dash-3d-tmp/phd
2
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Work

Browse Source
This commit is contained in:
Thomas Forgione
2019-10-03 12:03:42 +02:00
parent aeb78691db
commit a71b0be28b
15 changed files with 485 additions and 520 deletions
Download Patch File Download Diff File Expand all files Collapse all files

339
src/dash-3d/client.tex
View File

@@ -6,6 +6,78 @@ In this section, we specify a DASH NVE client that exploits the preparation of t
The generated MPD file describes the content organization so that the client gets all the necessary information to make educated decisions and query the 3D content it needs according to the available resources and current viewpoint.
A camera path generated by a particular user is a set of viewpoint $v(t_i)$ indexed by a continuous time interval $t_i \in [t_1,t_{end}]$.
\fresh{}
All DASH clients are built from the same basic bricks, as shown in Figure~\ref{d3:dash-scheme}:
\begin{itemize}
\item the \emph{access client}, which is the module that deals with making requests and receiving responses;
\item the \emph{segment parsers}, which decodes the data downloaded by the access client, whether it be materials, geometry or textures;
\item the \emph{control engine}, which analyses the bandwidth to dynamically adapt to it;
\item the \emph{media engine}, which renders the multimedia content and the user interface to the screen.
\end{itemize}
\begin{figure}[ht]
\centering
\begin{tikzpicture}
% Server
\draw[rounded corners=5pt,fill=Pink] (-10, 0) rectangle (-3, 7.5);
\node at (-9, 7) {Server};
% Segments
\begin{scope}[shift={(0.5,0.5)}]
\foreach \x in {0,...,3}
{
\draw [fill=Bisque](\x/2-7.5, 1.5-\x/2) rectangle (\x/2-5.5, 6-\x/2);
\node at (\x/2-6.5, 5.5-\x/2) {\fcolorbox{black}{LightBlue}{Segment}};
\node at (\x/2-6.5, 4.75-\x/2) {\fcolorbox{black}{LightBlue}{Segment}};
\draw [fill=LightBlue] (\x/2-6.5, 3.825-\x/2) circle (2pt) {};
\draw [fill=LightBlue] (\x/2-6.5, 3.325 -\x/2) circle (2pt) {};
\draw [fill=LightBlue] (\x/2-6.5, 2.825 -\x/2) circle (2pt) {};
\node at (\x/2-6.5, 2-\x/2) {\fcolorbox{black}{LightBlue}{Segment}};
}
\end{scope}
% MPD
\draw[fill=LightBlue] (-9.5, 6.5) rectangle (-7.5, 0.5);
\node at(-8.5, 3.5) {MPD};
% Client
\draw[rounded corners=5pt, fill=LemonChiffon] (-2, 0) rectangle (3, 7.5);
\node at (-0.5, 7) {DASH client};
% Access client
\draw[fill=PaleGreen] (-1.5, 0.5) rectangle (2.5, 1.5);
\node at (0.5, 1) {Access Client};
% Media engine
\draw[fill=PaleGreen] (-1.5, 5.5) rectangle (2.5, 6.5);
\node at (0.5, 6) {Media Engine};
% Control engine
\draw[fill=PaleGreen] (-1.5, 2) rectangle (0.25, 5);
\node[align=center] at (-0.625, 3.5) {Control \\ Engine};
% Segment parser
\draw[fill=PaleGreen] (0.75, 2) rectangle (2.5, 5);
\node[align=center] at (1.625, 3.5) {Segment \\ Parser};
% Access client to server
\draw[double arrow=5pt colored by RoyalBlue and white] (-3.25, 1.0) -- (-1.0, 1.0);
% Access client to control engine
\draw[double ended double arrow=5pt colored by RoyalBlue and white] (-0.625, 1.25) -- (-0.625, 2.5);
% Acces client to segment parser
\draw[double arrow=5pt colored by RoyalBlue and white] (1.625, 1.25) -- (1.625, 2.5);
% Segment parser to media engine
\draw[double arrow=5pt colored by RoyalBlue and white] (1.625, 4.5) -- (1.625, 5.75);
\end{tikzpicture}
\caption{DASH client-server architecture\label{d3:dash-scheme}}
\end{figure}
\copied{}
The DASH client first downloads the MPD file to get the material (.mtl) file containing information about all the geometry and textures available for the entire 3D model.
