diff --git a/src/app_layout.cpp b/src/app_layout.cpp
index b5d25af..59ed670 100644
--- a/src/app_layout.cpp
+++ b/src/app_layout.cpp
@@ -203,29 +203,34 @@ void App::init_sidebar()
     layers->on_layer_opacity_changed = [this](Node*, int idx, float value) {
         canvas->m_canvas->m_layers[canvas->m_canvas->m_order[idx]]->m_opacity = value;
         canvas->m_canvas->m_unsaved = true;
+        canvas->m_canvas->draw_merge();
         title_update();
     };
 
     layers->on_layer_visibility_changed = [this](Node*, int idx, bool visible) {
         canvas->m_canvas->m_layers[canvas->m_canvas->m_order[idx]]->m_visible = visible;
         canvas->m_canvas->m_unsaved = true;
+        canvas->m_canvas->draw_merge();
         title_update();
     };
     
     layers->on_layer_alpha_lock_changed = [this](Node*, int idx, bool locked) {
         canvas->m_canvas->m_layers[canvas->m_canvas->m_order[idx]]->m_alpha_locked = locked;
         canvas->m_canvas->m_unsaved = true;
+        canvas->m_canvas->draw_merge();
         title_update();
     };
     
     layers->on_layer_blend_mode_changed = [this](Node*, int idx, int mode) {
         canvas->m_canvas->m_layers[canvas->m_canvas->m_order[idx]]->m_blend_mode = mode;
         canvas->m_canvas->m_unsaved = true;
+        canvas->m_canvas->draw_merge();
         title_update();
     };
 
     layers->on_layer_highlight_changed = [this](Node*, int idx, bool highlight) {
         canvas->m_canvas->m_layers[canvas->m_canvas->m_order[idx]]->m_hightlight = highlight;
+        canvas->m_canvas->draw_merge();
     };
     if (auto* button = layout[main_id]->find<NodeButtonCustom>("btn-stroke"))
     {
diff --git a/src/canvas.cpp b/src/canvas.cpp
index 466cea0..c9f4d0c 100644
--- a/src/canvas.cpp
+++ b/src/canvas.cpp
@@ -143,11 +143,13 @@ void Canvas::clear(const glm::vec4& c/*={0,0,0,1}*/)
     snap_history({ 0, 1, 2, 3, 4, 5 });
     m_layers[m_current_layer_idx]->clear(c);
     m_unsaved = true;
+    draw_merge();
 }
 void Canvas::clear_all()
 {
     for (auto& l : m_layers)
         l->clear({0, 0, 0, 0});
+    draw_merge();
 }
 void Canvas::snap_history(const std::vector<int>& planes)
 {
@@ -196,9 +198,9 @@ void Canvas::stroke_end()
     }
     else
     {
+        m_show_tmp = false;
         stroke_commit();
         m_current_stroke = nullptr;
-        m_show_tmp = false;
     }
 }
 void Canvas::stroke_cancel()
@@ -465,6 +467,7 @@ void Canvas::stroke_draw()
     }
 
     m_dirty = true;
+    std::array<bool, 6> merge_faces;
 
     GLint vp[4];
     GLfloat cc[4];
@@ -538,6 +541,7 @@ void Canvas::stroke_draw()
             if (P.size() < 3)
                 continue;
             m_dirty_face[i] = true;
+            merge_faces[i] = true;
             m_tmp[i].bindFramebuffer();
             auto rect = stroke_draw_samples(i, P);
             m_tmp[i].unbindFramebuffer();
@@ -594,11 +598,13 @@ void Canvas::stroke_draw()
 
