Capturing and Displaying Camera Pictures

Cameras connected to SmartThings can use the imageCapture Capability, along with the Carousel Tile, to capture and view images. SmartThings-connected cameras are either LAN- or Cloud-Connected; this document outlines the steps to capture and display images for both.

Image Capture Capability

Add support for the imageCapture Capability by including it in the Device Handler’s metadata:

metadata {
    definition(name: "My Camera Device", namespace: "MyNamespace", author: "My Name") {
        capability "Image Capture"
        // other definition metadata...

The Image Capture Capability defines one attribute, “image”, and one command, take(). The Carousel Tile can be used to display images and allow the user to manually take a photo, as discussed next.

Tiles for taking and viewing pictures

Add tiles to allow the viewing and taking of images:

tiles {
    standardTile("image", "device.image", width: 1, height: 1, canChangeIcon: false, inactiveLabel: true, canChangeBackground: true) {
        state "default", label: "", action: "", icon: "", backgroundColor: "#FFFFFF"

    carouselTile("cameraDetails", "device.image", width: 3, height: 2) { }

    standardTile("take", "device.image", width: 1, height: 1, canChangeIcon: false, inactiveLabel: true, canChangeBackground: false) {
        state "take", label: "Take", action: "Image Capture.take", icon: "", backgroundColor: "#FFFFFF", nextState:"taking"
        state "taking", label:'Taking', action: "", icon: "", backgroundColor: "#00A0DC"
        state "image", label: "Take", action: "Image Capture.take", icon: "", backgroundColor: "#FFFFFF", nextState:"taking"

    main "image"
    details(["cameraDetails", "take"])

The carouselTile is where the images will be displayed, and the “take” tile allows users to capture images. Note that both are associated with the "image" attribute; this association allows the images to be taken and displayed properly.

Capture and display images

The take() command of the Image Capture Capability is responsible for capturing the image. Follow the protocol-specific instructions for implementing this command method below.

LAN-connected cameras

LAN-connected devices can capture images using HubAction, store them using storeTemporaryImage(), and display them with the Carousel Tile.

The take() command will issue a request to take a picture via a HubAction. The response from the device will be sent to the Device Handler’s parse() method, where it can then be moved to longer-lasting storage using storeTemporaryImage(). storeTemporaryImage() also emits the “image” event, causing the Carousel Tile to be updated with the new image.

Here’s an example of the take() command (details of the request will be specific to each device):

def take() {
    def host = getHostAddress()
    def port = host.split(":")[1]

    def path = "/some/path/"

    def hubAction = new physicalgraph.device.HubAction(
        method: "GET",
        path: path,
        headers: [HOST:host]

    hubAction.options = [outputMsgToS3:true]

    return hubAction

* Utility method to get the host addresses
private getHostAddress() {
    def parts = device.deviceNetworkId.split(":")
    def ip = convertHexToIP(parts[0])
    def port = convertHexToInt(parts[1])
    return ip + ":" + port

Some things to note about the implementation of the take() command:

  1. The specific path, method, and headers of the HubAction will vary for each device. Consult the device manufacturer’s documentation for this information.
  2. Make sure to specify hubAction.options = [outputMsgToS3: true]. This will result in the image being stored (temporarily). We will move the image to longer-lasting storage next.
  3. Remember to return the HubAction from the command method, otherwise it will not be executed!

Once we’ve made the request in the take() command method, the response from the device will be sent to the Device Handler’s parse() method. This response will contain a tempImageKey, which is the key of the photo just taken.

def parse(String description) {

    def map = stringToMap(description)

    if (map.tempImageKey) {
        try {
            storeTemporaryImage(map.tempImageKey, getPictureName())
        } catch (Exception e) {
            log.error e
    } else if (map.error) {
        log.error "Error: ${map.error}"

    // parse other messages too

private getPictureName() {
    return java.util.UUID.randomUUID().toString().replaceAll('-', '')

parse() does the following:

  1. Checks the response to see if tempImageKey was sent. If it was, this means that this is the image response from our take() command.
  2. Calls storeTemporaryImage() with the tempImageKey and a name for the picture. The name must be unique per device instance, contain only alphanumeric, “-”, “_”, and ”.” characters. This will move the image from temporary storage to a location where the image will be stored for 365 days, before being permanently deleted.

storeTemporaryImage() also emits the “image” event, which is the attribute our Carousel Tile is associated with. This is what allows the image to be displayed in the tile.

Cloud-connected cameras

The take() command will issue an HTTP request to the third-party service to capture the image, and store the resulting image bytes using storeImage().

Below is a simplified example (A real application will need to handle authentication with the third-party, as well as additional error handling):

def take() {
    def params = [
        uri: "https://some-uri",
        path: "/some/path"

    try {
        httpGet(params) { response ->
            // we expect a content type of "image/jpeg" from the third party in this case
            if (response.status == 200 && response.headers.'Content-Type'.contains("image/jpeg")) {
                def imageBytes =
                if (imageBytes) {
                    def name = getImageName()
                    try {
                        storeImage(name, imageBytes)
                    } catch (e) {
                        log.error "Error storing image ${name}: ${e}"

            } else {
                log.error "Image response not successful or not a jpeg response"
    } catch (err) {
        log.debug "Error making request: $err"


def getImageName() {
    return java.util.UUID.randomUUID().toString().replaceAll('-','')


Only synchronous HTTP requests are supported when using the Carousel Tile.

The take() command above does the following:

  1. Makes a request to a URI that will return an image response. A real integration would need to provide authorization information on the request. This would typically be an OAuth token obtained through the setup process, as documented here.
  2. If the response is successful and its Content-Type is our expected content, it gets the image bytes from
  3. Stores the image using storeImage(), using a name generated from a UUID. The name of the image is required to be unique for each device instance.

storeImage() will emit the “image” event, which causes the Carousel Tile to be updated with the new image.


httpGet() will serialize the response data for images into a ByteArrayInputStream, which is why we can pass the response body to storeImage().

Retrieving an image

If you need to retrieve the byte representation of an image stored with storeImage() or storeTemporaryImage(), use getImage(). This will return the bytes of the image in a ByteArrayInputStream.

// Image with "some-name" that was previously stored
ByteArrayInputStream img = getImage("some-name")

Image size limits

Images are limited to a maximum of one megabyte.

storeImage() will throw an InvalidParameterException if this limit is exceeded.

Attempting to capture an image exceeding this maximum using HubAction will result in the message sent to parse() containing an error response:

def parse(String description) {
    def map = stringToMap(description)

    if (map.error) {
        log.error "error: ${map.error}"
    } else if (map.tempImageKey) {

Allowed image name characters

Image names are restricted to alphanumeric, “-”, “_”, and ”.” characters.

An InvalidParameterException is thrown by storeTemporaryImage() and storeImage() if the name contains other characters.

Image storage duration

Images stored via a HubAction are stored for 24 hours, after which it is deleted (this is why we use storeTemporaryImage() to move images captured by a HubAction).

Images stored via storeImage() or storeTemporaryImage() are available to clients for seven days, and stored by SmartThings for 365 days, after which it is deleted.

Supported image formats

storeImage() supports both JPEG and PNG image formats. The content type can be specified when calling storeImage():

storeImage("some-image-name", imgBytes, "image/png")

The content type of "image/jpeg" is the default.

Images captured via a HubAction and stored with storeTemporaryImage() must be in JPEG format.

In either case, there is no need to include the file extension (e.g., ".jpg" or ".png" in the image name).