Display system

Abstract

The display system comprises an image generating unit 10; an output modulator 22 positioned after the image generating unit for selectively allowing light to pass; and acontroller. The controller is configured to: control the image generating unit to two images into respective eyeboxes for each eye. The two images are displayed by scanning at least one respective first and second emission bands across the image plane generated by the image generating unit each image band in cluding a portion of the image. The output modulator is controlled in synchrony with the scanning of the at least one first emission band and the scanning of the at least one second emission band. Also claimed are methods of displaying an image and a further display.

Description

Technical Field The present invention relates to display systems which can present a different image to each eye, such as an autostereoscopic display. Background Displays are known and can present a different image to each of a viewer’s left and right eyes. Some such displays are known as autostereoscopic displays and may include a parallax barrier so that each of a viewer’s eyes views a respective, different image that the brain then combines into a single image. WO 2005 / 112474 discusses an autostereoscopic display which can be combined with active or passive parallax barriers. The concept of a “scanning barrier” with a plurality of openable apertures is discussed and applied to a “multi-view” system. In a multi-view system, a plurality of images corresponding to different viewpoints are displayed so that each eye of a viewer sees a different one of the multi-view images and the brain can then reconstruct a three-dimensional image. As the eye position changes a viewer perceives different ones of the multi-view images giving the impression of looking around a three-dimensional scene / object rather than having a fixed viewpoint. Multiview systems have the advantage of working across a variety of viewing positions and not requiring additional viewer hardware, such as filtering glasses. However, as the number of views increases the requirements for image processing become large and the number of views, n, displayed simultaneously reduces the horizontal resolution of each image by a factor of n. EP-4016166 A2 discusses an augmented reality device. An eye tracking system is used to determine the position of a viewer’s eyes. Images for each eye are displayed sequentially and directed to the position of the relevant one of a user’s eyes by a spatial mask in conjunction with a lens array. An image for a left eye is displayed on the whole display and the spatial mask is opened at the position where the lens array directs the image only to the left eye. Next, an image for a right eye is displayed on the whole display and spatial mask is opened at the position where the lens array directs the image only to the right eye. EP-3554069 Al discusses an image projection apparatus for a vehicle that includes a display, at least one second optical member and a controller. The display displays an image of a polygon as viewed from a virtual camera. The second optical member projects the image onto a first optical member and makes a virtual image of the image visible to a viewer. The controller dynamically corrects the polygon in accordance with the position of both eyes of the viewer. It would be desirable to provide an improved display which can present different images to a viewer’s left and right eyes. Summary According to a first aspect of the present invention, there is provided a display system comprising an image generating unit, an output modulator and a controller. The output modulator is positioned after the image generating unit and is for selectively allowing light to pass. The controller is configured to: control the image generating unit to display a first image for viewing by a left eye of the viewer and a second image for viewing by a right eye of the viewer; and control the output modulator, such that: the first image is visible from a first eyebox corresponding to the left eye and the first image is not visible from a second eyebox corresponding to the right eye, and the second image is visible from the second eyebox and not visible from the first eyebox. The first image is displayed by scanning at least one first emission band across an image plane generated by the image generating unit, the first emission band consisting of a portion of the first image. The second image is displayed by scanning at least one second emission band across the image plane generated by the image generating unit, the second emission band consisting of a portion of the second image. The output modulator is controlled in synchrony with the scanning of the at least one first emission band and the at least one second emission band. Scanning the first emission band and the second emission band in synchrony with the output modulator enables the positions at which the emission band can be seen by a viewer to be controlled, which may allow a more compact, power efficient, light efficient and / or simple design. For example, reduced processing is required compared to a multiview system producing many images from different viewpoints, and the full resolution of the screen may be used. Alternatively, or additionally, lower frame rates may be used on pixellated display devices (such as LCDs). The construction may also be simpler, for example no additional optics may be required in addition to, or incorporated within, the output modulator for directing the image in a particular direction. Any suitable image generating unit (also referred to as a picture generating unit), can be used. Some image generating units inherently create an image by scanning, such as a CRT or laser-scanning system. Other image-generating units, such as those comprising LCD, LCoS, DMD, PLM, OLED and Micro LED devices, can be controlled so that only a portion of the image-generating unit is active, also referred to as emitting, at any one time. The present disclosure applies both to image-generating units that generate images without depth cues for the eye to focus on (sometimes referred to as “conventional” displays) and light field displays, such as 4-dimensional light field and Computer-Generated Hologram (CGH) displays, which display images with phase information to allow the eye to focus naturally at different depths in the image. Likewise, a wide variety of output modulators can be used, including: Liquid Crystal (LC) based output modulators, where an LC cell can be transmissive orblocking; mechanical shutters that can open, close and / or move; and mirror-based modulators, which can direct light into or out of a transmission path to a viewer. In some cases, the output modulator may also be referred to as a spatial light modulator or an active shutter system. The output modulator may define at least one transmission edge and / or at least one transmission band. A transmission band, comprising two edges, one from transmissive to blocking and one from blocking to transmissive may have benefits of blocking stray light and the effects of sunlight. In some examples the output modulator may comprise a plurality of regions which can selectively allow light to pass, in this case a movable aperture may be formed by selecting which of the plurality of regions allow light to pass. In other examples, the output modulator may comprise a movable aperture, such as an aperture of fixed position in an optical barrier, perhaps a slit formed in an opaque material, where the optical barrier is physically moved to move the aperture. In further examples, the output modulator may comprise an edge in an optical barrier, such as an element, flag or member formed from an opaque material, where the optical barrier is physically moved to move the edge. The image generating unit may generate an image in a first plane and the output modulator may be positioned in a second plane, separated from the first plane. In this way, the output modulator can control which parts of the image plane a particular viewing position can see. In some examples the first plane is at infinity (such as through suitable optics following the image generating unit). When the first plane is at infinity, scanning horizontally across the image also scans the viewing angle. The skilled person will understand that similar techniques can be used when the first plane is not at infinity, providing that the first and second planes are separate. In some examples where the image plane is not at infinity, image generating units with control over both ray angle and position may be used, such as CGH and 4D lightfield. The scanning of the emission band is preferably the scanning of a substantially vertical emission band horizontally across the display from the position of the viewer. This differs from conventional raster displays which tend to scan a substantially horizontal emission band vertically across the display. In some examples, a boundary between the first and second images may be scanned, thereby achieving emission bands which move across the image plane. For example, one emission band increases in area while the other emission band decreases in area. In other examples, there may be an unused portion of the image plane during the scanning at least one first emission band and the scanning at least one second emission band. In this way, the at least one first and / or second emission band does not occupy the whole of the image plane. In the case of a display device comprising a pixel array, this corresponds to a portion of the pixel array not being illuminated. An unused portion can reduce the effect of crosstalk. An unused portion can also increase the overall power or light efficiency by stopping emission from regions of the image plane that cannot reach any eyebox in some transient configuration of the output modulator. The unused portion may be contiguous with an emission band. Where there are two or more emission bands, an unused portion may be positioned between a first emission band and a second emission band. In some examples, the at least one emission band occupies less than 25%, less than 10%, or less than 5% of an image plane generated by the image generating unit at any one time. The at least one emission band may also occupy at least 1% of the image plane at any one time. For example, when a horizontal resolution is n pixels, the at least one emission band may have a width of at least n / 100 pixels. The first and second eyeboxes may be predetermined based on knowledge of average geometry of human faces such as a typical inter-pupillary distance (IPD) of an adult human and a predetermined viewing position. In that case, the viewer may position themselves physically in a “sweet spot” for viewing the display, or mechanical adjustment can be provided to adjust an optical path so that the left and right images are substantially positioned where a user’s eyes are. In some examples, the display system may be an autostereoscopic display system. Over a time interval, the scanning at least one first emission band and the scanning at least one second emission band may be substantially continuous and the output modulator is then controlled by an activation function with one or more limits moving substantially continuously. Here, reference to “substantially continuous” is used to include continuous changes which are quantised, such as a location expressed as a continuous function and quantised to a display element. The reference to “over a time interval” is because at times the emission band and activation function may be discontinuous, that is they may “jump” from one location to another, such as between images to be displayed or during a blanking period. The display system may advantageously comprise an eye tracking system for determining a position of a viewer’s eyes. The first eyebox and the second eyebox are then based on the determined position. This can allow the different eyeboxes and / or any autostereoscopic nature of the display system to be experienced over a wider range of viewing positions and / or without requiring mechanical adjustment. The addressable eyebox (which is the total area over which the eyeboxes may be positioned for the display system) may be expanded by including a ray-splitting element between the image generating unit and the output modulator. Suitable ray-splitting elements include a Diffractive Optical Element (DOE) and a replicating waveguide. In this case, the output modulator preferably extends over the whole viewable area, such as over all replications generated by the ray-splitting element. In some examples, the first image has a first correction applied and the second image has a second correction applied, the second correction being different from the first correction. This can be useful when the first and second images are viewed via a separate element which has a geometry that changes with position. For example, a vehicle Heads-Up Display (HUD) may use a windshield as an optical combiner to combine the displayed images with the real-world. Windshield curvature changes with position and so a left image may be distorted by the windshield in a different way than a right image because they are incident on different parts of the windshield. Such distortion may reduce image quality. In some examples where correction is applied, an image positioned at infinity may be displayed using different left and right images because of the different corrections applied. Example corrections that may be applied include one or more of distortion correction to compensate the windshield curvature, and an affine transformation to preserve parallel lines. In other examples, such as CGH and 4D light field displays, the correction may comprise aberration correction (possibly including distortion correction and wavefront phase correction). The controller may be configured to apply a blanking period between the display of the first and the second image based on a response time of the output modulator. A blanking period (also referred to as a blanking interval) may be useful when the output modulator has a relatively long response time to change from transmissive to blocking