Hi there! You are currently browsing as a guest. Why not create an account? Then you get less ads, can thank creators, post feedback, keep a list of your favourites, and more!
Test Subject
Original Poster
#1 Old 8th Feb 2006 at 7:18 PM Last edited by demon432 : 14th Feb 2006 at 5:35 PM. Reason: Update version.
Default GMDC TOOL v1.0a for solve problems with MilkShape meshes.
If You has some problems with meshes, created/edited into MilkShape 1.7.7 and later, exported by Wes_H's UniMeshTool v4.0x, You may try this tool before start lenghty iterations with welding and smoothing into MilkShape.
This tool allow automatic smart adjusting all seams normals without corrupting texture maps and try to solve problems if importing morps from some original meshes completed incorrectly.
Look on attached demo image for details.
Please, post here all questions, notes and warnings.
Screenshots
Attached files:
File Type: zip  GmdcTool.zip (72.5 KB, 175 downloads)
Advertisement
Lab Assistant
#2 Old 10th Feb 2006 at 2:06 AM
Thanks so much for this!! Should be a handy little tool (basically is saving me for like 10 of my meshes!!!) but anyways I’m a bit lost on how it works... could you explain more maybe say something on how you open it. Its probably easy but I’m just too stupid, but what ever. Thanks so much!



Andrew Shields My personal portfolio

Andrew Shields Facebook Page <3
Test Subject
Original Poster
#3 Old 10th Feb 2006 at 7:40 PM
Now version 1.0a uploaded - this windowed version more simply to use and has more functionality.
Lab Assistant
#4 Old 10th Feb 2006 at 9:28 PM
Well I went ahead and tried it, but with no luck. I think I might be unclear on the instructions. What I did was
I imported my. SimPE file and then export it again as.5gd file. Then I opened GMDC tool and loaded the file with the BW adjust. And then after that it didn't do anything but loaded a screen that stated some info on the file. I assumed it did its thing and I exited. Then I replaced the old GMCD with the new one and ran bodyshop. But to find that nothing had changed. Same old hands bleeding through the hips.



Andrew Shields My personal portfolio

Andrew Shields Facebook Page <3
Test Subject
Original Poster
#5 Old 10th Feb 2006 at 10:11 PM Last edited by demon432 : 10th Feb 2006 at 10:22 PM.
Very strange... Really requested operation immediately executed after file selection, and selected file rewrited with new information. I check this tool imeediately before uploading, and take screenshots from this tests. Possible reason of this situation - package file was not saved from SimPE after GMDC replacing was made...
For test of MMap and BW was completed and GMDC data updated, look on process statistic after first operation and repeat this operation on same file. If first operaton finished with non-zero references number for MMap or non-zero weights for BW, then incorrect data was really found and was corrected. Then repeated operatoin on same file must show zero refs or weights values.
(This test are not applicable for smoothing - this operation do not check unaligned normals, and always align all seamed vertexes! As a result, smoother will always print non-zero number of aligned normals!).
If not - this is abnormal situation: file was not succesfully updated, but this situation can be detected by program and error message on status window can be printed!
Please try all this steps and be sure what package file really saved from SimPE.
Also You may check this soft with original Maxis meshes (on screenshots original afBodySwimwear). If original file converted OK (and that can be!), but Your mesh continue has a artefacts, please mail me this strange .5gd file on [email protected] for I can check this situation and update algorithm.
Lab Assistant
#6 Old 11th Feb 2006 at 3:10 PM
I just want to so thank you so much!!!

You have no idea how big this is for me and LOT of other people. I don’t know why there’s not like 3 pages of THANK YOU SOOO Much on here. But I think your amazing and so is your program!!

As for the mesh I don’t know why it didn't work, it’s just a strange mesh I remade it and now it works good.



