Patent Application
20240340330
The present application is concerned with spatially unequal streaming such as occurring in virtual reality (VR) streaming.
VR streaming typically involves transmission of a very high-resolution video. The resolving capacity of the human fovea is around 60 pixels per degree. If transmission of the full sphere with 360°×180° is considered, one would end up by sending a resolution of around 22 k×11 k pixels. Since, sending such high resolution would lead to tremendously high bandwidth requirements, another solution is to send only the viewport shown at the Head Mounted Displays (HMDs), which have FoV of 90°×90°: leading to around a 6 k×6 k pixels video. A trade-off between sending the whole video at the highest resolution and sending only the viewport is to send the viewport at high resolution and some neighboring data (or the rest of the spherical video) at lower resolution or lower quality.
In a DASH scenario, an omni-directional video (aka spherical video) can be offered in such a way that the mixed resolution or mixed quality video described before is controlled by the DASH client. The DASH client only needs to know information that describes how the content is offered.
One example could be to offer different representations with different projections that have asymmetric characteristics, such as different quality and distortion for different parts of the video. Each representation would correspond to a given viewport and would have the viewport encoded with a higher quality/resolution than the rest of the content. Knowing the orientation information (direction of the viewport for which the content has been encoded with a higher quality/resolution) the DASH client can chose one or another representation dynamically to match the viewing direction of the user at any time.
A more flexible option for a DASH client to select such asymmetric characteristic for the omni-directional video would be when the video is split into several spatial regions, with each region being available at different resolution or quality. One option could be to split it into rectangular regions (aka tiles) based on a grid, but other options could be foreseen. In such a case, the DASH client would need some signaling about the different qualities into which the different regions are offered and it could download the different regions at different qualities so that the viewport shown to the user is at a better quality than the other non-shown content.
In any of the previous cases, when user interaction happens and the viewport is changed, the DASH client will need some time to react to user movement and download the content in such a way that matches the new viewport. During the time between the user moves and the DASH client adapts its requests to match the new viewport, the user will see in the viewport some regions in high quality and low quality simultaneously. Though the acceptable quality/resolution difference is content dependent, the quality the user sees is in any case degraded.
Thus, it would be favorable to have a concept at hand which alleviates, or renders more efficient, or even increases the visible quality for the user with respect to partial presentation of spatial scene content streamed by adaptive streaming.
Thus, the object of the present invention to provide concepts for streaming spatial scene content in a spatially unequal manner so that the visible quality for the user is increased, or the processing complexity or necessary bandwidth at the streaming retrieval site is decreased, or to provide concepts for streaming spatial scene content in a manner enlarging the applicability to further application scenarios.
This object is achieved by the subject matter of the pending independent claims.
A first aspect of the present application is based on the finding that streaming media content pertaining to a temporally-varying spatial scene such as a video in a spatially unequal manner may be improved in terms of visible quality at comparable bandwidth consumption and/or computational complexity at a streaming reception site if the media segments selected and retrieved and/or a signalization obtained from the server, provides the retrieving device with hints on a predetermined relationship to be complied with by qualities at which different portions of the temporally-varying spatial scene are encoded into the selected and retrieved media segments. Otherwise, the retrieving device may not know beforehand as to which negative impact the juxtaposition of portions encoded at different quality into the selected and retrieved media segment may have on the overall visible quality experienced by the user. Information contained in the media segments and/or a signalization obtained from the server such as, for instance, within a manifest file (media presentation description) or additional streaming related control messages from server to client such as SAND messages, enable the retrieving device to appropriately select among the media segments offered at the server. In this manner, virtual reality streaming or partial streaming of video content may be made more robust against quality degradation as it could otherwise occur owing to an inadequate distribution of the available bandwidth on to this spatial section of the temporally-varying spatial scene presented to the user.