At time instance $t_i$, the DASH client decides to download the appropriate segments containing the geometry and the texture to generate the viewpoint $v(t_{i+1})$ for the time instance $t_{i+1}$.
@@ -131,3 +203,270 @@ s^{\texttt{GREEDY}}_i= \argmax{s \in \mathcal{S} \backslash \mathcal{B}_i \cap \
\label{d3:greedy}
\end{equation}
\fresh{}
\subsection{JavaScript client\label{d3:js-implementation}}
In order to be able to evaluate our system, we need to collect traces and perform analyses on them.
Since our scene is large, and since the system we are describing allows navigating in a streaming scene, we developed a web client that implements our utility metrics and policies.
\subsubsection{Media engine}
Of course, in this work, we are concerned about performance of our system, and we will not be able to use the normal geometries described in Section~\ref{f:geometries}.
However, the way our system works, the way changes happen to the 3D content is always the same: we only add faces and textures to the model.
Therefore, we made a class that derives BufferGeometry, and that makes it more convenient for us.
\begin{itemize}
\item It has a constructor that takes as parameter the number of faces: it allocates all the memory needed for our buffers so we do not have to reallocate it which would be inefficient.
\item It keeps track of the number of faces it is currently holding: it can then avoid rendering faces that have not been filled and knows where to put new faces.
\item It provides a method that adds a face to the geometry.
\item It also keeps track of what part of the buffers has been transmitted to the GPU\@: THREE.js allows us to set the range of the buffer that we want to update and we are able to update only what is necessary.
\end{itemize}
\paragraph{Our 3D model class.\label{d3:model-class}}
As said in the previous subsections, a geometry and a material a bound together in a mesh.
This means that we are forced to have has many meshes as there are materials in our model.
To make this easy to manage, we made a \textbf{Model} class, that holds everything we need.
We can add vertices, faces, and materials to this model, and it will internally deal with the right geometries, materials and meshes.
In order to avoid having many models that have the same material which would harm performance, it automatically merges faces that share the same material in the same buffer geometry, as shown in Figure~\ref{d3:render-structure}.
\begin{figure}[ht]
\centering
\begin{tikzpicture}
\node[align=center] at(1.5, -1) {DASH-3D\\Structure};
\node at(-1.5, -3) {\texttt{seg0.obj}};
\draw[fill=Pink] (0, -2) rectangle (3, -3);
\node at(1.5, -2.5) {Material 1};
\draw[fill=PaleGreen] (0, -3) rectangle (3, -4);
\node at(1.5, -3.5) {Material 2};
\node at(-1.5, -6) {\texttt{seg1.obj}};
\draw[fill=Pink] (0, -5) rectangle (3, -7);
\node at(1.5, -6) {Material 1};
\node at(-1.5, -9) {\texttt{seg2.obj}};
\draw[fill=PaleGreen] (0, -8) rectangle (3, -9);
\node at(1.5, -8.5) {Material 2};
\draw[fill=LightBlue] (0, -9) rectangle (3, -10);
\node at(1.5, -9.5) {Material 3};
\node[align=center] at (7.5, -1) {Renderer\\Structure};
\node at(10.5, -3.5) {Object 1};
\draw[fill=Pink] (6, -2) rectangle (9, -5);
\node at(7.5, -3.5) {Material 1};
\node at(10.5, -7) {Object 2};
\draw[fill=PaleGreen] (6, -6) rectangle (9, -8);
\node at(7.5, -7) {Material 2};
\node at(10.5, -9.5) {Object 3};
\draw[fill=LightBlue] (6, -9) rectangle (9, -10);
\node at(7.5, -9.5) {Material 3};