and / or vice versa (changing from blocking to transmissive). For example, a Liquid Crystal output modulator may take more than 1 ms to change from transmissive to blocking. Including a blanking period may reduce cross-talk. In this disclosure, crosstalk is used to mean where a right eye can view an emission band of a left image and vice versa. In some examples, the scanning direction across the image plane is different for the first emission band and the second emission band. In this way, the first image is displayed by scanning in a different direction than the second image. This may be particularly advantageous when the transition time of the output modulator from transmissive to blocking is different from the transition time from blocking to transmissive. In that case having the faster response time at the leading or trailing edge of a scanning transmission band provided by the output modulator may reduce cross-talk. For example, the scanning direction may be such that the leading edge of the transmission band moves towards the region of the output modulator that must be opaque. In that case, the faster response time is at the leading edge of the transmission band when the transition from blocking (opaque) to transmissive is faster, or at the trailing edge of the transmission band when the transition from transmissive to blocking (opaque) is faster. The first emission band and the second emission band may be scanned simultaneously at different positions on the image plane. In this way, the first image and the second image can be scanned simultaneously. For example, it may be possible to display at least one first emission band of the first image simultaneously with at least one second emission band of the second image at different positions on the image plane. The output modulator can then be controlled so that each eye sees emission bands associated with the correct image. In one example, a transmission band comprising two edges in the output modulator may be shared by both images, so that the left eye sees one emission band and not the other, and vice versa. The at least one of the first image or the second image may comprise a plurality of sub-images which are displayed sequentially. For example, the sub-images may be of different colours and / or with different noise characteristics when the image generating unit exhibits quantisation noise. Quantisation noise may be present with an image generating unit comprising a binary amplitude modulator such as a DMD, or a light field image generating unit, such as a CGH or 4D Light field image generating unit. According to a second aspect, there is provided a method of displaying an image, such as an autostereoscopic image, comprising a first image for viewing by a left eye of a viewer and a second image for viewing by a right eye of the viewer. The method comprises: scanning a first emission band of the first image across an image plane; scanning a first transmission edge of an output modulator in synchrony with the scanning of the first emission band, the first transmission edge positioned such that the first emission band is visible to the left eye and not the right eye; scanning a second emission band of the second image across the image plane separately from the first emission band; and scanning a second transmission edge of the output modulator in synchrony with the scanning of the second emission band, the second transmission edge positioned such that the second emission band is visible to the right eye and not the left eye. Scanning both emission bands and transmission edges in this way may allow a display system to be more compact and or require fewer processing resources, as discussed above for the first aspect. In some examples, at least part of the image plane is unused during the scanning of the first emission band and the scanning of the second emission band. The method may comprise: receiving information of a position of a viewer’s left eye and right eye; and determining the relative position of the first emission band and the first transmission edge, and the relative position of the second transmission edge and second emission band, based on the position of the viewer’s left eye and right eye. The information on the position may be predetermined, such as from an initial calibration, or received from an eye-tracking system. In some examples, the scanning the first emission band and the scanning a second emission band together have a duration within a display refresh time. This may reduce flicker in the displayed image. At least one limit of an activation function for the output modulator may then travel with a velocity such that a transition distance in transmissivity of the output modulator is short enough for the display system to display images with substantial isolation between an eyebox of the first image and an eyebox of the second image. This may allow output modulators with finite transition times to maintain isolation between the left and right images. The method may comprise replicating an exit pupil of an image generating unit which is scanning the first emission band and the second emission band in an optical path before the scanning the first transmission edge and the scanning the second transmission edge. The replication may be in any suitable way, such as using a DOE or a replicating waveguide. The method may comprise: applying a first correction to the first image; and applying a second geometry to the second image. The second correction is different from the first correction. For example, the first and second corrections may correct for a geometry of a combiner in the optical path between the output modulator and the viewer’s eye. Suitable corrections were discussed above for the first aspect. A blanking period may be applied between the scanning of the first emission band and the scanning of the second emission band based on a response time of the output modulator. This may be useful when the output modulator has a reasonably long transition time from transmissive to blocking or vice versa. The first emission band may be scanned in a first direction across the image plane and the second emission band may be scanned in a second direction, which is different from the first direction, across the image plane. The second direction may be opposite to the first direction. In one example the first direction is left and the second direction is right, with respect to a viewer. In this case, the first and second transmissive edges are also scanning in different directions on the output modulator. The scanning a first emission band of the first image and the scanning a second emission band of the second image may be simultaneous at different positions on the image plane. In this case, the first and second transmissive edges are scanning simultaneously in the same direction at different positions on the output modulator. The first and second transmission edges may form a single transmission band of the output modulator in some cases. In other examples, the scanning an emission band of the first image and the scanning an emission band of the second image may be sequential. At least one of the first image or the second image may comprise a plurality of sub-images which are scanned sequentially across the image plane. According to a third aspect, there is provided a display system, for example an autostereoscopic display system, comprising: an image generating unit; an output modulator positioned after the image generating unit and for selectively allowing light to pass; and a controller configured to implement the method of the above-described second aspect, with or without the optional features also described. The display systems of this disclosure are well-suited to vehicle applications for a heads-up display. As briefly mentioned above, the systems and methods described herein may be applied to a wide range of image generating units, also referred to as picture generating units and display system architectures. The scanning of an emission band can be produced in various ways, including: • using of a raster display device, with slow direction substantially horizontal. This is opposite of many devices, such as CRTs which have the slow direction substantially vertical. • using fast image display device, such as a device having an array of pixels. This may be a self-emissive display device such as Organic Light Emitting Diode (OLED) or microLED, a non-self-emissive display device, such as a Digital Micromirror Device, DMD. • using of a fast high resolution CGH device, such as a Phase Light Modulator (PLM). • scanning the illumination of a pixelated display device comprising an array of pixels. In this case, the scanning may be done by use of a fast CGH display device, such as a "pupil-plane" DMD or PLM, this is described in more detail later with reference to Figures 13 and 14; and • using two pixellated display devices, each with independent scanning illumination, and the corresponding image planes are combined, such as via a beam splitter used in reverse. This allows scanning of two emission bands for different eyes simultaneously when using a slower display device, such as an LCD. In a fourth aspect of the disclosure, an image generating unit for scanning emission bands is provided. Such a device may be used for the image generating unit of Figures 9, 11 and 12 described below. The image generating unit comprises: one or more light sources, a rotating mirror and a scattering surface. The one or more light sources and rotating mirror are arranged to generate an emission band at the scattering surface, and the scattering surface is reimaged onto an image plane generated by the image generating unit. For example, the reimaging may be onto a passive device comprising an array of pixels, which could be transmissive or reflective. The image generating unit may also have one or more of the following features: the rotating mirror may rotate at a constant angular velocity; the image generating unit may comprise a static reflective surface to achieve bidirectional scanning; the scattering surface may be a luminescent converter or a diffuser; the output of the one or more light sources may be modulated during the scan according to the content of the displayed image. In a fifth aspect of the disclosure, scanning emission bands may be provided holographically. An SLM illuminates a passive (reflective or transmissive) display device comprising an array of pixels. The scanning illumination is achieved by the SLM displaying a sequence of holograms. This may be used to illuminate full emission bands, or the holograms may be calculated according to the content of the image for improved light efficiency. Such a device may be used with the image generating units of Figures 13, 14 and 15 described below. An image generating unit according to this aspect comprises one or more light sources and an SLM, the SLM is configured to display a plurality of computer generated holograms, and the plurality of computer generated holograms direct light from the one or more sources to sequentially scan emission bands across an image plane. Each hologram may be calculated according to the content of the image. For example, it may distribute light only where needed within the emission band scanned across the image plane i.e. where the image content is. The image generating unit may comprise a dynamic diffuser. In a sixth aspect of the disclosure, bidirectional raster scanning can be used for emission bands. This may be used with the image generating unit of Figure 16 described below. In this aspect, an image generating unit comprises a raster scanning device, and the raster scanning device has a slow horizontal scanning axis. A “slow horizontal scanning axis” refers to axis where scanning is slower, for example that each column (vertical) axis is scanned before changing the row (horizontal axis). The horizontal scanning axis of the raster scanning device may be bidirectional. In a seventh aspect of the disclosure, an image generating unit comprises a movable scanning element to provide the scanning emission band. This may be used with the image generating unit of Figure 17 described below. In this aspect, the image generating unit comprises one or more light sources, a scanning element (such as a steerable mirror) and one or more passive pixellated display devices. The one or more light sources and scanning element are arranged to generate a scanning emission band at the one or more passive pixellated display devices. The image generating unit may have a diffusing surface positioned between the scanning element and the one or more display devices; the one or more light sources and scanning element are arranged to generate a scanning emission band at the diffusing surface, which is then reimaged at the display device. In any of the above fourth to seventh aspects, the image generating unit may comprise an homogeniser. Further features and advantages of the invention will become apparent from the following description of preferred embodiments of the invention, given by way of example only, which is made with reference to the accompanying drawings. Brief Description of the Drawings Figure 1 shows a diagrammatic representation of how transmissivity of a liquid crystal cell varies over time in response to an activation signal; Figure 2 shows a diagrammatic representation of an output modulator; Figure 3 shows a diagrammatic representation of an activation function applied to the barrier of Figure 2; Figures 4A is a diagrammatic representation of a display system including a pupil replicating waveguide; Figure 4B is a diagrammatic representation of a display system having an output modulator with a transmission band according to an example; Figure 4C is a diagrammatic representation of a display system having an output modulator with a transmission edge according to an example; Figure 5A shows a diagrammatic representation of the activating function of Figure 3 applied to an output modulator for a single emission band; Figure 5B shows a diagrammatic representation of an activating function applied to an output modulator for a plurality of emission bands; Figure 6 shows a diagrammatic representation of ray angles