diff --git a/pax-chassis-web/Cargo.toml b/pax-chassis-web/Cargo.toml
index 0ae77e3ad..804dc54b9 100644
--- a/pax-chassis-web/Cargo.toml
+++ b/pax-chassis-web/Cargo.toml
@@ -13,7 +13,7 @@ include = ["/src", "/interface/public"]
 crate-type = ["cdylib", "rlib"]
 
 [features]
-default = ["console_error_panic_hook"]
+default = ["console_error_panic_hook", "gpu"]
 designtime = ["dep:pax-designtime", "pax-runtime/designtime"]
 gpu = []
 
diff --git a/pax-pixels/src/render_backend/stencil.rs b/pax-pixels/src/render_backend/stencil.rs
index 507c5bb78..b8d9c0544 100644
--- a/pax-pixels/src/render_backend/stencil.rs
+++ b/pax-pixels/src/render_backend/stencil.rs
@@ -18,6 +18,8 @@ pub struct StencilRenderer {
     stencil_layer: u32,
     width: u32,
     height: u32,
+    stencil_bind_group: wgpu::BindGroup,
+    _stencil_bind_group_layout: wgpu::BindGroupLayout,
 }
 
 #[repr(C)]
@@ -27,15 +29,39 @@ pub struct Vertex {
 }
 
 impl StencilRenderer {
-    pub fn new(device: &Device, width: u32, height: u32, _globals: &wgpu::Buffer) -> Self {
+    pub fn new(device: &Device, width: u32, height: u32, globals: &wgpu::Buffer) -> Self {
         let shader = device.create_shader_module(wgpu::ShaderModuleDescriptor {
             label: Some("Stencil Shader"),
             source: wgpu::ShaderSource::Wgsl(include_str!("stencil.wgsl").into()),
         });
 
+        let stencil_bind_group_layout =
+            device.create_bind_group_layout(&wgpu::BindGroupLayoutDescriptor {
+                entries: &[wgpu::BindGroupLayoutEntry {
+                    binding: 0,
+                    visibility: wgpu::ShaderStages::VERTEX_FRAGMENT,
+                    ty: wgpu::BindingType::Buffer {
+                        ty: wgpu::BufferBindingType::Uniform,
+                        has_dynamic_offset: false,
+                        min_binding_size: None,
+                    },
+                    count: None,
+                }],
+                label: Some("stencil_bind_group_layout"),
+            });
+
+        let stencil_bind_group = device.create_bind_group(&wgpu::BindGroupDescriptor {
+            layout: &stencil_bind_group_layout,
+            entries: &[wgpu::BindGroupEntry {
+                binding: 0,
+                resource: globals.as_entire_binding(),
+            }],
+            label: Some("stencil_bind_group"),
+        });
+
         let pipeline_layout = device.create_pipeline_layout(&wgpu::PipelineLayoutDescriptor {
             label: Some("Stencil Pipeline Layout"),
-            bind_group_layouts: &[],
+            bind_group_layouts: &[&stencil_bind_group_layout],
             push_constant_ranges: &[],
         });
 
@@ -62,7 +88,7 @@ impl StencilRenderer {
             primitive: wgpu::PrimitiveState {
                 topology: wgpu::PrimitiveTopology::TriangleList,
                 strip_index_format: None,
-                front_face: wgpu::FrontFace::Cw,
+                front_face: wgpu::FrontFace::Ccw,
                 cull_mode: None,
                 polygon_mode: wgpu::PolygonMode::Fill,
                 unclipped_depth: false,
@@ -150,7 +176,7 @@ impl StencilRenderer {
             usage: BufferUsages::VERTEX | BufferUsages::COPY_DST,
         });
 
