你好Michael ，我重复了您的操作，感谢，学习到了新的技巧。我的操作的确是您说的一样。事实上使用者们希望依靠投影就能够成功，而不使用其他的技巧。因为投影（或最近距离投影）和相交的曲线在视觉上几乎一样，准确的说无论看起来有没有差别，都是完全平滑的，所以希望能够直接成功。

附上模型附件（ps：不习惯上传模型附件的原因可能是论坛允许附件大小太小了，我更习惯上传图片吧）

（Hello Michael, I have repeated your operation. Thank you for learning new skills. My actions are indeed the same as what you said. In fact, users hope to succeed through projection without using other techniques. Because projection (or closest distance projection) and intersecting curves are almost identical visually, to be precise, they are completely smooth regardless of whether they appear different or not, so we hope to achieve direct success.
Attached model attachments (PS: The reason why I am not used to uploading model attachments may be that the forum allows attachment sizes to be too small, I am more accustomed to uploading picture）

ps，我知道Transform > Deform > Twist 的方法（PS, I know the method of Transform>Deform>Twist）


而我需要实现的效果如图片：Expected final effect.3dm - MoI.jpg所展示的，模型文件Expected final effect.3dm。边缘有个较大的fillet而这只能通过Transform—Deform—Twist的方式近似实现或network修补，Construct > Curve > Isect to intersec也难以进行较大fillet因为断裂的边缘经常被漏掉选择一类 的原因而我和部分使用者希望通过project>sweep（pipe）>boolean（diff）>fillet这样直接的方式完美的实现这类模型的实体建模，以减少，破碎面，断裂边的问题。
（And the effect I need to achieve is shown in the image: Expected final effect.3dm - MoI.jpg, the model file Expected final effect.3dm. There is a larger fill at the edge, which can only be approximated or network patched through Transform Transform Twist, Construct> Curve> It is also difficult to perform a large fill in the Isect to Intersect section because the broken edges are often missed. Therefore, some users and I hope to use Project> Sweep (pipe)> Boolean (diff)> Fillet's direct approach perfectly implements the solid modeling of such models to reduce the problems of broken surfaces and edges.）

当我将由blend生成的曲线连续度提高为G3,采用Curve>isect方式得到的管道对圆柱布尔进行（diff）就会出现细小的，破碎的面（Expected final effect G3.3dm）虽然选择边缘连续能够进行fillet，但希望直接使用project→sweep（pipe）达到Twist的没有破损面的效果。以避免因为细微选择差别，带来的运算结果完全不同的问题，以及后续的问题。

（When I increase the continuity of the curve generated by blend to G3 and use the Curve>isect method to obtain a pipeline for cylinder Boolean (diff), small and broken surfaces (Expected final effect G3.3dm) will appear. Although selecting edge continuity can fill, I hope to directly use Project → sweep (pipe) to achieve Twist's effect of no damaged surfaces. To avoid the problem of completely different calculation results caused by subtle selection differences, as well as subsequent problems.）

表达可能并不准确，问题的根本是：如果只有project生成的曲线是准确的，那么moi能否生成一条完整可用的管道？而不是使用任何近似的方式。毕竟这仅仅是一种例子。
（The expression may not be accurate, and the root of the problem is: if only the curve generated by the project is accurate, can moi generate a complete and usable pipeline? Instead of using any approximate method. After all, this is just an example.）

由此衍生的问题包括同类曲线边缘的倒角：个人认为都属于同类的问题（The resulting problems include chamfering at the edges of curves of the same type: I personally believe that they all belong to the same type of problem）