|
|
Boolian-caused artifacts
About Truespace Archives
These pages are a copy of the official truespace forums prior to their removal somewhere around 2011.
They are retained here for archive purposes only.
Boolian-caused artifacts // Archive: Tech Forum
Post by janetb // Mar 15, 2007, 7:28am
|
janetb
Total Posts: 35
|
I have a recurring TS problem which has become somewhat debilitating and was hoping someone out there might be able to shed light on the matter....I do a lot of cut-out letters in my jewelry work, and find the cut-outs often screw up the Solid Mode Display (in the workspace) of my models---it works fine up to a certain number of cuts (many) and then the next cut sets up artifacts which cover over parts of the earlier cut-out parts....The problem is that it renders fine, so I don't know if there is anything to be 'fixed'....I need to see the metals in the realtime workspace while I am designing---to see how the different layers look when layered in different combinations. So I need to see the layers showing thru the cuts. I also like doing a plain old screen capture to send the results to clients....I use the chrome type 'cheat' material for gold and silver so I don't have to mess around with lighting and environments. It's usually very convincing if you get it at the right angle.
In the present model, I tried doing it a million ways....In the past I solved the problem by boolian adding or gluing all the cutters together and doing it all in one cut. This time nothing worked. I also tried cutting a plane to be extruded afterwards instead of starting with a squashed cylinder---but that made my solid mode material disappear altogether on the workspace while the icon in the library remained a lovely shiny gold :confused: !!!! In all cases it works fine most of the way and then suddenly screws up with the next cut. In all cases they render fine and each cutter individually works fine. It seems in every way the geometry is OK. So how do I fix the workspace display? I'm curious to see if perhaps other computers don't have the artifacts I am getting.....
Another matter---my gold material looks vibrant and shiny on the plane disk but dull and darkish on the top face of the squashed cylinder. Why would that be? Is it the same on other computers?
Any suggestions would be much appreciated....I have a major design opportunity and would like to send the client a lot of alternative combinations of the various parts and materials.....
Many thanks!
Janet
A scene including parts as attached a the end of this post. |
Post by janetb // Mar 15, 2007, 7:47am
|
janetb
Total Posts: 35
|
I should have mentioned I have TS6.0 SSE and 2KPro.
Janet |
Post by Bobbins // Mar 15, 2007, 8:29am
|
Bobbins
Total Posts: 506
|
The problem is in the geometry of the disc. When you perform the boolean operation tS is triangulating the disc and having to guess at where to make the triangulations. The artifacts you see in the solid display is the result of tS trying to do that triangulation guesswork for you. In order to connect the resulting vertexes together some of the triangles are having to be placed over the top of the boolean, so it looks like the bolean subtract hasn't worked (it has, it's just covered over by a face). If you triangulate the face of the disc yourself, or quad divide it etc. before the boolean operation then you can help tS along.
As for the dull looking gold texture, I'd guess that what you are seeing is the reflection of the background colour. All metal, glass and other reflective textures must have something in the scene to reflect or they will always look 'wrong'. You might need to put a white plane above or to the sides of your object, or assign a background texture to the scene in the render properties so that there is something reflected in the object otherwise you will just see the background colour. |
Post by Chester Desmond // Mar 15, 2007, 9:32am
|
Chester Desmond
Total Posts: 323
|
IT's kind of hard to explain, but maybe this thread will help.
http://forums.caligari.com/discus/messages/1580/6658.html?
In the case of your object, the places where you've booleaned are all kind of just floating inside the face, with no attachment to the outside. TS doesn;t know how to display this because of the co-planar concept discussed in the thread I linked. I guess it must keep up until your last bit of cuts and then it freaks out.
Judging by your object, you have turned off triangulate in the Boolean options (which in this case is a good thing) because it is not a mess of lines as triangulating would have made.
If your end goal is to have this object display correctly in solid mode, you could try the following. Add some lines as seen in my first pic...this will 'isolate' the area from the rest of the object and give some more accurate data to the realtime display, ie it won;t have to look across the whole object to try and find surface info, it will only have to look in the little area we've set up. Those are 6 separate lines I added and make sure they are connected to a vertex of the outside circumference.
Do your boolean sub using the setting I've shown (make SURE delete edges is OFF) and it should give you the result in pic 2.
In the third pic, I assigned planar UV to the top and it seems to have improved the look of the metal shader. |
Post by spacekdet // Mar 15, 2007, 9:37am
|
spacekdet
Total Posts: 1360
|
Yep, I'd also suggest subdividing the top and bottom faces of the cylinder before you start the boolean operations. Untic the 'Delete Edges' checkbox.