against position of the transmission region; Figures 7A to 7F are diagrammatic representations of various schemes for displaying stereo images; and Figure 8 depicts a flowchart of a method of displaying first and second images according to an example. Figures 9A and 9B are diagrammatic representations of example image generating units comprising a rotating polygon mirror; Figures 10A to 10F are diagrammatic representations of an example bidirectional scanning using a polygon mirror and static mirror; Figure 11 is a three-dimensional representation of a rotating polygon mirror system with a linear array of illumination sources; Figure 12 is a diagrammatic representation of varying the power of illumination sources depending on an image to be displayed. Figures 13A to 13D show diagrammatic representations of example image generating units using a DMD for beam steering; Figures 14A to 14C show a flow chart of an example method of operation of an image generating unit using a DMD for beam steering; Figures 15A to 15E are diagrammatic representations of image processing resulting from the method of Figures 14A to 14C for use with the image generating units of Figures 13 A to 13D; Figure 16 is a diagrammatic representation of a beam scanning trajectory in a further example image generating unit; Figure 17 is a diagrammatic representation of another image generating unit using a steerable mirror; and Figures 18A and 18B are schematic diagrams depicting example arrangements of elements after the image generating unit. Detailed Description A display system, for example an autostereoscopic display system, according to embodiments described herein comprises at least one image generation unit, an output modulation component and an optional eye tracking system. The display system presents separate images to two or more eyeboxes (corresponding to two or more eye pupils) within an addressable eyebox. An addressable eyebox is the area within which images are visible to an eye. In automotive heads-up displays, the addressable eyebox may measure 130mm horizontally and 50mm vertically. In general, each image is displayed as one or more bands of light, scanning across an image plane, while simultaneously an amplitude modulation pattern, comprising one or more transmitting bands and one or more blocking bands, is scanned across a second plane that is separate from the image plane. The display system controls the temporal location of the transmitting bands so that light reaches the targeted eye and light is blocked from reaching the incorrect eye. For example, the temporal location of the transmitting bands may be controlled in response to the temporal location of the image emission band and the eye pupil position, as detected by the eye tracking system, when present. Output modulator First, the operation of an example output modulator will be discussed. Figure 1 shows how a liquid crystal (LC) cell may respond to an activation signal. Plot (a) shows the value of an activation signal U over time. The activation signal may be a voltage applied to the liquid crystal material. Plot (a) shows that the liquid crystal is activated at time to and deactivated at time ti. Plot (b) shows the response of the LC cell in terms of its transmissivity of light. The turn-on time of the LC cell is Ton and the turnoff time is to^. Starting from time to the LC cell changes continuously from opaque to transparent. A period Ton after time to, the LC cell reaches its maximum transparency. Starting from time ti the LC cell changes continuously from transparent to opaque. A period after time ti, the LC cell returns to its maximum opacity. (Note that this may be a simplification. In practice, after to, the cell may approach its maximum transparency asymptotically with a time constant proportional to Ton. Likewise, after ti the cell may approach its maximum opacity asymptotically with a time constant proportional to TOff) As an example, a liquid crystal pi-cell may have an alignment time of 0.2 ms and a relaxation time of 1.5 ms. (A pi-cell liquid crystal modulator with these characteristics is commercially available from Bolder Vision Optik, Inc. of Boulder, Colorado, USA, termed “Medium Pi-Cell Liquid Crystal Modulator”.) The aligned state may correspond to either the transmitting state or the opaque state, depending on the orientation of the cell polariser. In a first orientation of the cell polariser, the cell transition times may be Ton = 0.2 ms and to^ = 1.5 ms. In a second orthogonal orientation of the polariser, the cell transition times may be swapped. Figure 2 shows a diagrammatic representation of an output modulator in the form of a barrier comprising a linear array of LC cells. Each LC cell is a strip, of width w, that spans the clear aperture of the barrier in the y-direction. The LC cell has width w = 5 mm. in this example, other examples may have other dimensions. In this example, the edges of the linear array of LC cells are parallel. In other examples, the edges may not be parallel, for example they may be oriented in a “fan-like” pattern, such that they radiate from an imaginary point at a distance from the output modulator. This may be useful when using a curved windshield. The LC cells may be independently controlled by the application of independent voltages to each cell. As discussed with reference to Figure 1, an activation function U(x, t) specifies how the array of LC cells is controlled over time. The function T(x, t) specifies the response of the array of LC cells in terms of their transmissivity over time. Figure 3 shows a diagrammatic representation of an activation function U(x, t) applied to the barrier of Figure 2, and the resulting pattern of transmissivity T(x, t). The activation function is discrete in space, due to the finite number of the LC cells and consequently the transmissivity may be stepped / pixellated, but the functions are shown varying continuously. The activation function defines a transmissive band comprising two transmission edges moving in the positive x direction (the direction indicated by the arrow 2) with velocity v. (As used herein, a “transmission edge” is a transition from blocking / opaque to transmissive and vice versa.) The interval of the moving band is defined by the limits p(t) and q(t). Both derivatives are equal to the velocity of the band: dp^t) dq(t) ----=----= v dt dt The resulting transmissivity profile T(x, t) rises and falls over distances don and d0^ due to the finite turn-on and turn-off times of the cells and due to the finite width of the cells. don = VTon + tV doff = VToff + w (Note that in this discussion the rise and fall times are shown as linear for ease of explanation, in practice these are likely to follow an exponential curve.) The hatched regions 4, 6 in the transmissivity profile represent partial transmissivity. The white region 8 in the transmissivity profile represents high transmissivity. The LC cell shown in Figure 1 has a turn on time, Ton, shorter than the turn off time, ^off- For this reason, the distance over which the transmission band turns on, don, is shorter than the distance over which it turns off, dOff. If the direction of travel is reversed, the shape of the transmission profile reverses in x. Example autostereoscopic display system with Pupil replicating waveguide An example autostereoscopic display system including a pupil replicating waveguide will now be described. While the pupil replicating waveguide is optional, it can expand a viewable area in a compact manner and is well suited to use in a Heads-Up Display (HUD). Figure 4A shows an image generating unit 10, a pupil replicating waveguide 12 (also referred to as an image-replicating waveguide), and left and right eyeboxes 14a, 16a of a viewer. A single ray 18 from the image generating unit 10 is shown entering the waveguide 12 and multiple copies 20 are emitted. Generally the optical surface of the waveguide 12 that emits the multiple copies 20 is known as the output coupler. The rays may be parallel or approximately parallel. The replicated rays reach both the left and right eyeboxes, so it is not possible to display a different image to each eye. Note that an image generating unit will, in general, not emit a single ray as shown, but will emit rays that fill an exit pupil. The image generating unit may display an image plane at infinity, where each pixel at a field within the image produces a beam of collimated rays that fill an exit pupil. Figure 4B shows the system of Figure 4A with an output modulator in the form of a barrier 22 near the waveguide 12 output. The barrier 22 has a single transmissive band 24 and is opaque elsewhere. The transmissive band 24 allows replicated rays 20 to reach the target eyebox 14b and hence the left eye, while the opaque bands prevent rays from reaching the excluded eyebox 16b and hence the right eye. The dotted rays 20’ are those that would reach the right eye without the barrier present. Figure 4B illustrates how the positioning of a transmissive band 24 allows rays 20 from a certain direction to reach the target eye, but the opaque region prevents rays 20’ reaching the other eye. If rays from a different direction are to reach the left eye, the transmissive band may move to a new position. Likewise, the transmissive band may shift to allow rays to reach the right eye. The ray angle and transmissive band may vary over time in such a way that the target eye is exposed to the full range of field angles in an image. In Figure 4B we have shown a transmissive band surrounded by opaque regions, but in some examples the left-hand opaque region could be transmissive. In this case, the output modulator would have a single transmission edge, not two. Figure 4C shows such a system with an output modulator 23 having a single transmission edge 25. The construction is otherwise the same as Figure 4B. Generally, a pupil replicating waveguide requires an input image plane at infinity. An image generating unit containing a physical array of pixels, such as LCD, OLED, LCoS, DMD, microLED, may generate an image plane at infinity by having a collimating / condenser lens after the array so that the array is located in the front focal plane of the lens. In this case, each pixel of the display produces a beam of collimated rays out of the image generating unit. A laser beam scanning (LBS) display has no physical array of pixels and generally produces an image plane at infinity by default. Figures 4A, 4B and 4C show a single-depth image generating unit, also sometimes referred to as a “conventional” image generating unit. Light field displays, such as 4-dimensional light field and Computer-Generated Hologram (CGH) displays, can also be used with the methods and principles of this disclosure. Light field displays require additional design conditions, for example that the replication pitch of the image replicating combiner (in other words, the distance between each consecutive ray 20) is approximately equal to the width of the exit pupil of the image generating unit and greater than half the diameter of the eyebox 14b. Further information on the design and operation of such a system can be found in WO 2023 / 057543, incorporated herein by reference in its entirety for all purposes. Scanning operation and viewing areas Figure 5 A shows diagrammatic representation of the activating function of Figure 3 applied to an output modulator for a single emission band. The top part of Figure 5 A is the same as Figure 3. Below this, an output modulator in the form of a barrier 122 and left and right eyeboxes 114, 116 of a viewer, at time t. The positions of the eyeboxes (x[, zL) and (xR, zR\ relative to a reference point on the barrier 122 (denoted by a circle 126), have been determined by an eye tracking system (not shown). The width of both eyeboxes, Ax, may be large enough to accommodate any measurement errors of the eye position by the eye tracking system, as well as errors due to latency of the eye position measurement, particularly during rapid movement of the viewer’s head. An image generating unit (not shown) emits an image as a band of light scanning across the image plane. A scanning angle 0(t) defines the centre angle of the band, and a bandwidth A0 defines the extent. Therefore, the band consists of ray angles in the range 0(t) ± A9 / 2. Over the period that an image is scanned, ts, the scanning velocity may be a constant, cm 39(t) field of view (a) =------= --------------- dt ts The rays propagate through the waveguide and emerge at the output surface. A highly transmissive band at the barrier 122 allows the target eyebox for the image, in this case for the left eye, to receive the band of light. An opaque region of the barrier 122 blocks the band of light from reaching the excluded eyebox for the right eye. The dotted rays 120b indicate the rays that must be blocked. The region of partial transmissivity with width don is narrow enough for the required transition between high and low transmissivity. Figure 5A contains plots of the activation function U(x, t) applied to the barrier, and the resulting pattern of transmissivity Tfx, t) (the same as in Figure 3). The transmissivity may reach (or nearly reach) its maximum in the region of the barrier that emits rays to reach the left eyebox. Similarly, the transmissivity may be minimised (or nearly minimised) in the region of the barrier that blocks rays from reaching the right eyebox. At the time, t, shown, the scanning angle and hence the band of rays, as shown in Figure 5A, is rotating clockwise, i.e. m >0. The activation function and transmissivity are moving left to right, in the direction of the arrow shown. The activation function and transmissivity must track the scanning angle 9(t), which means that the velocity as defined in Figure 3 may be v = cvzL. The distance over which the barrier transmissivity turns on may now be specified as don COZL^on T tV where w is the width of an LC cell of the output modulator (depicted in Figure 2). This distance, don, must be short enough for the system to display images with sufficient optical isolation between the two eyeboxes. This is a geometric condition (apparent from the rays drawn in Figure 5A) that may be expressed by