIG. Figure 3 shows the different distributions of active and opaque zones under the first two lenses, representing odd-numbered and even-numbered lenses. In particular, the portion of the pixel panel under viewpoint 1 of the first lens corresponds to an opaque zone under the second lens, and the adjacent opaque portion of viewpoint 1 under the first lens corresponds to viewpoint 2 under the second lens. If the portion of the pixel panel under the first lens (representative of the odd lenses) is superimposed with the portion of the pixel panel under the second lens (representative of the even lenses), it will be clearly seen that the active zones of the even lenses correspond to the opaque zones of the odd lenses, and vice versa. Figure 5 shows a screen similar to Figure 3, with the only difference that each pixel of the screen is assigned a different image viewpoint, thus making it possible to display images with 42 viewpoints (N equals 7 and M equals 6). It will be seen that in the first basic block in the first column shown, viewpoints 42, 41, 40, 39, 38, and 37 are assigned to the six pixels that make up that basic block. In the adjacent basic block in the second column, viewpoints 12, 11, 10, 09, 08, and 07 are assigned to the six pixels of that basic block. In the following basic block in the third column, viewpoints 24, 23, 22, 21, 20, and 19 are assigned to the six pixels of that basic block, with this arrangement occurring in seven basic blocks arranged under the first lens across the first 12 rows of the pixel panel. These seven basic blocks are indicated schematically by dashed rectangles enclosing six consecutive pixels in the column direction.

[0042] In other words, in each of the seven basic blocks, one of the 42 viewpoints is assigned to one of the pixels of that basic block, and five subsequent viewpoints in the sequence of 42 viewpoints are assigned to the remaining pixels in that basic block, respectively. In the first basic block of the first column, viewpoint 37 can be considered the main viewpoint, and viewpoints 38, 39, 40, 41, and 42 can be considered intermediate viewpoints made up of subsequent viewpoints in the sequence of 42 viewpoints. Thus, under the first lens shown, the first 12 rows of the screen allow for 42 viewpoints of the image to be displayed. Note that viewpoint 42 is located in the second column, row 13 under the first lens. Viewpoint 12, in the first row, second column, is located in the third column, row 13, and so on for each of the other viewpoints. A similar pattern is repeated across the screen for each lens (not shown for clarity). Figure 4 shows a screen designed to display up to 42 viewpoints, with a vertical subpixel orientation. The screen therefore has a lenticular array tilted at an angle equivalent to 4.76° with respect to the column direction of the screen. The seven main viewpoints are distributed over four consecutive rows under the lens (as opposed to 12 rows in the horizontal subpixel version), as they are addressable by subpixels rather than whole pixels.

[0043] Thus, it can be seen that viewpoint 1 is coded by the white subpixel of the first pixel, located in the upper left corner of the image. Viewpoint 1 is coded by the light gray subpixel of the pixel at row 5, column 2 of the screen, located below the lens, and by the black subpixel of the pixel at row 9, column 3 of the screen. Similar relationships hold for all seven viewpoints in the image. Similar to the example of a screen in horizontal sub-pixel mode, it is possible to either use fewer viewpoints with higher resolution, or to "tile" intermediate viewpoints, depending on the stereoscopic depth in the scene. Regardless of the embodiment, the lenticular array has angles that allow each viewpoint to be distributed both vertically and horizontally. These viewpoints are referred to as "main viewpoints" when the same viewpoint is located only in pixels belonging to a vertical basic block, and as "intermediate viewpoints" when multiple viewpoints with slightly different positions are "tiled" within the same block to complement the main viewpoint. The intermediate viewpoints can be thought of as filler viewpoints that fill in the discrete disparity jumps (changes in disparity) between the main viewpoints, allowing the viewer to experience a continuous parallax sensation. These intermediate viewpoints are used to smooth out the disparity jumps that may be perceived between the main viewpoints.

