Hey everybody. I don't make meshes (yet), generally I just repaint. I have (finally) noticed errors connected to using ims on repaint uploads to the DLS and LOD looks to be a requirement. I have old content changed from pm to im thankfully with PEV's converter. My question is: Is there or will there be a conversion utility out there that would take in my im and build the one two or three degraded (De-polyied) ims as output, needed to properly fill in my missing LODs if I choose to upgrade my single im content to future LOD im content? Any information about this would be appreciated. Seems it would be helpful and useful for keeping old stuff up to date and also a reason and an assist, to go to higher versions. If this is mentioned or exists somewhere else, can you please point me there. Thank you very much.
Is there an IM to LOD IMs Conversion Utility
- Thread starter jclinton
- Start date
Short answer....nope
shaneturner12
Tutorial Creator
From what I understand, LOD related things need to be done at the creation stage of an asset.
Shane
Shane
BuilderBob
New member
There are ways to reduce the number of polys in a mesh automatically but they produce extremely variable results, as you find when you try it in 3dsMax, GMax or Blender. I have included such calculations in my QuickShadows program but again the results can be OK down to very poor at times.
Thank you for the answer. I conclude that those who are creating LOD meshes have to build each individually a multiple of times. It is a problem waiting to be solved, a "resolution" waiting to be uncovered. I guess it isn't as easy as dividing all lines by two and re-drawing, in multiple iterations, until reaching size desired with the complexity self-reducing as any "line" reaches "single point." Just thinking. Thanks again.
Increasing the poly count by sub-dividng each face into four new ones could be done automatically, don't know why though. Going the other way would pose a difficult problem. Which faces do you combine into one without loosing the overall look of the object. Doing it manually lets you inject some intelligence into the selection. I suppose it could be possible to program enough rules into an automated process to replicate that but I get the feeling that paint would dry faster.
I think many people, including some creators, would like to have that available, I know I would.
I think many people, including some creators, would like to have that available, I know I would.
johnwhelan
Well-known member
Generally speaking the process is build the most complex detailed mesh then knock details off it so once you've done it a couple of times its not that much effort. Take the source rename it, knock bits off. That way you retain the same mapping which is a major consideration. The old .pm meshes were processed on the CPU, the newer .im meshes especially in TS2009 and later get sent down to the GPU which in general have more spare capacity so just switching to .im will normally increase the frame rates. To get an improvement with lod you need to drop 500 polys or more. A lot of the older assets aren't especially high poly. You can also use :cull to drop bits off such as rotating wheels, couplings.
To be honest by the time you've built the model many just want to put it out the door so the lod side of things doesn't happen as often as it should.
Cheerio John
Actually it's pretty easy to do automatically if you are using Blender (Max may well have something similar) in that you just copy the mesh, apply a decimate modifier and move a slider to reduce the poly count. The latest Blender versions also retain the mapping on the reduced meshes which makes life a lot easier. Unfortunately a lot of people here want to use GMax which can't do neat tricks like this.
Paul
Thank you for adding to the discussion. So far, if I'm going to start building the "im" for TRAINZ, I am looking for the tool that functionally can do what TRAINZ requires. It sounds like Blender so far. Would anyone like to comment a little farther. In thinking more about how this multiple LOD would be done I thought it would be a fractal thing. Instead of the fractal/defractal-izer running off into space. You ask it to check it's source and it's ending and then you say, now project the "halfway" in between. Then you say Oh very good, now project me halfway between each of those two halfways and reference the halfway point. Half again if you want more tweeners. So your ending source at first would be a single point, and your source would be your beginning wire cage model. The second step up from a point is a line the third step up into 3d is the three sided pyramid with base. If I reference my source in cube, rectangle, pyramid, sphere, ellipse or twists and turns of those objects the program would go halfway between all of those points and determine the source object and redraw it. The only difficulty I see is when and how to remove a facet. A rule could be a degree measurement of all connected planes to their one shared point if all degrees approach 180 then assume flat and remove point. Just thinking. "im" oh that "im." A new function in the next modelling software release. Just thinking. On a different note, I note TRAINZ calls me a TRAINZ Spotter. I think I may be a TRAINZ viewer and a Trainz pointer, don't know about that spot thing.
Thanks again everybody for commenting. So far I guess, I'm TRAINZ spotting Blender.
Thanks again everybody for commenting. So far I guess, I'm TRAINZ spotting Blender.