Graphics modifiers

In addition to the Canvas composable, Compose has several useful graphics Modifiers which aid in drawing custom content. These modifiers are useful because they can be applied to any composable.

Drawing modifiers

All drawing commands are done with a drawing modifier in Compose. There are three main drawing modifiers in Compose:

The base modifier for drawing is drawWithContent, where you can decide the drawing order of your Composable and the drawing commands issued inside the modifier. drawBehind is a convenient wrapper around drawWithContent which has the drawing order set to behind the content of the composable. drawWithCache calls either onDrawBehind or onDrawWithContent inside of it - and provides a mechanism for caching the objects created in them.

Modifier.drawWithContent: Choose drawing order

Modifier.drawWithContent lets you execute DrawScope operations before or after the content of the composable. Be sure to call drawContent to then render the actual content of the composable. With this modifier, you can decide the order of operations, if you want your content to be drawn before or after your custom drawing operations.

For example, if you wished to render a radial gradient on top of your content to create a flashlight keyhole effect on the UI, you could do the following:

var pointerOffset by remember {
    mutableStateOf(Offset(0f, 0f))
}
Column(
    modifier = Modifier
        .fillMaxSize()
        .pointerInput("dragging") {
            detectDragGestures { change, dragAmount ->
                pointerOffset += dragAmount
            }
        }
        .onSizeChanged {
            pointerOffset = Offset(it.width / 2f, it.height / 2f)
        }
        .drawWithContent {
            drawContent()
            // draws a fully black area with a small keyhole at pointerOffset that’ll show part of the UI.
            drawRect(
                Brush.radialGradient(
                    listOf(Color.Transparent, Color.Black),
                    center = pointerOffset,
                    radius = 100.dp.toPx(),
                )
            )
        }
) {
    // Your composables here
}

Figure 1: Modifier.drawWithContent used on top of a Composable to create a flashlight type UI experience.

Modifier.drawBehind: Drawing behind a composable

Modifier.drawBehind lets you perform DrawScope operations behind the composable content that is drawn on screen. If you take a look at the implementation of Canvas, you might notice that it is just a convenient wrapper around Modifier.drawBehind.

To draw a rounded rectangle behind Text:

Text(
    "Hello Compose!",
    modifier = Modifier
        .drawBehind {
            drawRoundRect(
                Color(0xFFBBAAEE),
                cornerRadius = CornerRadius(10.dp.toPx())
            )
        }
        .padding(4.dp)
)

Which produces the following result:

Text and a background drawn using Modifier.drawBehind
Figure 2: Text and a background drawn using Modifier.drawBehind

Modifier.drawWithCache: Drawing and caching draw objects

Modifier.drawWithCache keeps the objects that are created inside of it cached. The objects are cached as long as the size of the drawing area is the same, or any state objects that are read have not changed. This modifier is useful for improving performance of drawing calls as it avoids the need to reallocate objects (such as: Brush, Shader, Path etc.) that are created on draw.

Alternatively, you could also cache objects using remember, outside of the modifier. However, this is not always possible as you don't always have access to the composition. It can be more performant to use drawWithCache if the objects are only used for drawing.

For example, if you create a Brush to draw a gradient behind a Text, using drawWithCache caches the Brush object until the size of the drawing area changes:

Text(
    "Hello Compose!",
    modifier = Modifier
        .drawWithCache {
            val brush = Brush.linearGradient(
                listOf(
                    Color(0xFF9E82F0),
                    Color(0xFF42A5F5)
                )
            )
            onDrawBehind {
                drawRoundRect(
                    brush,
                    cornerRadius = CornerRadius(10.dp.toPx())
                )
            }
        }
)

Caching the Brush object with drawWithCache
Figure 3: Caching the Brush object with drawWithCache

Graphics modifiers

Modifier.graphicsLayer: Apply transformations to composables

Modifier.graphicsLayer is a modifier that makes the content of the composable draw into a draw layer. A layer provides a few different functions, such as:

  • Isolation for its drawing instructions (similar to RenderNode). Drawing instructions captured as part of a layer can be re-issued efficiently by the rendering pipeline without re-executing application code.
  • Transformations that apply to all the drawing instructions contained within a layer.
  • Rasterization for composition capabilities. When a layer is rasterized, its drawing instructions are executed and the output is captured into an offscreen buffer. Compositing such a buffer for subsequent frames is faster than executing the individual instructions, but it will behave as a bitmap when transforms like scaling or rotation are applied.

