|
|
| Post Number: 21
|
carnage 
Group: Member *
Posts: 435
Joined: Apr. 2007
|
 |
Posted on: Sep. 07 2007,17:31 |
|
   |
ratty. i had another though at what might be causing the problem between the rocks
you seem to be displaying the image quite stretched over the model. perhaps this is causing the pixels in the cracks to be interpolations on the steep sides resulting in a near flat pixel
perhaps testing with a higher resolution image or re UV map the model to tile the texture more might show if this is the case
|
 |
|
|
| Post Number: 22
|
ratty redemption
"understood..."
Group: Member *
Posts: 1351
Joined: Sep. 2006
|
|
Posted on: Sep. 07 2007,18:12 |
|
 |
good idea carnage, I'll try that later tonight.
in the mean time, here's a couple of shots of the latest tests I've been doing. I've tweaked some of the rock's height and tilted in the inner rocks, but not the borders.
 
notes: there is an improvement imo, but now I've got a seam error with the normal shading along the tiling edge. it's not really evident in these two shots but I need to fix it somehow.
|
|
|
|
| Post Number: 23
|
ratty redemption
"understood..."
Group: Member *
Posts: 1351
Joined: Sep. 2006
|
|
Posted on: Sep. 27 2007,23:24 |
|
|
among other things I've spent weeks trying to fix the shading problem with my rock cracks. what ever I tried in blender or psp7 the results would end up the with my cracks being smoothed at the bottom 
so today I've done this simple test, using various v shaped planes to see if there was anything wrong with the normals, and although I am a noobie with renderbumpflat, I'm wondering if these results indicate the .ase model is not exporting properly?
notes: the planes are drawn in blender with their vertex and face normals. I've also tried various uv mapping and the results after renderbumpflat seemed the same.
I think I'll also try posting this in goofos's d3w .ase export topic.
Edited by ratty redemption on Sep. 27 2007,23:34
|
|
|
|
| Post Number: 24
|
kat
Admin
Group: Admin
Posts: 1325
Joined: May 2006
|
|
Posted on: Sep. 28 2007,01:14 |
|
|
Eh?! That's not right, it shouldn't be doing that at all. It won't be an axis orientation problem would it? How big is the offending mesh? What happens if you have an intersection object over those areas? Might be due to not having enough vertexes to render against?
--------------
Admin
|
|
|
|
| Post Number: 25
|
ratty redemption
"understood..."
Group: Member *
Posts: 1351
Joined: Sep. 2006
|
|
Posted on: Sep. 29 2007,18:31 |
|
|
thanks a lot kat, I'll do some tests tonight and report back.
|
|
|
|
| Post Number: 26
|
ratty redemption
"understood..."
Group: Member *
Posts: 1351
Joined: Sep. 2006
|
|
Posted on: Sep. 30 2007,01:17 |
|
|
(kat @ Sep. 28 2007,01:14)
QUOTE It won't be an axis orientation problem would it?
goofos .ase > d3 renderbumpflat seems to need a -y orientation, I tested the others and they didn't work any better.
QUOTE How big is the offending mesh?
3200 blender units. originally the mesh was x100 smaller but that didn't look any different after renderbumpflat.
QUOTE What happens if you have an intersection object over those areas?
not sure what you mean? I do have a simple plane just behind them.
QUOTE Might be due to not having enough vertexes to render against?
I just tested this as well using a subsurf modifier and exporting 32,700 faces, still no difference.
I could email you or ul the .blend file if it would help you figure this out?
also goofos doesn't appear to be posting in this dw3 topics atm
|
|
|
|
| Post Number: 27
|
ratty redemption
"understood..."
Group: Member *
Posts: 1351
Joined: Sep. 2006
|
|
Posted on: Sep. 30 2007,01:29 |
|
|
just tried exporting with blender's .obj export script and d3 wouldn't load it, and when I try the .lwo script blender reports a python error, although I never got python installed with blender.
|
|
|
|
| Post Number: 28
|
kat
Admin
Group: Admin
Posts: 1325
Joined: May 2006
|
|
Posted on: Sep. 30 2007,03:15 |
|
|
Try grabbing xNormal and using that to render the normal map test.
http://www.xnormal.net/Downloads.aspx
--------------
Admin
|
|
|
|
| Post Number: 29
|
ratty redemption
"understood..."
