| # Flatland Magnification | |
| Magnification is an accessibility feature that allows clients to upscale rendered content | |
| as aid for people with visual impairments. For instance, a small region centered around | |
| the user’s cursor would be shown at a higher scale somewhere else on the display - and that | |
| area would move as the cursor itself moves. There is also the case of full screen magnification, | |
| where as the name suggests, the magnified region would be rendered over the whole display. | |
| ## Design | |
| Flatland actually has no explicit notion of "magnification" in its API. Instead, magnification | |
| can be achieved by taking advantage of the flexibility of the Flatland scenegraph, namely, the | |
| ability for a transform node to scale its subtree and for a node to have multiple parent nodes. | |
| ## Partial Magnification | |
| Flatland traverses the scene graph in a depth-first manner in order to generate the final data lists | |
| to be sent to the renderer. A node with two parents will be traversed twice, and thus will show up | |
| twice in the final data list, with each instance inheriting a different set of global data depending | |
| on which parent was traversed for that instance. Because render data is duplicated, it is important | |
| to be aware of performance. Content behind fullscreen renderables will also be culled, which helps | |
| with this performance concern. | |
| Thus for magnification, clients can set up a multiparented scene graph, where one parent contains a | |
| scale factor that magnifies all of the instances generated from Flatland traversing down its | |
| specific sub-hierarchy. This magnification node can be modified dynamically to suit the needs of the | |
| client. For instance, if the magnified region needs to follow the cursor, then the translation | |
| component of the magnification node can be updated based on the cursor position to create the effect | |
| of a moving magnified region on the screen. | |
| ## Fullscreen Magnification | |
| In the case of full-screen magnification, we can get by with a single parent node scaled | |
| up to the desired level of magnification. The layout information provided to clients for a | |
| particular view will not be affected by the scale factor applied to a parent magnification node. | |
|  | |
| ## Pseudocode | |
| Here is some example code demonstrating how to set up a scene for partial magnification. | |
| // Create transform IDs. | |
| const TransformId kIdRoot = {1}; | |
| const TransformId kIdParent1 = {2}; | |
| const TransformId kIdParent2 = {3}; | |
| const TransformId kIdChild = {4}; | |
| // Create the transforms. | |
| flatland->CreateTransform(kIdRoot); | |
| flatland->CreateTransform(kIdParent1); | |
| flatland->CreateTransform(kIdParent2); | |
| flatland->CreateTransform(kIdChild); | |
| // Scale up the magnification parent node. | |
| flatland->SetScale(kIdParent2, {20, 20}); | |
| // Setup the diamond parent hierarchy. | |
| flatland->SetRootTransform(kIdRoot); | |
| flatland->AddChild(kIdRoot, kIdParent1); | |
| flatland->AddChild(kIdRoot, kIdParent2); | |
| flatland->AddChild(kIdParent1, kIdChild); | |
| flatland->AddChild(kIdParent2, kIdChild); | |
| // Present flatland. | |
| flatland->Present(/*args*/); |