MapSplitter: *.gen.cfg / map sizes

vvHedgehog

20-06-2006 12:45:08

Hello everyone!
first, sorry for my English! :?

I gonna split my map through MapSplitter.
Here is my landscape4.gen.cfg :
GroupName=PLSM2

LandScapeFileName=landscape4
TextureStretchFactor=2
LandScapeExtension=raw
RawWidth=1351
RawHeight=2051

OutDirectory=\Work\3d\tests\2

Data2DFormat=HeightField

PageSize=513
TileSize=65
ColorMapName=landscape4_texture.jpg
ColorMapSplit=yes
HeightMap=yes

MiniMap=yes
MiniMapWidth=128
MiniMapHeight=128

Source heightmap is raw-file 1351x2051, source texture 2702x4103.
I wanted to get 3x5 pages of 513x513 heightmaps and 3x5 pages of 1024x1024 textures,
but I got 3x5 pages of 513x513 heightmaps (good :) ) and 3x5 pages of 512x512 textures, besides textures was split using first 1536x2560 pix instead of using whole picture.

Here is my summary questions:
1) am I doing something wrong? i is possible to use such arbitrary map and tex sizes?
2) option "TextureStretchFactor" is valid in this cfg? (I tried to change it but got no difference)
3) what is MapSpliter mechanism (algorithm) when working with textures?

tuan kuranes

20-06-2006 14:51:13

CVS or SDK ?

1) seems ok.
2) yes. it's valid in both config files.
3) it splits in pagesize*texturestretchfactor (1025) then scale it to nearest power of 2 (1024)

can you try wihtout heightmap splitting ? (HeightMap=no )

vvHedgehog

20-06-2006 15:42:20

CVS or SDK ?
SDK
Plugin_PagingLandScapeSceneManager2.dll date 15.05.2006
MapSplitter.exe date 15.05.2006

1) seems ok.
2) yes. it's valid in both config files.
3) it splits in pagesize*texturestretchfactor (1025) then scale it to nearest power of 2 (1024)

good

can you try wihtout heightmap splitting ? (HeightMap=no )
Result: no heightmaps and the same textures

also if I set
NormalMap=yes

MapSplitter crashes

....... creating heightmas and textures......
created : landscape4.4.2.png
normal Map Calc
Width : 1537 Height : 2561 Bpp : 3
landscape4.Nsmall.png
created : landscape4.Nsmall.0.0.png
created : landscape4.Nsmall.0.1.png
created : landscape4.Nsmall.0.2.png
created : landscape4.Nsmall.1.0.png
created : landscape4.Nsmall.1.1.png
created : landscape4.Nsmall.1.2.png
created : landscape4.Nsmall.2.0.png
created : landscape4.Nsmall.2.1.png
created : landscape4.Nsmall.2.2.png
created : landscape4.Nsmall.3.0.png
created : landscape4.Nsmall.3.1.png
created : landscape4.Nsmall.3.2.png
created : landscape4.Nsmall.4.0.png
created : landscape4.Nsmall.4.1.png
created : landscape4.Nsmall.4.2.png
stretchFactor : 2
Width : 1537 Height : 2561 Bpp : 3
landscape4.N.png
created : landscape4.N.0.0.png
created : landscape4.N.0.1.png
created : landscape4.N.0.2.png
created : landscape4.N.1.0.png
created : landscape4.N.1.1.png
created : landscape4.N.1.2.png
.....crashed here exception code 0xC0000005

tuan kuranes

20-06-2006 15:49:19

can you mail me cfg + map files so that I can reproduce the bug here ?

vvHedgehog

20-06-2006 15:55:25

If I use source texture 3072x5120 =1024*3x1024*5 (which I got from initial texture 2702x4103 by addind white extra pixels by Microsoft Paint)
then results what I need : 3x5 pages of 1024x1024 textures

but the point is to make all proccess of preparing landscape aotomatical (and i have no programm tools to edit jpg)
I thought MapSplitter add extra pixel for me as it does when processing heightmaps.

p.s.: OS : WinXP SP2, OGRE ver that cames with PLSM2 SDK

tuan kuranes

20-06-2006 16:03:49

so It may be a bug in the "scale it to nearest power of 2 (1024) " part of mapslitter...
I'll try with a 2702x4103 texture
I thought MapSplitter add extra pixel for me as it does when processing heightmaps.
I thought too. Maybe it does, but is fooled by a bad "nearest power of 2" computation.

vvHedgehog

20-06-2006 16:05:50

arhived data soursed weith 7.7 Mb is this acceptable? or i should split it?

tuan kuranes

20-06-2006 16:06:35

no problem.

vvHedgehog

20-06-2006 16:09:49

so It may be a bug in the "scale it to nearest power of 2 (1024) " part of mapslitter...
I'll try with a 2702x4103 texture
I thought too. Maybe it does, but is fooled by a bad "nearest power of 2"
computation.

maybe should use not "nearest power of 2" but "nearest power of 2 not smaller than actual size"

vvHedgehog

21-06-2006 07:42:46

can`t send email it`s too big. I`l try to split it.

vvHedgehog

23-06-2006 11:30:52

tuan, have you got the email?
is there a new build of SDK in a future?

tuan kuranes

23-06-2006 11:34:46

yes got it. Need some time to work on it, but I have a lots of things to do currently... so you'll have to be patient.

vvHedgehog

23-06-2006 11:39:40

yeah! i`m understand!
will wait anyway)

vvHedgehog

14-08-2006 12:37:52

up

tuan kuranes

22-08-2006 08:20:00

Did you check the bug again latest SDK ?

vvHedgehog

22-08-2006 09:07:25

not yet. I`ll check it soon.