Leaflet.DistortableImage

A Leaflet extension to distort images -- "rubbersheeting" -- for the MapKnitter.org (src) image georectification service by Public Lab. Leaflet.DistortableImage allows for perspectival distortions of images, client-side, using CSS3 transformations in the DOM.

Begin running (and contributing to) this codebase immediately with GitPod:

Advantages include:

It can handle over 100 images smoothly, even on a smartphone
Images can be right-clicked and downloaded individually in their original state
CSS3 transforms are GPU-accelerated in most (all?) browsers, for a very smooth UI
No need to server-side generate raster GeoTiffs, tilesets, etc. in order to view distorted imagery layers
Images use DOM event handling for real-time distortion
Full resolution download option for large images, using WebGL acceleration

Download as zip or clone the repo to get a local copy.

Also available on NPM as leaflet-distortableimage:

npm i leaflet-distortableimage

Compatibility with Leaflet versions

Compatible with Leaflet 1.0.0 and greater

MapKnitter Lite

Read more about MapKnitter going offline here: https://publiclab.org/mapknitter

Check out a prototype of the Mapknitter Lite project which enables you to:

create maps and stitch images
download maps to save them
store saved maps and image collections in the Internet Archive (archive.org)
host maps at archive.org for others to see

Demo

Check out this simple demo.

And watch this GIF demo:

To test the code, open index.html in your browser and click and drag the markers on the edges of the image. The image will show perspectival distortions.

For the additional features in the multiple image interface, open select.html and use shift + click on an image or shift + drag on the map to "multi-select" (collect) images. For touch screens, touch + hold the image.

Single Image Interface

The simplest implementation is to create a map with our recommended TileLayer, then create an L.distortableImageOverlay instance and add it onto the map.

// set the initial map center and zoom level
map = L.map('map').setView([51.505, -0.09], 13);

// adds a Google Satellite layer with a toner label overlay
map.addGoogleMutant();

map.whenReady(function() {
  // By default, 'img' will be placed centered on the map view specified above
  img = L.distortableImageOverlay('example.jpg').addTo(map);
});

Note: map.addGoogleMutant() is a convenience function for adding our recommended layer to the map. If you want a different baselayer, skip this line and add your preferred setup instead.

Options available to pass during L.DistortableImageOverlay initialization:

actions
corners
editable
fullResolutionSrc
mode
rotation
selected
suppressToolbar
tooltipText

Actions

actions (optional, default: [L.DragAction, L.ScaleAction, L.DistortAction, L.RotateAction, L.FreeRotateAction, L.LockAction, L.OpacityAction, L.BorderAction, L.ExportAction, L.DeleteAction], value: array)

If you would like to overrwrite the default toolbar actions available for an individual image's L.Popup toolbar, pass an array with the actions you want. Reference the available values here.

For example, to overrwrite the toolbar to only include L.OpacityAction and L.DeleteAction , and also add on an additional non-default like L.RestoreAction:

img = L.distortableImageOverlay('example.jpg', {
  actions: [L.OpacityAction, L.DeleteAction, L.RestoreAction],
}).addTo(map);

Corners

corners (optional, default: an array of LatLangs that position the image on the center of the map, value: array)

Allows you to set an image's position on the map manually (somewhere other than the center default).

Note that this can manipulate the shape and dimensions of your image.

The corners should be passed as an array of L.latLng objects in NW, NE, SW, SE order (in a "Z" shape).

They will be stored on the image. See the Quick API Reference for their getter and setter methods.

Example:

img = L.distortableImageOverlay('example.jpg', {
  corners: [
    L.latLng(51.52,-0.14),
    L.latLng(51.52,-0.10),
    L.latLng(51.50,-0.14),
    L.latLng(51.50,-0.10),
  ],
}).addTo(map);

// you can grab the initial corner positions
JSON.stringify(img.getCorners())
=> "[{"lat":51.52,"lng":-0.14},{"lat":51.52,"lng":-0.1},{"lat":51.5,"lng":-0.14},{"lat":51.5,"lng":-0.1}]"

// ...move the image around...

// you can check the new corner positions.
JSON.stringify(img.getCorners())
=> "[{"lat":51.50685099607552,"lng":-0.06058305501937867},{"lat":51.50685099607552,"lng":-0.02058595418930054},{"lat":51.486652692081925,"lng":-0.06058305501937867},{"lat":51.486652692081925,"lng":-0.02058595418930054}]"

// note there is an added level of precision after dragging the image

Editable

editable (optional, default: true, value: boolean)

Internally, we use the image load event to trigger a call to img.editing.enable(), which sets up the editing interface (makes the image interactive, adds markers and toolbar).

If you want to enable editing based on custom logic instead, you can pass editable: false and then write your own function with a call to img.editing.enable(). Other passed options such as selected: true and mode will still be applicable and applied then.

Note: when using the multiple image interface (L.DistortableCollection) this option will be ignored on individual L.DistortableImageOverlay instances and should instead be passed to the collection instance.

Full-resolution download

fullResolutionSrc (optional)

We've added a GPU-accelerated means to generate a full resolution version of the distorted image.

When instantiating a Distortable Image, pass in a fullResolutionSrc option set to the url of the higher resolution image. This image will be used in full-res exporting.

img = L.distortableImageOverlay('example.jpg', {
  fullResolutionSrc: 'large.jpg',
}).addTo(map);

Our project includes two additional dependencies to enable this feature, glfx.js and webgl-distort, both of which you can find in our package.json.

Mode

mode (optional, default: "distort", value: string)

This option sets the image's initial editing mode, meaning the corresponding editing handles will always appear first when you interact with the image.

Values available to pass to mode are:

distort (default): Distortion via individually draggable corners.
drag: Translation via individually draggable corners.
rotate: Rotation only.
scale: Resize only.
freeRotate: Combines the rotate and scale modes into one.
lock: Locks the image in place. Disables any user gestures, toolbar actions, or hotkeys that are not associated with mode. Exception: L.ExportAction will still be enabled.

In the below example, the image will be initialized with "freeRotate" handles:

img = L.distortableImageOverlay('example.jpg', {
  mode: 'freeRotate',
}).addTo(map);

If you select a mode that is removed or unavailable, your image will just be assigned the first available mode on initialization.

Limiting modes:

Each mode is just a special type of action, so to ensure that these are always in sync the modes available on an image instance can be limited by the actions available on it. To remove a mode, limit its corresponding action via the actions option during initialization. This holds true even when suppressToolbar: true is passed.

In the below example, the image will be initialiazed with 'freeRotate' handles, and limit its available modes to 'freeRotate' and 'scale'.

We also remember to add the normal toolbar actions we will want:

img = L.distortableImageOverlay('example.jpg', {
  mode: 'freeRotate',
  actions: [L.FreeRotateAction, L.ScaleAction, L.BorderAction, L.OpacityAction],
}).addTo(map);

Likewise, it is possible to remove or add actions during runtime (addTool, removeTool), and if those actions are modes it will remove / add the mode.

Rotation

rotation (optional, default: {deg: 0, rad: 0}, value: hash)

Set the initial rotation angle of your image, in degrees or radians. Set the unit as the key, and the angle as the value.

img = L.distortableImageOverlay('example.jpg', {
  rotation: {
    deg: 180,
  },
}).addTo(map);

Selected

selected (optional, default: false, value: boolean)

By default, your image will initially appear on the screen as unselected (no toolbar or markers). Interacting with it will make them visible.

If you prefer that an image initially appears as selected instead, pass selected: true.

Note: when working with the multi-image interface, only the last overlay you pass selected: true to will appear with editing handles and a toolbar.

Suppress Toolbar

suppressToolbar (optional, default: false, value: boolean)

To initialize an image without its L.Popup instance toolbar, pass it suppressToolbar: true.

Typically, editing actions are triggered through our toolbar interface. If disabling the toolbar, the developer will need to implement their own toolbar UI connected to our actions (WIP API for doing this)

Multiple Image Interface

Our DistortableCollection class builds on the single image interface to allow working with multiple images simultaneously.

The setup is relatively similar.

Although not required, you will probably want to pass corners to individual images when adding multiple or they will be positioned on top of eachother.

