diff --git a/doc/images/integrator_direct_both.jpg b/doc/images/integrator_direct_both.jpg
new file mode 100644
index 00000000..14900e08
Binary files /dev/null and b/doc/images/integrator_direct_both.jpg differ
diff --git a/doc/images/integrator_direct_bsdf.jpg b/doc/images/integrator_direct_bsdf.jpg
new file mode 100644
index 00000000..5f58df9f
Binary files /dev/null and b/doc/images/integrator_direct_bsdf.jpg differ
diff --git a/doc/images/integrator_direct_lum.jpg b/doc/images/integrator_direct_lum.jpg
new file mode 100644
index 00000000..a0411c99
Binary files /dev/null and b/doc/images/integrator_direct_lum.jpg differ
diff --git a/doc/macros.sty b/doc/macros.sty
index fa925f7c..e5e8c21c 100644
--- a/doc/macros.sty
+++ b/doc/macros.sty
@@ -59,6 +59,23 @@
 	\setlength\fboxrule\fboxrulebackup
 	\arrayrulecolor{black}
 	}
+
+\newcommand{\remarks}[1]{
+	\setlength\fboxrulebackup\fboxrule
+	\setlength\fboxrule{0.8pt}
+	\vspace{1em}
+	\begin{figure}[htbp]\centering
+	\fcolorbox{remarkframe}{remark}{
+		\parbox{\columnwidth}{
+		\vspace{.6em}\hspace{.6em}\parbox{\columnwidth-1.6em}{
+		\textbf{Remarks:}\small
+		\vspace{-.4em}
+		\begin{itemize}
+		#1
+		\end{itemize}
+		}}}\end{figure}
+	\setlength\fboxrule\fboxrulebackup
+	}
 	
 \newcommand{\renderings}[1]{
 	\begin{figure}[htbp!]
diff --git a/doc/main.tex b/doc/main.tex
index fa6e3aae..cb42b53d 100644
--- a/doc/main.tex
+++ b/doc/main.tex
@@ -77,6 +77,8 @@
 \definecolor{lstframe}{gray}{0.80}
 \definecolor{lstcomment}{gray}{0.5}
 \definecolor{lstattrib}{rgb}{0,0.34,0}
+\definecolor{remark}{rgb}{1.0, 0.9, 0.9}
+\definecolor{remarkframe}{rgb}{1.0, 0.7, 0.7}
 
 % Listings settings
 \lstset{
diff --git a/doc/section_integrators.tex b/doc/section_integrators.tex
index 555ed439..687c0603 100644
--- a/doc/section_integrators.tex
+++ b/doc/section_integrators.tex
@@ -81,8 +81,7 @@ a single scattering event (\figref{path-explanation}).
 
 \subsubsection*{Progressive versus non-progressive}
 Some of the rendering techniques in Mitsuba are \emph{progressive}.
-What this means is that they display a rough preview, which
-improves over time. Leaving them running indefinitely will
-continually reduce noise (e.g. in Metropolis Light Transport)
-or both noise and bias (e.g. in Progressive Photon Mapping).
-
+What this means is that they display a rough preview, which improves over time. 
+Leaving them running indefinitely will continually reduce noise (in unbiased algorithms
+such as Metropolis Light Transport) or noise and bias (in biased 
+rendering techniques such as Progressive Photon Mapping).
diff --git a/src/integrators/direct/direct.cpp b/src/integrators/direct/direct.cpp
index 4c2b80d2..333cc07e 100644
--- a/src/integrators/direct/direct.cpp
+++ b/src/integrators/direct/direct.cpp
@@ -20,12 +20,59 @@
 
 MTS_NAMESPACE_BEGIN
 
-/**
- * Direct-only integrator using multiple importance sampling and
- * the power heuristic. Takes a user-specifiable amount of luminaire
- * and BSDF samples By setting one of the strategies to zero, this 
- * class can effectively be turned into a luminaire sampling or
- * BSDF sampling-based integrator. Ignores participating media.
+/*! \plugin{direct}{Direct illumination integrator}
+ * \parameters{
+ *     \parameter{shadingSamples}{\Integer}{This convenience parameter can be 
+ *         used to set both \code{luminaireSamples} and \code{bsdfSamples} at 
+ *         the same time.}
+ *     \parameter{luminaireSamples}{\Integer}{Optional more fine-grained
+ *        parameter: specifies the number of samples that should be generated
+ *        using the direct illumination strategies implemented by the scene's
+ *        luminaires\default{set to the value of \code{shadingSamples}}}
+ *     \parameter{bsdfSamples}{\Integer}{Optional more fine-grained
+ *        parameter: specifies the number of samples that should be generated
+ *        using the BSDF sampling strategies implemented by the scene's
+ *        surfaces\default{set to the value of \code{shadingSamples}}}
+ * }
+ *
+ * \renderings{
+ *     \medrendering{Only BSDF sampling}{integrator_direct_bsdf}
+ *     \medrendering{Only luminaire sampling}{integrator_direct_lum}
+ *     \medrendering{BSDF and luminaire sampling}{integrator_direct_both}
+ *     \caption{
+ *         \label{fig:integrator-direct}
+ *         This plugin implements two different strategies for computing the
+ *         direct illumination on surfaces. Both of them are dynamically 
+ *         combined then obtain a robust rendering algorithm.
+ *     }
+ * }
+ *
+ * This integrator implements a direct illumination technique that makes use
+ * of \emph{multiple importance sampling}: for each pixel sample, the 
+ * integrator generates a user-specifiable number of BSDF and luminaire 
+ * samples and combines them using the power heuristic. Usually, the BSDF
+ * sampling technique works very well on glossy objects but does badly 
+ * everywhere else (\subfigref{integrator-direct}{a}), while the opposite
+ * is true for the luminaire sampling technique 
+ * (\subfigref{integrator-direct}{b}). By combining these approaches, one
+ * can obtain a rendering technique that works well in both cases
+ * (\subfigref{integrator-direct}{c}).
+ *
+ * The number of samples spent on either technique is configurable, hence 
+ * it is also possible to turn this plugin into an luminaire sampling-only 
+ * or BSDF sampling-only integrator.
+ *
+ * For best results, combine the direct illumination integrator with the
+ * low-discrepancy sample generator (\code{ldsampler}). Generally, the number
+ * of pixel samples of the sample generator can be kept relatively 
+ * low (e.g. \code{sampleCount=4}), whereas the \code{shadingSamples} 
+ * parameter of this integrator should be increased until the variance in
+ * the output renderings is acceptable.
+ *
+ * \remarks{
+ *    \item This integrator does not handle participating media or
+ *          indirect illumination.
+ * }
  */
 class MIDirectIntegrator : public SampleIntegrator {
 public: