Use CSS pixel units everywhere

[lilleyse]

• Point cloud size
• Styled point cloud size
• Point cloud eye dome lighting radius
• Polyline arrow
• Polyline dash
• Polyline shadow volume
• GL line width
• Model minimum pixel size
• Star Burst demo
• Frustum getPixelDimensions

#8083 changed Cesium to render at device resolution rather than upscaling on high DPI displays. The problem now is there are some options in Cesium whose unit is pixels but it's unclear if that means "CSS pixel" or "device pixel".

Everything should be in CSS pixels so that a scene doesn't look different on different displays. We should sweep through everything that has pixel units and multiply by frameState.pixelScale like we're doing for SSE.

For example point cloud maximum attenuation is considered to be "device pixels" and because of that points are smaller than in 1.60.

1.60 Sandcastle

master Local Sandcastle

master with devicePixelRatio correction Local Sandcastle

[lilleyse]

A workaround is to set viewer.resolutionScale = 1.0 / window.devicePixelRatio.

This will render the Cesium canvas in CSS pixels rather than device pixels and will apply upscaling for high dpi devices. This is the behavior in 1.60 and before.

[chris-cooper]

Are polylines on terrain already on your list @lilleyse ?

[lilleyse]

@chris-cooper I haven't created a full list yet but that's a good one to add.

[smills2929]

A workaround is to set viewer.resolutionScale = 1.0 / window.devicePixelRatio.

This will render the Cesium canvas in CSS pixels rather than device pixels and will apply upscaling for high dpi devices. This is the behavior in 1.60 and before.

@lilleyse
It isn't quite so simple. If you set the viewer resolutionscale and then drag your browser window (with a cesium viewer in it) from a high DPI screen to a low DPI screen with different scaling, it doesn't adjust it, so you get a horribly pixelated viewer.

[OmarShehata]

drag your browser window (with a cesium viewer in it) from a high DPI screen to a low DPI screen with different scaling

You should be able to detect this in your application, and you can re-update the resolution scale then. See https://stackoverflow.com/a/31482424

[lilleyse]

@smills2929 also check out some of the solutions in #8083:

• SSE Independent from Resolution Density & Cesium Renders at Screen Res #8083 (comment)
• SSE Independent from Resolution Density & Cesium Renders at Screen Res #8083 (comment)

There was some discussion about making this a configurable option in cesium. I think there's enough demand now that it's worth doing. I opened an issue which we'll target for the coming release: #8215.

[lilleyse]

@smills2929 you can now replace any workaround code with viewer.useBrowserRecommendedResolution = true. It will ship with Cesium 1.62.

[smills2929]

That's great to hear. Thanks!

…

On Thu, 26 Sep. 2019, 4:39 am Sean Lilley, ***@***.***> wrote: @smills2929 <https://github.com/smills2929> you can now replace any workaround code with viewer.useBrowserRecommendedResolution = true. It will ship with Cesium 1.62. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#8113>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AC5EEWSICANVQABFW5GTH53QLPEAHANCNFSM4IQFNEKA> .

[lilleyse]

@IanLilleyT since just the GL line width problem is left can you open a separate issue? This one can be closed.

[IanLilleyT]

The separate GL lines issue can be found here: #8319