     if (m_commit_delayed)
     {
-        stroke_commit();
-        m_current_stroke = nullptr;
         m_show_tmp = false;
         m_commit_delayed = false;
+        stroke_commit();
+        m_current_stroke = nullptr;
     }
+
+    draw_merge(merge_faces);
 }
 bool Canvas::point_trace(glm::vec2 loc, glm::vec3& ray_origin, glm::vec3& ray_dir,
     glm::vec3& hit_pos, glm::vec2& fb_pos, glm::vec3& hit_normal, int& out_plane_id)
@@ -864,7 +870,248 @@ void Canvas::stroke_commit()
     action->m_canvas = this;
     //action->m_stroke = std::move(m_current_stroke);
     ActionManager::add(action);
+
+    draw_merge();
 }
+
+void Canvas::draw_merge(std::array<bool, 6> faces /*= SIXPLETTE(false)*/)
+{
+    gl_state gl;
+    gl.save();
+    glViewport(0, 0, m_width, m_height);
+    auto ortho = glm::ortho<float>(-0.5f, 0.5f, -0.5f, 0.5f, -1.f, 1.f);
+    const auto& b = m_current_stroke->m_brush;
+
+    // check if any layer use blend, otherwise draw directly on main framebuffer
+    bool use_blend = false;
+    for (size_t i = 0; i < m_order.size(); i++)
+    {
+        auto layer_index = m_order[i];
+        use_blend |= m_layers[layer_index]->m_blend_mode != 0;
+    }
+
+    // if not using shader blend, use gl rasterizer blend
+    glDisable(GL_DEPTH_TEST);
+
+    for (int plane_index = 0; plane_index < 6; plane_index++)
+    {
+        if (!faces[plane_index])
+            continue;
+
+        m_layers_merge.m_rtt[plane_index].bindFramebuffer();
+
+        if (use_blend)
+        {
+            glDisable(GL_BLEND);
+            m_layers_merge.m_rtt[plane_index].clear();
+        }
+        else
+        {
+            ShaderManager::use(kShader::Checkerboard);
+            ShaderManager::u_int(kShaderUniform::Colorize, false);
+            ShaderManager::u_mat4(kShaderUniform::MVP, ortho);
+            m_plane.draw_fill();
+            glEnable(GL_BLEND);
+        }
+
+        for (int layer_index : m_order)
+        {
+            if (!(m_show_tmp && m_current_layer_idx == layer_index) &&
+                (!m_layers[layer_index]->m_visible ||
+                    m_layers[layer_index]->m_opacity == .0f ||
+                    !m_layers[layer_index]->m_dirty_face[plane_index]))
+                continue;
+            
+            if (use_blend)
+            {
+                m_merge_rtt.bindFramebuffer();
+                m_merge_rtt.clear();
+            }
+
+            if (m_current_stroke && m_current_mode == kCanvasMode::Erase && m_show_tmp && m_current_layer_idx == layer_index)
+            {
+                m_sampler.bind(0);
+                ShaderManager::use(kShader::CompErase);
+                ShaderManager::u_int(kShaderUniform::Tex, 0);
+                ShaderManager::u_int(kShaderUniform::TexStroke, 1);
+                ShaderManager::u_int(kShaderUniform::TexMask, 2);
+                //ShaderManager::u_vec2(kShaderUniform::Resolution, zw(m_box) / zoom);
+                ShaderManager::u_float(kShaderUniform::Alpha, m_layers[layer_index]->m_opacity);
+                //ShaderManager::u_int(kShaderUniform::Lock, m_layers[layer_index]->m_alpha_locked);
+                ShaderManager::u_int(kShaderUniform::Mask, m_smask_active);
+                ShaderManager::u_mat4(kShaderUniform::MVP, ortho);
+                glActiveTexture(GL_TEXTURE0);
+                m_layers[layer_index]->m_rtt[plane_index].bindTexture();
+                glActiveTexture(GL_TEXTURE1);
+                m_tmp[plane_index].bindTexture();
+                glActiveTexture(GL_TEXTURE2);
+                m_smask.m_rtt[plane_index].bindTexture();
+                m_plane.draw_fill();
+                m_smask.m_rtt[plane_index].unbindTexture();
+                glActiveTexture(GL_TEXTURE1);
+                m_tmp[plane_index].unbindTexture();
+                glActiveTexture(GL_TEXTURE0);
+                m_layers[layer_index]->m_rtt[plane_index].unbindTexture();
+            }
+            else if (m_current_stroke && m_show_tmp && m_current_layer_idx == layer_index)
+            {
+                m_sampler.bind(0);
+
+                glm::vec2 patt_scale = glm::vec2(b->m_pattern_scale);
+                if (b->m_pattern_flipx) patt_scale.x *= -1.f;
+                if (b->m_pattern_flipy) patt_scale.y *= -1.f;
+
+                ShaderManager::use(kShader::CompDraw);
+                ShaderManager::u_int(kShaderUniform::Tex, 0);
+                ShaderManager::u_int(kShaderUniform::TexStroke, 1);
+                ShaderManager::u_int(kShaderUniform::TexMask, 2);
+                ShaderManager::u_int(kShaderUniform::TexDual, 3);
+                ShaderManager::u_vec2(kShaderUniform::Resolution, Canvas::I->m_size);
+                ShaderManager::u_int(kShaderUniform::TexPattern, 4);
+                ShaderManager::u_float(kShaderUniform::Alpha, m_layers[layer_index]->m_opacity);
+                ShaderManager::u_int(kShaderUniform::Lock, m_layers[layer_index]->m_alpha_locked);
+                ShaderManager::u_int(kShaderUniform::Mask, m_smask_active);
+                ShaderManager::u_int(kShaderUniform::UseFragcoord, false);
+                ShaderManager::u_int(kShaderUniform::BlendMode, b->m_blend_mode);
+                ShaderManager::u_mat4(kShaderUniform::MVP, ortho);
+                ShaderManager::u_int(kShaderUniform::UseDual, b->m_dual_enabled);
+                ShaderManager::u_int(kShaderUniform::DualBlendMode, b->m_dual_blend_mode);
+                ShaderManager::u_int(kShaderUniform::UsePattern, b->m_pattern_enabled && !b->m_pattern_eachsample);
+                ShaderManager::u_vec2(kShaderUniform::PatternScale, patt_scale);
+                ShaderManager::u_float(kShaderUniform::PatternInvert, b->m_pattern_invert);
+                ShaderManager::u_float(kShaderUniform::PatternBright, b->m_pattern_brightness);
+                ShaderManager::u_float(kShaderUniform::PatternContrast, b->m_pattern_contrast);
+                ShaderManager::u_float(kShaderUniform::PatternDepth, b->m_pattern_depth);
+                ShaderManager::u_int(kShaderUniform::PatternBlendMode, b->m_pattern_blend_mode);
+                ShaderManager::u_vec2(kShaderUniform::PatternOffset, Canvas::I->m_pattern_offset);
+                ShaderManager::u_float(kShaderUniform::DualAlpha, b->m_dual_opacity);
+
+                glActiveTexture(GL_TEXTURE0);
+                m_layers[layer_index]->m_rtt[plane_index].bindTexture();
+                glActiveTexture(GL_TEXTURE1);
+                m_tmp[plane_index].bindTexture();
+                glActiveTexture(GL_TEXTURE2);
+                m_smask.m_rtt[plane_index].bindTexture();
+                glActiveTexture(GL_TEXTURE3);
+                if (b->m_dual_enabled)
+                    m_tmp_dual[plane_index].bindTexture();
+                glActiveTexture(GL_TEXTURE4);
+                b->m_pattern_texture ?
+                    b->m_pattern_texture->bind() :
+                    glBindTexture(GL_TEXTURE_2D, 0);
+                m_plane.draw_fill();
+                glActiveTexture(GL_TEXTURE3);
+                if (b->m_dual_enabled)
+                    m_tmp_dual[plane_index].unbindTexture();
+                glActiveTexture(GL_TEXTURE2);
+                m_smask.m_rtt[plane_index].unbindTexture();
+                glActiveTexture(GL_TEXTURE1);
+                m_tmp[plane_index].unbindTexture();
+                glActiveTexture(GL_TEXTURE0);
+                m_layers[layer_index]->m_rtt[plane_index].unbindTexture();
+            }
+            else
+            {
+                m_sampler.bind(0);
+                m_sampler_linear.bind(1);
+                ShaderManager::use(kShader::TextureAlphaSep);
+                ShaderManager::u_int(kShaderUniform::Tex, 0);
+                ShaderManager::u_int(kShaderUniform::TexA, 1);
+                ShaderManager::u_float(kShaderUniform::Alpha, m_layers[layer_index]->m_opacity);
+                ShaderManager::u_int(kShaderUniform::Highlight, m_layers[layer_index]->m_hightlight);
+                ShaderManager::u_mat4(kShaderUniform::MVP, ortho);
+
+                glActiveTexture(GL_TEXTURE0);
+                m_layers[layer_index]->m_rtt[plane_index].bindTexture();
+                glActiveTexture(GL_TEXTURE1);
+                m_layers[layer_index]->m_rtt[plane_index].bindTexture();
+
+                m_plane.draw_fill();
+
+                glActiveTexture(GL_TEXTURE1);
+                m_layers[layer_index]->m_rtt[plane_index].unbindTexture();
+                glActiveTexture(GL_TEXTURE0);
+                m_layers[layer_index]->m_rtt[plane_index].unbindTexture();
+            }
+
+            if (use_blend)
+            {
+                m_merge_rtt.unbindFramebuffer();
+            }
+
+            // draw the blended
+            if (use_blend)
+            {
+                m_sampler.bind(0);
+                m_sampler_linear.bind(1);
+                m_sampler.bind(2);
+
+                ShaderManager::use(kShader::TextureBlend);
+                ShaderManager::u_int(kShaderUniform::Tex, 0);
+                ShaderManager::u_int(kShaderUniform::TexA, 1);
+                if (!ShaderManager::ext_framebuffer_fetch)
+                    ShaderManager::u_int(kShaderUniform::TexBG, 2);
+                ShaderManager::u_int(kShaderUniform::BlendMode, m_layers[layer_index]->m_blend_mode);
+                ShaderManager::u_float(kShaderUniform::Alpha, 1.f);
+                ShaderManager::u_mat4(kShaderUniform::MVP, ortho);
+
+                glActiveTexture(GL_TEXTURE0);
+                m_merge_rtt.bindTexture();
+                glActiveTexture(GL_TEXTURE1);
+                m_merge_rtt.bindTexture();
+                if (!ShaderManager::ext_framebuffer_fetch)
+                {
+                    glActiveTexture(GL_TEXTURE2);
+                    m_merge_tex.bind();
+                    glCopyTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, 0, 0, m_width, m_height);
+                }
+
+                m_plane.draw_fill();
+
+                if (!ShaderManager::ext_framebuffer_fetch)
+                {
+                    glActiveTexture(GL_TEXTURE2);
+                    m_merge_tex.unbind();
+                }
+                glActiveTexture(GL_TEXTURE1);
+                m_merge_rtt.unbindTexture();
+                glActiveTexture(GL_TEXTURE0);
+                m_merge_rtt.unbindTexture();
+            }
+        }
+
+        glActiveTexture(GL_TEXTURE2);
+        m_merge_tex.bind();
+        glCopyTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, 0, 0, m_width, m_height);
+        
+        // draw the grid behind the layers using a temporary copy
+        if (use_blend)
+        {
+            glEnable(GL_BLEND);
+
+            //draw the grid
+            ShaderManager::use(kShader::Checkerboard);
+            ShaderManager::u_int(kShaderUniform::Colorize, false);
+            ShaderManager::u_mat4(kShaderUniform::MVP, ortho);
+            m_plane.draw_fill();
+
+            // draw the layers
+            m_sampler.bind(0);
+            glActiveTexture(GL_TEXTURE0);
+            m_merge_tex.bind();
+            ShaderManager::use(kShader::Texture);
+            ShaderManager::u_int(kShaderUniform::Tex, 0);
+            ShaderManager::u_mat4(kShaderUniform::MVP, ortho);
+            m_plane.draw_fill();
+            m_merge_tex.unbind();
+        }
+        
+        m_layers_merge.m_rtt[plane_index].unbindFramebuffer();
+    }
+
+    gl.restore();
+}
+
 void Canvas::stroke_update(glm::vec3 point, float pressure)
 {
     m_current_stroke->add_point(point, pressure);
@@ -876,11 +1123,11 @@ void Canvas::stroke_start(glm::vec3 point, float pressure)
     // need to commit this now before starting a new stroke
     if (m_current_stroke && m_commit_delayed)
     {
+        m_show_tmp = false;
+        m_commit_delayed = false;
         stroke_commit();
         m_current_stroke = nullptr;
         m_dual_stroke = nullptr;
-        m_show_tmp = false;
-        m_commit_delayed = false;
     }
     