A further aspect of the present invention is based on the finding that streaming of media content pertaining to a temporally-varying spatial scene such as a video in a spatially unequal manner such as using a first quality at a first portion and a second, lower quality at a second portion or with leaving a second portion being non-streamed, may be improved in visible quality and/or may be made less complex in terms of bandwidth consumption and/or computational complexity at the streaming retrieval side, by determining a size and/or position of the first portion depending on information contained in the media segments and/or a signalization obtained from the server. Imagine, for instance, the temporally-varying spatial scene would be offered at the server at a tile-based manner for tile-based streaming, i.e. the media segments would represent spectral temporal portions of the temporally-varying spatial scene each of which would be a temporal segment of the spatial scene within a corresponding tile of a distribution of tiles into which the spatial scene is sub-divided. In such a case, it is up to the retrieving device (client) to decide as to how to distribute the available bandwidth and/or computational power over the spatial scene, namely, at the granularity of tiles. The retrieving device would perform the selection of the media segments to the extent that a first portion of the spatial scene which follows respectively tracks a temporally-varying view section of the spatial scene, is encoded into the selected and retrieved media segments in a predetermined quality which may, for instance, be the highest quality feasible at the current bandwidth and/or computational power conditions. A spatially neighboring second portion of the spatial scene may, for instance, not be encoded into the selected and retrieved media segments, or may be encoded there into at a further quality, reduced relative to the predetermined quality. In such a situation, it is a computationally complex matter, or even not feasible, to compute a number/count of neighboring tiles, the aggregation of which completely covers the temporally-varying view section irrespective of the view section's orientation. Depending on the projection chosen so as to map the spatial scene onto the individual tiles, the angular scene coverage per tile may vary over this scene and the fact that the individual tiles may mutually overlap, even renders a computation of a count of neighboring tiles sufficient to cover the view section in spatial terms, irrespective of the view section's orientation, more difficult. Accordingly, in such a situation, the aforementioned information could indicate the size of the first portion as a count N of tiles or a number of tiles, respectively. By this measure, the device would be able to track the temporally-varying view section by selecting those media segments having the co-located aggregation of N tiles encoded there at the predetermined quality. The fact that the aggregation of these N tiles sufficiently covers the view section may be guaranteed by way of the information indicating N. Another example would be information contained in the media segments and/or a signalization obtained from the server, which is indicative of the size of the first portion relative to a size of the view section itself. For example, this information could somehow set a “safety zone” or prefetch zone around the actual view section in order to account for movements of the temporally-varying view section. The larger the speed at which the temporally-varying view section moves across the spatial scene, the larger the safety zone should be. Accordingly, the aforementioned information could be indicative of the size of the first portion in a manner relative to a size of the temporally-varying view section such as in an incremental or scaling manner. A retrieving device setting the size of the first portion according to such information would be able to avoid quality degradation which may otherwise occur owing to non-retrieved or low-quality portions of the spatial scene being visible in the view section. Here, it is irrelevant whether this scene is offered in a tile-based manner or in some other manner.
Related to the just-mentioned aspect of the present application, a video bit stream having a video encoded there into, may be made decodable at an increased quality if the video bit stream is provided with a signalization of a size of a focus area within the video onto which a decoding power for decoding the video should be focused. By this measure, a decoder which decodes the video from the bit stream, could focus, or even restrict, its decoding power onto the decoding of the video onto a portion having the size of the focus area signalized in the video bit stream thereby knowing, for instance, that the thus-decoded portion is decodable by the available decoding power, and spatially covers a wanted section of the video. For instance, the size of the focus area thus signalized could be selected to be large enough in order to cover the size of the view section and a movement of this view section taking the decoding latency in decoding the video into account. Or, put differently, a signalization of a recommended preferred view-section area of the video contained in the video bitstream could allow the decoder to treat this area in a preferred manner, thereby allowing the decoder to focus its decoding power accordingly. Irrespective of performing area-specific decoding power focusing, the area signalization may be forwarded to a stage selecting on which media segments to download, i.e. where to place and how to dimension the portion of increased quality.
The first and second aspects of the present application are closely related to a third aspect of the present application according to which the fact that a vast number of retrieving devices stream media content from a server, is exploited, so as to gain information which may subsequently be used in order to appropriately set the aforementioned types of information allowing to set the size, or size and/or position, of the first portion and/or appropriately set the predetermined relationship between the first and second quality. Thus, in accordance with this aspect of the present application, the retrieving device (client) sends-out log messages logging one of a momentaneous measurement or a statistical value measuring a spatial position and/or movement of the first portion, a momentaneous measurement or a statistical value measuring a quality of the temporally-varying spatial scene as far as is encoded into the selected media segments and as far as is visible in a view section, and a momentaneous measurement or statistical value measuring the quality of the first portion or a quality of the temporally-varying spatial scene as far as is encoded into the selected media segments and as far as is visible in a view section. Momentaneous measurements and/or statistical values may be provided with time information concerning the time the respective momentaneous measurement or statistical value has been obtained. The log messages may be sent to the server where the media segments are offered, or to some other device evaluating the inbound log messages so as to update, based thereon, current settings of the aforementioned information used to set the size, or size and/or position, of the first portion and/or derive the predetermined relationship based thereon.
In accordance with a further aspect of the present application, streaming media content pertaining to a temporally-varying spatial scene such as a video, in particular in a tile-based manner, is made more effective in terms of avoidance of unavailing streaming trials by providing a media presentation description which comprises at least one version at which the temporally-varying spatial scene is offered for tile-based streaming, with an indication of benefitting requirements for benefitting from the tile-based streaming the respective version of the temporally-varying spatial scene for each of the at least one version. By this measure, the retrieving device is able to match the benefitting requirements of the at least one version with a device capability of the retrieving device itself or of another device interacting with the retrieving device with respect to tile-based streaming. For instance, the benefitting requirements could relate to decoding capability requirements. That is, if the decoding power for decoding the streamed/retrieved media content would not suffice to decode all media segments needed to cover a view section of the temporally-varying spatial scene, then trying to stream and present the media content would be a waste of time, bandwidth and computational power and accordingly, it would be more effective to not try it in any case. The decoding capability requirements could, for instance, indicate a number of decoder instantiations necessitated for a respective version if, for instance, the media segments relating to a certain tile form a media stream such as a video stream, separate from media segments pertaining to another tile. The decoding capability requirement could, for instance, also pertain to further information such as a certain fraction of decoder instantiations needed to fit to a predetermined decoding profile and/or level, or could indicate a certain minimum capability of a user input device to move in a sufficiently fast manner a viewport/section via which the user sees the scene. Depending on the scene content, a low movement capability may not suffice for the user to look onto the interesting portions of the scene.
