+++ /dev/null
-<chapter id="chapter-dparams">
- <title>Dynamic Parameters</title>
-
- <sect1 id="section-dparams-getting-started">
- <title>Getting Started</title>
- <para>
- The Dynamic Parameters subsystem is contained within the
- <filename>gstcontrol</filename> library.
-
- You need to include the header in your application's source file:
- </para>
- <programlisting>
-...
-#include <gst/gst.h>
-#include <gst/control/control.h>
-...
- </programlisting>
- <para>
- Your application should link to the shared library <filename>gstcontrol</filename>.
- </para>
- <para>
- The <filename>gstcontrol</filename> library needs to be initialized
- when your application is run. This can be done after the the GStreamer
- library has been initialized.
- </para>
- <programlisting>
- ...
- gst_init(&argc,&argv);
- gst_control_init(&argc,&argv);
- ...
- </programlisting>
- </sect1>
-
- <sect1 id="section-dparams-creating">
- <title>Creating and Attaching Dynamic Parameters</title>
- <para>
- Once you have created your elements you can create and attach dparams to them.
- First you need to get the element's dparams manager. If you know exactly what kind of element
- you have, you may be able to get the dparams manager directly. However if this is not possible,
- you can get the dparams manager by calling <filename>gst_dpman_get_manager</filename>.
- </para>
- <para>
- Once you have the dparams manager, you must set the mode that the manager will run in.
- There is currently only one mode implemented called <filename>"synchronous"</filename> - this is used for real-time
- applications where the dparam value cannot be known ahead of time (such as a slider in a GUI).
- The mode is called <filename>"synchronous"</filename> because the dparams are polled by the element for changes before
- each buffer is processed. Another yet-to-be-implemented mode is <filename>"asynchronous"</filename>. This is used when
- parameter changes are known ahead of time - such as with a timelined editor. The mode is called
- <filename>"asynchronous"</filename> because parameter changes may happen in the middle of a buffer being processed.
- </para>
- <programlisting>
- GstElement *sinesrc;
- GstDParamManager *dpman;
- ...
- sinesrc = gst_element_factory_make("sinesrc","sine-source");
- ...
- dpman = gst_dpman_get_manager (sinesrc);
- gst_dpman_set_mode(dpman, "synchronous");
- </programlisting>
- <para>
- If you don't know the names of the required dparams for your element you can call
- <filename>gst_dpman_list_dparam_specs(dpman)</filename> to get a NULL terminated array of param specs.
- This array should be freed after use. You can find the name of the required dparam by calling
- <filename>g_param_spec_get_name</filename> on each param spec in the array. In our example,
- <filename>"volume"</filename> will be the name of our required dparam.
- </para>
- <para>
- Each type of dparam currently has its own <filename>new</filename> function. This may eventually
- be replaced by a factory method for creating new instances. A default dparam instance can be created
- with the <filename>gst_dparam_new</filename> function. Once it is created it can be attached to a
- required dparam in the element.
- </para>
- <programlisting>
- GstDParam *volume;
- ...
- volume = gst_dparam_new(G_TYPE_FLOAT);
- if (gst_dpman_attach_dparam (dpman, "volume", volume)){
- /* the dparam was successfully attached */
- ...
- }
- </programlisting>
- </sect1>
-
- <sect1 id="section-dparams-changing">
- <title>Changing Dynamic Parameter Values</title>
- <para>
- All interaction with dparams to actually set the dparam value is done through simple GObject properties.
- There is a property value for each type that dparams supports - these currently being
- <filename>"value_float"</filename>, <filename>"value_int"</filename> and <filename>"value_int64"</filename>.
- To set the value of a dparam, simply set the property which matches the type of your dparam instance.
- </para>
- <programlisting>
-#define ZERO(mem) memset(&mem, 0, sizeof(mem))
-...