-        let mut indices: [u16; 1024] = [0; 1024];
+        let indices: [u16; 1024] = [0; 1024];
         let indices_buffer = device.create_buffer_init(&wgpu::util::BufferInitDescriptor {
             label: Some("Stencil Indices"),
             contents: bytemuck::cast_slice(&indices),
@@ -189,6 +215,8 @@ impl StencilRenderer {
             width,
             height,
             stencil_layer: 0,
+            stencil_bind_group,
+            _stencil_bind_group_layout: stencil_bind_group_layout,
         }
     }
 
@@ -266,6 +294,7 @@ impl StencilRenderer {
 
             render_pass.set_pipeline(&self.stencil_pipeline);
             render_pass.set_vertex_buffer(0, self.vertices_buffer.slice(..));
+            render_pass.set_bind_group(0, &self.stencil_bind_group, &[]);
             render_pass.set_index_buffer(self.indices_buffer.slice(..), wgpu::IndexFormat::Uint16);
             render_pass.draw_indexed(0..(geometry.indices.len() as u32), 0, 0..1);
         }
@@ -298,6 +327,7 @@ impl StencilRenderer {
 
                 // Use the decrement pipeline and fullscreen quad to decrease all stencil values
                 render_pass.set_pipeline(&self.decrement_pipeline);
+                render_pass.set_bind_group(0, &self.stencil_bind_group, &[]);
                 render_pass.set_vertex_buffer(0, self.fullscreen_vertices_buffer.slice(..));
                 render_pass.draw(0..3, 0..1);
             }
diff --git a/pax-pixels/src/render_backend/stencil.wgsl b/pax-pixels/src/render_backend/stencil.wgsl
index a5f36fa19..081c574ae 100644
--- a/pax-pixels/src/render_backend/stencil.wgsl
+++ b/pax-pixels/src/render_backend/stencil.wgsl
@@ -5,7 +5,7 @@ struct Globals {
 };
 
 
-// @group(0) @binding(0) var<uniform> globals: Globals;
+@group(0) @binding(0) var<uniform> globals: Globals;
 
 struct VertexOutput {
     @builtin(position) clip_position: vec4<f32>,
@@ -16,12 +16,12 @@ fn vs_main(
     @location(0) position: vec2<f32>,
 ) -> VertexOutput {
 	var out: VertexOutput;
- //    var pos = position;
- //    pos /= globals.resolution;
- //    pos *= 2.0;
- //    pos -= 1.0;
- //    pos.y *= -1.0;
-    out.clip_position = vec4<f32>(position, 0.0, 1.0);
+    var pos = position;
+    pos /= globals.resolution;
+    pos *= 2.0;
+    pos -= 1.0;
+    pos.y *= -1.0;
+    out.clip_position = vec4<f32>(pos, 0.0, 1.0);
     return out;
 }
 
diff --git a/pax-pixels/src/render_context.rs b/pax-pixels/src/render_context.rs
index 6a94b7999..7e1646f86 100644
--- a/pax-pixels/src/render_context.rs
+++ b/pax-pixels/src/render_context.rs
@@ -194,7 +194,7 @@ impl<'w> WgpuRenderer<'w> {
         }
     }
 
-    pub fn push_transform(&mut self, transform: Transform2D) {
+    pub fn transform(&mut self, transform: Transform2D) {
         let new_ind = self.buffers.transforms.len();
         self.buffers.transforms.push(GpuTransform {
             transform: transform.then(&self.current_transform()).to_arrays(),
@@ -204,7 +204,24 @@ impl<'w> WgpuRenderer<'w> {
         self.transform_index_stack.push(new_ind);
     }
 