the inequality: dnri = coz,Tnj) + w <x' R — x', — Ar U I L Lull fx L A. zl+zr 2 △9 Alternatively, in terms of an interpupillary distance IPD = x'R — x'L, the condition is: coz, + w <IPD — AY Lull X zL + zR △9 If we assume that the optical path lengths from the eyes to the waveguide are equal (i.e. z = zL = zR) we may simplify this to: mzxon + w <IPD — Ax — zA0 This is one of the inequalities that may be used to design a display system. Extension to multiple emission bands Figure 5B shows diagrammatic representation of an activating function applied to an output modulator to allow a plurality of simultaneous emission bands. In the example of Figure 5B, there are three emission bands, it will be appreciated that other examples may have other numbers of emission bands, such as two, four, five, etc. A plurality of simultaneous emission bands may be useful when the display is capable of displaying an image across its whole area simultaneously and the scanning means that only part of the display is in use at any one time. For example, multiple emission bands may reduce a time required to scan an image. In Figure 5B, the output modulator has time from blocking to transmissive of di and a time from transmissive to blocking of d2. As shown, di <d2, but this discussion is not limited to that, the results apply regardless of whether di <d2 or di >d2. As with Figure 5A, a scanning angle 0(t) defines the centre angle of an emission band, and a bandwidth A0 defines the extent. Therefore, the band consists of ray angles in the range 0(t) ± A0 / 2. The plurality of emission bands are substantially evenly spaced and have an angular separation of <|) between consecutive emission based. In the case of Figure 5B, it is required that all the of the emission bands can be seen at a first eyebox and not seen at a second eyebox (in this case, left eyebox 114’ and right eyebox 116’, respectively). Extending the constraints allows the following equations and inequalities to be derived. First the angular velocity of the activation is considered: m = 0 / ts This defines that the angular velocity, co, of the activation function is dependent on the angular separation of the emission bands, ¢, and the scanning time ts. The activation function itself can be thought of as including an activation region for each of the plurality of emission bands, three in this example. The transition distance from blocking to transmissive, di, is given by: d1 = 6J Z + W Where n is the time for the output modulator to change from blocking to transmissive, z is the distance between the display and the viewing position and w is the width of an LC cell on the output modulator (or a parallax barrier). Similarly, the transition distance from transmissive to blocking, d2, is given by: d2 = a> z t2 + w Where T2 is the time for the output modulator to change from transmissive to blocking. There are now two inequalities governing the design of the system. First, the constraint from the single emission band derived above for Figure 5A is repeated: d2 C IPD — Ax — z A0 There is also an additional multi-band constraint of: d± <z 0 — IPD — Ax — z AO It will be understood that this inequality will set a limit on the number of emission bands for a particular display configuration because of the presence of the angular separation, ¢. Display of left and right images and blanking times Figure 6 shows a plot (a) showing the evolution of the ray angles over time and a plot (b) showing the position of light emission and transparency at the barrier output surface over time. Both plots cover the period of two refresh cycles of the display, where a refresh period is tr = 16.7 ms in this example. In one refresh cycle, images are displayed sequentially to the left and right eyeboxes. First 42 and second 44 areas in plot (a) define the angles displayed of the left and right images, respectively, i.e. the image emission bands. The bandwidth is indicated as A9. Each refresh cycle, which may have period tr, is partitioned into phases denoted Left Blanking (LB), Left Scanning (LS), Right Blanking (RB) and Right Scanning (RS). The image generation unit only emits light during the Left Scanning and Right Scanning phases. Over the period of each scanning phase (LS or RS) the image generating unit scans the full horizontal field of view of each image, which in this example is 0 = 10°. The first 42 and second 44 areas in plot (b) define the regions on the barrier output surface where emitted rays of the left and right images reach the left and right eyeboxes respectively. The dark gray regions 46 in plot (b) define the activation of the barrier transmissivity according to the activation function U(x, t). The light gray regions 48 in plot (b) indicates the partial and highly transmissive bands, where the barrier transmissivity is more than the minimum value, i.e. where T(x, t) >Tmin. By definition this region also contains everywhere that U(x, t) is activated (i.e. it also includes the dark gray regions 46). The activation function U(x, t) may be chosen such that the barrier has maximum transmissivity (light gray region of the barrier in Figure E) everywhere that rays are emitted towards the target eyebox (i.e. the blue 42 and red 44 regions). To avoid leakage of the rays into the wrong eyebox, the duration of both blanking phases LB and RB may be the turn-off time of the barrier, i.e. tb = Toff. The scan time of both scanning phases LS and RS may be the remainder of the refresh period 2t5 = tr — 2tb. The full field of view must be scanned in the scan time, so the minimum scanning velocity is: 0 co = — This is a second constraint that may be used to design a display system. The scanning angle during the left blanking and scanning phases (LB, LS) of the first refresh cycle may increase monotonically as 0(0 = 0L(O = co(t - TB) - | The scanning angle during the right blanking and scanning phases (RB, RS) of the first refresh cycle may decrease monotonically as 0 0(0 = 0R(O = — — U)(t — Tg — Tr / 2) After the first refresh cycle, the scanning angle may repeat with the periodicity of the refresh time tr . In order for the target images to be directed to the target eyebox, and to avoid leakage of images to the excluded eyebox, the display system must control the activation of the barrier transmissivity in synchronisation with the scanning angle. The activation limits p(t) and q(t) define the lower and upper bounds, respectively, of the activation regions 46 shown in Figure 6. During the left blanking and scanning phases (LB, LS), the barrier may be controlled according to the activation limits: p(t) = x'L-^- + zL tan[QL(t) - A0 / 2] q(t) = x’L+-^ + zL tan[SL(t) + A0 / 2] + don During the right blanking and scanning phases (RB, RS), the barrier may be controlled according to the activation limits: p(t) = x’R - y- + zr tan[QR(t) - A0 / 2] - don q(t) = x'R + ^- + zR tan[0R(t) + A0 / 2] Notice how in each case the leading edge is extended by the turn-on distance don to ensure high transmissivity of the barrier where rays exit towards the target eyebox. Example viewing schemes for stereo images Figures 7A to 7F are a sequence of diagrams showing various schemes for displaying stereo images in the autostereoscopic displays of this disclosure. A stereo image is an image showing a different image for the left and right eye so that the brain can reconstruct a 3-dimensional scene. Variables include: the number of image generation units, whether colour channels are displayed sequentially or concurrently / simultaneously, whether left and right eye scanning directions alternate, and whether the barrier has a fast turn-on or turn-off transition. This is just a sample of the possible variants and the present disclosure is not limited to these. In each of Figures 7 A to 7E, the upper plots show the angular output of the image generating unit. The lower plots show transmissive properties of the barrier and the regions where rays are emitted that reach the target eyebox. The gray bands on the lower plots represent the partial or highly transmissive bands. In Figures 7A to 7E, dotted lines represent the left eye; dashed lines represent the right eye. Figure 7A shows the scheme of Figure 6 for reference. It has one image generation unit, displaying left and right eye images sequentially and colour channels concurrently. The barrier has a fast opaque-to-transmissive transition (Ton <xoyy). Left and right eyes may be scanned in opposite directions to each other. Figure 7B shows a variant similar to Figure 7A except that the barrier has a fast transmissive-to-opaque transition (Ton >t0^). In this case it is preferable for the scanning angle to vary in the opposite directions during the scanning phases (LS, RS) compared with Figure 7A; i.e. that the scanning angle increases during the right scanning (RS) phase and decreases during the left scanning (LS) phase. This maintains the condition that the steeper transmissivity gradient remains between the regions that are required to be transmissive and opaque. Figure 7C shows a variant with left and right eye images shown concurrently, with all colour channels also concurrent. Left and right eyes may be scanned in the same direction. In this variant, both the turn on and turn off times Ton,TOff of the barrier are preferably low (e.g. below 1 ms). A barrier comprising an array of dual frequency liquid crystal cells may be suitable in this case. Showing emission bands of left and right eye images concurrently may not be possible by scanning illumination of a slow pixellated device, such as an LCD. In this case, two LCDs with independent scanning illumination can be used. The two images generated by the two LCDs are then combined, for example using a beamsplitter in reverse. Figure 7D shows a variant with one image generation unit. Left and right eye images are shown sequentially, and colour channels for each eye are also shown sequentially. The colour channels are depicted in the order red, green, blue but other orders can also be used. Left and right eyes may be scanned in opposite directions. Figure 7E shows a variant with two sequential passes of pluralities of simultaneous emission bands, such as the system shown in Figure 5B. It has one image generation unit, displaying left and right eye images sequentially and colour channels concurrently (that is to say, each emission band is full-colour). Left and right eyes may be scanned in opposite directions to each other. The upper plot shows three emission bands, scanning concurrently, with the emission bands covering different parts of the field of view. The lower plot shows three simultaneous transmissive regions corresponding to the three emission bands in the upper plot. The examples of Figures 7A to 7E include emission bands which do not occupy all of the image plane. Unused portions of the image plane are depicted by the black portions of the upper plots, which show emission band output in terms of angle 0. The example of Figure 7F uses the whole of the image plane, but scans two emission bands 704 and 706 separated by a boundary 702. The emission band 704, depicted by a white area on the upper plot, corresponds to an image for a left eye, and the emission band 706, depicted by a dot-filled area of the upper plot, corresponds to an image for a right eye. The lower plot depicts the transmissive regions. It can be seen that at any one time there is a single transmissive region, represented by a white line, so that over a first time interval, a viewer’s left eye can see a portion 708 of the emission band 704, and hence the left image. Over a second time interval, a viewer’s right eye can see a portion 710 of the emission band 706, and hence the right image. The example of Figure 7F is not suitable for raster-based displays but could be implemented by a suitably fast image display device including self-emissive displays such as microLED and non-self-emissive displays, such as a DMD. Another way to implement a scanning boundary between images on the image plane is to: 1. Display the left and right images on two separate display devices (such as LCDs), 2. Scan the illumination of each display device to achieve the scanning emission bands 704, 706 shown in Figure 7F. 3. Combine the images of the two display devices, with e.g. a beam splitter. There are other variants not shown with different combinations of these attributes. While the examples of Figures 7A to 7F show the scanning of a transmission band, this may equally describe the scanning of one or more transmission edges. In the example of Figure 5 A, Figure 6 and Figure 7A, the positioning (or motion) of the leading edge of the activation function 46 is important, whereas there is additional freedom in the positioning of the trailing edge of the activation function. The position of the leading edge in Figure 5A is defined by the activation limit q(t), because the display system is in the left scanning (LS) phase. The positioning of the transition on the output modulator 122, from opaque to transmissive over the distance don, is important because it ensures sufficient optical isolation between the two eyeboxes. Hence, the position of the activation limit g(t), which controls the positioning of the transition from opaque to transmissive, is important. In the example of Figure 7B, the positioning of the trailing edge of the activation function is important and there is flexibility in the positioning of the leading edge. In the example of Figure 7C, however, the positioning of both the leading and trailing edges is important. For example, in Figures 5A and 6, the freedom of the positioning the trailing edge of the activation function 46, can be expressed by turning the equations for the trailing activation limits into inequalities: p(t) <x'L — + zL tan[BL(t) — A0 / 2] for left blanking and scanning phases (LB, LS) q(t) >x'R + + zr tan[QR(t) + A0 / 2] for right blanking and scanning phases (RB, RS) In some examples the trailing edges may be stationary, for example fixed values ^min> ^max)- p(t) = xmin for left blanking and scanning phases (LB, LS) q(t) = xmax for right blanking and scanning phases (RB, RS) The fixed values (xmui’xmax) may be the left and right edges of the output modulator shown in Figure 2. In this