Here is an example with two images:

// 1. Instantiate map
// 2. Instantiate images but this time *dont* add them directly to the map
img = L.distortableImageOverlay('example.jpg', {
  corners: [
    L.latLng(51.52, -0.14),
    L.latLng(51.52,-0.10),
    L.latLng(51.50, -0.14),
    L.latLng(51.50,-0.10),
  ],
});

img2 = L.distortableImageOverlay('example.jpg', {
  corners: [
    L.latLng(51.51, -0.20),
    L.latLng(51.51,-0.16),
    L.latLng(51.49, -0.21),
    L.latLng(51.49,-0.17),
  ],
});

// 3. Instantiate an empty `DistortableCollection` group
imgGroup = L.distortableCollection().addTo(map);

// 4. Add the images to the group
imgGroup.addLayer(img);
imgGroup.addLayer(img2);

Note: You must instantiate a blank collection, then dynamically add layers to it like above. This is because DistortableCollection internally uses the layeradd event to enable additional editing features on images as they are added, and it is only triggered when they are added dynamically.

Options available to pass during L.DistortableCollection initialization:

actions
editable
supressToolbar

Actions

actions (optional, default: [L.ExportAction, L.DeleteAction, L.LockAction, L.UnlockAction], value: array)

Overrwrite the default toolbar actions for an image collection's L.Control toolbar. Reference the available values here.

For example, to overrwrite the toolbar to only include the L.DeleteAction:

imgGroup = L.distortableCollection({
  actions: [L.DeleteAction],
}).addTo(map);

To add / remove a tool from the toolbar at runtime, we have also added the methods addTool(action) and removeTool(action).

Editable

editable (optional, default: true, value: boolean)

See editable.

Suppress Toolbar

suppressToolbar (optional, default: false, value: boolean)

Same usage as suppressToolbar, but for the collection group's L.Control toolbar instance.

This provides the developer with the flexibility to keep the popup toolbars, the control toolbar, both, or neither.

For ex.

// suppress this images personal toolbar
img = L.distortableImageOverlay('example.jpg', {
  suppressToolbar: true,
  corners: [
    L.latLng(51.52, -0.14),
    L.latLng(51.52,-0.10),
    L.latLng(51.50, -0.14),
    L.latLng(51.50,-0.10),
  ],
});

// suppress the other images personal toolbar
img2 = L.distortableImageOverlay('example.jpg', {
  suppressToolbar: true,
});

// suppress collection toolbar accessed during multi-image selection
imgGroup = L.distortableCollection({
  suppressToolbar: true,
}).addTo(map);

Tooltip Text

tooltipText (optional, default: '', value: string) This provides the flexibility to add tooltip text to every image placed on the tile layer.

For ex.

// Sets up tooltip text for an image, the text is displayed when mouse is placed on it
img = L.distortableImageOverlay(
       'example.jpg', 
       {tooltipText: 'Sample text'}
);

UI and functionalities

Currently it supports multiple image selection and translations, and WIP we are working on porting all editing tools to work for it, such as opacity, etc. Image distortions (via modes) still use the single-image interface.

A single toolbar instance (using L.control) renders the set of tools available to use on collections of images.

collect:

Collect an indvidiual image with shift + click.
Or for touch devices, touch + hold (aka longpress).
Collect multiple images at once with shift + drag (Uses our L.Map.BoxCollector).

decollect:

In order to return to the single-image interface, where each L.popup toolbar only applies actions on the image it's attached to, you must toggle all images out of collection with shift + click / touch + hold, or...
...Click on the map or hit the esc key to quickly decollect all.

Toolbar Actions (& Keybindings)

Single Image Interface

Default tools

L.BorderAction (b)
- Toggles a thin border around the overlay.
L.DeleteAction (backscpace, delete)
- Permanently deletes the image from the map. Uses a confirm() modal dialog.
- windows backspace / mac delete
L.DistortAction (d)
- Sets distort mode.
L.DragAction
- Sets drag mode.
L.ExportAction (e)
L.FreeRotateAction (f)
- Sets freeRotate mode.
L.LockAction (l, u)
- Toggles between lock mode and the initially set default mode (distort by default).
L.OpacityAction (o)
L.RotateAction (r):
- Sets rotate mode.
L.ScaleAction (s):
- Sets scale mode.