     m_pattern_offset = m_current_brush->m_pattern_rand_offset ?
@@ -947,6 +1194,7 @@ void Canvas::layer_add(std::string name)
     m_layers.back()->create(m_width, m_height, name);
     m_order.push_back(idx);
     m_current_layer_idx = idx;
+    draw_merge();
 }
 void Canvas::layer_remove(int idx) // m_order index
 {
@@ -958,10 +1206,12 @@ void Canvas::layer_remove(int idx) // m_order index
     m_layers.erase(m_layers.begin() + n);
     m_order.erase(m_order.begin() + idx);
     m_current_layer_idx = m_order[std::min<int>((int)m_layers.size() - 1, idx)];
+    draw_merge();
 }
 void Canvas::layer_order(int idx, int pos) // m_order index
 {
     std::swap(m_order[idx], m_order[pos]);
+    draw_merge();
 }
 void Canvas::layer_merge(int source_idx, int dest_idx) // m_layer index
 {
@@ -1054,6 +1304,7 @@ void Canvas::layer_merge(int source_idx, int dest_idx) // m_layer index
     action->m_stroke = std::move(m_current_stroke);
     ActionManager::add(action);
 */
+    draw_merge();
 }
 void Canvas::resize(int width, int height)
 {
@@ -1073,9 +1324,12 @@ void Canvas::resize(int width, int height)
 #endif
         m_tex2[i].create(width, height, GL_RGBA8);
     }
+    m_layers_merge.resize(width, height);
     for (auto& l : m_layers)
         l->resize(width, height);
     m_smask.create(width, height, "mask");
+    m_merge_rtt.create(width, height);
+    m_merge_tex.create(width, height);
     m_unsaved = true;
 }
 
@@ -1088,10 +1342,13 @@ void Canvas::destroy()
         m_tex[i].destroy();
         m_tex2[i].destroy();
     }
+    m_layers_merge.destroy();
     for (auto& l : m_layers)
         l->destroy();
     m_smask.destroy();
     m_mixer.destroy();
+    m_merge_rtt.destroy();
+    m_merge_tex.destroy();
 }
 
 bool Canvas::create(int width, int height)
@@ -1105,15 +1362,18 @@ bool Canvas::create(int width, int height)
         m_tmp[i].create(width, height, -1, GL_RGBA8);
         m_tmp_dual[i].create(width, height, -1, GL_RGBA8);
         m_tex[i].create(width, height, GL_RGBA8);
-        m_sampler_brush.create();
 #else
         m_tmp[i].create(width, height, -1, GL_RGBA32F);
         m_tmp_dual[i].create(width, height, -1, GL_RGBA32F);
         m_tex[i].create(width, height, GL_RGBA32F);
-        m_sampler_brush.create(GL_LINEAR, GL_CLAMP_TO_BORDER);
 #endif
         m_tex2[i].create(width, height, GL_RGBA8);
     }
+#if defined(__IOS__) || defined(__ANDROID__)
+    m_sampler_brush.create();
+#else
+    m_sampler_brush.create(GL_LINEAR, GL_CLAMP_TO_BORDER);
+#endif
     m_sampler.create(GL_NEAREST);
     m_sampler.create(GL_LINEAR);
     m_sampler_nearest.create(GL_NEAREST);
@@ -1129,11 +1389,15 @@ bool Canvas::create(int width, int height)
     m_brush_shape.create();
     m_mesh.create();
     m_brush_mix.create(8, 8);
+    m_layers_merge.create(width, height, "merge");
     for (auto& l : m_layers)
         l->create(width, height, "");
     m_smask.create(width, height, "mask");
     //m_smask.clear({1, 1, 1, 1});
+    m_merge_rtt.create(width, height);
+    m_merge_tex.create(width, height);
     m_unsaved = true;
+    draw_merge();
     return true;
 }
 
@@ -2198,6 +2462,17 @@ bool Canvas::project_open_thread(std::string file_path)
     fclose(fp);
     LOG("project restore from %s", file_path.c_str());
     