\node[minimum width=0.5cm,minimum height=2cm] (O1) at (6.25, -3.5) {};
\draw[-{Latex[length=3mm]}, color=FireBrick] (3, -2.5) -- (O1);
\draw[-{Latex[length=3mm]}, color=FireBrick] (3, -6) -- (O1);
\node[minimum width=0.5cm,minimum height=2cm] (O2) at (6.25, -7) {};
\draw[-{Latex[length=3mm]}, color=DarkGreen] (3, -3.5) -- (O2);
\draw[-{Latex[length=3mm]}, color=DarkGreen] (3, -8.5) -- (O2);
\node[minimum width=3cm,minimum height=2cm] (O3) at (7.5, -9.5) {};
\draw[-{Latex[length=3mm]}, color=RoyalBlue] (3, -9.5) -- (O3);
\node at (1.5, -10.75) {$\vdots$};
\node at (7.5, -10.75) {$\vdots$};
\end{tikzpicture}
\caption{Reordering of the content on the renderer\label{d3:render-structure}}
\end{figure}
\subsubsection{Access client}
In order to be able to implement our DASH-3D client, we need to implement the access client, which is responsible for deciding what to download and download it.
To do so, we use the strategy pattern, as shown in Figure~\ref{d3:dash-loader}.
We have a base class named \texttt{LoadingPolicy} that contain some attributes and functions to keep data about what has been downloaded that a derived class can use to make smart decisions, and exposes a function named \texttt{nextSegment} that takes two arguments:
\begin{itemize}
\item the MPD, so that a strategy can know all the metadata of the segments before making its decision;
\item the camera, because the best segment depends on the position of the camera.
\end{itemize}
The greedy, greedy predictive and proposed policies from the previous chapter are all classes that derive from \texttt{LoadingPolicy}.
Then, the main class responsible for the loading of segments is the \texttt{DashLoader} class.
It uses \texttt{XMLHttpRequest}s, which are the usual way of making HTTP requests in JavaScript, and it calls the corresponding parser on the results of those requests.
The \texttt{DashLoader} class accepts as parameter a function that will be called each time some data has been downloaded and parsed: this data can contain vertices, texture coordinates, normals, materials or textures, and they can all be added to the \texttt{Model} class that we described in Section~\ref{d3:model-class}.
\begin{figure}[ht]
\centering
\begin{tikzpicture}[scale=0.65]
\draw (0, 0) rectangle (5, -4);
\draw (0, -1) -- (5, -1);
\node at (2.5, -0.5) {DashClient};
\node[right] at (0, -1.5) {\scriptsize loadNextSegment()};
\draw (5, -2) -- (8, -2);
\draw (8, 0) rectangle (14, -4);
\draw (8, -1) -- (14, -1);
\node at (11, -0.5) {LoadingPolicy};
\node[right] at (8, -1.5) {\scriptsize nextSegment(mpd, camera)};
\draw (1, -6) rectangle (7, -10);
\draw (1, -7) -- (7, -7);
\node at (4, -6.5) {Greedy};
\node[right] at (1, -7.5) {\scriptsize nextSegment(mpd, camera)};
\draw (8, -6) rectangle (14, -10);
\draw (8, -7) -- (14, -7);
\node at (11, -6.5) {GreedyPredictive};
\node[right] at (8, -7.5) {\scriptsize nextSegment(mpd, camera)};
\draw (15, -6) rectangle (21, -10);
\draw (15, -7) -- (21, -7);
\node at (18, -6.5) {Proposed};
\node[right] at (15, -7.5) {\scriptsize nextSegment(mpd, camera)};
\draw[-{Triangle[open, length=3mm, width=3mm]}] (4, -6) -- (4, -5) -- (11, -5) -- (11, -4);
\draw (11, -6) -- (11, -5) -- (8, -5);
\draw (18, -6) -- (18, -5) -- (8, -5);
\end{tikzpicture}
\caption{Class diagram of our DASH client\label{d3:dash-loader}}
\end{figure}
\subsubsection{Performance}
In JavaScript, there is no way of doing parallel computing without using \emph{web workers}.
