Extracting simple models

[Tosyk]

I don't understand why finding faces is in the first step?

for me the easiest process is something like this:

1. step
- put vertex block start
- put vertex count nubmer
- choose some options depends on how strider work/look (float/half-float)

-------------at this point I can see vertex cloud---------------

2. step
- put face block start
- put face count number
- choose some options depends on how face strider work/look (byte word/dword)

---------------at this point I can see full mesh-----------------

3. step
- put uv info
- put vertex count number from 1 step
- choose some options depends on how vertex strider work/look

----------at this point I can see full mesh with uv-------------


h2o is a great tool but its ui looks overwhelmed and inconsistent — it had steps though a lot of options is just floating around. and some ui elements are hidden and is not clear what buttons related where. even 2 simple lines separated 3 steps could make ui clearer.

[shakotay2]

First of all many thanks for your reply! I was just reformatting/editing my post, not seeing that you answered already, sorry for that.

I don't understand why finding faces is in the first step?

well, because it's the simplest part.
Finding the scrambled alphabet is easy, you can't deny that seriously.
(Simply review forums post; first successful action of most newbies dealing with hex2obj is finding the face indices.)

-------------at this point I can see vertex cloud---------------

Agreed. Look here: viewtopic.php?f=16&t=18031&p=140097&hil ... ud#p140097

It's not forbidden to start with "point clouding". (I should mention it in the tutorial, that's correct, thanks!)

3. step
- put uv info
- put vertex count number from 1 step
- choose some options depends on how vertex strider work/look

----------at this point I can see full mesh with uv-------------

I see, you like it structured. In the first years of hex2obj I didn't care too much for uvs.
Presenting the mesh was the main goal.
So taking care for the uvs at the end of the process (although "placed" at step 2) has "historical reasons".
Also you don't need to put vertex count (for uvs) in hex2obj because it uses it automatically (which is ok for simple models).

it had steps though a lot of options is just floating around. and some ui elements are hidden and is not clear what buttons related where. even 2 simple lines separated 3 steps could make ui clearer.

I've mentioned those "floating" options in my edit. (yeah, this was a matter of laziness)

I'll see what I can change concerning the ui arrangement, but don't expect it too soon.
Also I don't think that I'll change the general approach because it made sense for me for hundreds of (simple) models now.

Thank you for your post, really appreciated!

[Tosyk]

First of all many thanks for your reply! I was just reformatting/editing my post, not seeing that you answered already, sorry for that.

I don't understand why finding faces is in the first step?

well, because it's the simplest part.
Finding the scrambled alphabet is easy, you can't deny that seriously.
(Simply review forums post; first successful action of most newbies dealing with hex2obj is finding the face indices.)

I do agree — I found that "alphabet" myself many years ago. Unfortunately knowing that info will lead nowhere within h2o, to be honest and that's why "step 1" felt really confusing from the very beginning

-------------at this point I can see vertex cloud---------------

Agreed. Look here: viewtopic.php?f=16&t=18031&p=140097&hil ... ud#p140097

It's not forbidden to start with "point clouding". (I should mention it in the tutorial, that's correct, thanks!)

yeah but it's "step 3"! and start from "step 3" is a lot more confusing than lack of feedback from "step 1" and "step 2"

3. step
- put uv info
- put vertex count number from 1 step
- choose some options depends on how vertex strider work/look

----------at this point I can see full mesh with uv-------------

I see, you like it structured. In the first years of hex2obj I didn't care too much for uvs.
Presenting the mesh was the main goal.
So taking care for the uvs at the end of the process (although "placed" at step 2) has "historical reasons".
Also you don't need to put vertex count (for uvs) in hex2obj because it uses it automatically (which is ok for simple models).

if I don't need to put vertex count for uv the ui should show me this in inactive field in which I will see same vertex count as I found for vertex buffer. See, that's why I always started from h2o tutorial :/

it had steps though a lot of options is just floating around. and some ui elements are hidden and is not clear what buttons related where. even 2 simple lines separated 3 steps could make ui clearer.

I've mentioned those "floating" options in my edit. (yeah, this was a matter of laziness)

I'll see what I can change concerning the ui arrangement, but don't expect it too soon.
Also I don't think that I'll change the general approach because it made sense for me for hundreds of (simple) models now.

Thank you for your post, really appreciated!

I understand that h2o's steps it's conventionality and I can fill data in any order but for any ui - especially advanced one - the visual feedback is very important - I would say the most important. If ui confusing user from the very first step the user will drop it but if the ui with every its step will give user some feedack - a result - it will give a better understanding of ui and reduce amount of questions on how to use it.

Anyways, all you do here is absolutely briliant. I'm following the forum and know all the local gurus

p.s.: sorry for too sirious threat, I understand this thing is just for fun, but wanted to leave here my thoughts

[shakotay2]

I do agree — I found that "alphabet" myself many years ago. Unfortunately knowing that info will lead nowhere within h2o

well, once again, this steps leads you to the maximum face index which is generally identical with the vertex count. The vertex count (vc) together with the size of the vertex data block (sizeOfVD) leads you to the FVF size.
(where FVFsize= sizeOfVD/vc, but not always, that's one reason why newbies fail.)

yeah but it's "step 3"! and start from "step 3" is a lot more confusing than lack of feedback from "step 1" and "step 2"

well, maybe I should talk of "recommended" order of steps but I see, trying to convince people that hex2obj is a helper tool, if they don't feel so doesn't make too much sense.