Andrew Shields My personal portfolio

Andrew Shields Facebook Page <3
Alchemist
#7 Old 13th Feb 2006 at 6:58 AM
Quote: Originally posted by demon432
If You has some problems with meshes, created/edited into MilkShape 1.7.7 and later, exported by Wes_H's UniMeshTool v4.0x, You may try this tool before start nighmare with welding and smoothing into MilkShape.
This tool allow automatic smart adjusting all seams normals without corrupting texture maps, solve problems with incorrect importing morps from some original meshes and allow automatic restore 100% summary bone weights assignments for underweighted vertexes - this remove visual artefacts like faces collisions on some exported meshes.
Look on attached demo image for details.
Please, post here all questions, notes and warnings.


Well, I sat and waited and though about what to do. Certainly, good meshing tools are needed for Sims2 modding. But I have decided in this case, silence is not golden when someone is promoting their own efforts by spreading incorrect information.

First, as I detailed to you in other messages, both private and posted, the morphs are not incorrectly imported on dual morph objects. You have based your conclusion on a cursory examination of the order of the flag bytes in the output MorphMap section. As I explained before, your analysis did not go far enough to determine that the morph data that was once in mixed sections in the input gmdc has been relocated during import in a fashion that segregates morph 1 data to one section, and morph 2 data to another. Thus, references that were once to section 2 have been changed to section 1.

The game rendering software can handle the data arranged either way. What is telling is that the rendered morphs are correct. A fat morph, for either a single or dual morph entity is rendered identical between the original mesh and an imported/exported mesh. I have retested to make sure my process is correct. You are wrong in your analysis, and should say so.

The automatic bone re-weighting you supply is also available in one of the add-on plugins I have written, independent of your posting. I have elected to not implement that automatically in the importer because I believe that an import should be as true to the data sourceas possible, within the constraints of MilkShape, which allows me only a single byte for each weight. The error is in the floating point to integer conversions. The way I have it arranged, you can elect to reweight underweight vertices (which can be automatically selected by another add-on plugin tool) manually, the way you want them, or automatically. Far better, in my opinion, to have the choice, and to be able to elect to make it without having to click on boxes in a dialog box for every import and export. I dislike even the dialog box on the importer for choosing "Create Blend Groups?", but it is a major decision point in the mesh importing process.

I am curious if your tool reweights vertices with 4 bone assignments? In all cases, these are always underweighted, as the weight for the 4th assignment is the difference between the sum of the three weights and 1.0 (100%). Adding additional weighting to the three weights to fully weight these vertices will in effect make the 4th assignment unutilized, which itself will cause incorrect rendering.

I think that, given you have studied the GMDC file format well enough to write the program that you did, you could find some more constructive ways to use your talent than to recklessly throw bombs at other people's work. Both Blender and Wingz are open-source programs that a lot of people here would like to be able to use for Sims2 meshing. Or you could write your own set of plugins for MilkShape and post them as an alternative to mine. I don't care.. I am not trying to 'fence off' an area here for my own, but I do object when you promote your work by incorrectly disparaging mine.

Good luck on your endeavors, and you can thank all the people that have been here for a while and spent lots of time deciphering the GMDC format and posting our results on the Wiki site here for you to study and learn from.

<* Wes *>

If you like to say what you think, be sure you know which to do first.
Test Subject
Original Poster
#8 Old 13th Feb 2006 at 9:52 AM
wes_h
This tool are not soft-diversion against Your UniMeshToll. This is simply additional workaround before You can release new updates of Your beautiful set. History of this tool very simply - since autumn 2004 I want tools, which can help me for editing morph datas for TS2 meshes. Now You build v4.0X UniMeshTool set, which allow to me realize my mind. I found only one trouble - importing original afBodySwimwear_tslocator_gmdc mesh morphing IS INCORRECT with v4.04 . I post to You about this. After You ignoring this and saying, what v4.04 soupport morph ordering fully correct (may be, but my troubles with original afBodySwimwear_tslocator_gmdc afbodyswimwear stay alive) - I have no chance to fast solving this trouble than build my small utility. Now Your UniMeshTool v4.04 with my workaround tool allow to me realise all my plans, which I want more than year.

