# Distortion on Bézier-curves **URL:** \ **Category:** Development\ **Created:** [September 15, 2026, 7:36pm UTC](https://forum.collaboraonline.com/t/distortion-on-bezier-curves/5039 "2026-09-15T19:36:00Z")\ **Posts on this page:** 4\ **Page:** 1
**Author:** ![Regina](https://avatars.discourse-cdn.com/v4/letter/r/d78d45/32.png) [@Regina](https://forum.collaboraonline.com/u/Regina)\ **Post date:** [September 15, 2026, 7:36pm UTC](https://forum.collaboraonline.com/t/distortion-on-bezier-curves/5039/1 "2026-09-15T19:36:00Z")
Hi all, there is an ongoing discussion on bug report tdf#173555 about how distortion of a Bézier-curve should work. Currently the distortion affects the anchor and control points of a Bézier-curve. The request is, that instead it should affect the points of the curve themselves. What is your opinion? ---
**Author:** ![knordback](https://avatars.discourse-cdn.com/v4/letter/k/65b543/32.png) [@knordback](https://forum.collaboraonline.com/u/knordback)\ **Post date:** [October 9, 2026, 3:54am UTC](https://forum.collaboraonline.com/t/distortion-on-bezier-curves/5039/2 "2026-10-09T03:54:34Z")
Hi Regina. This is interesting. I’m definitely no expert here, but I’m inclined to agree with you that this shouldn’t be called a bug. Some users may expect a pointwise warp of the curve, as is being requested, while others may expect the current behaviour of just mapping the control points, which keeps the result within the space of Béziers. In an ideal world it would be possible to choose from either method, but I suspect that that’s a low priority. Kurt Nordback Contractor Senior Software Engineer Collabora Productivity | St Johns Innovation Centre, Cowley Road, Cambridge, CB4 0WS, UK Registered in England & Wales no 08644931 > **[Home](https://www.collaboraonline.com/)** > > CollaborativeOpen-SourceDocumentEditing Controlled by You Collabora Online is a powerful online document editing suite. Integrate into your own infrastructure or access via one of our trusted hosting partners. Explore the FREE Demo... ---
**Author:** ![Novastart](https://dub1.discourse-cdn.com/flex013/user_avatar/forum.collaboraonline.com/novastart/32/2480_2.png) [@Novastart](https://forum.collaboraonline.com/u/Novastart)\ **Post date:** [October 20, 2026, 3:11am UTC](https://forum.collaboraonline.com/t/distortion-on-bezier-curves/5039/3 "2026-10-20T03:11:45Z")
For what it’s worth, the bug report has moved on since this thread started: a patch for tdf#173555 (“make function in \_xpoly/b2dpolygontools a real warp”) was pushed to LibreOffice master on Oct 8, and the bug says it will be in 27.2.0, so it can already be tried in a daily build. On the question itself, I’d lean towards the pointwise behaviour being what people expect, mainly because of the two cases shown in the video on the bug. A curve whose control points sit only on corners or on edges that don’t move isn’t warped at all, and two curves that look identical can come out differently depending on where their nodes and handles happen to be. Users only see the outline, so a result that depends on invisible handle positions is hard to explain. Mapping only the control points is exact for affine transforms (move, scale, rotate, shear), so nothing needs to change there. It’s only a real four-corner (bilinear) distortion where the exact result generally stops being a cubic Bézier, which is why subdivide-then-warp has to add nodes. A sensible tolerance for that subdivision, so simple shapes don’t turn into dozens of segments, looks like the main thing to watch. I’m coming at this from the whiteboard side (I work on Novastart, an infinite canvas workspace), where people drag shape corners constantly and judge the result purely by how the outline looks. ---
**Author:** ![mmeeks](https://dub1.discourse-cdn.com/flex013/user_avatar/forum.collaboraonline.com/mmeeks/32/371_2.png) [@mmeeks](https://forum.collaboraonline.com/u/mmeeks)\ **Post date:** [October 20, 2026, 4:32pm UTC](https://forum.collaboraonline.com/t/distortion-on-bezier-curves/5039/4 "2026-10-20T16:32:41Z")
> [@Novastart](#): > > I’m coming at this from the whiteboard side (I work on Novastart, an infinite canvas workspace), where people drag shape corners constantly and judge the result purely by how the outline looks. Ooh - sounds interesting; welcome to the forum. I’ve often thought that it would be lovely to re-work some of our OLE object type pieces to make small bits of Office’y functionality we could embed into an infinite canvas. A spreadsheet table with full formula support, or a chart, or whatever 🙂 Would love to see what we can do together; do mail me: [michael.meeks@collabora.com](mailto:michael.meeks@collabora.com)