diff --git a/tests/test_wgpu_native_basics.py b/tests/test_wgpu_native_basics.py
index fa662d31..61a65625 100644
--- a/tests/test_wgpu_native_basics.py
+++ b/tests/test_wgpu_native_basics.py
@@ -481,5 +481,82 @@ def test_limits_are_not_legal():
     assert not are_limits_wgpu_legal({"max-bind-group": 8})
 
 
+def test_bind_group_resources():
+    # make sure every resource type can be created in a bind group.
+    device = wgpu.utils.get_default_device()
+
+    sampler = device.create_sampler()
+    texture1 = device.create_texture(
+        label="texture1",
+        size=(8, 8, 1),
+        format=wgpu.TextureFormat.rgba8unorm,
+        usage=wgpu.TextureUsage.TEXTURE_BINDING | wgpu.TextureUsage.COPY_DST,
+    )
+    texture2 = device.create_texture(
+        label="texture2",
+        size=(8, 8, 1),
+        format=wgpu.TextureFormat.rgba8unorm,
+        usage=wgpu.TextureUsage.TEXTURE_BINDING | wgpu.TextureUsage.COPY_DST,
+    )
+    texture_view = texture2.create_view()
+
+    buffer = device.create_buffer(
+        size=256,
+        usage=wgpu.BufferUsage.UNIFORM,
+    )
+
+    entires = [
+        wgpu.structs.BindGroupEntry(binding=0, resource=sampler),
+        wgpu.structs.BindGroupEntry(binding=1, resource=texture1),
+        wgpu.structs.BindGroupEntry(binding=2, resource=texture_view),
+        wgpu.structs.BindGroupEntry(
+            binding=3,
+            resource=buffer,
+        ),
+        wgpu.structs.BindGroupEntry(
+            binding=4,
+            resource=wgpu.structs.BufferBinding(
+                buffer=buffer,
+            ),
+        ),
+        # maybe external texture one day
+    ]
+
+    # maybe we should write a bit of compute shader so we can use auto layout mode instead.
+    layout_entries = [
+        wgpu.BindGroupLayoutEntry(
+            binding=0,
+            visibility=wgpu.ShaderStage.COMPUTE,
+            sampler=wgpu.structs.SamplerBindingLayout(),
+        ),
+        wgpu.BindGroupLayoutEntry(
+            binding=1,
+            visibility=wgpu.ShaderStage.COMPUTE,
+            texture=wgpu.structs.TextureBindingLayout(),
+        ),
+        wgpu.BindGroupLayoutEntry(
+            binding=2,
+            visibility=wgpu.ShaderStage.COMPUTE,
+            texture=wgpu.structs.TextureBindingLayout(),
+        ),
+        wgpu.BindGroupLayoutEntry(
+            binding=3,
+            visibility=wgpu.ShaderStage.COMPUTE,
+            buffer=wgpu.structs.BufferBindingLayout(),
+        ),
+        wgpu.BindGroupLayoutEntry(
+            binding=4,
+            visibility=wgpu.ShaderStage.COMPUTE,
+            buffer=wgpu.structs.BufferBindingLayout(),
+        ),
+    ]
+
+    layout = device.create_bind_group_layout(entries=layout_entries)
+
+    bind_group = device.create_bind_group(layout=layout, entries=entires)
+
+    assert bind_group
+
+
 if __name__ == "__main__":
     run_tests(globals())
diff --git a/wgpu/_classes.py b/wgpu/_classes.py
index adb19a5c..09ed8b12 100644
--- a/wgpu/_classes.py
+++ b/wgpu/_classes.py
@@ -965,7 +965,7 @@ def create_bind_group(
             layout (GPUBindGroupLayout): The layout (abstract representation)
                 for this bind group.
             entries (list): A list of `structs.BindGroupEntry`s. The ``resource`` field
-                is either `GPUSampler`, `GPUTextureView` or `structs.BufferBinding`.
+                is either `GPUBuffer`, `GPUSampler`, `GPUTexture`, `GPUTextureView` or `structs.BufferBinding`.
 
         Example entry dicts:
 
diff --git a/wgpu/backends/wgpu_native/_api.py b/wgpu/backends/wgpu_native/_api.py
index c59d8dd0..a4398585 100644
--- a/wgpu/backends/wgpu_native/_api.py
+++ b/wgpu/backends/wgpu_native/_api.py
@@ -1699,9 +1699,10 @@ def create_bind_group(
         entries: Sequence[structs.BindGroupEntryStruct],
     ) -> GPUBindGroup:
         c_entries_list = []
+        _keep_alive = []
         for entry in entries:
             check_struct("BindGroupEntry", entry)
-            # The resource can be a buffer, sampler, texture view, or buffer descriptor
+            # The resource can be a buffer, sampler, texture, texture view, or buffer descriptor
             resource = entry["resource"]
             if isinstance(resource, GPUBuffer):
                 # H: nextInChain: WGPUChainedStruct *, binding: int, buffer: WGPUBuffer, offset: int, size: int, sampler: WGPUSampler, textureView: WGPUTextureView
@@ -1727,7 +1728,11 @@ def create_bind_group(
                     sampler=resource._internal,
                     textureView=ffi.NULL,
                 )
-            elif isinstance(resource, GPUTextureView):
+            elif isinstance(resource, (GPUTextureView, GPUTexture)):
+                if type(resource) is GPUTexture:
+                    # also see https://github.com/pygfx/wgpu-py/issues/825
+                    resource = resource.create_view()
+                    _keep_alive.append(resource)
                 # H: nextInChain: WGPUChainedStruct *, binding: int, buffer: WGPUBuffer, offset: int, size: int, sampler: WGPUSampler, textureView: WGPUTextureView
                 c_entry = new_struct(
                     "WGPUBindGroupEntry",
@@ -1767,6 +1772,7 @@ def create_bind_group(
 
         # H: WGPUBindGroup f(WGPUDevice device, WGPUBindGroupDescriptor const * descriptor)
         id = libf.wgpuDeviceCreateBindGroup(self._internal, struct)
+        del _keep_alive
         return GPUBindGroup(label, id, self)
 
     def create_pipeline_layout(