About alternatives chiose - my tool is not mandatory to use. It is only try to fast workaround for some troubles, which I look, before You can solve this on more professional level. This tool help to me for release my own mesh projects now and may be can help to other for fast altering of original meshes. All, who need, may try this tool and check results - if this results are suit, this can save some time, if result tottally bad - simply discard this tool it and continue with manual editind. This tool are not mandatory and is not claim to professional tool level - this is only workaround for someone like me.

About morph mapping now: please, if You can, compare of imported (by v4.04) fatbot morphs of originals afBodySwimwear_tslocator_gmdc and afBodyNaked_tslocator_gmdc meshes. They really different. Why? I do not know, but my MorphMap workaround can solve this to me.

I do not want to compromise of Yors work, I simply try to help to mesh-alterers.
If i wrong, I'm sorry. If You want, I may remove software and close thread.

demon432
Admin of Randomness
retired moderator
#9 Old 13th Feb 2006 at 7:26 PM Last edited by tiggerypum : 13th Feb 2006 at 7:36 PM.
Demon, I am interested in the seam issue (as I'm sure others are) but I am confused about the morphing issue - is that an 'automatic' fix, or is that something we can choose or not with your tool? Have you created meshes with new parts on them and then tried your 'fix'? Seeing it work with a simple unmodifed mesh is not really enough for me to understand if it will work for what I'll be doing, which is adding and deleting vertices to the mesh.

Also I'll note that a few of our experienced meshers have used the unimesh plugin to create meshes that are behaving as they expect for the fat and pregnancy morphs -- you are the only person who says the morph isn't right - it seems that Wes' new gdmcs are behaving as the meshers expect them to. People have been making fat and pregnant meshes with Wes' plugins (bodychop before unimesh) for some time. As bodychop did have some frustrating limitations, I am thrilled that Wes has gone and worked on something totally new to try and approach those limitations - even if we're now waiting on some more Milkshape fixes to get everything working as well as we'd like.

Wes also did put out an underweight bone locator and even an auto fixer for those of us who want it, so what I am interested in is the seam adjusting alone - I mean at least a person could try that and see if it works, but I would not want any other things adjusted, or it gets too confusing for me as a user to know which tool is giving me problems.

So can I adjust only the seams with this tool, without it changing anything else? At least until we get some more of the Milkshape issues resolved, this might be an interesting work around.
Scholar
#10 Old 13th Feb 2006 at 10:10 PM
I think maybe people have gotton too accustomed to trying to "weld" up the seams, from using the earlier BodyChop plugins and/or the .smd exporter/importer.

If you simply un-check "autosmooth" on the groups page before importing the mesh, it's not necessary to weld those seams at all, and the original Maxis normals will be preserved - meaning no visible seams (with no extra work at all) - the seams will look exactly as they did on the original Maxis mesh.
Test Subject
Original Poster
#11 Old 13th Feb 2006 at 10:27 PM Last edited by demon432 : 13th Feb 2006 at 10:41 PM.
tiggerypum
MMap tool I added for solve one sort of problems, which I found when try to import original afBodySwimwear_tslocator_gmdc from Maxis packages. For this file Fatbot set was imported incorrectly, and I try to solwe this problem.
For best viewing of this issue simply try to import original afBodySwimwear_tslocator_gmdc and afBodyNaked_tslocator_gmdc, and compare tatbot meshes. They can be viewed as identical, but really not. MMap tool can help me for normal importing afBodySwimwear_tslocator_gmdc.
Alchemist
#12 Old 14th Feb 2006 at 12:03 AM
I don't want to fuel the fire here any more.