Add-on tools

These may be added using addTool(), like this:

distortableImageLayer.editing.addTool(L.StackAction);

L.RestoreAction
- Restores the image to its natural dimensions, scale, rotation, and location on the map.
L.StackAction (q, a)
- Switch an image's overlap compared to neighboring images back and forth into view. Employs bringToFront() and bringToBack() from the Leaflet API.
L.GeolocateAction (WIP)

Multiple Image Interface

Defaults:

L.ExportAction (e)
L.DeleteAction (backscpace, delete)
- Permanently deletes a collection of images from the map.
L.LockAction (l)
- Sets lock mode for a collection of images.
L.UnlockAction (u)
- Unsets lock mode for a collection of images.

Quick API Reference

L.Map

We have extended Leaflet's L.Map to include a convenience method for this library:

addGoogleMutant(opts? <Mutant options>): this

Adds a Google Mutant layer with location labels according to our recommended setup.
Mutant options: {[mutantOpacity][, maxZoom][, minZoom][, labels][, labelOpacity][, doubleClickLabels]}
- mutantOpacity (default 0.8, value: number 0..1)
- maxZoom (default: 18, value: number 0..21)
- minZoom (default: 0, value: number 0..maxZoom)
- labels (default: true, value: boolean)
- labelOpacity (default: 1, value: number 0, 1)
- doubleClickLabels (default: true, value: boolean)
Mutant options are saved on the map and accessible during runtime as map.mutantOptions.

And the following custom handlers:

doubleClickLabels: this

Allows toggling label visibility on map dblclick.
Enabled by default on #addGoogleMutant unless the options labels: false or doubleClickLabels: false are passed to it.

If labels: false passed, removed from map altogether.
If there are labels present but doubleClickLabels: false was passed, just disabled and can always be enabled during runtime via Leaflet's Handler API.

Disables the map's default doubleClickZoom handler when enabled.

boxCollector: this

Overrides the map's default boxZoom handler. To use boxZoom instead, pass the options { boxCollector: false, boxZoom: true } to the map on initialization.
Allows multiple images to be collected when shift + draging on the map for the multiple image interface.

We have slightly changed a default Leaflet handler:

doubleClickZoom: this

This handler may not be enabled (and will return false) while the doubleClickLabels handler is enabled.
This handler and doubleClickLabels time and fire a custom singleclick event on map click.

Our "doubleClick" handlers mentioned above use a custom singleclick event to run logic on map dblclick while allowing the images on the map to remain selected. You can read more about the implications of this and how to disable it on our wiki "singleclick event".

L.DistortableImageOverlay

An individual image instance that can have transformation methods called on it and can be "selected".

getCorner(idx <number 0..3>): LatLng

Returns the coordinates of the image corner at index.

getCorners(): 4 [LatLng, LatLng, LatLng, LatLng]

Returns the coordinates of the image corners in NW, NE, SW, SE order.

setCorner(idx <number 0..3>, LatLng): this

Updates the coordinates of the image corner at index to LatLng and, where applicable, marker and toolbar positioning.
We use this internally for distort mode.

setCorners(LatLngCorners): this

Same as #setCorner, but takes in a "corners" object made up of LatLngs to update all 4 corners with only one UI update at the end.
We use this internally for image translation, rotation, and scaling.

LatLngCorners: { keys: <number 0..4>, values: LatLng }
ex.

var scaledCorners = {};
var i;
var p;

for (i = 0; i < 4; i++) {
  p = map
    .project(img.getCorner(i))
    .subtract(center)
    .multiplyBy(scale)
    .add(center);
  scaledCorners[i] = map.unproject(p);
}

img.setCorners(scaledCorners);

setCornersFromPoints(PointCorners): this

Same as #setCorners, but takes in a "corners" object made up of Points instead of LatLngs.
PointCorners: { keys: <number 0..4>, values: Point }

getCenter(): LatLng

Returns the center (centroid) of the image.

getAngle([unit = 'deg'] <string>): Number

Returns the image's rotation angle in units, or in degrees by default.
Number will always be >= 0.

unit (optional, default: 'deg', value: string 'deg'|'rad')
ex.

img.getAngle();
img.getAngle('deg');
img.getAngle('rad');

setAngle(angle <number>, [unit = 'deg'] <string>): this

Sets the image's rotation to angle in units, or in degrees by default.

unit (optional, default: 'deg', value: string 'deg'|'rad')
ex.