A further aspect of the present invention pertains to an extension of streaming of media content pertaining to temporally-varying spatial scenes. In particular, the idea in accordance with this aspect is that a spatial scene may in fact not only vary temporally but also in terms of at least one further parameter suggest, for instance, views and a position, view depth or some other physical parameter. The retrieving device may use adaptive streaming in this context by, depending on a viewport direction and the at least one further parameter, computing addresses of media segments, the media segments describing a spatial scene varying in time and the at least one parameter, and retrieving the media segments using the computed addresses from a server.
The above-outlined aspects of the present application and their advantageous implementations which are the subject of the dependent claims, may be combined individually or all together.
Preferred embodiments of the present application are set forth below with respect to the Figures among which
In order to ease the understanding of the description of embodiments of the present application with respect to the various aspects of the present application,
Further details with respect to server 20, client 10 and the way the spatial content 30 is offered at server 20 is illustrated in
In particular, as shown in
As shown in
Client 10 of
The media segments thus retrieved by client device 40 are forwarded by the latter to the one or more decoders 42 for decoding. In the example of
Thus, as depicted in
It is again emphasized that even the example of a tile-based streaming illustrated in
Finally, differing from a tile-based streaming concept, according to which the media segments which may be individually retrieved by device 40 from server 20, relate to tiles 50 into which scene 30 is spatially subdivided, the media segments offered at server 20 may alternatively, for instance, each having the scene 30 encoded thereinto in a spatially complete manner with a spatially varying sampling resolution, however, having sampling resolution maximum at different spatial positions in scene 30. For instance, that could be achieved by offering at the server 20 sequences of segments 54 relating to a projecting of the scene 30 onto truncated pyramids the truncated tip of which would be oriented into mutually different directions, thereby leading to differently oriented resolution peaks.
Further, as to optionally present stream modifier 38, it is noted that same may alternatively be part of the client 10, or same may even be positioned inbetween, within a network device via which client 10 and server 20 exchange the signals described herein.
After having explained rather generally the system of server 20 and client 10, the functionality of client device 40 with respect to an embodiment in accordance with a first aspect of the present application as described in more detail. To this end, reference is made to
As has also been discussed above, the selector 56 is configured to perform the selection so that the selected media segments have at least a spatial section of the temporally-varying spatial scene and encoded thereinto. The spatial section may continuously cover the complete scene spatially.
Further, selector 56 performs the selection such that the selected media segments have section 62 encoded thereinto in a manner of spatially unequal quality. To be more precise, a first portion 64, indicated by hatching in
It is, not possible for device 40 to assess as to which negative quality degradation may result from preliminarily presenting to the user reduced quality scene content along with the scene content within portion 64 which is of the higher quality, to the user. In particular, a transition between these two qualities results which may be clearly visible to the user. At least, such transitions may be visible depending on the current scene content within section 28. The severity of the negative impact of such a transition within the view of the user is a characteristic of the scene content as offered by server 20 and may not be forecast by device 40.
Accordingly, in accordance with the embodiment of
The media segments selected and finally retrieved by retriever 60 are finally forwarded to the one or more decoders 42 for decoding.
In accordance with a first example, for instance, the signaling mechanism embodied by information 68 involves information 68 indicating to device 40, which may be a DASH client, which quality combinations are acceptable for the offered video content. For example, the information 68 could be a list of quality pairs that indicate to the user or device 40 that the different regions 64 and 66 can be mixed with a maximum quality (or resolution) difference. Device 40 may be configured to inevitably use a certain quality level such as the highest one offered at sever 10, for portion 64 and derive quality levels at which portion 66 may be coded into the selected media segments from information 68 wherein same be contained in form of a list of quality levels for portion 68, for instance.
The information 68 could indicate an endurable value for a measure of a difference between the quality of portion 68 and the quality of portion 64. As a “measure” of the difference in quality, a quality index of the media segments 58, by way of which the same are distinguished in the media presentation description and by way of which the addresses of the same are computed using the computational rule described in the media presentation description, may be used. In MPEG-DASH, the corresponding attribute indicating the quality would be, for instance, at @qualityRanking. Device 40 could take the restriction in selectable quality level pairs at which portions 64 and 66 may be coded into the selected media segments into account in performing the selection.
However, instead of this difference measure, the difference in quality could alternatively be measured, for instance, in bit rate difference, i.e., an endurable difference in bit rate at which portions 64 and 66 are encoded into the corresponding media segments, respectively, assuming that the bit rate usually monotonically increases with increasing quality. The information 68 could indicate allowed pairs of options for qualities at which portions 64 and 66 are encoded into the selected media segments. Alternatively, the information 68 simply indicates allowed qualities for coding portion 66, thereby indirectly indicating allowed or endurable quality differences assuming that main portion 64 is encoded using some default quality such as, for instance, the highest quality possible or available. For instance, information 68 could be a list of acceptable representation IDs or could indicate a minimum bit rate level with respect to the media segments concerning portion 66.