case, the transmission band is simply a single transmission edge. In the example of Figure 6, the edges of the output modulator may be xmin = 0 mm and xmax = 400 mm. Indeed, in Figure 6 at times t = 0 ms and t = 8.3 ms, the trailing edges of the activation function 46 are shown at these stationary values due to the limited size of the output modulator. In other words, in Figure 6 in the intervals 0 ms <t <1 ms, and 8.3 ms <t <9.3 ms, the output modulator is seen to be scanning a transmission edge, and this behaviour (where the trailing edges are fixed at either end of the output modulator) may be extended to all time intervals. Methods of operation Figure 8 depicts an example method of operation to display an image, for example an autostereoscopic image. A first image is for display to a viewer’s left eye and a second, different image is for display to a viewer’s right eye. For example, the first and second image may be views of the same three-dimensional scene from different viewpoints corresponding to the viewer’s left and right eyes, so that the viewer perceives a 3-dimensional scene. The left and right images may have had a correction applied to account for any distortion introduced by a windshield or other combiner. In some examples the left and right images are the same view of a 2-dimensional scene but are different because different corrections are applied. Figure 8 shows a simple example of displaying the first and second images. At block 802, an emission band of the first image is scanned across an image plane, such as by using an image generating unit that generates images by scanning, or by operating an LCoS, DMD, micro LED, LCD or OLED device so that an emission band is scanned across the image plane. The physical plane of pixels may be reimaged by a lens so that the image plane appears at infinity. This is a common configuration for an image generating unit to be used with a pupil replicating waveguide and has the property that each pixel then corresponds to a particular ray angle at the output modulator. In synchrony with the scanning of the emission band, a transmission band of an output modulator is scanned at block 804. In Figure 8, the synchronicity is depicted by a dashed, double-headed arrow. The transmission band is positioned such that the emission band from step 802 is visible to the left eye and not the right eye. The principles discussed above the section “scanning operation and viewing areas” may be applied for example. Once display of the first image is complete, the method proceeds to block 806, where a further emission band of the second image is scanned across the image plane, while at 808 a further transmission band of the output modulator is scanned in synchrony with the scanning of the further emission band. The further transmission band is positioned such that the further emission band is visible to the right eye and not the left eye. The method may be repeated for additional images (such as in the case of moving or changing images for display) or the same images (such as when information does not change over time or changes intermittently, so the same image is displayed for a time until the image changes). Many variations can be applied to this method. For example, the display of the first image and / or the second image may involve the display of multiple sub-images. Example sub-images include Red, Green and Blue sub-images, such as when the image generating unit contains a single LCoS or DMD device (and colours are displayed sequentially), or when the image generating unit is a CGH image generating unit. The sub-images may be displayed sequentially (so all the sub-images from the first image are displayed followed by all the sub-images of the second image) or alternately (so a subimage of the first image is followed by a sub-image of the second image and so on until all the sub-images are displayed). Although the blocks 802, 804 and 806, 808 are shown as being sequential in Figure 8, they may be at least partially concurrent in other examples. This may be the case when one or more transmission bands can be transmitted simultaneously and positioned to allow viewing of the correct emission band while blocking other emission bands for both the left and right eye. In this case the period for in which the first image and the second image is displayed may not be entirely the same, for example they may overlap to avoid potential crosstalk between the images. Worked Example Applying the design principles above, a display system can be designed. An example set of parameters for an automotive heads-up display using a transmission band are: 1. Interpupillary distance, IPD = 60 mm 2. Optical path length from waveguide to eyebox, z = 1000 mm 3. Eyebox width, Ax = 20 mm 4. Horizontal field of view, 0 = 10° 5. Bandwidth, A0 = 1.6° 6. Output modulator cell width, w = 5 mm 7. Output modulator cell turn-on period, Ton = 0.2 ms 8. Output modulator cell turn-off period, royy = 2 ms 9. Display refresh period, = 16.7 ms From these parameters the following characteristics can be derived using the relationships defined above: 1. Blanking period, tb = 2 ms 2. Scanning period, = 6.3 ms 3. Scanning velocity, m = 1.6° / ms 4. Output modulator band velocity, v = 27.6 mm / ms It will be appreciated that this is only one example and other examples are possible. Example illumination systems The principles in the present disclosure can be applied to a wide range of different illumination systems. In other words, they are not limited to any one particular illumination system architecture. Examples include: 1. A 2D laser beam scanner (LBS), configured to scan vertical emission bands horizontally. 2. Display devices with self-emissive pixel arrays, such as OLED or micro LED arrays, configured to display images as dynamically scanned emission bands over the pixel array. 3. Fast high-resolution holographic devices, such as a PLM 4. Display devices with pixel arrays illuminated by a light source, such a DMD, LCD or LCoS device. In the case of fast high-resolution holographic devices (example 3 above), the scanning emission bands may be directly generated by the devices with high efficiency. An example of a fast high-resolution holographic device is a Phase Light Modulator (PLM). In this example, the image-generation unit comprising the PLM scans a plurality of sub-images sequentially across the image plane. In the case of display devices with pixel arrays illuminated by a light source (example 4 above), these may comprise one or more light sources, one or more beam scanners and one or more display devices. Generally, different scanning illumination sources can be used interchangeably with the different display devices. Example scanning illumination sources include: • RGB lasers, possibly with some homogenisation and despeckling • Single wavelength laser, such as a blue laser, with a luminescent converter • LEDs Example scanners include: • Rotating polygon mirror, possibly in combination with a static reflective surface to allow bidirectional scanning. • MEMs mirror • Galvo mirror, also referred to as a Galvo scanner. • 2D light source arrays (e.g. LEDs), reimaged at (or in the case of LCDs, as a backlight for) the display device and controlled to produce the scanning emission bands • Fast Spatial Light Modulator (SLM), such as a Digital Micromirror Device (DMD), a Ferroelectric Liquid Crystal on Silicon (FLCoS) SLM or a Phase Light Modulator (PLM). PLMs are commercially available from Texas Instruments. Some examples may simultaneously scan, despeckle and direct light. Light may be directed toward the bright regions of the image for improved efficiency. The primary display devices requiring external illumination include: • DMD • LCoS • TFT-LCD In some examples, a single display device may output all colour components of a colour image. In other examples, there may be three display devices, one for each colour component or colour channel. In yet further examples, the different colour components of the image may be displayed sequentially on a single display device. Some particular examples with combinations of these elements will now be described, with reference to Figures 9 to 16, it will be appreciated that other combinations are possible and this disclosure is not limited to these combinations. First Image generating system example Figure 9A shows a schematic layout of an example image generating unit in which three display devices (each comprising an array of pixels) are illuminated by a scanning band of light. The display devices may be TFT-LCD devices, but other examples may contain DMD or LCoS devices. The scanning band of light is generated by a scanning emission band emitted from a luminescent converter, which is then reimaged onto the pixel array of each display device. In more detail, Figure 9A shows one or more laser sources 902 with an associated collimator 904 generating collimated beams that are directed towards a rotating polygon mirror 906. Such rotating mirrors are commonly found in barcode scanners and laser printers. In this example, reflected light from the polygon mirror 906 reflects off one or more static mirrors 908 before hitting a luminescent converter 910. The combination of rotating polygon mirror 906 and static mirror 908 change the unidirectional scanning action from the rotating mirror alone to become a bi-directional scanning action, as may be required for the operation of an autostereoscopic display. This bi-directional scanning is explained in more detail later, with reference to Figures 10 and 11. Other examples may omit the static mirror for a unidirectional scanning. In this example, the luminescent converter 910 is illuminated by a laser wavelength close to 450nm (blue in colour). The luminescent converter generates white light which is split into the three primary colours (red, green and blue) and routed to a TFT-LCD for each colour-component of the displayed image. The luminescent converter may comprise a ceramic inorganic luminescent material (such as those available from Schott AG, Germany) or a matrix of quantum dots. The latter may have a narrower spectrum of emission wavelengths and may be preferrable when the display system includes a pupil-replicating waveguide with diffractive couplers. The luminescent converter 910 is thereby illuminated with a scanning band of light by the action to the rotating polygon mirror 906. The luminescent converter 910 reemits substantially white light that scans in a corresponding way to the laser light incident on the luminescent converter. The dynamic pattern of illumination at the luminescent converter 910 is reimaged onto each of the LCD devices shown, by means of one or more focussing elements. In this example, the optical relay comprises a pair of lenses 912, 914 in the optical path. As shown, a single first lens 912 is in the optical path for all components of the substantially white light. Three second lenses 914r, 914g, 914b are then provided in the path for each of the red, green and blue components separated from the white light. Light is separated using first and second low pass beam splitters 916, 918. The first low pass beam splitter 916 has a pass band set so that red light passes, reaching a first side of first mirror 920, while green and blue light is directed towards second mirror 922. Continuing to describe the red light path, the red light is reflected by the first side of the first mirror 920 towards a third mirror 924 which reflects the light through a second lens 914r towards a fourth mirror 926 whereupon it passes through a LCD 928r for modulating the red light. Returning to the light which was reflected by the first low pass beam splitter 916, the second mirror 922 reflects the light, which now contains green and blue components, to towards the second low pass beam splitter 918. The second low pass beam splitter 918 has a pass band set so that green light passes while blue light is directed towards a second side of the first mirror 920. Continuing with the green light path, the now green light passes through a second lens 914g before being reflected by a fifth mirror 930 towards a LCD 928g for modulating the green light. Following now the blue light after the second low pass beam splitter 918, it passes through a second lens 914b before being reflected by a second side of the first mirror 920 towards a LCD 928b for modulating the blue light. The lenses 912, 914r, 914b, 914g act to magnify the image of the source at the LCDs 928r, 928b, 928g. The emission from the luminescent converter is Lambertian, meaning that light is emitted with equal radiance in all directions. Efficiency is improved if a wide cone of light rays is collected by the relay optics. However, optical aberrations are generally smaller when each pixel of the display devices emits a narrow cone of rays. It may be beneficial for the relay optics to collect light with a low f-number (e.g. less than f / 1) and to illuminate the display devices with a higher f-number (e.g. greater than f / 2). This is achieved by designing relay optics 912, 914r, 914b, 914g with magnification. Consequently, the reimaged illumination pattern at the display device (LCD 928r, 928b, 928g in this example) may be a magnified image (such as 2* or 4x) of the emission at the luminescent converter 910. As shown in Figure 9A, a variety of emission band sources are shown emitting cones of rays from the luminescent converter. These represent different positions of the emission band as the illumination is scanned across surface. At the corresponding reimaged emission band sources at the LCD devices, the cones of rays have smaller included angles due to the magnification in the relay optics. Finally, the modulated red, green and blue light is combined by combiner 932, which is an X-cube RGB combiner in this example, before passing through a final collimation lens 934. The display devices (LCD 928r, 928b, 928g) are all located in the front focal plane of the collimation lens 934. The exit pupil of the image generating unit is located at the back focal plane of the lens 934. Preferably the exit pupil of the image generating unit is positioned at the entrance pupil of the waveguide discussed above with reference to