+    m_current_layer_idx = 0;
+    m_current_stroke = nullptr;
+    m_dual_stroke = nullptr;
+    m_show_tmp = false;
+    m_smask_active = false;
+    m_smask_mode = 0;
+    m_dirty = false;
+    m_commit_delayed = false;
+    m_dirty_stroke = false;
+    memset(m_dirty_face, 0, sizeof(bool) * 6);
+    memset(m_pick_ready, 0, sizeof(bool) * 6);
     m_unsaved = false;
     m_newdoc = false;
     if (App::I.layout.m_loaded)
@@ -2205,11 +2480,18 @@ bool Canvas::project_open_thread(std::string file_path)
         App::I.async_start();
         pb->destroy();
         gl.save();
+        draw_merge();
         App::I.async_update();
         gl.restore();
         App::I.title_update();
         App::I.async_end();
     }
+    else
+    {
+        App::I.async_start();
+        draw_merge();
+        App::I.async_end();
+    }
     return true;
 }
 
@@ -2393,6 +2675,8 @@ void Canvas::draw_objects_direct(std::function<void(const glm::mat4& camera, con
     glViewport(vp[0], vp[1], vp[2], vp[3]);
     glClearColor(cc[0], cc[1], cc[2], cc[3]);
     glActiveTexture(GL_TEXTURE0);
+
+    draw_merge();
 }
 
 void Canvas::draw_objects(std::function<void(const glm::mat4& camera, const glm::mat4& proj, int i)> observer, Layer& layer)
@@ -2483,6 +2767,8 @@ void Canvas::draw_objects(std::function<void(const glm::mat4& camera, const glm:
     glViewport(vp[0], vp[1], vp[2], vp[3]);
     glClearColor(cc[0], cc[1], cc[2], cc[3]);
     glActiveTexture(GL_TEXTURE0);
+
+    draw_merge();
 }
 
 void Canvas::draw_objects(std::function<void(const glm::mat4& camera, const glm::mat4& proj, int i)> observer)
@@ -2794,6 +3080,7 @@ void ActionStroke::undo()
             LOG("undo invalid box size (%d, %d)", (int)box_sz.x, (int)box_sz.y);
         }
     }
+    m_canvas->draw_merge();
 }
 
 size_t ActionStroke::memory()
diff --git a/src/canvas.h b/src/canvas.h
index f117c22..c90e7d8 100644
--- a/src/canvas.h
+++ b/src/canvas.h
@@ -167,6 +167,7 @@ public:
     std::unique_ptr<Stroke> m_dual_stroke;
     bool m_show_tmp = false;
     std::vector<std::unique_ptr<Layer>> m_layers;
+    Layer m_layers_merge;
     std::vector<int> m_order;
     std::vector<glm::vec2> m_plane_shape[6]; // screen space projection of the plane
     glm::mat4 m_plane_unproject[6] = SIXPLETTE(glm::mat4(1));
@@ -184,6 +185,8 @@ public:
     Texture2D m_brush_mix;
     Texture2D m_tex[6];
     Texture2D m_tex2[6];
+    RTT m_merge_rtt;
+    Texture2D m_merge_tex;
     bool m_pick_ready[6];
     std::unique_ptr<glm::u8vec4[]> m_pick_data[6] = SIXPLETTE(nullptr);
     static glm::vec3 m_plane_origin[6];
@@ -247,6 +250,7 @@ public:
     void stroke_end();
     void stroke_cancel();
     void stroke_commit();
+    void draw_merge(std::array<bool, 6> faces = SIXPLETTE(true));
     void clear(const glm::vec4& color = { 1, 1, 1, 0 });
     void clear_all();
     void pick_start();
diff --git a/src/node_canvas.cpp b/src/node_canvas.cpp
index 808dee8..2a992f4 100644
--- a/src/node_canvas.cpp
+++ b/src/node_canvas.cpp
@@ -81,10 +81,6 @@ void NodeCanvas::draw()
     m_canvas->m_box = box;
     m_canvas->m_vp = c;
 
-    //         auto plane_mvp = proj * camera * transform *
-    //             glm::scale(glm::vec3(sz, 1));
-
-
     m_sampler.bind(0);
     m_sampler.bind(1);
     m_sampler.bind(2);
@@ -114,21 +110,11 @@ void NodeCanvas::draw()
         m_canvas->m_plane_shape[plane_index] = m_canvas->face_to_shape2D(plane_index);
     }
 
-    // check if any layer use blend, otherwise draw directly on main framebuffer
-    bool use_blend = false;
-    for (size_t i = 0; i < m_canvas->m_order.size(); i++)
-    {
-        auto layer_index = m_canvas->m_order[i];
-        use_blend |= m_canvas->m_layers[layer_index]->m_blend_mode != 0;
-    }
+    bool draw_merged = !(m_canvas->m_current_mode == kCanvasMode::Camera);
 
-    if (use_blend)
-    {
-        m_cache_rtt.bindFramebuffer();
-        m_cache_rtt.clear({ 1, 1, 1, 0 });
-    }
-    else
+    if (draw_merged)
     {
+        glDisable(GL_BLEND);
         // draw the grid
         for (int plane_index = 0; plane_index < 6; plane_index++)
         {
@@ -141,243 +127,303 @@ void NodeCanvas::draw()
             ShaderManager::u_int(kShaderUniform::Colorize, false);
             ShaderManager::u_mat4(kShaderUniform::MVP, plane_mvp);
             m_face_plane.draw_fill();
-        }
-    }
 
