|
View previous topic :: View next topic
|
| Author |
Message |
Alfred Scott
Joined: 15 Apr 2004
Posts: 652
Location: Richmond, VA
|
 Posted: Wed Sep 01, 2004 9:14 am Post subject: Flood Tool Problems Examples Wanted |
    |
|
I'm working on the WildTools Flood tool, and I'd love to have some files from people where the tool is misbehaving. If possible, throw as much extra stuff in the drawing away, and put something in the drawing to indicate where you are clicking with the tool.
Please send your files to me at
alfred@seqair.com
Let's see if we can get some serious reliability into this tool.
Alfred
|
|
| Back to top |
|
 |
David Scott
Joined: 16 Apr 2004
Posts: 206
Location: Orkney Islands, Scotland
|
| Posted: Thu Sep 02, 2004 4:40 am Post subject: |
|
|
File sent to you today Alfred, with illustration of different problems with both Flood and PowerCADD Paint Bucket tools. (copy sent to Bill)
David
(file attached here as well)
|
|
| Back to top |
|
|
David Scott
Joined: 16 Apr 2004
Posts: 206
Location: Orkney Islands, Scotland
|
| Posted: Thu Sep 02, 2004 4:42 am Post subject: |
|
|
Sorry file didn't attach first time
| Description: |
|
Download |
| Filesize:
|
20.02 KB
|
| Downloaded:
|
232 Time(s)
|
|
|
| Back to top |
|
|
ms
Joined: 14 Apr 2004
Posts: 613
Location: Naples, Florida
|
| Posted: Fri Sep 03, 2004 5:25 am Post subject: |
|
|
David identified a problem I noticed but couldn't see what was happening [the thin line]. Thanks.
Also, he identified a major problem, and that's accuracy. The WT Flood tool does a really good job with finding vertices on simple polygons, but it's not good enough. In critical cases I always construct the polygons the old fashioned way.
The problem is easily demonstrated on irregular shapes that contain both convex and concave perimeter elements. I wonder if some user-control over the distance between approximating vertices internally constructed by the tool would be useful. [Or I could be completly wrong about how I imagine that the thing works].
–ms
looks like we dodged the second hurricane here in Naples, but those poor souls around Orlando-Arcadia, for example, are gonna get it again. It's a very poor part of Florida. A Cat 4 storm is something to take seriously.
|
|
| Back to top |
|
|
Alfred Scott
Joined: 15 Apr 2004
Posts: 652
Location: Richmond, VA
|
| Posted: Fri Sep 03, 2004 6:49 am Post subject: |
|
|
I'm working with Derek on the Flood tool, and we have an internal debate going over what the tool should be doing.
From Derek's standpoint (he's an engineer, poor thing), lines are infinitely thin and the Flood tool should only trace lines that actually intersect.
I have the tool looking slightly beyond the end of a line to see if another line meets it within the radius of the line weight.
I also have the tool seeing lines that come very close to touching it, so it can turn right on a line that nearly crosses it, but when the 'ink' of the pen touch each other.
Maybe this should be a user choice.
Alfred
|
|
| Back to top |
|
|
ms
Joined: 14 Apr 2004
Posts: 613
Location: Naples, Florida
|
| Posted: Fri Sep 03, 2004 7:47 am Post subject: |
|
|
Gee, and Derek always seems so OK. Go figure.
As to his view on Flood, there was a long conversation about this with Bill sometime ago, and I took the same position, and envisioned the thing doing exactly what you propose, which is to ONLY hit vertices, and to look beyond the end of a line in an attempt to find a place to go.
The exception to this of course is to all objects that are not lines or line segments. As I imagine the way the tool works, when confronting an arc, for example, it would place points on the curve of the arc at some distance apart until it came to the end of the arc. The spacing of these points is critical to accuracy, especially if the object is heavily convex or concave.
As I recall the discussion Bill was opposed to this concept but I cannot recall his thinking. My point was to have the tool look for points contained in a pixel radius defined by the snap pixel number.
On your last poiint, 'I also have the tool seeing lines that come very close to touching it, so it can turn right on a line that nearly crosses it, but when the 'ink' of the pen touch each other', I don't understand this as it seems included in your idea of looking beyond the end of a line [or any object I suppose?].
–ms
|
|
| Back to top |
|
|
Derek
Joined: 13 Apr 2004
Posts: 570
Location: Melbourne, Australia
|
| Posted: Fri Sep 03, 2004 2:31 pm Post subject: |
|
|
Just for the record, here was my last email to Alfred:
Alfred,
We can probably argue these points forever and I don't want to do this.
I think the difference is in the definition of what drawings are and this varies from discipline to discipline.
For mechanical drawings, lines are only a representation of something that is, in fact, infinitely thin: typically the edge of something. In these cases, things (edges) either touch or they don't. The line thickness used to represent the edge is irrelevant and should be ignored.
For architectural and artistic drawings, lines form part of the drawing and line weights are used for effect. This is also how the artist visualises the object in their mind so that two lines touching as a result of their thicknesses are indeed touching.
So should the Flood tool consider the 'on paper' ink or the theoretical model of what the lines represent? Both have merit and to each his own. Maybe we need two tools...
Derek.
So I don't think it's fair to say I'm against one or the other. It is just that I mostly do mechanical drawings.
Alfred, I really like the idea of having the option to "Consider Line Weights" or not.
An alternative to considering line weights or snap pixels would be to work with the precision set in the Units tab of the Drawing Setup. For example, let's say you have the precision set to 1/4". You could have the Flood Tool look for anything within that distance but not beyond.
BTW: to my mind, "Considering Line Weights" actually means considering lines to be thin rectangles with their widths the same as the line weights and with infinitely thin edges 
One thing is for sure, everything will be very messy when we get Transparency happening in PowerCADD... how transparent can a line get before the liquid of the flood seeps through?
|
|
| Back to top |
|
|
ms
Joined: 14 Apr 2004
Posts: 613
Location: Naples, Florida
|
| Posted: Sat Sep 04, 2004 4:34 am Post subject: |
|
|
I draw lots of plans in both catergories and I don't think I ever thought of a line as anything more than a two-dimensional object. 'Consider line thickness' is going to be really confusing and I don't see why I would ever want to do this.
If the line thickness has meaning in the drawing context then it's not a line but some sort of object.
Anyway, the bounding line is always on 'top' of the fill color.
–ms
|
|
| Back to top |
|
|
Alfred Scott
Joined: 15 Apr 2004
Posts: 652
Location: Richmond, VA
|
| Posted: Tue Sep 28, 2004 11:45 am Post subject: |
|
|
I've now finished a rework of the Flood tool, and it's dramatically improved. It will be in the next version of WildTools, and we may have it available 'for testing' at some point in the current shipping version.
Right now, the Flood tool can flood all examples everyone sent me, plus a lot of torture-test examples. If you think you have an un-floodable anything, send an example to me at alfred@seqair.com
We did discuss the concept of having a "Consider Line Weights" option. In the re-write of the tool, I began with that as an option, but it quickly became apparent that the tool was completely worthless when created this way. It failed more often than it worked, even with simple examples.
Alfred
|
|
| Back to top |
|
|
|
blogger
del.icio.us
digg
slashdot
