[Bobboau]

I copied and pasted something that had some weird symbols (something from wikipedia, that's were I was doodleing with geodecics) but I removed all that, so I didn't think about it. that's all I could imagine would be the problem.

[Woolie Wool]

One thing I would definitely want is model rescale, where you can resize a model to a desired size and the program calculates the correct POF data by multiplying/dividing the original POF data by the scale factor.

[Kazan]

if you change the geometry PCS will rebuild the affected subobject(s) completely not just hack the coordinates

[Woolie Wool]

if you change the geometry PCS will rebuild the affected subobject(s) completely not just hack the coordinates

Could you please elaborate for someone (me) who is not very familiar with modeling outside the features afforded by ModelView's editor?

[Kazan]

if you change the geometry PCS will rebuild the affected subobject(s) completely not just hack the coordinates

Could you please elaborate for someone (me) who is not very familiar with modeling outside the features afforded by ModelView's editor?

well.. for starters modelview doesn't let you touch most of the data of a POF... PCS2 is a combination of PCS1's ability to edit EVERYTHING and modelview's ease of use (you should see a lot of influence from modelview in the interface i'd think).

PCS1 and Modelview when they load a file - it stays in POF-format in memory... this is not true for PCS2

all model files, once loaded into PCS2 are in a format known as PMF - PCS Model Format - it's PCS2's own personal format.  So when you open a .cob/.scn it gets loaded by the COBHandler code, then converted to a PMF - when you load a POF the same is true.

PMF is flexible, easy to extend, easy to manipulate.

All of the data in a POF is easy to manipulate except one thing - the geometry - the geometry in a POF is stored in what is called a "Binary Space Partitioning Tree" - or BSP Tree.  This is very useful for collision detection, and it used to be very useful for making rendering faster (pre-video cards having Hardware Transform Clipping and Lighting). 

BSP trees are not easy, nor safe, to manipulate.  It is not difficult however (in terms of CPU usage) to read the data out of this (As is done when converting the POF into a PMF) nor computationally very difficult to rebuild that tree.

When PCS2 loads a POF it cache's the BSP however - so if it doesn't have to make a new BSP tree for that geometry, it doesn't do so - (just to keep consistentcy.. don't need to recompile the BSP on [V] models.. their BSP compiler was even better than mine). 

So it only rebuilds if you change the geomtry - and then only for the subobjects that the geometry changed on.

[Woolie Wool]

And how does this tie into my idea of changing a model's size and adjusting things like gunpoints and docking paths to fit?

[Kazan]

And how does this tie into my idea of changing a model's size and adjusting things like gunpoints and docking paths to fit?

it requires rebuilding the BSP... PCS2 can easily do it... i was just saying it requires a BSP recompile (nothing major) when you write out the POF

[Woolie Wool]

OK, is the ability to rescale the geometry present in current builds of PCS2?

[Bobboau]

no, and it's a second phase feature, we haven't gotten all of the basic stuff done yet, that is geometry editing, that's a big huge feature on it's own.

[Bobboau]

so, any requests?
particularly any sort of feature that would automatically make some sort of data for you, or figure it out in some way.

[Kazan]

did i make the subobject context rendering work (ala modelview style) or did you?

[edit]
oh.. i'm going to reintegrate BSP Sortnorm/boundbox debugging like was present in PCS1 (since it let me find so many errors)

here is how it works

when you load POF compiled with [V]'s BSPGEN, PCS1 of high enough version to have compiler 1.3.4, or a PCS2 build it "caches the bsp" - so it doesn't have to rebuild it unless you change the geometry.

when you save a file to POF it also cache's the results of the compilation (As of 20 seconds ago) - in either case "can_bsp_cache" is set to true.

in this case i need a button to become usable (a toggle button) "BSP Debug" and then i'll do a context render of the sortnorms and bound boxes

[edit]
I'd also like the UI To display the polycount of an object, and the polycount of all it's children summed, and the total of those two

[Bobboau]

I'll set a bool in the model to true, you can check that in the renderer, and either use the current active model or apply it to everything.

[Kazan]

I'll set a bool in the model to true, you can check that in the renderer, and either use the current active model or apply it to everything.

kk - BSP debug will have the same context scoping as lods/models

oh... on that note - make child nodes of header for each LOD and we'll tie that into context rendering

[Bobboau]

oh, you mean like
header
   +(LOD0-LOD4 and debris)
and we use that rather than the currently active model to determine which LOD to draw?



and PCS_Model::draw_bsp will be the flag for your BSP drawing, in addition there is an event BSP_RENDER_CHANGE which will get sent every time the check box is changed (use GetInt on the event for the value), although I doubt you'll need to use that.

[Water]

Dragging one sub-object on to another seems to turn the target into a group heading. Instead of both being children of a new group.

What is the maximum poly count for sub-objects in pcs2?  (when saving a pof) It seems the same as pcs1.
Once you get past the limit on a sub-object the file usually doesn't save past the 1 meg mark. If the offending sub-object is divided it saves fine. 

Also would it be possible to have a drop box to set front, side, top etc view? Modelview has them down the side. 

[Kazan]

What is the maximum poly count for sub-objects in pcs2?  (when saving a pof) It seems the same as pcs1.

how much memory you have in your computer.

PCS1 and PCS2 do not have the same BSP compiler

PCS1 had a fixed destination buffer for the BSP data (4megs or so)
PCS2 builds a tree in memory and can calculate the buffer size from that.

PCS2's BSP compiler should also be more accurate... hence why it takes longer.

[IPAndrews]

Go, go dynamic memory allocation.

[Col. Fishguts]

Can I just throw in that an option import 3DSMax ASCII files would still be awesome. They have per-polygon smoothing groups stored in them and are relatively easy to parse.

format reference pls

I did send you a simple example file last time this was discussed. I can put it online when I get home later.

I don't think there's an official format reference document.

EDIT: Ok I've found something online. A reference for the ASE format on the UnrealWiki: http://www.unrealwiki.com/wiki/ASE_File_Format

The interesting part with comments from the Unreal modders starts further down.
I may be repeating myself, but: If PCS2 would support ASE import it would be the ultimate modding tool, since ASE is an open format. Everbody could then write simple exporters for their favourite modeling tool.

[Kazan]

hmmm ASE

i can get to that eventually..... if soemone wants to volunteer to code it now i'll add you to CVS repo

[edit]
oh bob.. if you need a fast and dirty draw a bounding box on the screen there is a function defined in pcs_pof_bspfuncs.h OpenGL_RenderBox(vector3d min, vector3d max) as of my next commit

[Kazan]

hey bob.. can you change the checkboxes

split the "BSP debug" into "Sortnorms" and "Boundboxes" and add a checkbox for "Render children"

oh.. and i've discovered that PCS2 is outputting bad snorms and bboxes