I am not angry, nor am I concerned about 'competition'. But I cannot find the problem that Demon432 says is there. It WAS there in BodyChop, and was there until V4.03 in UniMesh. Morague was the one that pointed the problem out. I spent a bunch of time comparing models in the game (UniMesh morphs versus Maxis originals) because of Demon432's missives, and cannot see where the CURRENT plugins are wrong. All I can see is that he has noticed that the data is rearranged when exported, and that some vertices formerly flagged as a '2' are now '1', and some formerly '1' are flagged '2'. However, if you look carefully at the data values for those vertices, you will find that they were swapped, too, making the whole package consistent.

Certainly, some allowance for him should be granted as far as English as a second language, but calling the mesh editing process using my plugins a 'nightmare' is inflammatory rhetoric, even if unintentional.

I do think Demon432 has some good ideas, and he clearly has worked hard at understanding the GMDC data format. I think if he applies those skills at solving some real shortcomings, he can produce some useful things no one has yet tackled. Things like a new manual editor or a new selection editor, or a visual normal adjuster tool, or a good mesh previewer tool (within MilkShape) would all help the modders here a lot.

<* Wes *>

If you like to say what you think, be sure you know which to do first.
Alchemist
#13 Old 14th Feb 2006 at 12:24 AM
Quote: Originally posted by Dr Pixel
I think maybe people have gotton too accustomed to trying to "weld" up the seams, from using the earlier BodyChop plugins and/or the .smd exporter/importer.

If you simply un-check "autosmooth" on the groups page before importing the mesh, it's not necessary to weld those seams at all, and the original Maxis normals will be preserved - meaning no visible seams (with no extra work at all) - the seams will look exactly as they did on the original Maxis mesh.


You are dead right, Dr. Pixel. I think the arms actually look smoother on some of my exports than they do on the original Maxis meshes.

A couple of notes:

In older versions of MilkShape, when the model was exported to a plugin, all duplicate vertices and duplicate normals were combined by MilkShape before the plugin ever saw them.

Likewise, when a plugin passed a model to MilkShape, duplicate vertices and normals were combined before displaying the imported model.

So the original Maxis seams and normals were already tinkered with coming and going. This works OK on plain models like the xxBodyNaked and afBodySwimsuit, but a lot of the clothing meshes have seams intentionally created to support seperate normals at specific locations such as the skirt hemlines where the faces turn 'under' the shirt, skirt, shorts, etc., allowing seperate inside and outside rendering to be done. When welded, these locations collapse into a single normal (1 vertice = 1 normal in Maxis GMDC format), making the transition very sharp (>270 degrees) and causing unwanted shadowing issues.

Other seams (rows of duplicated vertices) are necessary because that is where the front and back halves of the texture map meet. Again, 1 vertice = 1 UV mapping pair, so the only way to have a texture map in two pieces is to have two seams arranged in a fashion where the two halves abut tightly.

The number of extra vertices taken up by the seams is small, less than 10% of the models I have looked at. Efforts at reducing vertex counts this small will have no noticeable or measurable effect on game performance. Reworking the mesh to eliminate a single seam by total UV remapping is a pretty steep price to pay for a 5% vertex reduction (and no face count reduction), and now there is still a seam that needs to go somewhere. At least with two seams the can fall along the locations where clothing designs normally have seams anyway.

<* Wes *>

If you like to say what you think, be sure you know which to do first.
Admin of Randomness
retired moderator
#14 Old 14th Feb 2006 at 12:25 AM
Thank you Dr. Pixel. So basically, if we do things properly to begin with, we should never see seams as shown in the first screenshot. Well, that's a relief, as that seemed to be the only parts of the unimesh discussion I was still confused about!
Alchemist
#15 Old 14th Feb 2006 at 12:46 AM
Quote: Originally posted by tiggerypum
Thank you Dr. Pixel. So basically, if we do things properly to begin with, we should never see seams as shown in the first screenshot. Well, that's a relief, as that seemed to be the only parts of the unimesh discussion I was still confused about!