Here is a minitut on a subdividing procedure (done for tS4, but the general idea is there).
http://home.earthlink.net/~geneg/pictures/VioletTut.jpg (http://home.earthlink.net/%7Egeneg/pictures/VioletTut.jpg) |
Post by janetb // Mar 15, 2007, 11:16am
|
janetb
Total Posts: 35
|
Hi All!
I had just written a rather lengthy letter to Bobbins and was getting a screen capture of my vibrant metal piece--adjusted by rotating the UV projection...:-)...I saw the new posts coming into my email, and when I clicked on the link to get back to the forum, I lost my letter to Bobbins...:-(.....So I will respond to all now, but just a brief summary of the lost letter.....
I didn't think the extra bits were faces---don't faces have to have vertices? If they were proper faces, they wouldn't be artifacts, no? Also the fact that the object rendered perfectly made me feel it wasn't bad geometry. That is many paragraphs boiled down to three short sentences...:-).....
In any case I tried triangulating and then again subdividing and got exactly the same results as before, so that was not the problem.....
As regards the dull texture, everything you say, Bobbins, is correct---for rendering. But the idea here was not to render but to do a screen capture using a 'cheat metal'---they have built-in 'reflections', so you use them in a totally empty scene. Rendering is a whole other story....
Chester, your post was fascinating....Success is so nice...:-).....
Regarding the co-planar issue, I had chopped of the top of the disk with a cube to make the surface flat and it didn't help, so that wasn't the problem....The old fix for such things was to subtract a cube---no one could ever explain how it worked, but it did...But here it didn't....
My Boolian settings were like yours, but Delete Edges was ON and Keep Material was ON (did you add the material later?).....I thought the fewer edges the better, and thus was very interested to see your additional lines and settings. Did you try it with your lines and with Delete Ddges On?---I wonder if the added lines perhaps were enough to solve the problem....I can't wait to try it!
As regards the dull surface, as I said above, I too had adjusted it with the UV projections. It's such a blast---you can get really good effects. I'll post this now---before I lose it---and get the pitcure in a bit later....
Janet
[ |
Post by janetb // Mar 15, 2007, 11:37am
|
janetb
Total Posts: 35
|
This is a screen capture of the TS workspace...... |
Post by janetb // Mar 15, 2007, 11:42am
|
janetb
Total Posts: 35
|
Chester--
What size are the image files yu posted?? What are the pixel dimensions?
Janet |
Post by Chester Desmond // Mar 15, 2007, 12:05pm
|
Chester Desmond
Total Posts: 323
|
I didn't think the extra bits were faces---don't faces have to have vertices? If they were proper faces, they wouldn't be artifacts, no? Also the fact that the object rendered perfectly made me feel it wasn't bad geometry. That is many paragraphs boiled down to three short sentences...:-).....
In any case I tried triangulating and then again subdividing and got exactly the same results as before, so that was not the problem.....
It has to do with the way the info is passed to the video card I believe. Usually the video card looks at the faces defined by points and shows it correctly but ,like Bobbins mentioned, behind the scenes, TS is "connecting" this array of points which don't really join together in a meaningful way and is sort of picking some to "join" to be able to display a face. Because it can't find a nice straight path across the face, because it is interrupted by all the vertices you added by booleaning out the symbols, it sort of lays the nice straight line over top and renders a false face over top of the existing face. We have to add the lines on both sides because the same concept will happen on both sides.
Imagine a little car driving on a frozen pond. on a smooth surface it can easily drive straight across, but punch a few holes in the ice and sudddenly the car is swerving all over to try and get across..imagine it is painting a line as it goes to show the path across the ice...now imagine truespace is using this painted line to pass info to the video card of how to display the face.
Chester, your post was fascinating....Success is so nice...:-).....
Regarding the co-planar issue, I had chopped of the top of the disk with a cube to make the surface flat and it didn't help, so that wasn't the problem....The old fix for such things was to subtract a cube---no one could ever explain how it worked, but it did...But here it didn't....
My Boolian settings were like yours, but Delete Edges was ON and Keep Material was ON (did you add the material later?).....I thought the fewer edges the better, and thus was very interested to see your additional lines and settings. Did you try it with your lines and with Delete Ddges On?---I wonder if the added lines perhaps were enough to solve the problem....I can't wait to try it!
Thanks for the compliment.
