Looking for Feedback

Good idea, Mark. One thing I have noticed is that the extend to surface hardly ever works. I keep geting the message: `up-to surface is invalid'. Often what I end up doing is extend by a length past where I wanted to go and then I trim with that same surface that was called invalid.

I could be using this wrong, but since it never seems to do what I am looking for-I just don't use it for extending to a surface.

I also have problems with one of the items on your list. That is different edges expending in different directions ot in unexpected directions.

jk

Interesting. I had always just accepted these issues, assuming they were caused somehow by corrupt surface definition.

Just this morning I used extend face and it worked (create the feature in the feature manager tree) and then the whole surface disappeared. No, it's not the video card. I couldn't even select it anymore. The strangest thing...

I've always wished there were a 'reverse' option in the Extend command. That is to say, instead of adding to the surface, it would trim it by the said amount.

Michael,

As a work around I have in the past used Face Curves to create a curve to trim to.
You'll need to figure out the trim length as a percentage of the full length of the surface.

HTH,
Muggs

Mark,

A little late to the party, it looks like. I've recently done a project for a team SW is sponsoring. I used a lot of extends in it. Extends frustrated me a little when you input a distance to extend, say .1, and on one side it extends .001 and in the middle it extends 4. I understand that the closer it is to a rectangular UV the better, but when things are that simple, you don't need to extend.

I don't like getting V notches in extended edges.

I really don't like it when an extend curls. The extend should assume you want to continue the same sort of curvature that the face already has. I get extends that on one end of an edge will get a sharp U and the other end are smooth the way you want it.

Mike Wilson said he wanted the extend to work the other way too, to remove part of the face. I think that's a great idea, especially given the hard times that trim gives me.

I've also had an extend that flipped the extended side to the opposite UV direction. So if I extended the north side of a rectangle, it later extended the south side instead. Weird, kind of the way trims fail. I think Will Smith observed this as well.

I can't post my example, but the situations aren't all that rare. I usually like to avoid extend, but in this case it was better than the alternatives.

I get the split edges all the time, but have found that changing up the combinations of the edges selected for extend will often take care of this. So if I've got 12 edges, I'll skip some of them in one extend feature, then get them in another extend feature. Sometimes this solves the issue, sometimes not--it depends on the combination of which edges are selected in which feature(s).

What kills me is when I've got three surfaces, with one of those surfaces destined to end when extended. Think of a surface on top, a surface on the bottom, with the one in the middle coming eventually to a point within the extend feature. What I'd like to see is a rational way of extend dealing with this. Instead, it hoses the surfaces. It should allow the mid-surface to come to a point and end, with the top/bottom surfaces knitting together after that point, but I get odd hourglass-shaped flips, etc. after.

I can share some files if they're kept confidential/internal. I have two "seed" surface-heavy parts with all sorts of software bugs exhibited within--primarily with trims failing/forgetting surfaces or simply refusing to function anymore. The dumb thing is that the involved surfaces will trim just fine if I roll back just before the failed trim--works exactly as the original trim with no errors. So I have to later carve out that failed feature and patch up my downstream knits, offsets, whatever--with no rational explanation as to why my original trim failed. (I was up past midnight "healing" my model of these very surface bugs just last night--I love it when that happens ;-)

You know, last night I tried the method of deselecting everything and then reselecting the surfaces and what was to be kept in the trim, but even that wasn't working (!!!). I admit I was cussing well after 11:00PM, fully knowing about the downstream problems of losing a trim feature way up in the tree like that. For one of them (one that didn't work with the deselect/reselect trick) I rolled back to before the trim and it still wouldn't let me create a mutual trim, fresh. So I did two trim operations (non-mutual) and that worked. Just had to roll down the tree slowly (several hundred features) and fix the Knits and other items downstream as I went along. When at the end, I was able to delete the silly failing trim feature.

In this case, the gig was hourly. Still, how do I charge a client for this sort of crap? With this project, the client has to take the fall, since they specified v2008 (doh!), but I really don't like having my clients paying for debugging time. (And I certainly don't like paying for it myself through sleep deprivation.)