img.setAngle(180);
img.setAngle(180, 'deg');
img.setAngle(Math.PI, 'rad');

rotateBy(angle <number>, [unit = 'deg'] <string>): this

Rotates the image relative to its current angle by angle in units, or in degrees by default.

unit (optional, default: 'deg', value: string 'deg'|'rad')
ex.

img.rotateBy(180);
img.rotateBy(180, 'deg');
img.rotateBy(Math.PI, 'rad');

scaleBy(factor <number>): this

Scales the image by the given factor and calls #setCorners.
A scale of 0 or 1 will leave the image unchanged - but 0 causes the function to automatically return.
A negative scale will invert the image and, depending on the factor, change its size.
Ex. img.scaleBy(0.5)

restore(): this

Restores the image to its natural dimensions, scale, rotation, and location on the map.

isSelected(): Boolean

Returns true if the individual image instance is selected.

select(): this

Selects an individual image instance.
If its editing handler is disabled or the multiple image interface is on (imgGroup.anyCollected() === true), does not select and instead just returns undefined.
Internally invoked on image click.

deselect(): this

Deselects an individual image instance.
If its editing handler is disabled, does not deselect and instead just returns undefined.
Internally invoked on map click and image collect (shift + click).

L.DistortableImageOverlay.Edit

A handler that holds the keybindings and toolbar API for an image instance. It is always initialized with an instance of L.DistortableImageOverlay. Besides code organization, it provides the ability to enable and disable image editing using the Leaflet API.

Note: The main difference between the enable / disable runtime API and using the editable option during initialization is in runtime, neither individual image instaces nor the collection group get precedence over the other.

enable(): this

Sets up the editing interface (makes the image interactive).
Called internally by default (editable), but unlike the option it can be used in runtime and is not ignored if there is a collection group. In fact...
...An individual image can be enabled while the group is disabled. i.e. calling img.editing.enable() after imgGroup.editing.disable() is valid. In this case, the single image interface will be available on this image but not the multi-image interface.

disable(): this

Deselects the image, and disables its editing interface (makes it non-interactive).
Called internally by default on image deletion.
An individual image can be disabled while the group is enabled.

enabled(): Boolean

Returns true if editing on the individual image instance is enabled.
img.editing.enabled()

hasMode(mode <string>): Boolean

Returns true if the image has the passed mode.

getMode(): String

Returns the current mode of the image.

getModes(): Hash

Returns all the modes available on the image.

nextMode(): this

Sets the mode of the image to the next one in the modes array by passing it to #setMode.
If the image's editing interface is not enabled or modes only has 1 mode, it will instead return undefined and not update the image's mode.
We use this internally to iterate through an image's editing modes easily on dblclick, but you can call it programmatically if you find a need. Note that dblclick also selects the image (given it's not disabled and the collection interface is not on).

setMode(mode <string>): this

Sets the mode of the image to the passed one given that it is in the modesarray, it is not already the current mode, and the image editing interface is enabled. Otherwise, does not set the mode and instead just returns undefined.

L.DistortableCollection

A collection instance made up of a group of images. Images can be "collected" in this interface and a "collected" image is never also "selected".

isCollected(img <DistortableImageOverlay>): Boolean

Returns true if the passed L.DistortableImageOverlay instance is collected, i.e. its underlying HTMLImageElement has a class containing "selected".

anyCollected(): Boolean

Returns true if any L.DistortableImageOverlay instances are collected.

Retrieve image from Json file containing image property set. The property set can be used to instantiate new imageOverlays.

recreateImagesFromJsonUrl(string): {avg_cm_per_pixel , imgCollectionProps<{}>}

Returns imageCollectionIbject if successful or empty object if unsuccessful

Example
// 1. Instantiate an empty distortableCollection
imgGroup = L.distortableCollection().addTo(map);

// 2. Get property set for each of the images
const imageCollectionObj = await map.imgGroup.recreateImagesFromJsonUrl(jsonDownloadURL);

Note: jsonDownloadUrl must be in either of these formats:
i. https://archive.org/download/mkl-2-2/mkl-2-2.json (for json files generated from Mapknitter-Lite)
- "mkl-2-2" is the identifier provided by Internet Archive after a file is uploaded to the service (i.e., archive.org)
- "mkl-2-2.json" name of the Json file
ii. https://archive.org/download/mapknitter/--10.json (for json files from legacy mapknitter.org)
- "mapknitter" is the path for all the legacy Json files and must be present in the URL
- "--10.json" is th name of the Json file