In using my method, having DELETE EDGES on would delete the lines I added after the first piece was subtracted, bringing me back to square one for the next piece; turning it off allowed my little area to be stay defined (using my car example, it is easy to drive in a straight line across a small area that has been roped off from the hole-filled rest of the pond). I had turned off 'keep material' so it wouldn't leave blue behind.
The object itself, not to be cruel, is very unstable and probably wouldn't survive being exported outside TS. If you are just using TS you should be fine. There are too many floating faces and dead ends for a program to figure out what to connect to what. Just try triangulating or subdividing the finished product to see the mess it leaves behind in trying to make it a "good geometry" object; incidentally I just tried it and by using the 'triangulate' tool (after all the booleaning was done) it got rid of the display artifacts\errors and now looks not bad in solid mode but in wireframe it looks like a disaster (see pic below-top left shows the problem you are having, top right- solid mode after triangulation, bottom- wire of triangulated object -yecch-) If you are interested in having a more structurally sound object, you might consider dividing the disc into quadrants (liek a pie) so each symbol has it's own little area to be booleaned out of. There will still be geometry errors but they won't be across the whole object (ie it should avoid the phantom faces you see in solid mode)
Looking forward to seeing the picture. I can see how in your business you might want thigns to look nice and shiny! For my own interest, why don't you use renders to show your clients what their product will look like, or is the solid mode for your own preference while modelling? |
Post by Chester Desmond // Mar 15, 2007, 12:08pm
|
Chester Desmond
Total Posts: 323
|
That looks good for a solid display, gives a good impression of metallicity.
My images are a few different sizes, just randomly cropped screen grabs. You can right click them and view their properties I believe. |
Post by jamesmc // Mar 15, 2007, 12:50pm
|
jamesmc
Total Posts: 2566
|
Since the design template is flat, why not make it in a 2D graphics program, then export it out as an AI file?
You can then import it into tS as an AI file.
Granted there won't be as much manipulation of the object as it will have some artifacts, but to me it is much easier to do it that way than booleans.
For instance, this design was for part of a piano I did. It's a fairly intricate design and exported as AI, then imported as AI into tS.
I can be much more precise in intricate detail in 2D programs than I can in tS. |
Post by kena // Mar 16, 2007, 9:33pm
|
kena
Total Posts: 2321
|
what I used to do in TS6 when I had to subtract a lot of objects from a single other one was to first align all the objects where they needed to be, then I would join them together as a single object. Subtract that single mult-object form the single circular object, and I didn't get the problem you described. Then again, I tend to find strange solutions to problems :D |
Post by stan // Mar 17, 2007, 8:39am
|
stan
Total Posts: 1240
|
another way to add geometry for boolean ops when using a cylinder is the magic ring. by a click and drag of the top handle will produce the vertices..drag left click for the ones that join in the center and drag right click for the radial ones
always glue all drill objects togeather too.:) |
Post by janetb // Mar 17, 2007, 10:41am
|
janetb
Total Posts: 35
|
Hi Kena and Stan,
As I said in my original post, I did try 'joining' all the cutters first and then subtracting the 'multi-object'! I joined first by gluing before subtracting, and then started over again boolian-joining them before subtracting. Both methods created the artifacts---that's why I posted...:-)....
The Magic Ring gives way too many additional lines for the desired # of longitude lines---your sample with fewer lines was not a round enough circumference...:-(....I guess I could delete a bunch of them. But it would be much easier to draw in a few diameters. On the basis of Chester's solution, probably wouldn't need too many....Maybe even just dividing the disk into quarters...(?)....
Stan---in your sample you failed to cut the problem cut--the right side myrtle leaves! I also had no problems up to the point of your sample...:-)....
J. |
Post by janetb // Mar 17, 2007, 10:57am
|
janetb
Total Posts: 35
|
Jamesmc---
I had originally done all of this in .ai, so I first tried importing a simple shape in and cutting it out. 1) It lost all the beziers so I couldn't do any adjusting, and 2) I got really weird result when boolian cutting ..So I did it all over again in TS.....I guess you are saying you do the cutting out in Illustrator and then importing? I will give that a try. Will report results. Thanks.
J. |
Post by janetb // Mar 17, 2007, 11:01am
|
janetb
Total Posts: 35
|
Chester--
I'll get back to yu on your long poost....I tried doing it like you did in the .pdf, but although it improved, I didnt come out as good as yours. Were you cutting into a circular plane and then extruding, or cutting into a flattened cylinder?