Figure 4A. While Figure 9A depicts one particular arrangement of optical paths, including mirrors at 45° to the optical path, other arrangements may be used in other examples. The luminescent converter 910 may be forward-emitting (transmissive) as in Figure 9A or it may be back-emitting. An alternative configuration using a back-emitting luminescent converter 910’ is shown as in Figure 9B, in which case the incident laser light from the collimator may be directed off-axis and the converted light collected on-axis. Figure 9B includes otherwise the same elements as Figure 9A, and also depicts an alternative arrangement after the first lens 912 which is generally a mirror image of that in Figure 9A. Once again, any suitable arrangement can be used. Figures 10A to 10F show how bidirectional scanning illumination may be achieved with a single moving element that rotates with substantially constant angular velocity in one direction. A collimated beam from a laser source 1002 (for example a laser diode, collimated by a lens) is directed towards a rotating polygon mirror 1004. As depicts the polygon mirror 1004 rotates clockwise. For one range of polygon mirror angles, the reflected beam travels directly to the luminescent converter 1006 via a homogeniser 1008 (which may be omitted in some examples) and scans in a first direction as the polygon mirror rotates. The homogeniser may help to turn the light from a laser into a source suitable for forming an emission band, by increasing the angular subtense of a source in the vertical image direction. This may also be included in embodiments using lasers and either a raster scanner or scanning element. In this range of angles, depicted in Figures 10A, 10B and 10C, the light is not incident on static mirror 1010. For a second range of polygon mirror angles, depicted in Figures 10D, 10E and 10F, the reflected beam is reflected off the static mirror 1010 and then directed towards the luminescent converter 1006. This doubly-reflected beam scans in a second direction, opposite to the first, as the polygon mirror rotates. In other examples, additional static mirrors may be added, causing each face of the polygon mirror to generate discrete ranges of singly-reflected, doubly-reflected, triply-reflected etc rays, so that, as the incident beam crosses each face of the polygon mirror, the beam at the luminescent converter makes multiple traversals of the surface. In the example of Figures 10A to 10F, the polygon mirror has 6 reflective faces. Other examples may have other numbers of reflective faces, such as 8, 12, or more. Likewise, other examples may have polygon mirrors with an odd number of faces. Figure 11 depicts a 3-dimensional representation of the polygon mirror arrangement of Figures 10A to 10F, and shows how a linear array 1012 of single wavelength (such as blue) laser sources may be arranged to generate a scanning vertical emission band at a luminescent converter 1006. The luminescent converter 1006 is preferably reimaged to the display device. Therefore, in this example the vertical direction is defined by the correspondence between locations on the luminescent converter and locations on the image plane. The homogenizer 1008 may improve the homogeneity of the emission band in the vertical direction, and may be omitted in some examples. The construction of Figure 11 can also apply to other sources than laser sources (such as more broadband illumination sources including LEDs) and other colours than blue. Figure 12 is a diagrammatic representation of varying power of illumination sources depending on a scene to be displayed. When a vertical array of illumination sources is used, such as the laser array of Figure 11, the output of each laser source may be modulated dynamically according to the content of the displayed image, which may improve contrast and power efficiency. A scene has bright and dark areas. As the illumination band is scanned across the scene, the power of each source is modulated according to the content of the image. The pixels of a display device are configured to show a scene 1102. (In Figure 12 the scene is shown in negative for clarity, so a dark area corresponds to a brighter part of the scene.) At the locations of objects 1104 in the scene 1102, the pixels are set to be substantially emitting (or transmissive or reflecting depending on the display device type), and substantially blocking elsewhere. The ellipses 1106 indicate the vertical positions of the illumination light sources corresponding to light emitted by the vertical array 1012 of light sources in Figure 11. Together these light sources scan across the pixel array as an emission band. The schematic diagram of traces 1108 in Figure 12 represents how the output power of each illumination source may be modulated as the emission band is scanned across the image. Each trace shows the output modulated between low and high powers. The output of each source is generally high as the light source scans across a bright object 1104 in the image 1102. Second image generating unit example Figure 13A shows an optical layout of an image generating unit that comprises a Digital Micromirror Device (DMD) 1302 as a beam steering element for a laser 1304 and collimating lens to illuminate a LCD 1306, a TFT LCD in this example, where an image is formed. In this example, DMD is a binary-amplitude SLM, meaning that each pixel of the DMD can be either off or on. Computer-Generated Holograms are displayed on the DMD at a high frame rate (e.g. >600Hz) causing an emission band to be scanned across the LCD device. Figures 13A, 13B and 13C show how a different hologram at the DMD can direct a beam to a different area of a LCD. The computer-generated holograms displayed on the DMD can be determined in any suitable manner. The holograms may be predetermined, for example corresponding to a predetermined scanning pattern. Here the hologram is performing a beam steering function, it is not generating a holographic image for display. In other examples, the holograms may be generated based on an image or scene to be displayed, corresponding to both a predetermined scanning pattern and the scene to be displayed. In that case, the holograms displayed at the DMD may distribute the light only where needed for the scene to be displayed by the LCD. This may allow brighter images and / or improve the energy efficiency for a given brightness of displayed image because less power is absorbed at the LCD in regions corresponding to dark areas of the image. Here the hologram is not only performing a beam steering function, it is also generating holographic portion of an image for display, that may be combined with a calculated array of pixel transparencies at the LCD to form an image for display. The holographic portion of the image may be lower resolution than the LCD. Figures 13A, 13B and 13C show a single-wavelength configuration for simplicity and to show the different components and beam paths more clearly. The system is depicted up to the waveguide entrance pupil 1308. An output modulator, such as that described with reference to Figure 2, is positioned after the waveguide. Figure 13A shows the DMD 1302 configured with a hologram that directs an illumination band to a first region of the LCD 1306. The light source is a collimated laser 1304, which is directed at the DMD 1302 at an oblique angle. Light is deflected at the DMD 1302 towards a lens 1310, a Fourier lens in this example, and then towards the LCD 1306. Preferably the DMD 1302 is located in the front focal plane of lens 1310 and the LCD 1306 is located in the back focal plane or the Fourier plane. A first micro lens array 1312, with lenses aligned with the pixels of the LCD 1306, converges incident light through the pixels, which improves transmission efficiency. The LCD 1306 is configured to show an image to be displayed. The hologram at the DMD and pixel values at the LCD 1306 may be chosen together so that the target image is displayed with minimal light absorbed at the LCD, i.e. with improved efficiency. A second micro lens array 1314 acts to converge rays from the LCD pixels so that, after passing through a collimation lens 1316, the exit pupil 1308 matches the entrance pupil size of a waveguide pupil expander (for example). The focal length of the second micro lens array 1314 may be less than the focal length of the first micro lens away 1312, causing the diameter of the exit pupil to be increased. Generally, the exit pupil diameter should be greater than the ray replication spacing of the subsequent pupil-replicating waveguide (not illustrated). Light from the LCD is directed towards a collimation lens, which may be positioned so that the LCD is positioned at the front focal plane of the lens. The image plane, as viewed from the exit pupil 1308, is then positioned at infinity, as is generally required for an image coupled into a pupil-replicating waveguide. Figures 13B and 13C show the DMD 1302 configured to illuminate different regions of the LCD. This demonstrates how a sequence of holograms presented by the DMD may act to scan an emission band across the LCD within a scan time. Figures 13 A, 13B and 13C represent a sample from a sequence of configurations that together make a sweep of the emission band across the LCD, so that every point on the image formed by the LCD 1306 is illuminated by at least one hologram in the sequence. In this example, the motion of the emission band is not fully continuous because the scanning is achieved with a discrete set of holograms presented by the DMD. However, there is a clear direction of motion of the emission band across the image plane. Even though the DMD frames are discrete, and the output modulator pixels are discrete, the changes between discrete, consecutive configurations are small and appear substantially continuous to a viewer. This may improve stereo isolation (i.e. reduce crosstalk between a viewer’s left and right eyes). In some examples, the emission band may be scanned in small discrete steps so that there is overlap, such as at least 10%, at least 20%, at least 30%, at least 40%, at least 50% overlap, at least 70% or at least 90%, in the location of the emission band between consecutive DMD frames. In this case the holograms for groups of consecutive DMD frames may be chosen so that, after incoherent summation, they minimize the perceived effect of the holographic speckle noise that is present in the individual hologram light distributions at the LCD. In this way, a high frame rate DMD may act partly as a dynamic diffuser or despeckler. There are many known methods of reducing perceived holographic speckle noise by displaying a number of holographic images within a period of persistence of vision. One widely known method is One Step Phase Retrieval (OSPR). In other examples, a collimated beam produced by a laser 1304 may be conditioned by a dynamic diffuser to increase the etendue of the beam. The resulting image projected at the LCD is convolved by an optical blurring function so that it contains fewer higher spatial frequencies and consequently reduced perceived holographic speckle noise. Examples of dynamic diffusers are given in WO 2021 / 245407 Al, incorporated herein by reference for all purposes. Dynamic diffusers are described in particular with reference to Figures 5, 6 and 7 of WO 2021 / 245407 AL The arrangements of Figures 13A, 13B and 13C can also be used for colour images using time-sequential display of each colour channel, for example by using a fullcolour LCD, such as an LCD with a colour filter. For example, during a sweep, while the illumination band is targeted at the region of the LCD shown in Figure 13 A, red, green and blue illumination systems may be cycled. As the hologram is wavelengthdependent, a different hologram may be displayed on the DMD for each of the red, green and blue wavelengths to target the given region of the LCD, in synchrony with the cycling of the illumination systems. This is possible because the DMD has a much faster response time than the LCD. Alternatively, if only a monochrome LCD is available, each colour channel may be scanned sequentially, such as scanning from the beam as shown in Figure 13A to that shown in Figure 13C with a single colour, before repeating with the next colour. This variant is portrayed in Figure 7D. Figure 13D depicts an alternative configuration to allow multiple wavelengths to be used simultaneously by an optical arrangement similar to that of Figures 9A and 9B of the first example. The red, green and blue lasers operate simultaneously and are split into different optical paths by an optical arrangement comprising two low-pass optical beam splitters, a system of mirrors, and an X-cube combiner to provide a colour image. Unlike the example of Figure 9A and 9B, the pairs of lenses are not required because the DMD plane is not reimaged (relayed) onto the LCD planes. Instead, a Fourier lens and collimation lens are provided before the first low-pass beam splitter and after the X-cube combiner, respectively. Fig. 13D shows how the collimated beams of different wavelengths illuminate different patches of the DMD. The DMD presents a separate hologram, for each patch / wavelength, simultaneously; thereby scanning the three wavelengths simultaneously. While Figures 13A, 13B, 13C and 13D have been described as including micro lens arrays, other examples may omit the micro lens array. In that case, some other way of filling the entrance pupil of the waveguide is provided. For example a larger DMD may be used and / or a diffuser provided after the LCD. Additionally, while a DMD may be used as the beam steering element, other SLMs can be used, such as a Phase Light Modulator (PLM), which may be more light efficient. Furthermore, displays devices such as amplitude LCoS or DMDs may be used instead of LCDs. Figures 14A to 14C are flow charts of an example method for use with the second image generating unit example of Figures 13 A to 13D, where a DMD or other spatial light modular (SLM) can be used to direct light and also form a low resolution version of an image for display through use of holograms. Figure 14A is an overview of the whole method, while Figure 14B is one of example of