-    // if not using shader blend, use gl rasterizer blend
-    use_blend ? glDisable(GL_BLEND) : glEnable(GL_BLEND);
-    glDisable(GL_DEPTH_TEST);
-
-    const auto& b = m_canvas->m_current_stroke->m_brush;
-
-    for (size_t i = 0; i < m_canvas->m_order.size(); i++)
-    {
-        auto layer_index = m_canvas->m_order[i];
-        for (int plane_index = 0; plane_index < 6; plane_index++)
-        {
-            if (!(m_canvas->m_show_tmp && m_canvas->m_current_layer_idx == layer_index) && 
-                (!m_canvas->m_layers[layer_index]->m_visible ||
-                m_canvas->m_layers[layer_index]->m_opacity == .0f ||
-                !m_canvas->m_layers[layer_index]->m_dirty_face[plane_index]))
-                continue;
-
-            if (use_blend)
-            {
-                m_blender_rtt.bindFramebuffer();
-                m_blender_rtt.clear();
-            }
-
-            int z = (int)(m_canvas->m_order.size() - i);
-            auto plane_mvp_z = proj * camera * 
-                glm::scale(glm::vec3(z + 1)) * 
-                glm::eulerAngleYXZ(yaw, pitch, roll) *
-                m_canvas->m_plane_transform[plane_index] * 
-                glm::translate(glm::vec3(0, 0, -1));
-            
-            if (m_canvas->m_current_stroke && m_canvas->m_current_mode == kCanvasMode::Erase && m_canvas->m_show_tmp && m_canvas->m_current_layer_idx == layer_index)
-            {
-                m_sampler.bind(0);
-                ShaderManager::use(kShader::CompErase);
-                ShaderManager::u_int(kShaderUniform::Tex, 0);
-                ShaderManager::u_int(kShaderUniform::TexStroke, 1);
-                ShaderManager::u_int(kShaderUniform::TexMask, 2);
-                //ShaderManager::u_vec2(kShaderUniform::Resolution, zw(m_canvas->m_box) / zoom);
-                ShaderManager::u_float(kShaderUniform::Alpha, m_canvas->m_layers[layer_index]->m_opacity);
-                //ShaderManager::u_int(kShaderUniform::Lock, m_canvas->m_layers[layer_index]->m_alpha_locked);
-                ShaderManager::u_int(kShaderUniform::Mask, m_canvas->m_smask_active);
-                ShaderManager::u_mat4(kShaderUniform::MVP, plane_mvp_z);
-                glActiveTexture(GL_TEXTURE0);
-                m_canvas->m_layers[layer_index]->m_rtt[plane_index].bindTexture();
-                glActiveTexture(GL_TEXTURE1);
-                m_canvas->m_tmp[plane_index].bindTexture();
-                glActiveTexture(GL_TEXTURE2);
-                m_canvas->m_smask.m_rtt[plane_index].bindTexture();
-                m_face_plane.draw_fill();
-                m_canvas->m_smask.m_rtt[plane_index].unbindTexture();
-                glActiveTexture(GL_TEXTURE1);
-                m_canvas->m_tmp[plane_index].unbindTexture();
-                glActiveTexture(GL_TEXTURE0);
-                m_canvas->m_layers[layer_index]->m_rtt[plane_index].unbindTexture();
-            }
-            else if(m_canvas->m_current_stroke && m_canvas->m_show_tmp && m_canvas->m_current_layer_idx == layer_index)
-            {
-                m_sampler.bind(0);
-                
-                glm::vec2 patt_scale = glm::vec2(b->m_pattern_scale);
-                if (b->m_pattern_flipx) patt_scale.x *= -1.f;
-                if (b->m_pattern_flipy) patt_scale.y *= -1.f;
-                
-                ShaderManager::use(kShader::CompDraw);
-                ShaderManager::u_int(kShaderUniform::Tex, 0);
-                ShaderManager::u_int(kShaderUniform::TexStroke, 1);
-                ShaderManager::u_int(kShaderUniform::TexMask, 2);
-                ShaderManager::u_int(kShaderUniform::TexDual, 3);
-                ShaderManager::u_vec2(kShaderUniform::Resolution, Canvas::I->m_size);
-                ShaderManager::u_int(kShaderUniform::TexPattern, 4);
-                ShaderManager::u_float(kShaderUniform::Alpha, m_canvas->m_layers[layer_index]->m_opacity);
-                ShaderManager::u_int(kShaderUniform::Lock, m_canvas->m_layers[layer_index]->m_alpha_locked);
-                ShaderManager::u_int(kShaderUniform::Mask, m_canvas->m_smask_active);
-                ShaderManager::u_int(kShaderUniform::UseFragcoord, false);
-                ShaderManager::u_int(kShaderUniform::BlendMode, b->m_blend_mode);
-                ShaderManager::u_mat4(kShaderUniform::MVP, plane_mvp_z);
-                ShaderManager::u_int(kShaderUniform::UseDual, b->m_dual_enabled);
-                ShaderManager::u_int(kShaderUniform::DualBlendMode, b->m_dual_blend_mode);
-                ShaderManager::u_int(kShaderUniform::UsePattern, b->m_pattern_enabled && !b->m_pattern_eachsample);
-                ShaderManager::u_vec2(kShaderUniform::PatternScale, patt_scale);
-                ShaderManager::u_float(kShaderUniform::PatternInvert, b->m_pattern_invert);
-                ShaderManager::u_float(kShaderUniform::PatternBright, b->m_pattern_brightness);
-                ShaderManager::u_float(kShaderUniform::PatternContrast, b->m_pattern_contrast);
-                ShaderManager::u_float(kShaderUniform::PatternDepth, b->m_pattern_depth);
-                ShaderManager::u_int(kShaderUniform::PatternBlendMode, b->m_pattern_blend_mode);
-                ShaderManager::u_vec2(kShaderUniform::PatternOffset, Canvas::I->m_pattern_offset);
-                ShaderManager::u_float(kShaderUniform::DualAlpha, b->m_dual_opacity);
-
-                glActiveTexture(GL_TEXTURE0);
-                m_canvas->m_layers[layer_index]->m_rtt[plane_index].bindTexture();
-                glActiveTexture(GL_TEXTURE1);
-                m_canvas->m_tmp[plane_index].bindTexture();
-                glActiveTexture(GL_TEXTURE2);
-                m_canvas->m_smask.m_rtt[plane_index].bindTexture();
-                glActiveTexture(GL_TEXTURE3);
-                if (b->m_dual_enabled)
-                    m_canvas->m_tmp_dual[plane_index].bindTexture();
-                glActiveTexture(GL_TEXTURE4);
-                b->m_pattern_texture ?
-                    b->m_pattern_texture->bind() :