J. |
Post by stan // Mar 17, 2007, 12:01pm
|
stan
Total Posts: 1240
|
here is another sample showing the object with less radial vertice rings.[it may not even need the radial ring of vertices] all drill objects are glued into one object as before:) |
Post by janetb // Mar 19, 2007, 4:25am
|
janetb
Total Posts: 35
|
Chester,
The .pdf was from another thread...Sorry.....I meant your explanation in your post above...I did it again, but it doesn't come aout as good as yours...See below...Did you use the cut-out disk in my scene or did you do it from scratch? If the latter, how did you set up the disk (cylinder?) prior to cutting....You must have done something different from me (I didn't do any thing---it's just a squashed cylinder....)
Stan,
That certainly looks good---it was all in one cut with all the cutters glued? What are the settings for the cylinder? Thanks.
J. |
Post by stan // Mar 19, 2007, 5:08am
|
stan
Total Posts: 1240
|
Janet, yes all drill objects are glued togeather into one
here are the settings..done on a two sided plane.. I just used the handle of the magic ring , with a left click and drag for the main vertices and right click and drag for the radial ones... ..just make sure the setting is still at 0.0 for the main vertices., but you create the vertices.. 2.0 for the radial ones will create one ring of vertices...
the settings for the chrome shader are in the last post :) |
Post by Chester Desmond // Mar 19, 2007, 5:19am
|
Chester Desmond
Total Posts: 323
|
Hi Janet, I used the objects you provided in your scene. I added the lines to the disc which was already mostly booleaned by you. Not sure why you'd get different results unless your lines were added to different verts than mine were.
I tried to join 'corners' of the cut out areas. Here's your scene with the lines added. |
Post by janetb // Mar 19, 2007, 7:09am
|
janetb
Total Posts: 35
|
OK Stan---
I see you edited the settings on your post :D ....I had printed out the original post and just spent the last few hours trying to understand it and get it to work....:o ....I was just about to post the results and ask why on earth you would use 0.1 for the Conic Angle and 4 for the Spherical Latitude Division and how that could possibly work...:( .....So I was pleased to see the re-adjusted settings and will go try that....:) I'm looking forward to perfect results...Have to go to a Pilates class, so I will continue when I return....
J. |
Post by stan // Mar 19, 2007, 11:24am
|
stan
Total Posts: 1240
|
yes I changed the settings, I forgot to check the z measure after..when using a cylinder it added to the z measure. with a double or single sided plane it works . once created select faces of one side of the double sided plane and sweep. turn 'backside" off when using rectangle selection with a double sided plane or just select all the faces on a single sided plane and sweep.[single sided is the way to go ]
with the plane primitive just creating a circle creates the vertices..just the radial one needs to be added by entering 2 in the box shown in last post .
sorry for the confusion, didn't have time before to explain properly either:rolleyes: |
Post by janetb // Mar 22, 2007, 5:06am
|
janetb
Total Posts: 35
|
1) I was able to do the cut-outs with Stan's 'radialled' disk with the cutters boolian-joined but not with the cutters glued. Then I noticed I had Identity (Boolian Property Panel) set at 58 and Stan had it set at 1! Quite a difference!! So I set it to 1 and it worked with the glued cutters...:-)....Can someone say something about how to know how to set that property? Chester---what setting did you use for Identity?
2) I set up a chrome material according to Stan's picture, but, as you can see, it came out much darker....How can that be? At first I assumed it was Stan's lighting set-up, but that shouldn't affect the preview in the Material Editor....Why would it be so dark? Stan--what were your settings for the properties below "Specular color"---that panel was closed in your screen capture.....
3) I tried both Chester's and Stan's techniques for creating the disk. In the end they both enabled me to cut everything out with proper see-thru holes but there were some artifacts on the surface of the solid part in both cases. It was improved with the UV editor, but what really got rid of them all was simply more light! I had been using Texture lighting, and when I changed it to White Lighting (both in the standard TS library), it pretty much cleared away the artifacts! Some positions brought them back. Also, making the disk thicker (ie vertical scaling) also screwed things up.... Stan and Chester--What are your lighting set-ups?
4) Why did my 159KB image load into this forum at 37.5KB??
J. |
Post by janetb // Mar 22, 2007, 5:14am
|
janetb
Total Posts: 35
|
I see Chester's Identity setting was 50. Why did it have to be so much lower to allowed glued cutters to work on Stan's radial disk? (Boolian-joined cutters worked OK with Identity at 50.)
