[Mark Bell]

I've had this issue in the past and it's back again. I've downloaded a 3D model of a Pink Trumpet tree which is in bloom so there are some green leaves with the remainder in flower. The material editor shows each material correctly and maps out the unused portion leaving the correct form and image of a leaf or flower. The problem is when assigning the materials to the model. I've gone through the various UV mapping options and none seem to display the leaf or flowers correctly. The model appears okay at a distance but when zooming in closer it's clear most of the polygons look strange and haven't applied the material UV's correctly.

I've uploaded the MXS file of the tree which has the materials embedded if anyone wishes to have a look and see what I'm doing wrong.
.
https://app.box.com/s/uvgv5o852gom2xbt8akrwh7ilqxs8zfz
.


Thanks in advance.

[Andreas Hopf]

When you import the downloaded OBJ file of the tree, check if the UVs look correct in textured decal mode. If they don't, there's nothing you can do within MS.

[Forester]

I'm with Andreas on this one. The first thing to check is to look at the UV mapping of one of the textures in a conventional 3d model-building program to determine whether or not the model-builder built correct UV maps on the model itself. But, you've probably tried this, yes? Because you provided us with only the *.mxs, none of us can determine whether the model's UV maps are correct. Any of us would have to first look at the original model.

A thing I have seen in the past is that a model-builder correctly constructs UV maps for a very high rez model, but then makes a low poly version, without building a new set of UV maps for the lower rez version. Since a UV map is binding pixels to vertices, when a bunch of vertices are deleted, the UV map gets wonky and never applies well to the lower rez model. This is real common for models of plants that are made with multiple level-of-detail ("LOD") generation. The lower LOD's may look fine at medium distance, but they weren't intended for closeup views. Too big a mismatch between the intended UV mapping, and the actual UV mapping.

Alternatively, your tree model may simply be a low resolution model that has too few polygons for any decent application of flower and leaf textures. Again, we'd have to start by looking at the original model, and its original UV maps. But, since you are an experienced model-builder, you've probably already done this.

Mark, these are my immediate suspicions, ... because I've never had any problems with Maxwell mis-applying a texture to a correctly UV mapped model. The only problems I've ever experienced with any of the versions of Maxwell have been with poorly made models.

If you can post the original model, without getting into copyright trouble, I'd be happy to take a look at its UV maps and polygonal structure in Maya for you. Just to rule out the model-being the cause of your problems. Or, if you're concerned with copyright trouble, I can give you my email or a dropbox location to temporarily transfer the original model, so that I could look at it for you. (I searched for a 3d model of that particular tree to try to obtain it myself, but its not anything I could find. Although I found at least five nice equivalent trees, including the very nice, FREE set of Xfrog versions on CGTrader. I looked closely at those latter, and they map just fine in Maxwell.)

[Mark Bell]

Hi Andreas, Forester,

Thanks for your responses. I think you're both right here. Normally, Studio is pretty good at mapping UV's and it only takes a few clicks to get things right. I've been working in Texture Decal mode to see the UV maps and it's all random. Because Studio doesn't support SKP or Vray direct import format the tree had to be converted to another format then brought into Studio. This is where all the mapping likely got broken. In some instances in the past they do import with UV's Locked and it worked. Unlock them and it all falls apart. In the absence of a good objects library from NL, we have to accept whatever we can find. It would be a good addition if MR could have some extra import formats to help here. In this instance, the correct species (needed for a DA to Planning we're working on), needs to be used and the one below is what was found (450K polys in SKP and Vray formats). The trunk was quick to assign a displacement material but the leaves and flowers are the problem. The render supplied shows everything mapping correctly.

Here's the web site and download page (free) - https://www.facebook.com/3dsmaxfiles/po ... 411592255/

Thanks again.

[Forester]

Ok, I will attempt to download that one (some difficulty there) and see if I can "help" the model a little. Oh, OK, I got it......

Meanwhile, Mark, would any of the tree in this free Xfrog version work for you ?
https://www.cgtrader.com/free-3d-models ... euia-rosea

[Forester]

Oh, that mapping is horrible !!!!! (Whoops - that is just looking at the tree trunk. The flowers and leaves seem to be OK. Although certainly not ideal.)

I think I can help fix the UV mapping for that model, though. Give me an hour.

DISREGARD the above - see below.

[Forester]

Mark, try this version of your tree.

https://www.dropbox.com/scl/fi/3l5co0z7 ... a2yna&dl=0

For this version, I did not remap the original. I simply got hold of the Simplot SKP convertor for Maya, brought that tree into Maya using that particular convertor, re-built the materials and applied them, and then took a look at the results. They look like this.



The UV mapping of everything is correct, as nearly as I can tell. So, the tree in that Dropbox link above is an *.fbx version that will import into Maxwell, and when you apply the materials, it should look the same as in these pictures.

With regard to re-building this tree in some way, frankly if you asked me to do it, I would run to Speedtree and see if I couldn't have a decent version in about 2-3 hours. When the Simplot convertor is finished with its work (making a conversion that can be imported into Maya), every vertex that should be a single vertex, actually is from three to six co-incident ("fully overlapping") vertices. So, this tree would be a nightmare to try to clean up and make better. Also, while the flowers and leaves are mapped within a proper UV space, the trunk is not. It is all over the place - most polys of which lie way outside any acceptable UV space. When I tried to clean it up, Maya stalled out on me two or three times. Frankly, I could build a much better tree (as far as the flowers and leaves go) faster than I could clean up this one.

You should note that while Flower #2 (the paler, white one) is arranged in a circle (see the renders above), the more dramatic pink flower is just flat panels that look like a bunch of flatpanels in closeup. (This is why I could build you a similiar, but better flowered tree in Speedtree.) Also, the leaves and flowers on this tree intersect too frequently to be really good for a closeup.

I'm trying to finish a project today and tomorrow, but if I get to where I need a break, I might try to whip out a similiar tree for you in Speedtree.

[Forester]

Just for comparison, here is a render of XFrog's Tree #1.



Its materials are two-sided, so its a little bit more work.

[Mark Bell]

Hi Forester,

Thank you tremendously for all this~! It worked, (the remapped version of the tree you uploaded) and everything now maps and renders correctly. So thanks again for your help and quick response on this. My shout next time

[Forester]

OK, good to know. We'll leave this issue there, then.

Happy Rendering!