I bought the most excellent Aston Martin model which used to be on sale here prior to the "Great Purge Of Those Things Which May Infringe Copyright/Trademark". Now, this is a beautiful model, but Oh, Dear Lord, it is so poly heavy that I can hear Poser grunt with the strain when it's loaded. Most of the time I don't need such a detailed model so I'd like to know if there's an easy way (read - Totally Stupid With Modelling Apps) to cut down on the poly count. I ain't even bothered if the finished result is posable; I just need something I can use without having to fire up the Cray 2 every time I use it. :D TIA.

book, I'd like to know this as well. Thanks,

Most modellers have a "decimate" command, but it's usually brain-dead, meaning you end up removing the wrong polygons, causing the AM to look awful. The modeller probably kept them as low as possible, but that's the price of accurate detail.

Well, I'm not criticising the modeller or their skills but I'm certain that a hell of a lot could be removed without sacrificing too much detail. Example - Dudley's motors are usually a bit too low poly for my liking, this one is a little too high. What can I say, I'm a fussy arse. :) So, somewhere between the two would be ideal. I take your points, though, Mateo. Oh well, time to crank up the Cray. :)

Attached Link: http://www.vizup.com/

You can try Vizup. I've downloaded but never tried it. The results the show look very impressive. I don't see it mentioned but if I recall correctly it only works with VRML format models. I'm not sure whether that's input, output or both. That may have changed in the latest version. If not, you'll have to convert your .obj to .wrl and back after reduction. There should be a free converter on the web somewhere that will do that. ----------------- Some time later: ----------------- OK downloaded the latest version and did a very quick and dirty test. Poser I was reminded exports VMRL (.wrl). I exported the Casual Woman and loaded her into Vizup with no problem. I set the reduction for 15% and saved the file. You'll need to find a way to convert the reduced file to obj since Poser exports but doesn't doesn't read it--I used 3D Exploration. The file reduced file (both .wrl. and the .obj were larger than the originals??? The .obj imports into Poser fine only one material but the UVmapping is intact, maybe a conversion issue. Image: Left=Reduced - Right=original P4CW .obj file. The model is reduced in complexity as you possibly see--you can actually see it reduce inside Vizup. Loading both .obj files into UVMapper, I get: Original 15405 Facets/14447 Vertices Reduced 22910 Facets/12787 Vertices So I don't know what to say facets, which I assume equate to polygons went up but vertices went down roughly 15% not quite. I didn't read the Vizup help so I may be missing something. Also, the exported .obj probably has the normal lines which could be stripped for Poser. I'll have to play with this some more, it may have possibilities. Give it a try. OK down

Interesting... Well, models are converted to triangles so the polygon count is doubled before reducing. Basically it isn't worth doing on quad polygon models as that you end up with a lower quality version at a higher polygon count, but if you have a triangulated model, this thing is way cool. This would be great for reducing the polygon count of trangulated dynamic cloth models and cloth props. Love the info. Thanks for a cool thread, Anton

Thanks for the info. The model appears to be triangles, so I guess it will work just fine. And... Howdy, Anton, how you keeping, guy?

:) Howdy back. Great thanks. Summer is coming so..even better. :)

Glad it may help someone. I must have downloaded three or four versions of Vizup, meaning to try it 'one of these days,' so this was a good excuse. Since it's freeware and only about 800 Kb, can't hurt to give it a shot. Of course, somebody will be griping that Vick 3 don't work no more reduced to 1200 polygons :-)

Grab their sample models while you're there, a nice plane and a house. Before I saw them, I thought .wrl would only support crappy lorez models for online viewing. Live and learn. BTW, Vizup will let you select parts of the model and reduce them by different amounts or not all so you should hopefully be able to keep detail where you need it and squash the other parts.

Message edited on: 04/14/2005 17:00

Something to keep in mind as well.. if the model has any morphs at all, they will be broken when you downsize a mesh. Lightwave has a number of different poly reducing plugins, and each yields varying results, from useable, to very messy and worthless.

Download Melody from the Nvidia website. It reduces polygons for .obj, also creates normal maps based on the high-resolution mesh and fits them to the lower res product.

Message edited on: 04/14/2005 17:15

I have to echo Gareee here. People WILL be complaining about V3 after they reduce polygons. When you reduce, you remove points and/or create new ones. When you remove/create points on a mesh, the morphs will be useless (all of them!). Keep that in mind... The only way to avoid this would be to reduce polygons while retaining both the order and number of vertices - which means that the UV vertices will need to remain intact as well. But for anything else, no problems with polygon reduction.

