diff --git a/DynObj.cpp b/DynObj.cpp
index 58ae59fa..b2c9de30 100644
--- a/DynObj.cpp
+++ b/DynObj.cpp
@@ -1943,7 +1943,8 @@ TDynamicObject::Init(std::string Name, // nazwa pojazdu, np. "EU07-424"
                         0.0f );
                 }
                 ++compartmentindex;
-            } while( submodel != nullptr );
+            } while( ( submodel != nullptr )
+                  || ( compartmentindex < 2 ) ); // chain can start from prefix00 or prefix01
         }
     }
     // wyszukiwanie zderzakow
@@ -4014,8 +4015,13 @@ void TDynamicObject::LoadMMediaFile(std::string BaseDir, std::string TypeName,
                 }
                 else {
                     // Ra: tu wczytywanie modelu ładunku jest w porządku
-                    if( false == asLoadName.empty() ) {
-                        mdLoad = TModelsManager::GetModel( asLoadName, true ); // ladunek
+                   if( false == asLoadName.empty() ) {
+                        // try first specialized version of the load model, vehiclename_loadname
+                        mdLoad = TModelsManager::GetModel( BaseDir + TypeName + "_" + MoverParameters->LoadType + ".t3d", true );
+                        if( mdLoad == nullptr ) {
+                            // if this fails, try generic load model
+                            mdLoad = TModelsManager::GetModel( asLoadName, true );
+                        }
                     }
                 }
             }
diff --git a/renderer.cpp b/renderer.cpp
index e66f60b5..6f531b8b 100644
--- a/renderer.cpp
+++ b/renderer.cpp
@@ -484,7 +484,7 @@ opengl_renderer::Render_pass( rendermode const Mode ) {
                 ::glEnable( GL_SCISSOR_TEST );
                 setup_matrices();
                 ::glEnable( GL_POLYGON_OFFSET_FILL ); // alleviate depth-fighting
-                ::glPolygonOffset( 2.f, 2.f );
+                ::glPolygonOffset( 1.f, 1.f );
                 // render
                 // opaque parts...
                 setup_drawing( false );
@@ -663,6 +663,7 @@ opengl_renderer::setup_pass( renderpass_config &Config, rendermode const Mode, f
         }
 
         case rendermode::shadows: {
+
             // calculate lightview boundaries based on relevant area of the world camera frustum:
             // ...setup chunk of frustum we're interested in...
             auto const znear = 0.f;
@@ -695,18 +696,43 @@ opengl_renderer::setup_pass( renderpass_config &Config, rendermode const Mode, f
                 point = lightviewmatrix * point;
             }
             bounding_box( frustumchunkmin, frustumchunkmax, std::begin( frustumchunkshapepoints ), std::end( frustumchunkshapepoints ) );
-            // ...use the dimensions to set up light projection boundaries
+            // quantize the frustum points with 50 m resolution, to reduce shadow shimmer on scale/orientation changes
+            frustumchunkmin = 50.f * glm::floor( frustumchunkmin * ( 1.f / 50.f ) );
+            frustumchunkmax = 50.f * glm::ceil( frustumchunkmax * ( 1.f / 50.f ) );
+            // ...use the dimensions to set up light projection boundaries...
             // NOTE: since we only have one cascade map stage, we extend the chunk forward/back to catch areas normally covered by other stages
-            auto const quantizationstep = ( Global::shadowtune.depth + 1000.f ) / m_shadowbuffersize;
-            frustumchunkmin.x -= std::remainder( frustumchunkmin.x, quantizationstep );
-            frustumchunkmin.y -= std::remainder( frustumchunkmin.y, quantizationstep );
-            frustumchunkmax.x -= std::remainder( frustumchunkmax.x, quantizationstep );
-            frustumchunkmax.y -= std::remainder( frustumchunkmax.y, quantizationstep );
             camera.projection() *=
                 glm::ortho(
                     frustumchunkmin.x, frustumchunkmax.x,
                     frustumchunkmin.y, frustumchunkmax.y,
                     frustumchunkmin.z - 500.f, frustumchunkmax.z + 500.f );
+/*
+            // fixed ortho projection from old build, for quick quality comparisons
+            camera.projection() *=
+                glm::ortho(
+                    -Global::shadowtune.width, Global::shadowtune.width,
+                    -Global::shadowtune.width, Global::shadowtune.width,
+                    -Global::shadowtune.depth, Global::shadowtune.depth );
+            camera.position() = Global::pCameraPosition - glm::dvec3{ Global::DayLight.direction };
+            if( camera.position().y - Global::pCameraPosition.y < 0.1 ) {
+                camera.position().y = Global::pCameraPosition.y + 0.1;
+            }
+            viewmatrix *= glm::lookAt(
+                camera.position(),
+                glm::dvec3{ Global::pCameraPosition },
+                glm::dvec3{ 0.f, 1.f, 0.f } );
+*/
+            // ... and adjust the projection to sample complete shadow map texels:
+            // get coordinates for a sample texel...
+            auto shadowmaptexel = glm::vec2 { camera.projection() * glm::mat4{ viewmatrix } * glm::vec4{ 0.f, 0.f, 0.f, 1.f } };
+            // ...convert result from clip space to texture coordinates, and calculate adjustment...
+            shadowmaptexel *= m_shadowbuffersize * 0.5f;
+            auto shadowmapadjustment = glm::round( shadowmaptexel ) - shadowmaptexel;
+            // ...transform coordinate change back to homogenous light space...
+            shadowmapadjustment /= m_shadowbuffersize * 0.5f;
+            // ... and bake the adjustment into the projection matrix
+            camera.projection() = glm::translate( glm::mat4{ 1.f }, glm::vec3{ shadowmapadjustment, 0.f } ) * camera.projection();
+
             break;
         }
         case rendermode::reflections: {
diff --git a/version.h b/version.h
index 9aaf2c1a..740b7596 100644
--- a/version.h
+++ b/version.h
@@ -1,5 +1,5 @@
 #pragma once
 
 #define VERSION_MAJOR 17
-#define VERSION_MINOR 1102
+#define VERSION_MINOR 1112
 #define VERSION_REVISION 0