A web worker is a script in JavaScript that runs in the background, on a separate thread and that can communicate with the main script by sending and receiving messages.
Since our system has many tasks to do, it seems natural to use workers to manage the streaming without impacting the framerate of the renderer.
However, what a worker can do is very limited, since it cannot access the variables of the main script.
Because of this, we are forced to run the renderer on the main script, where it can access the HTML page, and we move all the other tasks to the worker (the access client, the control engine and the segment parsers), and since the main script is the one communicating with the GPU, it will still have to update the model with the parsed content it receives from the worker.
Using the worker does not so much improve the framerate of the system, but it reduces the latency that occurs when receiving a new segment, which can be very frustrating since in a single thread scenario, each time a segment is received, the interface freezes for around half a second.
A sequence diagram of what happens when downloading, parsing and rendering content is shown in Figure~\ref{d3:sequence}.
\begin{figure}[ht]
\centering
\begin{tikzpicture}
\node at(0, 1) {Main script};
\draw[->, color=LightGray] (0, 0.5) -- (0, -17.5);
\node at(2.5, 1) {Worker};
\draw[->, color=LightGray] (2.5, 0.5) -- (2.5, -17.5);
\node at(10, 1) {Server};
\draw[->, color=LightGray] (10, 0.5) -- (10, -17.5);
% MPD
\draw[color=blue] (0, 0) -- (2.5, -0.1) -- (10, -0.5);
\draw[color=blue, fill=PaleLightBlue] (10, -0.5) -- (10, -1.5) -- (2.5, -2) -- (2.5, -1) -- cycle;
\node[color=blue, rotate=5] at (6.25, -1.25) {Download};
\node[color=blue, above] at(1.25, 0.0) {Ask MPD};
\draw[color=blue, fill=LightBlue] (2.375, -2) rectangle (2.625, -2.9);
\node[color=blue, right=0.2cm] at(2.5, -2.45) {Parse MPD};
\draw[color=blue] (2.5, -2.9) -- (0, -3);
\draw[color=blue, fill=LightBlue] (-0.125, -3.0) rectangle(0.125, -3.5);
\node[color=blue, left=0.2cm] at (0.0, -3.25) {Update model};
% Ask segments
\begin{scope}[shift={(0, -3.5)}]
\draw[color=red] (0, 0) -- (2.5, -0.1);
\node[color=red, above] at(1.25, 0.0) {Ask segment};
\draw[color=red, fill=Pink] (2.375, -0.1) rectangle (2.625, -1);
\node[color=red, right=0.2cm] at (2.5, -0.55) {Compute utilities};
\draw[color=red] (2.5, -1) -- (10, -1.5);
\draw[color=red, fill=PalePink] (10, -1.5) -- (10, -2.5) -- (2.5, -3) -- (2.5, -2) -- cycle;
\node[color=red, rotate=5] at (6.25, -2.25) {Download};
\draw[color=red, fill=Pink] (2.375, -3) rectangle (2.625, -3.9);
\node[color=red, right=0.2cm] at(2.5, -3.45) {Parse segment};
\draw[color=red] (2.5, -3.9) -- (0, -4);
\draw[color=red, fill=Pink] (-0.125, -4.0) rectangle(0.125, -4.5);
\node[color=red, left=0.2cm] at (0.0, -4.25) {Update model};
\end{scope}
% Ask more segments
\begin{scope}[shift={(0, -8)}]
\draw[color=DarkGreen] (0, 0) -- (2.5, -0.1);
\node[color=DarkGreen, above] at(1.25, 0.0) {Ask segment};
\draw[color=DarkGreen, fill=PaleGreen] (2.375, -0.1) rectangle (2.625, -1);