a method to generate a hologram and Figure 14C is more detail of an image reconstruction step. Referring to Figure 14A, the method has inputs of a source image 1400 for display, a starting scan angle 0o and a scan angle increment 60. In this example, the source image is assumed to be an array of pixel luminance values. Next, a low-resolution modified version of the image 1400 is formed. The low-resolution modified image is obtained by processing image 1400 by a greyscale dilation operation 1402 with a suitable structuring element, followed by a down sampling operation 1404. Other low-resolution modifications may also be used, for example down-sampling without the grayscale dilation. The dilation may be useful to ensure that the whole of the areas required to be illuminated are illuminated by increasing the size of objects within the low resolution modified image. Some examples may use a greyscale dilation without the downsampling, although the higher resolution will increase the processing resources required to calculate the hologram to direct the light. At 1406, a mask is applied to the low-resolution modified image. The mask corresponds to the emission band for display at the starting scan angle 0o. For example, in the image scanning shown in Figure 6 the starting angle for the Left Scanning phase would be 0O = — 5°. A hologram is then generated corresponding to the masked low-resolution modified image at 1408. The hologram can be generated by any suitable means. One example is discussed below with reference to Figure 14B. By generating the hologram across the whole of the image, the effect of the mask is to direct light into the area required Once the hologram has been generated, it is stored in memory at 1410 for display as part of a sequence of holograms. In addition, because the hologram includes part of the image for display, a reconstruction of that part of the image is formed at 1412. This reconstruction can be used to adjust the image displayed on the LCD to compensate for holographic speckle noise in the illumination pattern on the LCD (which is a function of the random phase applied in Figure 14B). The reconstruction can use any suitable means, one example is discussed below with reference to Figure 14C. Blocks 1406, 1408, 1410, and 1412 are repeated for each of the N emission bands, illustrated in Figure 14 as blocks 14061 ... 1406n, 1408i ... 1408n, 1410i ... 141 On, and 14121 ... I4I2\. The masking operations {1406, 1406i. 14062... 1406n} produce emission band scanning angles {00,0O + 60, 0O + 260,... , 0O + N60} respectively. The scan angle increment may be 60 = Q / N, where 0 is the horizontal field of view. This repetition may be sequential, in parallel or any combination (fully parallel processing is depicted in Figure 14A). Each of the image reconstructions from blocks 1412, 1412i... 1412n is summed to give a total reconstructed image across all the sequence of holograms. The emission bands may overlap in some examples, so this summing may include contributions from more than one reconstruction at any particular point. An optical blurring function is applied at 1414. This may be an estimate of the blur introduced during display of the hologram which is convolved with the image. Then, at 1416, the blurred sum of the reconstructions is upscaled to match the display resolution of the LCD or of the original image. Any suitable upscaling can be used, with or without interpolation. Finally at 1418 the image to form on the LCD is calculated. In this example it is determined by dividing the original image 1400 by the upscaled image from 1416. Such a division is computationally simple, but other examples may determine the LCD image differently for improved image quality. For example, in regions of high luminance in the source image 1400, it may be preferable that the holographic speckle noise is not corrected. Rather, the LCD pixels may be set to near-maximum transmission. This ensures that bright regions of the image are as bright as possible, at the expense of increased noise in the image. This may be useful because the human eye is less sensitive to noise in bright regions. The stored holograms are then displayed in sequence on the DMD while the LCD displays the image determined at block 1418 throughout the display of all the holograms. Figure 14B is an example method of determining holograms at block 1406. This is just one example, specific to the display of a 2D (flat) holograms using a binary amplitude-modulating DMD. Other examples may be used including those with SLMs other than a binary amplitude modulating DMD. At block 1420, the masked image is converted to an amplitude representation by applying the square-root function to the luminance values. Next, at block 1422, the amplitude-converted image has random phase added. The image is then converted to the Fourier domain, such as by using an FFT, at block 1426 and the result is binary quantised for display on the DMD at 1428. Optionally, a focus correction (not shown) may also be applied between blocks 1426 and 1428, in order to accommodate slight shifts in the location of the LCD image plane. Figure 14C is an example of the image reconstruction of block 1412. The hologram is converted back to the image domain at block 1430, by taking an inverse Fourier transform, such as an inverse FFT, of the hologram data. If an optional focus correction has been applied, an opposite focus correction (not shown) would also be applied before block 1430. Then, at block 1432 the image data is converted to luminance data by squaring the modulus of the result of block 1430. Figures 15A to 15E are diagrammatic representations of image processing, following the method of Figures 14A to 14C, which can be used with the image generating units of Figures 13A to 13D to cause the DMD to distribute light at the LCD more efficiently. Figure 15A shows an image 1502 to be displayed to the viewer. The image 1502 is inverted so the dark regions (the dial 1504 and arrow 1506) denote lightemitting objects in the image. Figure 15B shows a low-resolution modified version of the image 1502 of Figure 15 A, the output of block 1404 in Figure 14A. The low-resolution modified image defines the target distribution of light at the LCD. Figures 15C, 15D and 15E show target distributions of light required at the LCD corresponding to examples from a sequence of DMD configurations that together make a sweep of the emission band across the LCD, after the mask has been applied at block 1406 of Figure 14A. (The full sequence is not shown.) The dotted lines in Figures 15C-E denote the width of a full emission band, but light is only required in the illuminated regions shown. Figures 15C to 15E are therefore target images for holograms displayed by the DMD during a scan across the image plane. The holograms themselves are calculated from these targets by a CGH algorithm, such as described above with reference to Figure 14B, to be presented by the DMD device. Advantages of this method include increased optical efficiency (because less light is absorbed by opaque pixels at the LCD) and in, some examples, the DMD acts as a dynamic diffuser, removing the need for a separate despeckling component in the optical path as is often required in projectors (image generating units) employing multi-mode lasers. While the operation described above with reference to Figures 14A to 14C and 15A to 15E included use of a hologram formed by a DMD to both scan or steer the illumination and form a low resolution copy of the image, other examples may use the DMD, or any other suitable SLM, to form holograms to scan or steer the image only. In that case, the holograms for display while an image is displayed on the LCD can be predetermined by using a constant brightness rectangle occupying the emission band. This reduces the processing resources required for display but will be less efficient as more the illumination will be blocked or filtered out by non-transmissive parts of the LCD. Third Image generating unit example This example illustrates laser beam scanning with slow horizontal scanning axis. Figure 16 shows a scanning trajectory across the image plane of a Laser Beam Scanning Picture Generating Unit (PGU). Laser beam scanning PGUs are commercially available and form images by scanning a laser beam (or laser beams) directly over the image plane, this can then be modulated by an output modulator such as that described above with reference to Figure 2. Figure 16A shows the trajectory for a first eye of a viewer. For example, this trajectory may correspond to the LS (left scanning) area 42 in Figure 6. Figure 16B shows a trajectory with the opposite direction for a second eye of the viewer. For example, this trajectory may correspond to the RS (right scanning) area 44 in Figure 6. Notice that the slow scan direction is horizontal rather than the vertical direction. LBS PGUs conventionally scan slowly in the vertical direction. The change in slow scanning direction improves the operation with an output modulator. Fourth Image generating unit example This example illustrates RGB lasers with illumination scanning by a steerable mirror. Figure 17 shows an image generating unit comprising an array of lasers 1702 emitting red, green and blue wavelengths; a steerable mirror 1704; a diffuser 1706 and a TFT-LCD 1708r, 1708g, 1708b for each of the wavelengths. The TFT-LCDs are illuminated by one or more scanning emission bands. Multiple laser sources of each wavelength may be arranged in a vertical array, similar to the array discussed above with reference to Figure 11. The output of the lasers is directed to the steerable mirror 1704 having controllable tilt. The steerable mirror may be, for example, a MEMS mirror such as those commercially available from Hamamatsu, or mirror galvanometer such as those commercially available from Thorlabs. Light reflected from the mirror 1704 is focussed by a lens 1710 onto the diffuser 1706 forming an illumination band. This illumination band at the diffuser is then reimaged onto the TFT-LCDs. As the steerable mirror 1704 rotates, the illumination band at the diffuser, and hence the TFT-LCDs, also moves. The diffuser expands the etendue of the laser sources, for example to fill an entrance pupil of a pupil expander (not shown). The steerable mirror 1704 performs a similar function to the rotating mirror 1004 describe above with reference to Figure 10, but the layout in Figure 17 has an advantage that ray angles at the LCDs do not depend on the instantaneous direction of the scanning. In systems such as that of Figure 17, the etendue of the illumination sources is smaller than the etendue in systems with wavelength conversion such as Figure 9A. This is beneficial for light efficiency, but it becomes important for the illumination at each point on an LCD to have repeatable incident direction. Downstream components from the image generating units Figures 18A and 18B are schematic diagrams depicting example arrangements of elements after the image generating unit, for example after the final elements depicted in Figures 9A, 9B, 13A, 13B, 13C, 13D and 17. Figure 18A is a schematic diagram of a display system which uses an optical combiner such as vehicle windshield combiner, to provide a Heads-Up Display (HUD), the image generating unit is shown schematically at box 1802. This represents the systems described above with reference to Figures 9A, 9B, 13A, 13B, 13C, 13D, 16 and 17. The output from all these systems is a collimated image. At 1806, a pupil expander is used to expand the area over which the image is visible, for example this may use image replication to replicate the pupil over a larger area. Any suitable pupil expander may be used, such an image-replicating waveguide. An output modulator 1808 is provided after the pupil expander 1806. The output modulator provides blocking and transmissive region(s) or areas, such as described above with reference to 22 and 122 in Figures 4 and 5. Next, light which passes through the transmission region(s) of the output modulator 1808 enters a windshield combiner 1810, such as a vehicle windshield. The windshield combiner combines the light from the output modulator with light from the real world. The display is visible in an “eye box” 1812 where a viewer’s pupil may be positioned. Figure 18B is a schematic diagram of a HUD, such as vehicle HUD, which uses a mirror-based system rather than the waveguide-based system of Figure 18A. Such a mirror-based system is typically larger than a waveguide-based system but may have a simpler construction. In the system of Figure 18B, the image generating unit 1802’ includes a system of mirrors to provide the HUD, mirror systems are widely used in automotive HUDs for example. The system of mirrors replaces the collimation lens 934 in Figures 9A and 9B. After leaving the system of mirrors, the light from the image generating unit continues through transmissive regions of the output modulator 1808’ to a combiner 1810’, such as vehicle windshield. The display is visible in an eyebox 1812’. Further simplification of the construction of the image generating unit is also possible in some examples of Figure 18B. For example, the relay optics 912-930 could be omitted when the image generator uses the construction of Figure 9B, with an RGB LCD directly adjacent to the phosphor and the output modulator following that. The above embodiments are to be understood as illustrative examples of the invention. Further embodiments of the invention are envisaged. For example, although a single emission band has been considered above, other embodiments may simultaneously scan two separate emission bands from one image (for one eye / target eyebox) with two transmission bands, concurrently. It is to be understood that any feature described in relation to any one embodiment may be used alone, or in combination with other features described, and may also be used in combination with one or more features of any other of the embodiments, or any combination of any other of the embodiments. Furthermore, equivalents and modifications not described above may also be employed without departing from the scope of the invention, which is defined in the accompanying claims.