-
- gfloat set_to_value;
- GstDParam *volume;
- GValue set_val;
- ZERO(set_val);
- g_value_init(&set_val, G_TYPE_FLOAT);
- ...
- g_value_set_float(&set_val, set_to_value);
- g_object_set_property(G_OBJECT(volume), "value_float", &set_val);
- </programlisting>
- <para>Or if you create an actual GValue instance:</para>
- <programlisting>
- gfloat set_to_value;
- GstDParam *volume;
- GValue *set_val;
- set_val = g_new0(GValue,1);
- g_value_init(set_val, G_TYPE_FLOAT);
- ...
- g_value_set_float(set_val, set_to_value);
- g_object_set_property(G_OBJECT(volume), "value_float", set_val);
- </programlisting>
-
- </sect1>
-
- <sect1 id="section-dparams-types">
- <title>Different Types of Dynamic Parameter</title>
- <para>
- There are currently only two implementations of dparams so far. They are both for real-time use so
- should be run in the <filename>"synchronous"</filename> mode.
- </para>
- <sect2>
- <title>GstDParam - the base dparam type</title>
- <para>
- All dparam implementations will subclass from this type. It provides a basic implementation which simply
- propagates any value changes as soon as it can.
- A new instance can be created with the function <filename>GstDParam* gst_dparam_new (GType type)</filename>.
- It has the following object properties:
- </para>
- <itemizedlist>
- <listitem><para><filename>"value_float"</filename>
- - the property to set and get if it is a float dparam
- </para></listitem>
- <listitem><para><filename>"value_int"</filename>
- - the property to set and get if it is an integer dparam
- </para></listitem>
- <listitem><para><filename>"value_int64"</filename>
- - the property to set and get if it is a 64 bit integer dparam
- </para></listitem>
- <listitem><para><filename>"is_log"</filename>
- - readonly boolean which is TRUE if the param should be displayed on a log scale
- </para></listitem>
- <listitem><para><filename>"is_rate"</filename>
- - readonly boolean which is TRUE if the value is a proportion of the sample rate.
- For example with a sample rate of 44100, 0.5 would be 22050 Hz and 0.25 would be 11025 Hz.
- </para></listitem>
- </itemizedlist>
- </sect2>
- <sect2>
- <title>GstDParamSmooth - smoothing real-time dparam</title>
- <para>
- Some parameter changes can create audible artifacts if they change too rapidly. The GstDParamSmooth
- implementation can greatly reduce these artifacts by limiting the rate at which the value can change.
- This is currently only supported for float dparams - the other types fall back to the default implementation.
- A new instance can be created with the function <filename>GstDParam* gst_dpsmooth_new (GType type)</filename>.
- It has the following object properties:
- </para>
- <itemizedlist>
- <listitem><para><filename>"update_period"</filename>
- - an int64 value specifying the number nanoseconds between updates. This will be ignored in
- <filename>"synchronous"</filename> mode since the buffer size dictates the update period.
- </para></listitem>
- <listitem><para><filename>"slope_time"</filename>
- - an int64 value specifying the time period to use in the maximum slope calculation
- </para></listitem>
- <listitem><para><filename>"slope_delta_float"</filename>
- - a float specifying the amount a float value can change in the given slope_time.
- </para></listitem>
- </itemizedlist>
- <para>
- Audible artifacts may not be completely eliminated by using this dparam. The only way to eliminate
- artifacts such as "zipper noise" would be for the element to implement its required dparams using the
- array method. This would allow dparams to change parameters at the sample rate which should eliminate
- any artifacts.
- </para>
-
- </sect2>
- <sect2>
- <title>Timelined dparams</title>
- <para>
- A yet-to-be-implemented subclass of GstDParam will add an API which allows the creation and manipulation
- of points on a timeline. This subclass will also provide a dparam implementation which uses linear
- interpolation between these points to find the dparam value at any given time. Further subclasses can
- extend this functionality to implement more exotic interpolation algorithms such as splines.