// 3. Iterate through each of the property sets, extract the imageURL, tooltipText and corners imageCollectionObj then place each of them on the tile map using:
image = L.distortableImageOverlay(imageURL,{tooltipText, corners});
map.imgGroup.addLayer(image);

L.DistortableCollection.Edit

Same as L.DistortableImage.Edit but for the collection (L.DistortableCollection) instance.

enable(): this

Sets up the multi-editing interface.
Called internally by default, see editable.
Calls each individual image's #enable method and then enables the multi-image interface.

disable(): this

Removes the editing interface (makes the image non-interactive, removes markers and toolbar).
Called internally by default on image group deletion, but can also be used for custom behavior.
Calls each individual image's #disable method and disables the multi-image interface.

enabled(): Boolean

Returns true if editing on the collection instance is enabled.
imgGroup.editing.enabled()

removeTool(action <EditAction>): this

Removes the passed tool from the control toolbar in runtime if the tool is present.
ex: imgGroup.removeTool(Deletes)

addTool(action <EditAction>): this

Adds the passed tool to the end of the control toolbar in runtime.

replaceTool(old <EditAction>), next <EditAction>)

Replaces the first parameter with the second parameter. Returns the parent object.

hasTool(action <EditAction>): Boolean

Returns true if the tool is present in the currently rendered control toolbar.

Additional Components

Keymapper

// add a position option with combinations of 'top', 'bottom', 'left' or 'right'
L.distortableImage.keymapper(map, {
  position: 'topleft',
});

Options:

position (optional, default: 'topright', value: string)

Adds a control onto the map which opens a keymapper legend showing the available key bindings for different editing / interaction options.

(WIP) Currently includes keybindings for all available actions and does not update yet if you use the actions API to limit available actions.

Custom Translations

You can translate the LDI toolbar buttons in your native language by providing a custom translation object to DistortableImageOverlay or DistortableCollection.

NOTE: If you don't specify a custom translation for a certain field, it will fallback to English.

These are the defaults:

var translation = {
  deleteImage: 'Delete Image',
  deleteImages: 'Delete Images',
  distortImage: 'Distort Image',
  dragImage: 'Drag Image',
  exportImage: 'Export Image',
  exportImages: 'Export Images',
  removeBorder: 'Remove Border',
  addBorder: 'Add Border',
  freeRotateImage: 'Free rotate Image',
  geolocateImage: 'Geolocate Image',
  lockMode: 'Lock Mode',
  lockImages: 'Lock Images',
  makeImageOpaque: 'Make Image Opaque',
  makeImageTransparent: 'Make Image Transparent',
  restoreImage: 'Restore Natural Image',
  rotateImage: 'Rotate Image',
  scaleImage: 'Scale Image',
  stackToFront: 'Stack to Front',
  stackToBack: 'Stack to Back',
  unlockImages: 'Unlock Images',
  confirmImageDelete: 'Are you sure? This image will be permanently deleted from the map.',
  confirmImagesDeletes: 'Are you sure? These images will be permanently deleted from the map.',
};

For confirmImagesDeletes you can pass a function that returns a string. This is useful for languages where noun form depends on the number:

var translation = {
  confirmImagesDeletes: function(n) {
    var cond = n%10 >= 2 && n%10 <= 4 && n%100 - n%10 !== 10;
    var str = 'Czy na pewno chcesz usunąć ' + n;
    if(cond) str += ' obrazy?';
    else str += ' obrazów?';
    return str;
  },
  // ...
}

L.DistortableImageOverlay

img = L.distortableImageOverlay('example.jpg', {
  translation: {
    deleteImage: 'Obriši sliku',
    distortImage: 'Izobliči sliku',
    dragImage: 'Pomjeri sliku',
    // ...
  },
}).addTo(map);

L.DistortableCollection

imgGroup = L.distortableCollection({
  translation: {
    deleteImages: 'Obriši slike',
    exportImages: 'Izvezi slike',
    // ...
  },
}).addTo(map);