Anyways, all you do here is absolutely briliant.

From your posts I don't feel so but anyways I think you're a dedicated follower of the Model Researcher, it's ok. You should use the tool, that gives you the best results.

[Tosyk]

Anyways, all you do here is absolutely briliant.

From your posts I don't feel so

I'm always excited with people knowledge and trying to input my best. But I see you helping people in every thread and that is really nice

but anyways I think you're a dedicated follower of the Model Researcher, it's ok. You should use the tool, that gives you the best results.

actually maxscript is the tool I understand better than any others

[Tosyk]

sorry for dp

I start to make a maxscript for MMB format of Division since python script is outdated: viewtopic.php?f=16&t=14374
Then I realized that at this point of my knowledge I'm able to read (and even debug) some python code and for the last few hours before sleep I interpreted full MMB header from python to the maxscript. though I can't understand some python functions:

first of all, what 'bs.read("9I")' mean? what is "9I" or "10I"?

Code: Select all

for j in range(meshNumGroups):
   if fileType == 9 or fileType == 10:
        meshGroupInfo.append(bs.read("9I"))
   elif fileType == 11:
        meshGroupInfo.append(bs.read("10I"))

second of all, where the script starts read vertex stride?

here's what I have so far:

division_v00_header.zip

[Acewell]

what 'bs.read("9I")' mean?

it means read nine unsigned integers for a total of 36 bytes (9 * 4)
if you do print(meshGroupInfo) after reading it should list the 9 integers in log.

[Tosyk]

what 'bs.read("9I")' mean?

it means read nine unsigned integers for a total of 36 bytes (9 * 4)
if you do print(meshGroupInfo) after reading it should list the 9 integers in log.

dank a bunch so it can be done by bs.seek(36, NOESEEK_REL) as well, correct?

now I only need to undestand a loop by which noesis read vertex/face buffers to build a mesh

[chrrox]

what 'bs.read("9I")' mean?

it means read nine unsigned integers for a total of 36 bytes (9 * 4)
if you do print(meshGroupInfo) after reading it should list the 9 integers in log.

dank a bunch so it can be done by bs.seek(36, NOESEEK_REL) as well, correct?

now I only need to undestand a loop by which noesis read vertex/face buffers to build a mesh

bs.seek(36, NOESEEK_REL) would skip those bytes and not read them.

__NPExample.txt shows 2 ways to read meshes.

style 1 like max scripts
meshes = []
for i in range(0, numMeshes):
for j in range(0, numIdx): - faces / triangles
idxList.append(bs.readInt())
for j in range(0, numPos): - vert position
posList.append(NoeVec3.fromBytes(bs.readBytes(12)))
mesh = NoeMesh(idxList, posList, meshName, meshMat)
for j in range(0, numNrm): - vert normals
mesh.normals.append(NoeVec3.fromBytes(bs.readBytes(12)))
for j in range(0, numUVs): - vert uv's
mesh.uvs.append(NoeVec3.fromBytes(bs.readBytes(12)))
meshes.append(mesh)

or the easier way
for i in range(0, numMeshes):
triangles = bs.readBytes(numIdx * 4)
positions = bs.readBytes(numPos * 12)
normals = bs.readBytes(numPos * 12) if numNrm == numPos else None
uvs = bs.readBytes(numPos * 12) if numUVs == numPos else None
rapi.rpgBindPositionBuffer(positions, noesis.RPGEODATA_FLOAT, 12)
rapi.rpgBindNormalBuffer(normals, noesis.RPGEODATA_FLOAT, 12)
rapi.rpgBindUV1Buffer(uvs, noesis.RPGEODATA_FLOAT, 12)
rapi.rpgCommitTriangles(triangles, noesis.RPGEODATA_INT, numIdx, noesis.RPGEO_TRIANGLE, 1)
rapi.rpgClearBufferBinds() #reset in case a subsequent mesh doesn't have the same components
mdl = rapi.rpgConstructModel()

[Tosyk]

chrrox, thanks for clarify this a bit

so what this line (meshGroupInfo.append(bs.read("9I"))) exactly do? can I read vertex buffer (12 bytes in stride) the same way? like (bs.read("3I"))?

it's unfortune that noesis doesn't have documentation for its python functions nor detailed/explained examples

[chrrox]

chrrox, thanks for clarify this a bit

so what this line (meshGroupInfo.append(bs.read("9I"))) exactly do? can I read vertex buffer (12 bytes in stride) the same way? like (bs.read("3I"))?

it's unfortune that noesis doesn't have documentation for its python functions nor detailed/explained examples