[0044] The present invention has been described using a fixed M within an image. In other words, the number of viewpoints is uniformly determined based on the target parallax within the image. However, it is also possible to use a screen configured to dynamically change the number of viewpoints (i.e., the integer M) within the same image depending on the parallax. For example, M can be fixed to 1 or 2 in areas with low three-dimensionality, and to 3 or 6 in areas with higher three-dimensionality. This makes it possible to limit the need to generate three-dimensional information using a large number of different viewpoints to only those zones of the image that are particularly three-dimensional.

Claims

1. A screen (10) for the autostereoscopic display, considered to be of photographic quality, of autostereoscopic images having MxN viewpoints ordered from 1 to MxN, comprising: N is an odd integer strictly greater than 1, M is an integer selected from 1, 2, 3, and 6; The screen comprises a panel (20) of pixels arranged in rows and columns, each pixel (26) consisting of a plurality of P active sub-pixels of different colors and an opaque inactive zone forming part of the black inter-row and / or inter-column areas of the panel; The screen further comprises a regular array of cylindrical lenses (30) mounted on the pixel panel; Each lens is inclined at an angle equal to arctan(1 / 12) relative to the screen columns, 1. A screen, comprising: an array having a pitch of N / 2 pixels in width, so that on the one hand the even and odd lenses (30) of the regular array can be distinguished by a different distribution of the active and black zones (24) of the pixels (26) of the panel arranged under each even and odd lens, and on the other hand each of the M×N viewpoints can be decoded, the M×N viewpoints being distributed in N vertical blocks of 6 pixels in height, consisting of M viewpoints (6 / M pixels per viewpoint), with each block being repeated periodically on the lens axis every 12 rows, if the subpixels of the panel are horizontal; and in P×N vertical blocks of 6 / P subpixels in height, consisting of M / P viewpoints (6 / M subpixels per viewpoint), with each block being repeated periodically on the lens axis every 12 / P rows, if the subpixels of the panel are vertical.

2. 2. The autostereoscopic display screen (10) of claim 1, wherein each pixel (26) of the pixel panel is composed of a plurality of P horizontal sub-pixels of different colors arranged next to each other in a column direction, such that each of the M×N viewpoints can be coded with 6 / M different vertical pixels repeated every 12 rows under each lens, where M is equal to 1, 2, 3 or 6.

3. 2. The autostereoscopic display screen (10) of claim 1, wherein each pixel (26) of the pixel panel is composed of a plurality of P vertical sub-pixels of different colors juxtaposed to one another in the row direction, such that each of the M×N viewpoints can be coded with 6 / M different vertical sub-pixels repeated every 12 / P rows under each lens, where M is equal to 3 or 6.

4. Screen according to any one of claims 1 to 3, characterized in that N is equal to 11.

5. A screen according to any one of claims 1 to 4, characterized in that P is equal to 3.

6. A screen according to any one of claims 1 to 5, characterized in that the pixel panel (20) is an 8K pixel panel.

7. 1. A method for photographic quality autostereoscopic display of an autostereoscopic image having M×N viewpoints ordered from 1 to M×N, comprising: N is an odd integer strictly greater than 1, and M is an integer selected from 1, 2, 3, and 6, said method comprising: - selection of an autostereoscopic display screen (10) according to any one of claims 1 to 6, - distribution of the MxN viewpoints under each lens (30), in N vertical blocks of 6 pixels high, consisting of M viewpoints (6 / M pixels per viewpoint) when the panel's subpixels are horizontal, each block being repeated periodically every 12 rows on the lens axis, or in PxN vertical blocks of 6 / P subpixels high, consisting of M / P viewpoints (6 / M subpixels per viewpoint) when the panel's subpixels are vertical, each block being repeated periodically every 12 / P rows on the lens axis. A method comprising:

8. 8. The method of claim 7, wherein each of the MxN viewpoints of the image is coded using 6 / M different vertical pixels within N elementary blocks of 6 pixels in height, where M is equal to 1, 2, 3 or 6, in a panel composed of horizontal sub-pixels.

9. 8. The method of claim 7, wherein each of the MxN viewpoints of the image is coded using 6 / M different vertical sub-pixels within PxN basic blocks of height 6 / P sub-pixels, where M is equal to 3 or 6, in a panel composed of vertical sub-pixels.

Images