Group: Member *
Posts: 1351
Joined: Sep. 2006
|
|
Posted on: Sep. 30 2007,13:58 |
|
|
thanks kat, will try that this afternoon.
ot: did you get my pm yesterday?
|
|
|
|
| Post Number: 30
|
ratty redemption
"understood..."
Group: Member *
Posts: 1351
Joined: Sep. 2006
|
|
Posted on: Oct. 19 2007,00:59 |
|
|
I haven't yet tried the xnormal renderer as kat suggested but I have made an improvement as I'm hoping you guys will agree.
to speed up my tests, I've been working on just one of the rocks as seen below. I've also been testing in d3 with a simple box map, without any displacement on the map's polys.
this is a solid view of the rock in blender. this has nearly 46,000 polys after triangulating.
notes: this rock is actually wedge shaped, although the normals don't really show that after rendering in d3 or when seen in game, so not sure what I can do to fix that.
here is the low poly mesh which has 511 faces including several quads.
notes: the yellow highlighted faces belong to my vertex group used in the smooth modifier. I've done tons of testing with various combinations of vertex groups smoothing and displacing the subsurf mesh (which was set to 3 times simple subdivision)
this is a sold view of just the subsurf and smooth modifiers applied.
notes: the reason for using a vertex group to control the smoothing was I didn't want the jagged edges of the cracks smoothed.
this shows the side and crack displacements, again using vertex groups mapped to their own modifiers.
notes: blender will run faster if not all the modifiers are enabled at the same time, as well as it being easier to tweak them like this. modifiers sharing the same vertex groups will mix their results.
here is the d3 renderbumpflat normals of the low poly mesh without any of the modifiers enabled.
notes: I was able to get sharp rendering of the crack normals, because I modeled a steep bevel at the bottom of the cracks as suggested by tintin from the d3 phobos team.
this is the high poly normals with all the modifiers enabled.
notes: because I was using the more organic uvw unwrap in blender, the smaller and more dense the polys of the base mesh, the more detail appears in the subsurf displacement. this might not always be desired, so if I had of used a planar project from view method, the displacement size would be more uniform.
this next shot is of crazybump's render displacement map for these normals. which I then used in psp7 to add the thin black lines of the cracks to my d3 rock's height (displacement) and diffuse maps.
notes: increasing the values of crazybump's fake occlusion and enhance detail sliders allowed more of a white background to be rendered, although it would of been possible to have a more grey looking displacement map if needed.
and here is a shot of the psp7 layers showing my previously rendered d3 heightmap and the one from crazybump.
like an inverted alpha channel, the multiply layer mode blends in only the darker pixels, which is why I wanted the crazybump render to have that white background.
this shot is from d3 showing my box test map with just the single rock normals and a white diffuse map.
notes: this isn't the intended scale as that rock will be a lot smaller in the final version, but it was good for seeing a high quality zoomed in render of the normals.
this one shows the same camera angles but now with my d3 heightmap enabled in my material shader.
notes: in this case the heightmap is the intended scale of the final, although I could of scaled it in the material file for these tests if needed. there is now a slight shading error again at the bottom of the cracks but not as much as before I beveled the crack polys. this is why I ended up with editing the height and diffuse maps to help minimize the error as much as possible.
and now we see the diffuse and specular also enabled in the material shader.
notes: in the final version I will use vertex painting on the map terrain model to blend in at least two variations of the diffuse and probably different height maps, but I didn't do that for this test map.
and lastly for now, a different camera angle showing a close up of the rock cracks.
notes: during the last few weeks I've also done some quick tests using blender's sculpt tool to model the high poly normal mesh, but I wanted to perfect the low poly, displace modifier method as much as possible, partly as the pc I'm using struggles with high poly sculpting.
Edited by ratty redemption on Oct. 19 2007,01:05
|
|
|
|
|
|
|