Transformations

Modifier.graphicsLayer provides isolation for its drawing instructions; for instance, various transformations can be applied using Modifier.graphicsLayer. These can be animated or modified without needing to re-execute the drawing lambda.

Modifier.graphicsLayer does not change the measured size or placement of your composable, as it only affects the draw phase. This means that your composable might overlap others if it ends up drawing outside of its layout bounds.

The following transformations can be applied with this modifier:

Scale - increase size

scaleX and scaleY enlarges or shrinks content in the horizontal or vertical direction, respectively. A value of 1.0f indicates no change in scale, a value of 0.5f means half of the dimension.

Image(
    painter = painterResource(id = R.drawable.sunset),
    contentDescription = "Sunset",
    modifier = Modifier
        .graphicsLayer {
            this.scaleX = 1.2f
            this.scaleY = 0.8f
        }
)

Figure 4: scaleX and scaleY applied to an Image composable
Translation

translationX and translationY can be changed with graphicsLayer, translationX moves the composable left or right. translationY moves the composable up or down.

Image(
    painter = painterResource(id = R.drawable.sunset),
    contentDescription = "Sunset",
    modifier = Modifier
        .graphicsLayer {
            this.translationX = 100.dp.toPx()
            this.translationY = 10.dp.toPx()
        }
)

Figure 5: translationX and translationY applied to Image with Modifier.graphicsLayer
Rotation

Set rotationX to rotate horizontally, rotationY to rotate vertically and rotationZ to rotate on the Z axis (standard rotation). This value is specified in degrees (0-360).

Image(
    painter = painterResource(id = R.drawable.sunset),
    contentDescription = "Sunset",
    modifier = Modifier
        .graphicsLayer {
            this.rotationX = 90f
            this.rotationY = 275f
            this.rotationZ = 180f
        }
)

Figure 6: rotationX, rotationY and rotationZ set on Image by Modifier.graphicsLayer
Origin

A transformOrigin can be specified. It is then used as the point from which transformations take place. All the examples so far have used TransformOrigin.Center, which is at (0.5f, 0.5f). If you specify the origin at (0f, 0f), the transformations then start from the top-left corner of the composable.

If you change the origin with a rotationZ transformation, you can see that the item rotates around the top left of the composable:

Image(
    painter = painterResource(id = R.drawable.sunset),
    contentDescription = "Sunset",
    modifier = Modifier
        .graphicsLayer {
            this.transformOrigin = TransformOrigin(0f, 0f)
            this.rotationX = 90f
            this.rotationY = 275f
            this.rotationZ = 180f
        }
)

Figure 7: Rotation applied with TransformOrigin set to 0f, 0f

Clip and Shape

Shape specifies the outline that the content clips to when clip = true. In this example, we set two boxes to have two different clips - one using graphicsLayer clip variable, and the other using the convenient wrapper Modifier.clip.

Column(modifier = Modifier.padding(16.dp)) {
    Box(
        modifier = Modifier
            .size(200.dp)
            .graphicsLayer {
                clip = true
                shape = CircleShape
            }
            .background(Color(0xFFF06292))
    ) {
        Text(
            "Hello Compose",
            style = TextStyle(color = Color.Black, fontSize = 46.sp),
            modifier = Modifier.align(Alignment.Center)
        )
    }
    Box(
        modifier = Modifier
            .size(200.dp)
            .clip(CircleShape)
            .background(Color(0xFF4DB6AC))
    )
}

The contents of the first box (the text saying “Hello Compose”) are clipped to the circle shape:

Clip applied to Box composable
Figure 8: Clip applied to Box composable

If you then apply a translationY to the top pink circle, you see that the bounds of the Composable are still the same, but the circle draws underneath the bottom circle (and outside of its bounds).

