利用者:Witt/GSoC2017/Reports/weekly09

提供: wiki
< 利用者:Witt‎ | GSoC2017‎ | Reports
2017年7月29日 (土) 03:57時点におけるwiki>Wittによる版 (Created page with "=Weekly Report #09 - Silhouette Operator= The second evaluation is now done. That means only one month left to finish the Silhouette Operator! Judging by the rate of progress I ...")
(差分) ← 古い版 | 最新版 (差分) | 新しい版 → (差分)
移動先: 案内検索

Weekly Report #09 - Silhouette Operator

The second evaluation is now done. That means only one month left to finish the Silhouette Operator! Judging by the rate of progress I will not be able to implement all the features and ideas I had within the Gsoc timeframe. Nevertheless achieving the number one goal: “Giving the sculptor the freedom to add new geometry contrary to just shaping the existing” is definitely within reach.

What has been done this week?

This week i started to finish the geometry generation by adding code to generate backsides as well. Now generating a silhouette creates a waterproof shape with the configured settings. I created a small video to demonstrate the impact of the different settings. You can find it here for now: https://www.dropbox.com/s/x12a62zf2ynevum/Basic_002_fs.mp4?dl=0


In the second half of the week I started with the next big step: Connecting new silhouettes to existing geometry. This poses a challenge for multiple reasons:

  • The intersection of both, the silhouette and the old geometry needs to be calculated. Depending on the geometry complexity this is really computational heavy.
  • The transition between the two shapes can be rather complex because minimal assumptions can be made about both shapes.
  • The topology of the transition is not trivial. Depending on the existing topology many different factors need to be taken into account. Vertex density, edge/loop flow orientation and varying face sizes or even ngons.

For the transition between the two shapes i decided to go with a more practical than precise approach. CAD-style precision is not the goal of the tool. Moreso the tool is supposed to give the artist the option to get a quick basemesh to start sculpting on. -Not a perfect shape.

I was able to get a good start tackling the first challenging factor. I reduced the intersection problem to a 2D-Polygon point intersection task. It is fast but gives inaccurate results if the shape in the z-axis is not primitive as is the case if smoothness is close to 1. But in practise this is a reasonable drawback to take. If the smoothness factor is high, the transition between the silhouette and old geometry is supposed to be blurred and smooth anyways. I finished detecting interior vertices and I am currently in the process to find the necessary edges and loops which will later be deleted and/or connected.