Janet |
Post by Chester Desmond // Mar 22, 2007, 5:51am
|
Chester Desmond
Total Posts: 323
|
To be quite honest, I have never used the identity setting; I just leave it at default (50). I very rarely use boolean operations so haven't had much need to mess with that. My understanding of identity is that it checks (in 1/100th of mm) to make sure two vertices won't be too close together when the operation is performed. The lower the value, the closer it will allow vertices to be to each other in the final object.
Looking into this did allow me to discover some neat functions when you hold ctrl while subtracting or unioning. It's almost like an eraser in subtract mode!
I think in the case of Stan's method there are a lot more vertices in the disc object, so the chances of vertices being co-incident (too close) is much higher, and so you need to make it more forgiving by lowering the identity.
As for the lighting, I would have just used whatever you had in the scene you sent. The White lighting setup is probably bright enough to kind of wash out the artifacts. |
Post by Chester Desmond // Mar 22, 2007, 9:17am
|
Chester Desmond
Total Posts: 323
|
On a related note, I was just working on a project with similar problems.. a 'bad' face that exhibited this same symptom in solid mode. You can see in the solid mode screenshot (1) that all the vertices are "floating" and not really connected as in the case of your disc. When I rendered it with Virtualight (2) it closed off the face (ie it looked just as wrong as in solid mode). Going further I exported it as an OBJ (with the Luuv plugin) and it said "there are holes in object. do you want to triangulate?" so I said yes and it gave me the results shown in (3). I also saved it as an OBJ without triangulating and as a .cob.
In pic 4 the objects are arranged .COB, triang. OBJ, nontriang. OBJ and rendered in Bryce 5. As you can see, the problem still exists in all 3 cases.
When rendered in TS it worked fine and had no problems with the hole (the object is essentially a shelled sphere which I lopped a corner off of)
http://forums1.caligari.com/truespace/attachment.php?attachmentid=4978&stc=1&d=1174516003
So it looks like the phantom\nonplanar face is not necessarily a TS Solid Mode issue as much as a general 3d issue and also confirms that the objects created like this are structurally unsound, although TS\Lightworks can handle them within TS. Something to keep in mind if we want to export to another app. |
Post by stan // Mar 23, 2007, 5:40am
|
stan
Total Posts: 1240
|
I didn't change any lighting from your test scene. I did use phong reflectance shader
luminace 0.37
diffuse 0.9
shininess 0.1
specular 0
as for the boolean I don't know how it actually works, but I don't think it has any thing to do with amount of vertices or how close they are.
I use 1 as a starting point when glued groups are used for union or subtraction. it doesn't matter if I'm booleaning to a simple plane or many verticed object..you need the vertices in a subtraction that I know, it seems that each sub object of the drill still gets done one at a time, so it gives truespace a place to put the hole without covering up another one it had
already done with a portion of a face..union is different because it doesn't make holes.
after the first boolean of a multi object drill, 1 might not work on consecutive attempts, another low number might, but after much trial and error I found it better to just boolean all at once because sometimes you can't complete your object otherwise.
for union ot many objects you can use a plugin called 'mesh collapse' or another called 'poly combine' to get a similar result to boolean but not exactly because they don't delete vertices where objects intersect.
as for the difference in settings between glued or unioned drills it might just be how truespace handles them in a slightly different way..the identity number seems to be a number in the boolean equation that can be altered to give different results when 50 doesn't work thats all I know..
Gord :cool: |
Post by Chester Desmond // Mar 23, 2007, 6:18am
|
Chester Desmond
Total Posts: 323
|
as for the boolean I don't know how it actually works, but I don't think it has any thing to do with amount of vertices or how close they are.
From the manual:
Identity: Identity is used for adjusting the tolerance that the Boolean operations will use when performing a Boolean operation. This value is a distance in 1/100 of a millimeter for identifying near-coincident vertices. In general, the default value of 50 should be used, but this sometimes needs to be adjusted to achieve a successful result. If a Boolean operation does not work, trueSpace may notify you that adjusting this value could allow the Boolean to succeed. |
Post by janetb // Mar 24, 2007, 10:28pm
|
janetb
Total Posts: 35
|
From the manual:
but this sometimes needs to be adjusted to achieve a successful result... trueSpace may notify you that adjusting this value could allow the Boolean to succeed.
The question in effect was: how to make an educated guess in adjusting an Identity number? What sorts of situations require what sort of number? Stan basically suggests starting with one and working your way up for multiple (glued objects)....Can anyone say anything more? It's a looong way between 1 and 50 :confused: ......
Janet |
|