However, a more gradual quality difference could alternatively be desired, wherein, instead of quality pairs, quality groups (more than two qualities) could be indicated, wherein, dependent on the distance to section 28, i.e., the viewport, the quality difference could be increased. That is, the information 68 could indicate the endurable value for the measure of a difference between the qualities for portion 64 and 66 in a manner depending on a distance to view section 28. This could be done by way of a list of pairs of a respective distance to the view section and a corresponding endurable value for the measure of the difference in quality beyond the respective distance. Below the respective distance, the quality difference has to be lower. That is, each pair would indicate for a corresponding distance that a part within portion 66, further away from section 28 than the corresponding distance, may have a quality difference to the quality of portion 64 exceeding the corresponding endurable value of this list entry.
The endurable value may increase within increasing distance to view section 28. The acceptance of the just discussed quality difference is often dependent on the time that these different qualities are shown to the user. For instance, content with a high quality difference might be acceptable if it is only shown for 200 microseconds, while content with a lower quality difference might be acceptable if it is shown for 500 microseconds. Therefore, in accordance with a further example, the information 68 could also include, in addition to the aforementioned quality combinations, for instance, or in addition to the allowed quality difference, a time interval for which the combination/quality difference may be acceptable. In other words, the information 68 may indicate an endurable or maximally allowed difference between the qualities of portions 66 and 64 along with an indication of a maximally allowed time interval for which portion 66 may be shown within the view section 28 concurrently with portion 64.
As already noted previously, the acceptance of quality differences depends on the content itself. For instance, the spatial position of the different tiles 50 has an influence on the acceptance. Quality differences in a uniform background region with low frequency signals are expected to be more acceptable than quality differences in a foreground object. Furthermore, the position in time also has an influence on the acceptance rate due to changing content. Therefore, according to another example, signals forming information 68 are sent to device 40 intermittently such as, for instance, per representation or period in DASH. That is, the predetermined relationship indicated by information 68 may be intermittently updated. Additionally and/or alternatively, the signaling mechanism realized by information 68 may vary in space. That is, the information 68 may be made spatially dependent such as, by way of an SRD parameter in DASH. That is, different predetermined relationships may be indicated by information 68 for different spatial regions of scene 30.
The embodiment of device 40, as described with respect to
In the following, a further embodiment for an advantageous implementation of device 40 is described. In particular,
That is, device 40 comprises a selector 56, a retriever 60 and a deriver 66. The selector 56 selects from the media segments 58 of the plurality 46 offered by server 20 and retriever 60 retrieves the selected media segments from the server.
In any case, selector 56 selects, similar to the description with respect to
In such a situation, it is also not easy to forecast by client 40 as to how large section 62 or portion 64 should be. Depending on the scene content, most users may act similarly in moving view section 28 across scene 30 and, accordingly, the same applies to the interval of view section 28 speeds at which view section 28 may presumably move across scene 30. Accordingly, in accordance with the embodiment of
That is, in accordance with the embodiments of
In accordance with an example, information 70 could, for instance, include a probability for a given movement speed of viewport of view section 28. Information 74 could, as already denoted above, result within the media presentation description made available for client device 40 which may, for instance, be a DASH client, or some in-band mechanisms may be used to convey information 74 such as event boxes, i.e., EMSG or SAND messages in case of DASH. The information 74 could also be included in any container format such as ISO file format or transport format beyond MPEG-DASH such as MPEG-2TS. It could also be conveyed in the video bitstream such as in SEI messages as described later. In other words, the information 74 may indicate a predetermined value for a measure of a spatial speed of view section 28. In this manner, the information 74 is indicative of the size of portion 64 in the form of a scaling, or in the form of an increment relative to a size of view section 28. That is, information 74 starts from some “base size” for portion 64 necessary to cover the size of section 28 and increases this “base size” appropriately such as incrementally or by scaling. For example, the aforementioned movement speed of view section 28 could be used to correspondingly scale the circumference of a current position of view section 28 so as to determine, for instance, the furthest positions of the circumference of view section 28 along any spatial direction feasible after this time interval, for example, determining the latency in adjusting the spatial location of portion 64 such as, for instance, the time duration of the temporal segments 54 corresponding to the temporal length of media segments 58. The speed times this time duration adds to the circumference of a current position of viewport 28, omni-directional, could thus result into such a worst case circumference and could be used to determine an enlargement of portion 64 relative to some minimum expansion of portion 64 assuming a non-moving viewport 28.
Information 74 may even be related to an evaluation of statistics of user behavior. Later on, embodiments are described which are suitable for feeding such an evaluation process. For instance, information 74 could indicate maximum speeds with respect to certain percentages of users. For example, information 74 could indicate that 90% of users move at a speed lower than 0.2 rad/s and 98% of users move at a speed lower than 0.5 rad/s. The information 74 or the messages carrying the same could be defined in such a way that probability-speed pairs are defined or a message could be defined that signals the maximum speed for a fixed percentage of users, e.g., always for 99% of the users. The movement speed signaling 74 could additionally comprise directional information, i.e., an angle in 2D or 2D plus depth in 3D also known as light field application. Information 74 would indicate different probability-speed pairs for different movement directions.