-                    glBindTexture(GL_TEXTURE_2D, 0);
-                m_face_plane.draw_fill();
-                glActiveTexture(GL_TEXTURE3);
-                if (b->m_dual_enabled)
-                    m_canvas->m_tmp_dual[plane_index].unbindTexture();
-                glActiveTexture(GL_TEXTURE2);
-                m_canvas->m_smask.m_rtt[plane_index].unbindTexture();
-                glActiveTexture(GL_TEXTURE1);
-                m_canvas->m_tmp[plane_index].unbindTexture();
-                glActiveTexture(GL_TEXTURE0);
-                m_canvas->m_layers[layer_index]->m_rtt[plane_index].unbindTexture();
-            }
-            else
-            {
-                m_sampler.bind(0);
-                m_sampler_linear.bind(1);
-                ShaderManager::use(kShader::TextureAlphaSep);
-                ShaderManager::u_int(kShaderUniform::Tex, 0);
-                ShaderManager::u_int(kShaderUniform::TexA, 1);
-                ShaderManager::u_float(kShaderUniform::Alpha, m_canvas->m_layers[layer_index]->m_opacity);
-                ShaderManager::u_int(kShaderUniform::Highlight, m_canvas->m_layers[layer_index]->m_hightlight);
-                ShaderManager::u_mat4(kShaderUniform::MVP, plane_mvp_z);
-                
-                glActiveTexture(GL_TEXTURE0);
-                m_canvas->m_layers[layer_index]->m_rtt[plane_index].bindTexture();
-                glActiveTexture(GL_TEXTURE1);
-                m_canvas->m_layers[layer_index]->m_rtt[plane_index].bindTexture();
-                
-                m_face_plane.draw_fill();
-                
-                glActiveTexture(GL_TEXTURE1);
-                m_canvas->m_layers[layer_index]->m_rtt[plane_index].unbindTexture();
-                glActiveTexture(GL_TEXTURE0);
-                m_canvas->m_layers[layer_index]->m_rtt[plane_index].unbindTexture();
-            }
-
-            if (use_blend)
-            {
-                m_blender_rtt.unbindFramebuffer();
-            }
-
-            // draw the blended
-            if (use_blend)
-            {
-                m_sampler.bind(0);
-                m_sampler_linear.bind(1);
-                m_sampler.bind(2);
-
-                ShaderManager::use(kShader::TextureBlend);
-                ShaderManager::u_int(kShaderUniform::Tex, 0);
-                ShaderManager::u_int(kShaderUniform::TexA, 1);
-                if (!ShaderManager::ext_framebuffer_fetch)
-                    ShaderManager::u_int(kShaderUniform::TexBG, 2);
-                ShaderManager::u_int(kShaderUniform::BlendMode, m_canvas->m_layers[layer_index]->m_blend_mode);
-                ShaderManager::u_float(kShaderUniform::Alpha, 1.f);
-                ShaderManager::u_mat4(kShaderUniform::MVP, glm::ortho(-1, 1, -1, 1));
-                
-                glActiveTexture(GL_TEXTURE0);
-                m_blender_rtt.bindTexture();
-                glActiveTexture(GL_TEXTURE1);
-                m_blender_rtt.bindTexture();
-                if (!ShaderManager::ext_framebuffer_fetch)
-                {
-                    glActiveTexture(GL_TEXTURE2);
-                    m_blender_bg.bind();
-                    glCopyTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, 0, 0,
-                        m_blender_bg.size().x, m_blender_bg.size().y);
-                }
-
-                m_face_plane.draw_fill();
-
-                if (!ShaderManager::ext_framebuffer_fetch)
-                {
-                    glActiveTexture(GL_TEXTURE2);
-                    m_blender_bg.unbind();
-                }
-                glActiveTexture(GL_TEXTURE1);
-                m_blender_rtt.unbindTexture();
-                glActiveTexture(GL_TEXTURE0);
-                m_blender_rtt.unbindTexture();
-            }
-
-#ifdef _DEBUG
-            // draw dirty area
-            {
-                auto bb = m_canvas->m_layers[layer_index]->m_dirty_box[plane_index] / (float)m_canvas->m_layers[layer_index]->w;
-                glm::vec2 bbmin = xy(bb);
-                glm::vec2 bbsz = zw(bb) - xy(bb);
-                ShaderManager::use(kShader::Color);
-                ShaderManager::u_vec4(kShaderUniform::Col, { 1, 0, 0, 1 });
-                ShaderManager::u_mat4(kShaderUniform::MVP, plane_mvp_z 
-                    * glm::translate(glm::vec3(bbmin * 2.f, 0))
-                    * glm::translate(glm::vec3(-1, -1, 0))
-                    * glm::scale(glm::vec3(bbsz, 1))
-                    * glm::translate(glm::vec3(1, 1, 0))
-                );
-                m_face_plane.draw_stroke();
-            }
-#endif
-        }
-    }
-    if (use_blend)
-    {
-        m_cache_rtt.unbindFramebuffer();
-    }
-
-    // draw the grid behind the layers using a temporary copy
-    if (use_blend)
-    {
-        glEnable(GL_BLEND);
-        
-        //draw the grid
-        for (int plane_index = 0; plane_index < 6; plane_index++)
-        {
-            auto plane_mvp = proj * camera *
-                glm::scale(glm::vec3(m_canvas->m_order.size() + 500.f)) *
+            int z = 1;
+            auto plane_mvp_z = proj * camera *
+                //glm::scale(glm::vec3(z + 1)) *
+                //glm::eulerAngleYXZ(yaw, pitch, roll) *
                 m_canvas->m_plane_transform[plane_index] *
-                glm::translate(glm::vec3(0, 0, -1.f));
+                glm::translate(glm::vec3(0, 0, -1));
+
+            m_sampler.bind(0);
+            m_sampler_linear.bind(1);
+            ShaderManager::use(kShader::TextureAlphaSep);
+            ShaderManager::u_int(kShaderUniform::Tex, 0);
+            ShaderManager::u_int(kShaderUniform::TexA, 1);
+            ShaderManager::u_float(kShaderUniform::Alpha, 1.f);
+            ShaderManager::u_int(kShaderUniform::Highlight, false);
+            ShaderManager::u_mat4(kShaderUniform::MVP, plane_mvp_z);
+
+            glActiveTexture(GL_TEXTURE0);
+            m_canvas->m_layers_merge.m_rtt[plane_index].bindTexture();
+            glActiveTexture(GL_TEXTURE1);
+            m_canvas->m_layers_merge.m_rtt[plane_index].bindTexture();
 