Attached Link: http://www.mootools.com

I found an easy to use utility at mootools.com. The 3D Photo Browser comes with Polygoncruncher as well as saving between 6 or 7 common formats I've gone past the trial period period so I can't test it fully but the controls are clear and it keeps textures. It also seems to require more computer power than I can throw at it.

In addition to these issues, keep in mind clothing and props are designed to work with the mesh as it was build.. reducing the mesh density will most likely create a lot of popthrough when using clothing. For static building car or prop type meshes, it can be useful. For characters, be prepared to almost rebuild them from the base up again. I reduced the Daz Martian Tripod's mesh to about a 3rd of it's original mesh density, and there is very little noticable difference.. but I also spend the better part of a full week finding out what parts could be reduced, and how much they could be reduced, before they were no longer pleasing or useful.

I have tried about twenty different polygon reducers and untriangulators. I am in the process of writing a tutorial on polygon reduction, with links to all the free reducers and untriangulators, and a review of each. If you want to spend money and do a really good job, the best in my opinion are: For triangulated models, the best reducer is Rational Reducer. It has the best algorythms and keeps the best detail out of all the untriangulators I tried. It has preview functions, mesh overlays and other tools to help you compare the results, and you can fine tune percentages. To untriangulate a model, I think Cinema4D does the best job. You can set the evaluation angle, and also untriangulate only certain parts or areas. Both of these cost money though. If anyone else has any better results than these, please let me know so that I can include it in the tutorial.

Polygon cruncher for lightwave 3d is pretty good, and quemloss3 for lightwave is also excellent.

Victoria 3 can be reduced by about 80% and still look quite good. (I tried it yesterday with an old demo version of vizup). Accutrans and cinema 4d can save vrml files to use with vizup.

Spanx, babies! I be reviewing all this info tomorrow. Cheers, chaps!

Wow, take a nap and things start buzzing :-) Yes, morphs will definitely be hosed and there may be other side effects as well depending on your use. While Vizup does retain the model's UV mapping, the caution you that textures may acquire some artifacts due to the reduction. Regarding morphs, clothing, etc., I think if you morphed, dressed and posed a figure and then reduced it it should work. That's what people would normally use Posette or one of the LoRes Millennium figures for and I don't see a real advantage off the top of my head but I'm sure someone might have a reason to want to do it. Normally though, I think you'd want to stick to props. Going in the opposite direction, a long time ago someone (Mobeius I think), created a version of Posette with a hires chest--subdivided in Max IIRC). He also used the same technique on some of her morphs to make them compatible with the new mesh. Someone once started advertising an application to increase the resolution of figures through subdivision and said that they were going to have compatible morphs available. I don't know whatever happened to it. I suppose, in theory, you could do the same thing in reverse, creating morphs compatible with a reduced figure. The reducer application would, I assume, have to be capable of doing the reduction in the exact same way every time to make it work. I have no idea whether Vizup can do that. In a thread the other day, someone wished for on-the-fly reduction in Poser. People sort of scoffed at the utility of such a function. I think someone said that Firefly's micropoly SubD provided essentially the same thing. I seem to recall reading about some applications that implemented it--maybe even gaming engines. It would be interesting to have a figure automatically reduced when moved into the background. Vizup seemed to reduce the P4 figure essentially instantly--don't know how long V3 would take.

Reducing v3 in vizup is quite fast, I did it with an older demo version and it only took a few minutes to do it on a 1 gigahertz computer.

"...only took a few minutes to do..." That would probably rule put anything approaching real time, even on a current SOTA system I suppose. Be neat to have it happen as you zoomed the camera in and out. Probably have to do it in hardware. While we're at it, might as well throw in dynamic crowd generation with variations in gender, age etc. based on parameters you enter. Randomize the clothing too using something like the built in clothing on the old Poser figures. Let's see, 'Computer, give me 50 fit Swedish females, between the ages of 18 and 45. What's that? No, no need to bother with clothes.'

Um..... on second thoughts, make it between the ages of 18 and 19. And put swords in their hands. Yes, in the temple, off course. Damn computers don't know anything about art, you will have to refine your parameters properly. And the area render should automatically render the breast area for your thumbnails.

Got it. We'll be using mocap data from the Swedish Bikini Team jumping rope to feed the AI algorithms in the Dynamic Mammary Movement (DMM) ASIC. I think the Temperature Variant Nipple Displacement (TVND) routines are done. Make sure you have an empty PCI-X slot and a drool cup handy.