In other words, information 74 may apply to a given time span such as, for instance, the temporal length of a media segment. It may consist of trajectory-based (x percentile, average user path) or velocity-based pairs (x percentile, speed) or distance-based pairs (x percentile, aperture/diameter/preferred) or area-based pairs (x percentile, recommended preferred area) or single maximal boundary values for path, velocity, distance or preferred area. Instead of relating the information to percentiles, a simple frequency ranking could be done according to most of the users move at a certain speed, second most users move at a further speed and so on. Additionally or alternatively, information 74 is not restricted to indicate the speed of view section 28, but could likewise indicate an advantageous area to be viewed respectively to direct the portion 62 and/or 64 which is sought to track view section 28 to, with or without an indication about statistical significance of the indication such as percentage of users having complied with that indication or indication of whether the indication coincides with the user viewing speeds/view sections having been logged most often, and with or without temporal persistence of the indication. Information 74 could indicate another measure of the speed of view section 28, such as a measure for a travelling distance of view section 28 within a certain period in time, such as within a temporal length of the media segments or, in more detail, the temporal length of temporal segments 54. Alternatively, information 74 could be signaled in a manner distinguishing between certain directions of movement into which view section 28 may travel. This pertains to both an indication of speed or velocity of view section 28 into a certain direction as well as the indication of traveled distance of view section 28 with respect to a certain direction of movement. Further, the expansion of portion 64 could be signaled by way of information 74 directly, either omni-directionally or in a manner discriminating different movement directions. Furthermore, all of the just outlined examples may be modified, in that the information 74 indicates these values along with a percentage of users for which these values suffice in order to account for their statistical behavior in moving view section 28. In this regard, it should be noted that the view speed, i.e., the speed of view section 28 may be considerable and is not restricted to speed values for a user head, for instance. Rather, the view section 28 could be moved depending on the user's eye movement, for instance, in which case the view speed may be considerably larger. The view section 28 could also be moved according to another input device movement such as according to the movement of a tablet or the like. As all these “input possibilities” enabling the user to move section 28 result in different expected speeds of view section 28, information 74 may even be designed such that it distinguishes between different concepts for controlling the movement of view section 28. That is, information 74 could indicate or be indicative of the size of portion 64 in a manner indicating different sizes for different ways of controlling the movement of view section 28 and device 40 would use the size indicated by information 74 for the correct view section control. That is, device 40 gains knowledge about the way view section 28 is controlled by the user, i.e., checks whether view section 28 is controlled by head movement, eye movement or tablet movement or the like and sets the size in accordance with that part of information 74 which corresponds to this kind of view section control.
Generally, the movement speed can be signaled per content, period, representation, segment, per SRD position, per pixel, per tile, e.g., on any temporal or spatial granularity or the like. The movement speed can also be differentiated in head movement and/or eye movement, as just outlined. Further, the information 74 about user movement probability may be conveyed as a recommendation about high resolution prefetch, i.e., video area outside user viewport, or spherical coverage.
The latter possibility of signaling the size of portion 64 is depicted in
In any case, at the very end of
With respect to
The motivation is as follows.
In order to be able to derive statistics, such as the most interesting regions or speed-probability pairs, as described previously, reporting mechanisms from users are required. Additional DASH Metrics to the ones defined in Annex D of ISO/IEC23009-1 are used.
One metric would be the FoV of Client as DASH Metric, where DASH clients send back to a Metric Server (it could be the same as the DASH server or another one) the characteristics of the end device in term of FoV.
One Metric would be ViewportList, where DASH clients send back to a Metric Server (it could be the same as the DASH server or another one) the viewport watched by each client in time. An instantiation of such a message could be as follows:
For the Viewport (region of interest) message, the DASH client could be asked to report whenever a Viewport change occurs, with potentially a given granularity (with or without avoiding reporting of very small movements) or with a given periodicity.
Such message could be included in the MPD as an attribute @reportViewPortPeriodicity or an element or descriptor. It could be also indicated out of band, such as with a SAND message or any other means.
Viewport can also be signalled on tile granularity.
Additionally or alternatively, log messages could report on other current scene related parameters changing responsive to user input, such as any of the parameters discussed below with respect to
Another metric would be the ViewportSpeedList, where DASH clients indicate the movement speed for a given viewport in time when a movement happens.
This message would be sent only if the client performs a viewport movement. However, the server could, as well as for the previous case, indicate that the message should be only sent if the movement is significant. Such a configuration could be something like @minViewportDifferenceForReporting signalling the size in pixels or angle or any other magnitude that needs to have changed for a message to be sent.
Another important thing for a VR-DASH service, where asymmetric quality is offered as described above, is to evaluate how fast users switch from an asymmetric representation or a set of unequal quality/resolution representations for a Viewport to another representation or set of representation more adequate for another viewport. With such a metric, Servers could derive statistics that help them to understand relevant factors that impact the QoE. Such a metric could look like:
Alternatively, the duration described before could be given as an average:
All of these metrics could additionally have the time at which the measurement has been performed, as for other DASH Metrics.