-            ShaderManager::use(kShader::Checkerboard);
-            ShaderManager::u_int(kShaderUniform::Colorize, false);
-            ShaderManager::u_mat4(kShaderUniform::MVP, plane_mvp);
             m_face_plane.draw_fill();
+
+            glActiveTexture(GL_TEXTURE1);
+            m_canvas->m_layers_merge.m_rtt[plane_index].unbindTexture();
+            glActiveTexture(GL_TEXTURE0);
+            m_canvas->m_layers_merge.m_rtt[plane_index].unbindTexture();
+        }
+    }
+    else
+    {
+        // check if any layer use blend, otherwise draw directly on main framebuffer
+        bool use_blend = false;
+        for (size_t i = 0; i < m_canvas->m_order.size(); i++)
+        {
+            auto layer_index = m_canvas->m_order[i];
+            use_blend |= m_canvas->m_layers[layer_index]->m_blend_mode != 0;
         }
 
-        // draw the layers
-        m_sampler.bind(0);
-        glActiveTexture(GL_TEXTURE0);
-        m_cache_rtt.bindTexture();
-        ShaderManager::use(kShader::Texture);
-        ShaderManager::u_int(kShaderUniform::Tex, 0);
-        ShaderManager::u_mat4(kShaderUniform::MVP, glm::ortho<float>(-1, 1, -1, 1));
-        m_face_plane.draw_fill();
-        m_cache_rtt.unbindTexture();
+        if (use_blend)
+        {
+            m_cache_rtt.bindFramebuffer();
+            m_cache_rtt.clear({ 1, 1, 1, 0 });
+        }
+        else
+        {
+            // draw the grid
+            for (int plane_index = 0; plane_index < 6; plane_index++)
+            {
+                auto plane_mvp = proj * camera *
+                    glm::scale(glm::vec3(m_canvas->m_order.size() + 500)) *
+                    m_canvas->m_plane_transform[plane_index] *
+                    glm::translate(glm::vec3(0, 0, -1));
+
+                ShaderManager::use(kShader::Checkerboard);
+                ShaderManager::u_int(kShaderUniform::Colorize, false);
+                ShaderManager::u_mat4(kShaderUniform::MVP, plane_mvp);
+                m_face_plane.draw_fill();
+            }
+        }
+
+        // if not using shader blend, use gl rasterizer blend
+        use_blend ? glDisable(GL_BLEND) : glEnable(GL_BLEND);
+        glDisable(GL_DEPTH_TEST);
+
+        const auto& b = m_canvas->m_current_stroke->m_brush;
+
+        for (size_t i = 0; i < m_canvas->m_order.size(); i++)
+        {
+            auto layer_index = m_canvas->m_order[i];
+            for (int plane_index = 0; plane_index < 6; plane_index++)
+            {
+                if (!(m_canvas->m_show_tmp && m_canvas->m_current_layer_idx == layer_index) && 
+                    (!m_canvas->m_layers[layer_index]->m_visible ||
+                    m_canvas->m_layers[layer_index]->m_opacity == .0f ||
+                    !m_canvas->m_layers[layer_index]->m_dirty_face[plane_index]))
+                    continue;
+
+                if (use_blend)
+                {
+                    m_blender_rtt.bindFramebuffer();
+                    m_blender_rtt.clear();
+                }
+
+                int z = (int)(m_canvas->m_order.size() - i);
+                auto plane_mvp_z = proj * camera * 
+                    glm::scale(glm::vec3(z + 1)) * 
+                    glm::eulerAngleYXZ(yaw, pitch, roll) *
+                    m_canvas->m_plane_transform[plane_index] * 
+                    glm::translate(glm::vec3(0, 0, -1));
+            
+                if (m_canvas->m_current_stroke && m_canvas->m_current_mode == kCanvasMode::Erase && m_canvas->m_show_tmp && m_canvas->m_current_layer_idx == layer_index)
+                {
+                    m_sampler.bind(0);
+                    ShaderManager::use(kShader::CompErase);
+                    ShaderManager::u_int(kShaderUniform::Tex, 0);
+                    ShaderManager::u_int(kShaderUniform::TexStroke, 1);
+                    ShaderManager::u_int(kShaderUniform::TexMask, 2);
+                    //ShaderManager::u_vec2(kShaderUniform::Resolution, zw(m_canvas->m_box) / zoom);
+                    ShaderManager::u_float(kShaderUniform::Alpha, m_canvas->m_layers[layer_index]->m_opacity);
+                    //ShaderManager::u_int(kShaderUniform::Lock, m_canvas->m_layers[layer_index]->m_alpha_locked);
+                    ShaderManager::u_int(kShaderUniform::Mask, m_canvas->m_smask_active);
+                    ShaderManager::u_mat4(kShaderUniform::MVP, plane_mvp_z);
+                    glActiveTexture(GL_TEXTURE0);
+                    m_canvas->m_layers[layer_index]->m_rtt[plane_index].bindTexture();
+                    glActiveTexture(GL_TEXTURE1);
+                    m_canvas->m_tmp[plane_index].bindTexture();
+                    glActiveTexture(GL_TEXTURE2);
+                    m_canvas->m_smask.m_rtt[plane_index].bindTexture();
+                    m_face_plane.draw_fill();
+                    m_canvas->m_smask.m_rtt[plane_index].unbindTexture();
+                    glActiveTexture(GL_TEXTURE1);
+                    m_canvas->m_tmp[plane_index].unbindTexture();
+                    glActiveTexture(GL_TEXTURE0);
+                    m_canvas->m_layers[layer_index]->m_rtt[plane_index].unbindTexture();
+                }
+                else if(m_canvas->m_current_stroke && m_canvas->m_show_tmp && m_canvas->m_current_layer_idx == layer_index)
+                {
+                    m_sampler.bind(0);
+                
+                    glm::vec2 patt_scale = glm::vec2(b->m_pattern_scale);
+                    if (b->m_pattern_flipx) patt_scale.x *= -1.f;
+                    if (b->m_pattern_flipy) patt_scale.y *= -1.f;
+                
+                    ShaderManager::use(kShader::CompDraw);
+                    ShaderManager::u_int(kShaderUniform::Tex, 0);
+                    ShaderManager::u_int(kShaderUniform::TexStroke, 1);
+                    ShaderManager::u_int(kShaderUniform::TexMask, 2);
+                    ShaderManager::u_int(kShaderUniform::TexDual, 3);
+                    ShaderManager::u_vec2(kShaderUniform::Resolution, Canvas::I->m_size);
+                    ShaderManager::u_int(kShaderUniform::TexPattern, 4);
+                    ShaderManager::u_float(kShaderUniform::Alpha, m_canvas->m_layers[layer_index]->m_opacity);
+                    ShaderManager::u_int(kShaderUniform::Lock, m_canvas->m_layers[layer_index]->m_alpha_locked);
+                    ShaderManager::u_int(kShaderUniform::Mask, m_canvas->m_smask_active);
+                    ShaderManager::u_int(kShaderUniform::UseFragcoord, false);
+                    ShaderManager::u_int(kShaderUniform::BlendMode, b->m_blend_mode);
+                    ShaderManager::u_mat4(kShaderUniform::MVP, plane_mvp_z);
+                    ShaderManager::u_int(kShaderUniform::UseDual, b->m_dual_enabled);
+                    ShaderManager::u_int(kShaderUniform::DualBlendMode, b->m_dual_blend_mode);
+                    ShaderManager::u_int(kShaderUniform::UsePattern, b->m_pattern_enabled && !b->m_pattern_eachsample);
+                    ShaderManager::u_vec2(kShaderUniform::PatternScale, patt_scale);
+                    ShaderManager::u_float(kShaderUniform::PatternInvert, b->m_pattern_invert);
+                    ShaderManager::u_float(kShaderUniform::PatternBright, b->m_pattern_brightness);
+                    ShaderManager::u_float(kShaderUniform::PatternContrast, b->m_pattern_contrast);