(meshGroupInfo.append(bs.read("9I"))) just split python functions to make them easier to understand.
1. meshGroupInfo.append()
append whats in the brackets to the list meshGroupInfo
2. bs.read("9I")
bs is he current file stream variable
read("9I")
read 9 ints in a row

so lets do an example

say I have 1 model with 3 meshes
each mesh has the info for
vert count, face count, material id
I would read it like
meshGroupInfo = []
for i in range(0,3):
meshGroupInfo.append(bs.read("3I"))
(mesh data could be here or later in a file)
(We will assume its later so I can call it like this)

for i in range(0,3):
print(meshGroupInfo[0], "vert count")
print(meshGroupInfo[1], "face count")
print(meshGroupInfo[2], "material id")

[Tosyk]

meshGroupInfo = []
for i in range(0,3):
meshGroupInfo.append(bs.read("3I"))
(mesh data could be here or later in a file)
(We will assume its later so I can call it like this)

so here you create a mechanism by which you'll read further info. in maxscript word: read 3 floats - one after another

for i in range(0,3):
print(meshGroupInfo[0], "vert count")
print(meshGroupInfo[1], "face count")
print(meshGroupInfo[2], "material id")

and here you apply created mechanism to the stride and read: 3 flots for vertex, 3 flots for face, 3 flots for mat ID

AND
these "vert count", "face count", "material id" is multipliers, correct?

edit:
and this code:

Code: Select all

			for j in range(meshNumGroups):
				if fileType == 9 or fileType == 10:
					meshGroupInfo.append(bs.read("9I"))
				elif fileType == 11:
					meshGroupInfo.append(bs.read("10I"))

means a rule for the stride size: for 9 or 10 version size would be 9 floats and for 11 version size would be 10 floats (1 additional float to the normal version). correct?
and maxscript example for that rule would be (right?):

Code: Select all

if vertexSize == 48 then
					(
						for x = 1 to vertexcount do
							(
								getPos = ftell f + vertexSize	
								dx=readFloat f
								dy=readFloat f
								dz=readFloat f
								
								null = readlong f
								
								x3=readHalfFloat f
								x4=readHalfFloat f
								x5=readHalfFloat f
								x6=readHalfFloat f
								
								tu=readHalfFloat f --first UV channel
								tv=readHalfFloat f --first UV channel
								
								x9=readshort f
								x10=readshort f
								x11=readshort f
								x12=readshort f --always the same
								
								bone1 = readbyte f #unsigned
								bone2 = readbyte f #unsigned
								bone3 = readbyte f #unsigned
								bone4 = readbyte f #unsigned	
								weight1 = readbyte f #unsigned
								weight2 = readbyte f #unsigned
								weight3 = readbyte f #unsigned
								weight4 = readbyte f #unsigned
								
								readfloat f
								
								w = (weight_data boneids:#() weights:#())
								maxweight = 0
								if(weight1 != 0) then maxweight = maxweight + weight1
								if(weight2 != 0) then maxweight = maxweight + weight2
								if(weight3 != 0) then maxweight = maxweight + weight3
								if(weight4 != 0) then maxweight = maxweight + weight4

								if(maxweight != 0) then
									(
										if(weight1 != 0) then
											(
												w1 = weight1 as float
												append w.boneids (bone1 + 1)
												append w.weights (w1 / 255.0)
											)

										if(weight2 != 0) then
											(
												w2 = weight2 as float
												append w.boneids (bone2 + 1)
												append w.weights (w2 / 255.0)
											)

										if(weight3 != 0) then
											(
												w3 = weight3 as float
												append w.boneids (bone3 + 1)
												append w.weights (w3 / 255.0)
											)

										if(weight4 != 0) then
											(
												w4 = weight4 as float
												append w.boneids (bone4 + 1)
												append w.weights (w4 / 255.0)
											)
									)

								append Weight_array w			
								fseek f getPos #seek_set
								append Vert_array[dx,dz,dy] 
								append UV_array[tu,tv*-1,0]
							)
					)

[thorh62]

Hello all,

first of all, thanks for providing this valuable tool .

I am facing a problem when running the latest version of hex2obj. I get two error messages "wordpad.exe" is missing
and also "glut32.dll" is missing. I am pretty sure that I have wordpad installed in my computer.

Maybe it is because I am using Windows 10?

Any advice about how to fix these errors? unfortunately these errors are preventing me from reading the tutorial (because I guess the tutorial might be using wordpad?) and also .obj files can not be created or previewed because I am missing the second file.

[shakotay2]

hex2obj exes are NOT working standalone, so
for zips with exe-only you need the files in this zip as a base, too:

---------- version 0.24e ----------
hex2obj_0.24e.zip

or ---------- >>> version 0.24c <<< ----------
hex2obj_0.24c.zip

Copy the lastest exe to the unzipped folder then.

( btw: wordpad.exe is not required, you can just double click hex2obj_tut.rtf )

[Tosyk]

I have good progress in the script for division. right now I'm trying to understand where Gh0stBlade getting mesh scaler. also I noticed that some meshes are not fit their skeleton just a bit. for example on this screenshot full model need to be scaled down fo 1% and moved by Y:-0.1 to fit the skeleton. I found that some of the head models need to be transformed the same way but with different numbers.



I would apreciate if anyone explain to me how this (viewtopic.php?p=118852#p118852) python reads scaler