In some cases, it could happen that if unequal quality content is downloaded and bad quality (or a mixture of good and bad quality) is shown for a long enough time (which could be only a couple of seconds) the user is unhappy and leaves the session. Conditional to leaving the session the user could send a message of the quality shown in the last x time-interval:
Alternatively, a max-quality difference could be reported or the max_quality and min_quality of the viewport
As became clear from the above description relating to
Imagine a tile-based streaming service using video with a cubic projection as depicted in
While the above description rather generally, inter alias, indicated a feedback or log message which indicates the quality at which the video is presented to the user in the viewport, in the following, a more specific and advantageous metric applicable in this regard shall be outlined. The metric described now could be reported back from client side and be referred to as Effective Viewport Resolution. It is supposed to indicate to the service operator the effective resolution in the viewport of the client. In case the reported effective viewport resolutions indicates that a user was only presented a resolution towards the resolution of the low-resolution tiles, a service operator could change the tiling configuration, resolution ratios or segment length accordingly to achieve a higher effective viewport resolution.
One embodiment would be the average pixel count in the viewport 28, measured in the projection plan where the pixel array of the tiles 50 as coded into the segments 58 lies. The measurement could distinguish between, or be specific for, horizontal direction 204 and vertical direction 206 in relation to the covered Field of View (FoV) of the viewport 28. The following table shows a possible example for suitable syntax and semantics which could be contained in log messages so as to signal the outlined viewport quality measure.
The breakdown into horizontal and vertical direction could be left off with using a scalar value for the average pixel count instead. Along with an indication of the aperture or size of the viewport 28 which might also be reported to the recipient of the log messages, namely evaluator 82, the average count is indicative of the pixel density within the viewport.
It may be advantageous to reduce the FoV considered for the metric to be smaller than the FoV of the viewport actually presented to the user, thereby excluding areas towards the boundaries of the viewport that are only used for peripheral vision and hence do not have an impact on subjective quality perception. This alternative is illustrated by a dashed line 202 which encircles pixels lying in such a central section of the viewport 28. The reporting of the considered FoV 202 for the reported metric in relation to the viewport's 28 overall FoV, might also be signalled to the log message recipient 82. The following table shows a corresponding extension of the previous example.
According to a further embodiment, the average pixel density is not measured by averaging the quality in a spatially uniform manner in the projection plane as effectively the case in the example described so far with respect to the examples containing EffectiveFoVResolutionH/V, but in a manner weighting this averaging in a non-uniform manner over the pixels, i.e. the projection plane. The averaging may be performed in a spherically uniform manner. As an example, averaging may be performed uniform with respect to sample points distributed as the circles 198 are. In other words, the averaging may performed by weighting the local densities by a weight decreasing in a quadratic manner with increasing local projection plane distance and increasing according to sine of local tilt of the projection against the line connecting with the view point. The message may include an optional (flag controlled) step to adjust for the inherent oversampling of some of the available projections (such as Equirectangular Projection), e.g. by using an uniform sphere sampling grid. Some projections do not have a big oversampling issue and forcing computing removal of oversampling might lead to unnecessary complexity issues. This must not be limited to Equirectangular Projection. Reporting does not need to distinguish horizontal and vertical resolution but can combine them. The following gives one embodiment.
Applying an equiangular uniformity in averaging, is illustrated in
In the following, an embodiment for a further sort of log message is described which is related to DASH clients 10 having multiple media buffers 300 as exemplarily depicted in
Here, representations would be offered at the server and advertised in the MPD for download, each of which relates to one tile 50, i.e. one section of the scene. Representations relating to one tile 50, but encoding this tile 50 at different qualities would be summarized in adaptation set which grouping is optional, but exactly this grouping is used for association to buffers. Thus, according to this example, there would be one buffer per tile 50, or in other words, per viewport (view section) encoding.
Another representation set and distribution would be
According to this example, each representation would cover the whole region, but the high quality region would focus onto one hemisphere, while a lower quality is used for the other. Representations merely differing in the exact qualities spent in this manner, i.e. equaling in higher quality hemisphere's location, would be collected in one adaptation set and are, according to this characteristic, distributed onto the buffers, here exemplarily six.
Thus, the following description assumes that such distribution onto buffers according to different viewport encodings, video sub-region such as tiles, associated with AdaptationSets or the like is applied.
The advantage therefrom is that buffer fullness of multiple media buffers on client-side can be reported with a metric and be identified and associated to a type of buffer. The association types are for example:
One embodiment of this invention is given in Table 1 that defines a metric for reporting buffer level status events for each buffer with identification and association.
A further embodiments using viewport depending encodings is as follows:
In a viewport-dependent streaming scenario, a DASH client downloads and prebuffers several media segments related to a certain viewing orientation (viewport). If the amount of prebuffered content is too high, and the client changes its viewing orientation, the portion of the prebuffered content to be played out after the viewport change is not presented and the respective media buffer is purged. This scenario is depicted in
Another embodiment would relate to a traditional video streaming scenario with multiple representation (quality/bitrates) of the same content and may be spatially uniform quality at which the video content is encoded.