\node[color=DarkGreen, right=0.2cm] at (2.5, -0.55) {Compute utilities};
\draw[color=DarkGreen] (2.5, -1) -- (10, -1.5);
\draw[color=DarkGreen, fill=PalePaleGreen] (10, -1.5) -- (10, -2.5) -- (2.5, -3) -- (2.5, -2) -- cycle;
\node[color=DarkGreen, rotate=5] at (6.25, -2.25) {Download};
\draw[color=DarkGreen, fill=PaleGreen] (2.375, -3) rectangle (2.625, -3.9);
\node[color=DarkGreen, right=0.2cm] at(2.5, -3.45) {Parse segment};
\draw[color=DarkGreen] (2.5, -3.9) -- (0, -4);
\draw[color=DarkGreen, fill=PaleGreen] (-0.125, -4.0) rectangle(0.125, -4.5);
\node[color=DarkGreen, left=0.2cm] at (0.0, -4.25) {Update model};
\end{scope}
% Ask even more segments
\begin{scope}[shift={(0, -12.5)}]
\draw[color=purple] (0, 0) -- (2.5, -0.1);
\node[color=purple, above] at(1.25, 0.0) {Ask segment};
\draw[color=purple, fill=Plum] (2.375, -0.1) rectangle (2.625, -1);
\node[color=purple, right=0.2cm] at (2.5, -0.55) {Compute utilities};
\draw[color=purple] (2.5, -1) -- (10, -1.5);
\draw[color=purple, fill=PalePlum] (10, -1.5) -- (10, -2.5) -- (2.5, -3) -- (2.5, -2) -- cycle;
\node[color=purple, rotate=5] at (6.25, -2.25) {Download};
\draw[color=purple, fill=Plum] (2.375, -3) rectangle (2.625, -3.9);
\node[color=purple, right=0.2cm] at(2.5, -3.45) {Parse segment};
\draw[color=purple] (2.5, -3.9) -- (0, -4);
\draw[color=purple, fill=Plum] (-0.125, -4.0) rectangle(0.125, -4.5);
\node[color=purple, left=0.2cm] at (0.0, -4.25) {Update model};
\end{scope}
\foreach \x in {0,...,5}
{
\draw[color=Goldenrod, fill=LemonChiffon] (-0.125, -\x/2) rectangle (0.125, -\x/2-0.5);
}
\node[color=Goldenrod, left=0.2cm] at (0.0, -1.5) {Render};
\foreach \x in {0,...,7}
{
\draw[color=Goldenrod, fill=LemonChiffon] (-0.125, -\x/2-3.5) rectangle (0.125, -\x/2-4);
\draw[color=Goldenrod, fill=LemonChiffon] (-0.125, -\x/2-8) rectangle (0.125, -\x/2-8.5);
\draw[color=Goldenrod, fill=LemonChiffon] (-0.125, -\x/2-12.5) rectangle (0.125, -\x/2-13);
}
\node[color=Goldenrod, left=0.2cm] at (0.0, -5.5) {Render};
\node[color=Goldenrod, left=0.2cm] at (0.0, -10) {Render};
\node[color=Goldenrod, left=0.2cm] at (0.0, -14.5) {Render};
\end{tikzpicture}
\caption{Repartition of the tasks on the main script and the worker\label{d3:sequence}}
\end{figure}
\subsection{Rust client\label{d3i:rust-implementation}}
However, a web client is not sufficient to analyse our streaming policies: many tasks are performed (such as rendering, and managing the interaction) and all this overhead pollutes the analysis of our policies.
This is why we also implemented a client in Rust, for simulation, so we can gather precise simulated data.
Our requirements are quite different that the ones we had to deal with in our JavaScript implementation.
In this setup, we want to build a system that is the closest to our theoretical concepts.
Therefore, we do not have a full client in Rust (meaning an application to which you would give the URL to an MPD file and that would allow you to navigate in the scene while it is being downloaded).