The first screenshot shows a pattern I only ever saw when I welded the seams. Due to the way the UV map is laid out in front and back halves, without duplicating vertices (which welding combines) at the juncture between the halves of the UV map, there will be a gap between the halves equal to one triangle strip in width.

A single vertex is limited by the data format to one UV coordinate. In order to have both halves 'meet', you need two colocated vertices, one with the front half edge UV, and another with the corresponding back half edge UV mapped to it.

Following Dr. Pixel's prescription for smoother models is good policy.

The artifacts in the remaining screenshots are from under/over weighted joints.

<* Wes *>

If you like to say what you think, be sure you know which to do first.
Lab Assistant
#16 Old 15th Feb 2006 at 2:14 PM
well wether or not this tool is right, it has helped me a lot. it has fixed almost every mesh that i have something wrong with. the onyl mesh iv had problems with is attached. for some reason i keep running it through the tool and it never seems to save its changes. could you please take a look at it and tell me what i can do to fix it?
Attached files:
File Type: zip  Extractedmesh.zip (356.0 KB, 13 downloads)



Andrew Shields My personal portfolio

Andrew Shields Facebook Page <3
Lab Assistant
#17 Old 15th Feb 2006 at 2:46 PM
Just a note on the fire in this thread.

Wes's new plug-ins are amazing. with them we are now able to do the, at one point, impossible. Although when you add new parts to the mesh there are work arounds to getting the shading just right. But from my experiences, a mesh is never truly finished and once I have the new objects merged with the body that’s when I see room for improvment. And at this point we can’t smooth all anymore unless we want seems. This is were the tool comes in. we can adjust things in milkshape export using wes_h plug-ins and then from there use the tool to fix minor shading errors.

I’m not going to lie I haven’t read all that’s posted here so I might be making a complete ass of myself right now but looking back at problems with the plug-ins I was having involving a dress, wes you told me there was basically no hope when I did what you said and it didn't work. I'm looking at the finished product and its looking amazing thanks to you AND demon482.



Andrew Shields My personal portfolio

Andrew Shields Facebook Page <3
Alchemist
#18 Old 16th Feb 2006 at 5:46 AM
Quote: Originally posted by Forwardmotionis
I’m not going to lie I haven’t read all that’s posted here so I might be making a complete ass of myself right now but looking back at problems with the plug-ins I was having involving a dress, wes you told me there was basically no hope when I did what you said and it didn't work. I'm looking at the finished product and its looking amazing thanks to you AND demon482.


I cannot say that his tool has no merit, as I do not know what he is doing with the normals, and there is certainly room for improvement there. The plugins can only deal with the data that is passed to them, and MilkShape has no native functions that deal with modifying normals, other than setting up smoothing groups and using the smoothing functions. Some of the high dollar mesh editors allow normals to be changed, but not MilkShape.

Now I have tried to build a plugin that worked similar to what he described he was doing, but I never released it because I didn't find it made anything better. Now it is very possible that the reason I didn't find it made anything better was because I didn't apply the right approach, and he has. However, I stand behind my observations about welding all the seams, and you need only review Warlokk's posts in the UniMesh thread to see what happened with welding, and how he left portions unwelded and cured the shadowing.

What I was protesting was several problems he cites that I cannot verify exist in the plugins, specifically the morphs being incorrect... my older plugins had this problem, but that was corrected in V4.03.

The automatic bone reweighting is also available as one of my milkshape plugins, and I think I recall some of your issues looked like under/over weighted bone assignments. But anything good it did to the normals at the seams would be an improvement over what you can get from MilkShape, as my plugins are merely a 'translator' between MilkShape and SimPE... they pass poor models the same way that they pass good ones. It's part of the challenge to the artist to learn how to make a model look good.

<* Wes *>

If you like to say what you think, be sure you know which to do first.
Back to top