The distribution could then look like:
That is, here, each representation would cover the whole scene for instance which may not be a panoramic 360 scene, but at different quality, namely a spatially uniform quality, and these representations would be distributed onto the buffers individually. All the examples set out in the last three tables should be treated as not limiting the manner at which segments 58 of representations offered at the server are distributed onto the buffers. Different ways exist and the rule may be based on membership of segments 58 to representations, membership of segments 58 to adaptation sets, the direction of locally increased quality of a spatially unequally coding of the scene into the representation which the respective segment belongs to, the quality at which a scene is encoded into the respective segment belongs to and so forth.
A client could maintain a buffer per representation and, upon experiencing a rise in available throughput, decide to purge the remaining low-quality/bitrate media buffer before playback and download high quality media segments of time durations within the already existing low-quality/bitrate buffer. Similar embodiments can be constructed for streaming based on tiles and viewport dependent encodings.
A service operator may be interested in understanding what amount and which data was downloaded without being presented as this introduces cost without gain on server-side and decreases quality on client side. Therefore, the invention is to provide a reporting metric that correlates the two events ‘media download’ and ‘media presentation’ to be easily interpreted. This invention avoids that verbosely reported information about each and every media segment download and playback status is to be analysed and allows efficient reporting of purging events only. The invention also includes identification of the buffer and association to the type as above. An embodiment of the invention is given in Table 2.
Although the internal structure of device 40 may differ from the one depicted in
Matcher 92 matches the benefitting requirements and thus, selects a corresponding version or rejects all versions, completely.
The benefitting requirements may, however, additionally concern profiles/levels which the one or more decoders 42 have to be able to cope with. For example, a DASH MPD includes multiple positions that allow for indicating profiles. Typical profiles describe attributes, elements that can be present at the MPD, as well as video or audio profiles of the streams that are offered for each representation.
Further examples for the benefiting requirements concern, for example, the client's side capability to move the viewport 28 across the scene. The benefiting requirements may indicate a necessary viewport speed which should be available for the user to move the viewport so to be able to really enjoy the offered scene content. The matcher would check, for instance, whether this requirement is met be the, for example, plugged-in user input device such as HMD 26. Alternatively, a set of “sufficient types of input devices” could be indicated by way of the benefiting requirements assuming that different types of input devices for moving a viewport are associated with typical movement speeds in orientation sense.
In a tiled streaming service of spherical video, there are too many configuration parameters that can be dynamically set, e.g. number of qualities, number of tiles. In a case where tiles are independent bitstreams that need to be decoded by separate decoders, if the number of tiles is too high, it might be impossible for hardware devices with few decoders to decode all bitstreams simultaneously. A possibility is to leave this as a degree of freedom and that a DASH device parses all possible representations and counts how many decoders are required to decode all the representations or a given number that cover the FoV of the devices and thus derives whether it is possible or not for that DASH client to consume the content. However, a more clever solution for interoperability and capability negotiations is to use signaling in the MPD that is mapped to a kind of profile that is used as a promise to the client that if a profile is supported, the offered VR content can be consumed. Such a signaling should be done in the form of a URN, such as urn::dash-mpeg::vr::2016, which can be packed either at the MPD level or at adaptation sets. This profiling would mean that N decoders at X profile are enough for consuming the content. Depending on the profiles the DASH client could ignore or accept an MPD or a part of an MPD (adaptation set). In addition, there are several mechanisms that do not include all information such as Xlink or MPD chaining, where few signaling for selection are made available. In such a circumstance the DASH client would not be able to derive whether it can consume the content or not. Exposing the decoding capabilities in terms of number of decoders and profile/level of each decoder by means of such an urn (or something similar) is required, such that the DASH client can now, whether performing Xlink or MPD Chaining or a similar mechanism, makes sense. The signaling could also mean different operation points such as N decoders with X profile/level or Z decoders with Y profile/level.
As an alternative, X may be a parameter defining a view depth, i.e. a distance from the scene center 100 in radial sense. While offering scene in different versions differing in view center portion X allows for the user “walking” through the scene, offering scene in different versions differing in view depth would allow for the user “radially zoom” back and forth through the scene.
For multiple, non-concentrical viewports the MPD thus may use a further signaling of the position of the current viewport. Signaling can be done on segment, representation or period level, or the like.
Non-concentric spheres: in order for a user to move, the spatial relationship of the different spheres should be signaled in the MPD. This could be done by means of coordinates (x,y,z) in arbitrary units in relation to the sphere diameter. In addition, the diameter of a sphere should be indicated for each sphere. A sphere could be “good-enough” to be used for a user that is in its center plus an additional space for which the content would be ok. If a user would move beyond that signaled diameter, another sphere should be used for showing the content.
Exemplary signaling of viewport could be done relative to a predefined center point in space. Each viewport is signaled relative to that center point. In MPEG-DASH, this can be signaled e.g. in the AdaptationSet element.
AdaptationSet
ViewPortPosition
ViewPortPosition
@viewport_x
@viewport_y
@viewport_z
@viewport_diameter
Lastly,
Although some aspects have been described in the context of an apparatus, it is clear that these aspects also represent a description of the corresponding method, where a block or device corresponds to a method step or a feature of a method step. Analogously, aspects described in the context of a method step also represent a description of a corresponding block or item or feature of a corresponding apparatus. Some or all of the method steps may be executed by (or using) a hardware apparatus, like for example, a microprocessor, a programmable computer or an electronic circuit. In some embodiments, one or more of the most important method steps may be executed by such an apparatus.