In order to be able to run simulations, we develop the bricks of the DASH client separately: the access client and the media engine are totally separated:
\begin{itemize}
\item the \textbf{simulator} takes a user trace as a parameter, it then replays the trace using specific parameters of the access client and outputs a file containing the history of the simulation (what files have been downloaded, and when);
\item the \textbf{renderer} takes the user trace as well as the history generated by the simulator as parameters, and renders images that correspond to what would have been seen.
\end{itemize}
When simulating experiments, we will run the simulator on many traces that we collected during user-studies, and we will then run the renderer program on it to generate images corresponding to the simulation.
We are then able to compute PSNR between those frames and the ground truth frames.
Doing so guarantees us that our simulator is not affected by the performances of our renderer.

2
src/dash-3d/conclusion.tex
View File

@@ -12,7 +12,7 @@ This way, our system answers, at least partially, all the open problems we menti
\item \textbf{It prepares and structures the content in a way that enables streaming}: all this preparation is precomputed, and all the content is structured, even materials and textures. Furthermore, textures are prepared in a multi-resolution manner, and even though multi-resolution geometry is not discussed here, the difficulty of integrating it in this system seem moderated: we could encode levels of detail in different representations and define a utility metric for each representation and the system should adapt naturally.
\item \textbf{We are able to estimate the utility of each segment} by exploiting all the metadata given in the MPD and by analysing the camera parameters of the user.
\item \textbf{We proposed a few streaming policies}, from the easiest to implement to the more complex, that are able to exploit the utility metrics we defined in order to guess the best decision.
\item \textbf{The implementation is efficient}: the content preparation allows a client to get all the information it needs from metadata and the server has nothing else to do than to serve files. Special attention has been granted to the client's performance, and explanations are given in Chapter~\ref{d3i}.
\item \textbf{The implementation is efficient}: the content preparation allows a client to get all the information it needs from metadata and the server has nothing else to do than to serve files. Special attention has been granted to the client's performance.
\end{itemize}
However, the work described in this chapter does not take any quality of experience metrics into account.

15
src/dash-3d/content-preparation.tex
View File

@@ -1,7 +1,7 @@
\copied{}
\section{Content preparation\label{d3:dash-3d}}
In this section, we describe how we preprocess and store the 3D data of the NVE, consisting of a polygon soup, textures, and material information into a DASH-compliant Media Presentation Description (MPD) file.
In this section, we describe how we pre-process and store the 3D data of the NVE, consisting of a polygon soup, textures, and material information into a DASH-compliant Media Presentation Description (MPD) file.
In our work, we use the \texttt{obj} file format for the polygons, \texttt{png} for textures, and \texttt{mtl} format for material information.
The process, however, applies to other formats as well.
@@ -14,6 +14,19 @@ This element does not apply to NVE, and we use a single \texttt{period} for the
Each \texttt{period} element contains one or more adaptation sets, which describe the alternate versions, formats, and types of media.
We utilize adaptation sets to organize a 3D scene's material, geometry, and texture.
\fresh{}
The piece of software that does the preprocessing of the model mostly consists in file manipulation and is written is Rust as well.
It successively preprocess the geometry and then the textures.
The MPD is generated by a library named \href{https://github.com/netvl/xml-rs}{xml-rs}, and that works like a stack:
\begin{itemize}
\item a structure is created on the root of the MPD file;
\item the \texttt{start\_element} method creates a new child in the XML file;
\item the \texttt{end\_element} method ends the current child and pops the stack.
\end{itemize}
This structure is passed along with our geometry and texture preprocessors that can add elements to the XML file as they are generating the corresponding files.
\copied{}
\subsection{Adaptation Sets}
When the user navigates freely within an NVE, the frustum at given time almost always contains a limited part of the 3D scene.
Similar to how DASH for video streaming partitions a video clip into temporal chunks, we segment the polygons into spatial chunks, such that the DASH client can request only the relevant chunks.