Anyone have any thoughts on these? I've been working with image entities in dwg files for a couple years now, so I can attest that the mixing of a local right-handed coordinate system with a global left-handed system is an absolute pain. Clipping is my main concern here. How should viewports and images be clipped? Using an invisible entity? I could see possibly an un-named layer or something (somehow the layer is a child of the viewport or image?) If you want to clip to a circle, you just refer to the circle. The other option is that the viewport or images has one "clip_entity:" slot where we nest some entity data-structure (e.g. circle, arc, etc.) That frees-up the issue about an entity not on a layer, etc. without having to use an entirely new syntax to express the clipping. --- # ./<ID>.yml # viewport entity ID: integer points: - # bottom-left corner - float # x - float # y - float # z (optional) - # top-right corner - float # x - float # y - float # z (optional) type: "viewport" layer: layer_ID clip: points: - - float # x - float # y - float # z (optional) - - float # x - float # y - float # z (optional) ... type: "polyline" closed: boolean color: [hex, "style"] object_style: object_style_ID linetype: [linetype_ID, "style"] So you just take whatever entity is clipping, but without the layer/style/color/linetype info and plug it in. BTW, the viewport needs more info (what is it viewing.) Is using a rectangle for viewports/images okay? Another option is using one point, plus rotation and size (height/width) values. --Eric -- A counterintuitive sansevieria trifasciata was once literalized guiltily. --Product of Artificial Intelligence