+                    ShaderManager::u_float(kShaderUniform::PatternDepth, b->m_pattern_depth);
+                    ShaderManager::u_int(kShaderUniform::PatternBlendMode, b->m_pattern_blend_mode);
+                    ShaderManager::u_vec2(kShaderUniform::PatternOffset, Canvas::I->m_pattern_offset);
+                    ShaderManager::u_float(kShaderUniform::DualAlpha, b->m_dual_opacity);
+
+                    glActiveTexture(GL_TEXTURE0);
+                    m_canvas->m_layers[layer_index]->m_rtt[plane_index].bindTexture();
+                    glActiveTexture(GL_TEXTURE1);
+                    m_canvas->m_tmp[plane_index].bindTexture();
+                    glActiveTexture(GL_TEXTURE2);
+                    m_canvas->m_smask.m_rtt[plane_index].bindTexture();
+                    glActiveTexture(GL_TEXTURE3);
+                    if (b->m_dual_enabled)
+                        m_canvas->m_tmp_dual[plane_index].bindTexture();
+                    glActiveTexture(GL_TEXTURE4);
+                    b->m_pattern_texture ?
+                        b->m_pattern_texture->bind() :
+                        glBindTexture(GL_TEXTURE_2D, 0);
+                    m_face_plane.draw_fill();
+                    glActiveTexture(GL_TEXTURE3);
+                    if (b->m_dual_enabled)
+                        m_canvas->m_tmp_dual[plane_index].unbindTexture();
+                    glActiveTexture(GL_TEXTURE2);
+                    m_canvas->m_smask.m_rtt[plane_index].unbindTexture();
+                    glActiveTexture(GL_TEXTURE1);
+                    m_canvas->m_tmp[plane_index].unbindTexture();
+                    glActiveTexture(GL_TEXTURE0);
+                    m_canvas->m_layers[layer_index]->m_rtt[plane_index].unbindTexture();
+                }
+                else
+                {
+                    m_sampler.bind(0);
+                    m_sampler_linear.bind(1);
+                    ShaderManager::use(kShader::TextureAlphaSep);
+                    ShaderManager::u_int(kShaderUniform::Tex, 0);
+                    ShaderManager::u_int(kShaderUniform::TexA, 1);
+                    ShaderManager::u_float(kShaderUniform::Alpha, m_canvas->m_layers[layer_index]->m_opacity);
+                    ShaderManager::u_int(kShaderUniform::Highlight, m_canvas->m_layers[layer_index]->m_hightlight);
+                    ShaderManager::u_mat4(kShaderUniform::MVP, plane_mvp_z);
+                
+                    glActiveTexture(GL_TEXTURE0);
+                    m_canvas->m_layers[layer_index]->m_rtt[plane_index].bindTexture();
+                    glActiveTexture(GL_TEXTURE1);
+                    m_canvas->m_layers[layer_index]->m_rtt[plane_index].bindTexture();
+                
+                    m_face_plane.draw_fill();
+                
+                    glActiveTexture(GL_TEXTURE1);
+                    m_canvas->m_layers[layer_index]->m_rtt[plane_index].unbindTexture();
+                    glActiveTexture(GL_TEXTURE0);
+                    m_canvas->m_layers[layer_index]->m_rtt[plane_index].unbindTexture();
+                }
+
+                if (use_blend)
+                {
+                    m_blender_rtt.unbindFramebuffer();
+                }
+
+                // draw the blended
+                if (use_blend)
+                {
+                    m_sampler.bind(0);
+                    m_sampler_linear.bind(1);
+                    m_sampler.bind(2);
+
+                    ShaderManager::use(kShader::TextureBlend);
+                    ShaderManager::u_int(kShaderUniform::Tex, 0);
+                    ShaderManager::u_int(kShaderUniform::TexA, 1);
+                    if (!ShaderManager::ext_framebuffer_fetch)
+                        ShaderManager::u_int(kShaderUniform::TexBG, 2);
+                    ShaderManager::u_int(kShaderUniform::BlendMode, m_canvas->m_layers[layer_index]->m_blend_mode);
+                    ShaderManager::u_float(kShaderUniform::Alpha, 1.f);
+                    ShaderManager::u_mat4(kShaderUniform::MVP, glm::ortho(-1, 1, -1, 1));
+                
+                    glActiveTexture(GL_TEXTURE0);
+                    m_blender_rtt.bindTexture();
+                    glActiveTexture(GL_TEXTURE1);
+                    m_blender_rtt.bindTexture();
+                    if (!ShaderManager::ext_framebuffer_fetch)
+                    {
+                        glActiveTexture(GL_TEXTURE2);
+                        m_blender_bg.bind();
+                        glCopyTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, 0, 0,
+                            m_blender_bg.size().x, m_blender_bg.size().y);
+                    }
+
+                    m_face_plane.draw_fill();
+
+                    if (!ShaderManager::ext_framebuffer_fetch)
+                    {
+                        glActiveTexture(GL_TEXTURE2);
+                        m_blender_bg.unbind();
+                    }
+                    glActiveTexture(GL_TEXTURE1);
+                    m_blender_rtt.unbindTexture();
+                    glActiveTexture(GL_TEXTURE0);
+                    m_blender_rtt.unbindTexture();
+                }
+
+    #ifdef _DEBUG
+                // draw dirty area
+                {
+                    auto bb = m_canvas->m_layers[layer_index]->m_dirty_box[plane_index] / (float)m_canvas->m_layers[layer_index]->w;
+                    glm::vec2 bbmin = xy(bb);
+                    glm::vec2 bbsz = zw(bb) - xy(bb);
+                    ShaderManager::use(kShader::Color);
+                    ShaderManager::u_vec4(kShaderUniform::Col, { 1, 0, 0, 1 });
+                    ShaderManager::u_mat4(kShaderUniform::MVP, plane_mvp_z 
+                        * glm::translate(glm::vec3(bbmin * 2.f, 0))
+                        * glm::translate(glm::vec3(-1, -1, 0))
+                        * glm::scale(glm::vec3(bbsz, 1))
+                        * glm::translate(glm::vec3(1, 1, 0))
+                    );
+                    m_face_plane.draw_stroke();
+                }
+    #endif
+            }
+        }
+        if (use_blend)
+        {
+            m_cache_rtt.unbindFramebuffer();
+        }
+
+        // draw the grid behind the layers using a temporary copy
+        if (use_blend)
+        {
+            glEnable(GL_BLEND);
+        
+            //draw the grid
+            for (int plane_index = 0; plane_index < 6; plane_index++)
+            {
+                auto plane_mvp = proj * camera *
+                    glm::scale(glm::vec3(m_canvas->m_order.size() + 500.f)) *
+                    m_canvas->m_plane_transform[plane_index] *
+                    glm::translate(glm::vec3(0, 0, -1.f));
+
+                ShaderManager::use(kShader::Checkerboard);
+                ShaderManager::u_int(kShaderUniform::Colorize, false);
+                ShaderManager::u_mat4(kShaderUniform::MVP, plane_mvp);
+                m_face_plane.draw_fill();
+            }
+
+            // draw the layers
+            m_sampler.bind(0);
+            glActiveTexture(GL_TEXTURE0);
+            m_cache_rtt.bindTexture();
+            ShaderManager::use(kShader::Texture);
+            ShaderManager::u_int(kShaderUniform::Tex, 0);
+            ShaderManager::u_mat4(kShaderUniform::MVP, glm::ortho<float>(-1, 1, -1, 1));
+            m_face_plane.draw_fill();
+            m_cache_rtt.unbindTexture();
+        }
     }
 
     glDisable(GL_DEPTH_TEST);