saved image silhouette width too fat

[Elaine Meng]

If I save a smaller image than the window the silhouettes become much too fat. Shouldn't they scale down proportionally? There is an attempt (hardware permitting) to do the converse when you save an image larger than the window, at least in Chimera. Maybe related to #686.

[Tom Goddard]

Silhouette width is specified in pixels. If an image is saved with the width specified at 100 pixels or 1000 pixels the silhouette width is the same fixed value in pixels. So with a smaller image the silhouette width is thicker relative to the model size. This is the intended behavior. It seems like what you want is that the silhouette width be in physical units (Angstroms) instead. That would give very poor rendering speed as the silhouette width would need to change as you zoom in and out, and with the width in physical units it would vary with depth with a perspective view. Even if it were fast I think it would be a worse choice. You would then either need to specify the silhouette width in some way.

The current behavior when you save an image and specify the pixel width, it gives you exactly what you would see if you resized your graphics window to that width. That is not the same as saving the image at the current size and then scaling to the new width for features that are sized in pixels such as silhouettes or labels with height specified in pixels.

[Tom Goddard]

I think in all case the user really does want the silhouette width and labels with size specified in pixels to appear at the same size relative to models in images. A more common case than making a tiny image is to save a much larger image (e.g. 3000 pixels) than is on the screen for a publication figure.

I think save image should adjust the pixel sizes of anything (silhouettes, labels) specified in pixels so the maintain their relative size. This complicates the code but is doable, and is what Chimera 1 does.

[Tom Goddard]

The current behavior with silhouettes having a fixed width in pixels, even in saved images that are saved at larger or smaller sizes than shown on screen seems ok. More complex code that scales the pixel sizes when saving an image is possible but adds complexity -- it could be added in the future.