- </para>
- </sect2>
- </sect1>
-
-</chapter>
<programlisting>
GstDParam *volume;
...
- volume = gst_dparam_new(G_TYPE_FLOAT);
+ volume = gst_dparam_new(G_TYPE_DOUBLE);
if (gst_dpman_attach_dparam (dpman, "volume", volume)){
/* the dparam was successfully attached */
...
<para>
All interaction with dparams to actually set the dparam value is done through simple GObject properties.
There is a property value for each type that dparams supports - these currently being
- <filename>"value_float"</filename>, <filename>"value_int"</filename> and <filename>"value_int64"</filename>.
+ <filename>"value_double"</filename>, <filename>"value_float"</filename>, <filename>"value_int"</filename> and <filename>"value_int64"</filename>.
To set the value of a dparam, simply set the property which matches the type of your dparam instance.
</para>
<programlisting>
#define ZERO(mem) memset(&mem, 0, sizeof(mem))
...
- gfloat set_to_value;
+ gdouble set_to_value;
GstDParam *volume;
GValue set_val;
ZERO(set_val);
- g_value_init(&set_val, G_TYPE_FLOAT);
+ g_value_init(&set_val, G_TYPE_DOUBLE);
...
- g_value_set_float(&set_val, set_to_value);
- g_object_set_property(G_OBJECT(volume), "value_float", &set_val);
+ g_value_set_double(&set_val, set_to_value);
+ g_object_set_property(G_OBJECT(volume), "value_double", &set_val);
</programlisting>
<para>Or if you create an actual GValue instance:</para>
<programlisting>
- gfloat set_to_value;
+ gdouble set_to_value;
GstDParam *volume;
GValue *set_val;
set_val = g_new0(GValue,1);
- g_value_init(set_val, G_TYPE_FLOAT);
+ g_value_init(set_val, G_TYPE_DOUBLE);
...
- g_value_set_float(set_val, set_to_value);
- g_object_set_property(G_OBJECT(volume), "value_float", set_val);
+ g_value_set_double(set_val, set_to_value);
+ g_object_set_property(G_OBJECT(volume), "value_double", set_val);
</programlisting>
</sect1>
It has the following object properties:
</para>
<itemizedlist>
+ <listitem><para><filename>"value_double"</filename>
+ - the property to set and get if it is a double dparam
+ </para></listitem>
<listitem><para><filename>"value_float"</filename>
- the property to set and get if it is a float dparam
</para></listitem>
<para>
Some parameter changes can create audible artifacts if they change too rapidly. The GstDParamSmooth
implementation can greatly reduce these artifacts by limiting the rate at which the value can change.
- This is currently only supported for float dparams - the other types fall back to the default implementation.
+ This is currently only supported for double and float dparams - the other types fall back to the default implementation.
A new instance can be created with the function <filename>GstDParam* gst_dpsmooth_new (GType type)</filename>.
It has the following object properties:
</para>
<listitem><para><filename>"slope_time"</filename>
- an int64 value specifying the time period to use in the maximum slope calculation
</para></listitem>
+ <listitem><para><filename>"slope_delta_double"</filename>
+ - a double specifying the amount a double value can change in the given slope_time.
+ </para></listitem>
<listitem><para><filename>"slope_delta_float"</filename>
- a float specifying the amount a float value can change in the given slope_time.
</para></listitem>
<title>Timelined dparams</title>
<para>
A yet-to-be-implemented subclass of GstDParam will add an API which allows the creation and manipulation
- of points on a timeline. This subclass will also provide a dparam implementation which uses linear
- interpolation between these points to find the dparam value at any given time. Further subclasses can
+ of points on a timeline. This subclass will also provide a dparam implementation which uses linear
+ interpolation between these points to find the dparam value at any given time. Further subclasses can
extend this functionality to implement more exotic interpolation algorithms such as splines.
</para>
</sect2>