Clip applied with translationY, and red border for outline
Figure 9: Clip applied with translationY, and red border for outline

To clip the composable to the region it's drawn in, you can add another Modifier.clip(RectangleShape) at the start of the modifier chain. The content then remains inside of the original bounds.

Column(modifier = Modifier.padding(16.dp)) {
    Box(
        modifier = Modifier
            .clip(RectangleShape)
            .size(200.dp)
            .border(2.dp, Color.Black)
            .graphicsLayer {
                clip = true
                shape = CircleShape
                translationY = 50.dp.toPx()
            }
            .background(Color(0xFFF06292))
    ) {
        Text(
            "Hello Compose",
            style = TextStyle(color = Color.Black, fontSize = 46.sp),
            modifier = Modifier.align(Alignment.Center)
        )
    }

    Box(
        modifier = Modifier
            .size(200.dp)
            .clip(RoundedCornerShape(500.dp))
            .background(Color(0xFF4DB6AC))
    )
}

Clip applied on top of graphicsLayer transformation
Figure 10: Clip applied on top of graphicsLayer transformation

Alpha

Modifier.graphicsLayer can be used to set an alpha (opacity) for the whole layer. 1.0f is fully opaque and 0.0f is invisible.

Image(
    painter = painterResource(id = R.drawable.sunset),
    contentDescription = "clock",
    modifier = Modifier
        .graphicsLayer {
            this.alpha = 0.5f
        }
)

Image with alpha applied
Figure 11: Image with alpha applied

Compositing strategy

Working with alpha and transparency might not be as simple as changing a single alpha value. In addition to changing an alpha, there is also the option to set a CompositingStrategy on a graphicsLayer. A CompositingStrategy determines how the content of the composable is composited (put together) with the other content already drawn on screen.

The different strategies are:

Auto (default)

The compositing strategy is determined by the rest of the graphicsLayer parameters. It renders the layer into an offscreen buffer if alpha is less than 1.0f or a RenderEffect is set. Whenever the alpha is less than 1f, a compositing layer is created automatically to render the contents and then draw this offscreen buffer to the destination with the corresponding alpha. Setting a RenderEffect or overscroll always renders content into an offscreen buffer regardless of the CompositingStrategy set.

Offscreen

The contents of the composable are always rasterized to an offscreen texture or bitmap before rendering to the destination. This is useful for applying BlendMode operations to mask content, and for performance when rendering complex sets of drawing instructions.

An example of using CompositingStrategy.Offscreen is with BlendModes. Taking a look at the example below, say you want to remove parts of an Image composable by issuing a draw command that uses BlendMode.Clear. If you do not set the compositingStrategy to CompositingStrategy.Offscreen, the BlendMode interacts with all the contents below it.

Image(painter = painterResource(id = R.drawable.dog),
   contentDescription = "Dog",
   contentScale = ContentScale.Crop,
   modifier = Modifier
       .size(120.dp)
       .aspectRatio(1f)
       .background(
           Brush.linearGradient(
               listOf(
                   Color(0xFFC5E1A5),
                   Color(0xFF80DEEA)
               )
           )
       )
       .padding(8.dp)
       .graphicsLayer {
           compositingStrategy = CompositingStrategy.Offscreen
       }
       .drawWithCache {
           val path = Path()
           path.addOval(
               Rect(
                   topLeft = Offset.Zero,
                   bottomRight = Offset(size.width, size.height)
               )
           )
           onDrawWithContent {
               clipPath(path) {
                   // this draws the actual image - if you don't call drawContent, it wont
                   // render anything
                   this@onDrawWithContent.drawContent()
               }
               val dotSize = size.width / 8f
               // Clip a white border for the content
               drawCircle(
                   Color.Black,
                   radius = dotSize,
                   center = Offset(
                       x = size.width - dotSize,
                       y = size.height - dotSize
                   ),
                   blendMode = BlendMode.Clear
               )
               // draw the red circle indication
               drawCircle(
                   Color(0xFFEF5350), radius = dotSize * 0.8f,
                   center = Offset(
                       x = size.width - dotSize,
                       y = size.height - dotSize
                   )
               )
           }

       }
)

By setting the CompositingStrategy to Offscreen, it creates an offscreen texture to execute the commands to (applying the BlendMode only to the contents of this composable). It then renders it on top of what is already rendered on screen, not affecting the content already drawn.