-    pub fn push_clipping(&mut self, path: Path) {
+    pub fn clip(&mut self, path: Path) {
+        // fine to transform on CPU - shouldn't be large meshes
+        // let path = path.transformed(&self.current_transform());
+        if self.buffers.primitives.len() > 0 {
+            self.render_backend.render_primitives(&mut self.buffers);
+            let CpuBuffers {
+                geometry,
+                primitives,
+                transforms: _,
+                colors,
+                gradients,
+            } = &mut self.buffers;
+            geometry.vertices.clear();
+            geometry.indices.clear();
+            primitives.clear();
+            colors.clear();
+            gradients.clear();
+        }
         let options = FillOptions::tolerance(self.tolerance);
         let mut geometry = VertexBuffers::new();
         let mut geometry_builder =
diff --git a/pax-runtime/src/engine/mod.rs b/pax-runtime/src/engine/mod.rs
index d3332d449..5923a693a 100644
--- a/pax-runtime/src/engine/mod.rs
+++ b/pax-runtime/src/engine/mod.rs
@@ -301,7 +301,6 @@ impl PaxEngine {
                 .unwrap_or(false)
             {
                 rcs.clear(i);
-                rcs.clip(i, Rect::new(0.0, 0.0, 0.5, 0.5).into_path(0.01)) // only for testing, REMOVE
             }
         }
         // This is pretty useful during debugging - left it here since I use it often. /Sam
diff --git a/pax-runtime/src/engine/pax_pixels_render_context.rs b/pax-runtime/src/engine/pax_pixels_render_context.rs
index 75314577c..cd2670ccd 100644
--- a/pax-runtime/src/engine/pax_pixels_render_context.rs
+++ b/pax-runtime/src/engine/pax_pixels_render_context.rs
@@ -99,12 +99,12 @@ impl RenderContext for PaxPixelsRenderer {
         self.with_layer_context(layer, |context| {
             log::debug!("pax_pixels_render_context path {:?}", path);
             let path = convert_kurbo_to_lyon_path(&path);
-            context.push_clipping(path);
+            context.clip(path);
         });
     }
     fn transform(&mut self, layer: usize, affine: kurbo::Affine) {
         self.with_layer_context(layer, |context| {
-            context.push_transform(Transform2D::from_array(
+            context.transform(Transform2D::from_array(
                 affine.as_coeffs().map(|v| v as f32),
             ))
         });
diff --git a/pax-std/src/core/frame.rs b/pax-std/src/core/frame.rs
index 64b9c95d7..9c2f074d9 100644
--- a/pax-std/src/core/frame.rs
+++ b/pax-std/src/core/frame.rs
@@ -98,7 +98,7 @@ impl InstanceNode for FrameInstance {
         for layer in 0..layers {
             //our "save point" before clipping — restored to in the post_render
             rcs.save(layer);
-            // rcs.clip(layer, transformed_bez_path.clone());
+            rcs.clip(layer, transformed_bez_path.clone());
         }
     }
 
diff --git a/pax-std/src/drawing/image.rs b/pax-std/src/drawing/image.rs
index 568eff15a..af076e5fd 100644
--- a/pax-std/src/drawing/image.rs
+++ b/pax-std/src/drawing/image.rs
@@ -220,7 +220,7 @@ impl InstanceNode for ImageInstance {
             let clip_path = kurbo::Rect::new(0.0, 0.0, container_width, container_height);
             rc.save(layer_id);
             rc.transform(layer_id, t_and_b.transform.into());
-            // rc.clip(layer_id, clip_path.into_path(0.01));
+            rc.clip(layer_id, clip_path.into_path(0.01));
             rc.draw_image(layer_id, &path, transformed_bounds);
             rc.restore(layer_id);
             self.initial_load
diff --git a/pax-std/src/drawing/path.rs b/pax-std/src/drawing/path.rs
index 810dbc64e..ca33e0f14 100644
--- a/pax-std/src/drawing/path.rs
+++ b/pax-std/src/drawing/path.rs
@@ -223,7 +223,7 @@ impl InstanceNode for PathInstance {
             //our "save point" before clipping — restored to in the post_render
 
             rc.save(layer_id);
-            // rc.clip(layer_id, transformed_clip_path.clone());
+            rc.clip(layer_id, transformed_clip_path.clone());
             rc.fill(layer_id, transformed_bez_path, &properties.fill.get());
             if properties
                 .stroke