Claims

1. A display system comprising:an image generating unit;an output modulator positioned after the image generating unit for selectively allowing light to pass; anda controller which is configured to:control the image generating unit to display a first image for viewing by a left eye of the viewer and a second image for viewing by a right eye of the viewer; andcontrol the output modulator, such that:the first image is visible from a first eyebox corresponding to the left eye and the first image is not visible from a second eyebox corresponding to the right eye, andthe second image is visible from the second eyebox and not visible from the first eyebox;wherein:the first image is displayed by scanning at least one first emission band across an image plane generated by the image generating unit, the first emission band consisting of a portion of the first image;the second image is displayed by scanning at least one second emission band across the image plane generated by the image generating unit, the second emission band consisting of a portion of the second image; andthe output modulator is controlled in synchrony with the scanning of the at least one first emission band and the scanning of the at least one second emission band.

2. The display system of claim 1, wherein there is an unused portion of the image plane during the scanning at least one first emission band and the scanning at least one second emission band.

3. The display system of claim 1 or 2, wherein over a time interval, the scanning at least one first emission band and the scanning at least one second emission band issubstantially continuous and the output modulator is controlled by an activation function with one or more limits moving substantially continuously.

4. The display system according to claim 1, 2 or 3, comprising an eye tracking system for determining a position of a viewer’s eyes, wherein the first eyebox and the second eyebox are based on the determined position.

5. The display system according to any preceding claim, comprising a ray splitting element between the image generating unit and the output modulator.

6. The display system according to any preceding claim, wherein the first image has a first correction applied and the second image has a second correction applied, the second correction being different from the first correction.

7. The display system according to any preceding claim, wherein the controller is configured to: apply a blanking period between the display of the first and the second image based on a response time of the output modulator.

8. The display system according to any preceding claim, wherein a scanning direction across the image plane is different for the first emission band and the second emission band.

9. The display system according to any of claims 1 to 7, wherein the first emission band and the second emission band are scanned simultaneously at different positions on the image plane.

10. The display system according to any preceding claims, wherein at least one of the first image or the second image comprise a plurality of sub-images which are displayed sequentially.

11. A method of displaying an image comprising a first image for viewing by a left eye of a viewer and a second image for viewing by a right eye of the viewer; the method comprising:scanning a first emission band of the first image across an image plane;scanning a first transmission edge of an output modulator in synchrony with the scanning of the first emission band, the first transmission edge positioned such that the first emission band is visible to the left eye and not the right eye;scanning a second emission band of the second image across the image plane separately from the first emission band;scanning a second transmission edge of the output modulator in synchrony with the scanning of the second emission band, the second transmission edge positioned such that the second emission band is visible to the right eye and not the left eye.

12. The method according to claim 11, where at least part of the image plane is unused during the scanning of the first emission band and the scanning of the second emission band.

13. The method according to claim 11 or 12, comprising:receiving information of a position of a viewer’s left eye and right eye; anddetermining the relative position of the first emission band and the first transmission edge, and the relative position of the second transmission edge and second emission band, based on the position of the viewer’s left eye and right eye.

14. The method according to claim 13, wherein the information of the position of the viewer’s left eye and right eye is received from an eye-tracking system.

15. The method according to any of claims 11 to 14, wherein:the scanning the first emission band and the scanning the second emission band together have a duration within a display refresh time; andat least one limit of an activation function for the output modulator travels with a velocity such that a transition distance in transmissivity of the output modulator is shortenough for the display system to display images with substantial isolation between an eyebox of the first image and an eyebox of the second image.

16. The method according to any of claims 11 to 15, comprising:replicating an exit pupil of an image generating unit which is scanning the first emission band and the second emission band in an optical path before the scanning the first transmission edge and the scanning the second transmission edge.

17. The method according to any of claims 11 to 16, comprisingapplying a first correction to the first image; and applying a second correction to the second image;wherein the second correction is different from the first correction.

18. The method according to any of claims 11 to 17, comprising:applying a blanking period between the scanning of the first emission band and the scanning of the second emission band based on a response time of the output modulator.

19. The method according to any of claims 11 to 18, wherein the scanning the first emission band is in a first direction across the image plane and the scanning the second emission band is in a second direction, which is different from the first direction, across the image plane.

20. The method according to any of claims 11 to 18, wherein the scanning the first emission band and the scanning the second emission band are simultaneous at different positions on the image plane.

21. The method according to any of claims 11 to 20, wherein at least one of the first image or the second image comprises a plurality of sub-images which are scanned sequentially across the image plane.

22. A display system comprising:an image generating unit;an output modulator positioned after the image generating unit in an optical path for selectively allowing light to pass; anda controller configured to implement the method of any of claims 11 to 21.

Images