Feature Layers are client side layers that access data and draw the features as vector graphics in the browser. Feature Layers can be used to create interactive visualizations. These visualizations are enhanced by some new renderers and new capabilities to existing renderers. The rotationInfo property of a renderer defines how marker symbols are rotated. You can use rotation to depict wind direction, vehicle heading, etc. Only applied to marker symbols (based on point data). The property type defines how the angle of rotation was measured. There are two ways you can define the angle: Note that this is different from a MarkerSymbol’s angle property, which defines rotation that is applied before the symbol is offset (using xoffset and yoffset). rotationInfo defines rotation that is applied after the symbol is offset (with the geographic point as the origin of rotation). View Samples: Renderer with proportional symbols and rotation | Class breaks with rotation and proportional symbols Based on data that represents non-distance quantity. For example: the traffic count of highways, the population of states, etc. Depending on how you want to visualize your data, there are four approach you may use to scale the symbols: The DotDensityRenderer visualizes a quantity in polygon with numerous randomly placed dots. The relative sparseness of the dots indicates the distribution of a discrete spatial phenomenon, such as population distribution map, crop production map, etc. The ScaleDependentRenderer defines multiple renderers for a layer, where each renderer is only applied to the layer at a specific scale range or zoom range. While other renderer types allow you to define graphics styling based solely on their attribute values, a scale dependent renderer allows you to define styling based on map scale as well. For each renderInfo object in rendererInfos, minScale and maxScale can be replaced with minZoom and maxZoom respectively, if your map has defined LODs (Levels of Detail). View Samples: Scale dependent renderer | Dot density renderer | Dot density with multiple classes | Renderer with graduated symbols for polygons The heatmaprenderer renders point data into a raster visualization that emphasizes areas of higher density or weighted values. The beta version of this was introduced at version 3.11 and officially added at 3.12. This renderer uses a Gaussian Blur technique to average the influence of each point out over the area determined by the blurRadius. A Gaussian blur uses a Gaussian, or Normal, distribution (also called a Bell-curve) to spread value out in vertical and horizontal directions. This averaging function is applied horizontally and vertically to produce a blurred area of influence instead of a single specific point. This process is repeated for each point. Each time the calculation is performed, the blurred values are added to the underlying pixels. The results are cumulative, so a pixel that has several points near it will have a higher value than a pixel with a single point in it. This accumulated value is called the intensity of the pixel. If needed, the intensity calculation can be further adjusted. Source.