Modifier.drawWithContent on an Image showing a circle indication, with the BlendMode.Clear inside app
Figure 12: Modifier.drawWithContent on an Image showing a circle indication, with the BlendMode.Clear and CompositingStrategy.Offscreen inside app

If you didn't use CompositingStrategy.Offscreen, the results of applying BlendMode.Clear clears all the pixels in the destination, regardless of what was already set– leaving the window’s rendering buffer (black) visible. Many of the BlendModes that involve alpha won't work as expected without an offscreen buffer. Note the black ring around the red circle indicator:

Modifier.drawWithContent on an Image showing a circle indication, with the BlendMode.Clear and no CompositingStrategy set
Figure 13: Modifier.drawWithContent on an Image showing a circle indication, with the BlendMode.Clear and no CompositingStrategy set

To understand this a bit further: if the app had a translucent window background, and you did not use the CompositingStrategy.Offscreen, the BlendMode would interact with the whole app. It would clear all of the pixels to show the app or wallpaper underneath, as in this example:

No CompositingStrategy set and using BlendMode.Clear with an app that has a translucent window background. The pink wallpaper is shown through the area around the red status circle.
Figure 14: No CompositingStrategy set and using BlendMode.Clear with an app that has a translucent window background. Notice how the pink wallpaper is shown through the area around the red status circle.

It's worth noting that when using CompositingStrategy.Offscreen, an offscreen texture that is the size of the drawing area is created and rendered back on screen. Any drawing commands that are done with this strategy, are by default be clipped to this region. The below code snippet illustrates the differences when switching to using offscreen textures:

@Composable
fun CompositingStrategyExamples() {
   Column(
       modifier = Modifier
           .fillMaxSize()
           .wrapContentSize(Alignment.Center)
   ) {
       /** Does not clip content even with a graphics layer usage here. By default, graphicsLayer
       does not allocate + rasterize content into a separate layer but instead is used
       for isolation. That is draw invalidations made outside of this graphicsLayer will not
       re-record the drawing instructions in this composable as they have not changed **/
       Canvas(
           modifier = Modifier
               .graphicsLayer()
               .size(100.dp) // Note size of 100 dp here
               .border(2.dp, color = Color.Blue)
       ) {
           // ... and drawing a size of 200 dp here outside the bounds
           drawRect(color = Color.Magenta, size = Size(200.dp.toPx(), 200.dp.toPx()))
       }

       Spacer(modifier = Modifier.size(300.dp))

       /** Clips content as alpha usage here creates an offscreen buffer to rasterize content
       into first then draws to the original destination **/
       Canvas(
           modifier = Modifier
               // force to an offscreen buffer
               .graphicsLayer(compositingStrategy = CompositingStrategy.Offscreen)
               .size(100.dp) // Note size of 100 dp here
               .border(2.dp, color = Color.Blue)
       ) {
           /** ... and drawing a size of 200 dp. However, because of the CompositingStrategy.Offscreen usage above, the
           content gets clipped **/
           drawRect(color = Color.Red, size = Size(200.dp.toPx(), 200.dp.toPx()))
       }
   }
}

CompositingStrategy.Auto vs CompositingStrategy.Offscreen - offscreen clips to the region, where auto doesn’t
Figure 15: CompositingStrategy.Auto vs CompositingStrategy.Offscreen - offscreen clips to the region, where auto doesn’t
ModulateAlpha

This composition strategy modulates the alpha for each of the drawing instructions recorded within the graphicsLayer. It won't create an offscreen buffer for alpha below 1.0f unless a RenderEffect is set, so it can be more efficient for alpha rendering. However, it can provide different results for overlapping content. For use cases where it is known in advance that content is not overlapping, this can provide better performance than CompositingStrategy.Auto with alpha values less than 1.