The signals occurring above such as the streamed signal, the MPD or any other of the mentioned signals can be stored on a digital storage medium or can be transmitted on a transmission medium such as a wireless transmission medium or a wired transmission medium such as the Internet.
Depending on certain implementation requirements, embodiments of the invention can be implemented in hardware or in software. The implementation can be performed using a digital storage medium, for example a floppy disk, a DVD, a Blu-Ray, a CD, a ROM, a PROM, an EPROM, an EEPROM or a FLASH memory, having electronically readable control signals stored thereon, which cooperate (or are capable of cooperating) with a programmable computer system such that the respective method is performed. Therefore, the digital storage medium may be computer readable.
Some embodiments according to the invention comprise a data carrier having electronically readable control signals, which are capable of cooperating with a programmable computer system, such that one of the methods described herein is performed.
Generally, embodiments of the present invention can be implemented as a computer program product with a program code, the program code being operative for performing one of the methods when the computer program product runs on a computer. The program code may for example be stored on a machine readable carrier.
Other embodiments comprise the computer program for performing one of the methods described herein, stored on a machine readable carrier.
In other words, an embodiment of the inventive method is, therefore, a computer program having a program code for performing one of the methods described herein, when the computer program runs on a computer.
A further embodiment of the inventive methods is, therefore, a data carrier (or a digital storage medium, or a computer-readable medium) comprising, recorded thereon, the computer program for performing one of the methods described herein. The data carrier, the digital storage medium or the recorded medium are typically tangible and/or non-transitionary.
A further embodiment of the inventive method is, therefore, a data stream or a sequence of signals representing the computer program for performing one of the methods described herein. The data stream or the sequence of signals may for example be configured to be transferred via a data communication connection, for example via the Internet.
A further embodiment comprises a processing means, for example a computer, or a programmable logic device, configured to or adapted to perform one of the methods described herein.
A further embodiment comprises a computer having installed thereon the computer program for performing one of the methods described herein.
A further embodiment according to the invention comprises an apparatus or a system configured to transfer (for example, electronically or optically) a computer program for performing one of the methods described herein to a receiver. The receiver may, for example, be a computer, a mobile device, a memory device or the like. The apparatus or system may, for example, comprise a file server for transferring the computer program to the receiver.
In some embodiments, a programmable logic device (for example a field programmable gate array) may be used to perform some or all of the functionalities of the methods described herein. In some embodiments, a field programmable gate array may cooperate with a microprocessor in order to perform one of the methods described herein. Generally, the methods are preferably performed by any hardware apparatus.
The apparatus described herein may be implemented using a hardware apparatus, or using a computer, or using a combination of a hardware apparatus and a computer.
The apparatus described herein, or any components of the apparatus described herein, may be implemented at least partially in hardware and/or in software.
The methods described herein may be performed using a hardware apparatus, or using a computer, or using a combination of a hardware apparatus and a computer.
The methods described herein, or any components of the apparatus described herein, may be performed at least partially by hardware and/or by software.
Some example solutions are listed below:
81. Streaming server for streaming media content pertaining a temporally varying spatial scene (30), configured to
115. Method for streaming media content pertaining a temporally varying spatial scene, comprising
The above described embodiments are merely illustrative for the principles of the present invention. It is understood that modifications and variations of the arrangements and the details described herein will be apparent to others skilled in the art. It is the intent, therefore, to be limited only by the scope of the impending patent claims and not by the specific details presented by way of description and explanation of the embodiments herein.
| Number | Date | Country | Kind |
|---|---|---|---|
| 16193601.8 | Oct 2016 | EP | regional |
| 17180403.2 | Jul 2017 | EP | regional |
This application is a continuation of U.S. application Ser. No. 17/935,824, filed on Sep. 27, 2022, which is a continuation of U.S. application Ser. No. 17/669,260, filed on Feb. 10, 2022, now U.S. Pat. No. 11,516,273, which is a continuation of U.S. patent application Ser. No. 16/819,957, filed Mar. 16, 2020, now U.S. Pat. No. 11,283,850, which is a continuation of U.S. patent application Ser. No. 16/381,912, filed Apr. 11, 2019, now U.S. Pat. No. 11,218,530, which is a continuation of International Application No. PCT/EP2017/075971, filed Oct. 11, 2017, which is incorporated herein by reference in its entirety, and additionally claims priority from European Application No. 16193601.8, filed Oct. 12, 2016, and from European Application No. 17180403.2, filed Jul. 8, 2017, which are also incorporated herein by reference in their entirety.
| Number | Date | Country | |
|---|---|---|---|
| Parent | 17935824 | Sep 2022 | US |
| Child | 18747320 | US | |
| Parent | 17669260 | Feb 2022 | US |
| Child | 17935824 | US | |
| Parent | 16819957 | Mar 2020 | US |
| Child | 17669260 | US | |
| Parent | 16381912 | Apr 2019 | US |
| Child | 16819957 | US | |
| Parent | PCT/EP2017/075971 | Oct 2017 | WO |
| Child | 16381912 | US |