Another example of different composition strategies is below - applying different alphas to different parts of the composables, and applying a Modulate strategy:

@Preview
@Composable
fun CompositingStratgey_ModulateAlpha() {
  Column(
      modifier = Modifier
          .fillMaxSize()
          .padding(32.dp)
  ) {
      // Base drawing, no alpha applied
      Canvas(
          modifier = Modifier.size(200.dp)
      ) {
          drawSquares()
      }

      Spacer(modifier = Modifier.size(36.dp))

      // Alpha 0.5f applied to whole composable
      Canvas(modifier = Modifier
          .size(200.dp)
          .graphicsLayer {
              alpha = 0.5f
          }) {
          drawSquares()
      }
      Spacer(modifier = Modifier.size(36.dp))

      // 0.75f alpha applied to each draw call when using ModulateAlpha
      Canvas(modifier = Modifier
          .size(200.dp)
          .graphicsLayer {
              compositingStrategy = CompositingStrategy.ModulateAlpha
              alpha = 0.75f
          }) {
          drawSquares()
      }
  }
}

private fun DrawScope.drawSquares() {

  val size = Size(100.dp.toPx(), 100.dp.toPx())
  drawRect(color = Red, size = size)
  drawRect(
      color = Purple, size = size,
      topLeft = Offset(size.width / 4f, size.height / 4f)
  )
  drawRect(
      color = Yellow, size = size,
      topLeft = Offset(size.width / 4f * 2f, size.height / 4f * 2f)
  )
}

val Purple = Color(0xFF7E57C2)
val Yellow = Color(0xFFFFCA28)
val Red = Color(0xFFEF5350)

ModulateAlpha applies the alpha set to each individual draw command
Figure 16: ModulateAlpha applies the alpha set to each individual draw command

Write contents of a composable to a bitmap

A common use case is to create a Bitmap from a composable. To copy the contents of your composable to a Picture, use the drawIntoCanvas method:

Column(
    modifier = Modifier
        .padding(padding)
        .fillMaxSize()
        .drawWithCache {
            // Example that shows how to redirect rendering to an Android Picture and then
            // draw the picture into the original destination
            val width = this.size.width.toInt()
            val height = this.size.height.toInt()

            onDrawWithContent {
                val pictureCanvas =
                    androidx.compose.ui.graphics.Canvas(
                        picture.beginRecording(
                            width,
                            height
                        )
                    )
                // requires at least 1.6.0-alpha01+
                draw(this, this.layoutDirection, pictureCanvas, this.size) {
                    this@onDrawWithContent.drawContent()
                }
                picture.endRecording()

                drawIntoCanvas { canvas -> canvas.nativeCanvas.drawPicture(picture) }
            }
        }

) {
    ScreenContentToCapture()
}

The snippet above draws the contents of the composable into the Picture object. To then convert the Picture into a Bitmap:

val bitmap = if (Build.VERSION.SDK_INT >= Build.VERSION_CODES.P) {
    Bitmap.createBitmap(picture)
} else {
    val bitmap = Bitmap.createBitmap(
        picture.width,
        picture.height,
        Bitmap.Config.ARGB_8888
    )
    val canvas = android.graphics.Canvas(bitmap)
    canvas.drawColor(android.graphics.Color.WHITE)
    canvas.drawPicture(picture)
    bitmap
}

The bitmap can then be saved to disk and shared. For more details, see the full snippet.

Custom drawing modifier

To create your own custom modifier, implement the DrawModifier interface. This gives you access to a ContentDrawScope, which is the same as what is exposed when using Modifier.drawWithContent(). You can then extract common drawing operations to custom drawing modifiers to clean up the code and provide convenient wrappers; for example, Modifier.background() is a convenient DrawModifier.

For example, if you wanted to implement a Modifier that vertically flips content, you can create one as follows:

class FlippedModifier : DrawModifier {
    override fun ContentDrawScope.draw() {
        scale(1f, -1f) {
            this@draw.drawContent()
        }
    }
}

fun Modifier.flipped() = this.then(FlippedModifier())

Then use this flipped modifier applied on Text:

Text(
    "Hello Compose!",
    modifier = Modifier
        .flipped()
)

Custom Flipped Modifier on Text
Figure 17: Custom Flipped Modifier on Text

Additional resources

For more examples using graphicsLayer and custom drawing, check out the following resources: