bad smoothing groups - solved

is there a set of rules that go along with making collision models?

im trying to make a staircase in milkshape and ive gotten the model and the sequnce part done, then i draw the boxes in a new file where i want the collisions to happen and assign each individual box to its own smoothing group, then i export it

when i go to compile it, i get a warning saying that my collision modl is bad and i should check the smoothing groups, then it ends up truncating the physics model, ive retried making this model 4 times over and i get the same result. the only way i can get it to work is if i only have one box, but i definately need more than just one box

ive made collision models before and theyve worked just fine, but i realized that they were only 1 box and not more like i am trying to now

---

the only problem with steam updating things every week is it doesent give the developers an incentive to playtest thier work, so we do it for them

each 'block' that the collision model uses needs to be assigned to its own smoothing group number. like with a set of stairs, you might give the 1st step a block with all faces assigned to SG1, the 2nd step would have a block with all faces assigned to SG2 etc.

in milkshapes smoothing groups, there are 2 buttons. one for selecting SG's, and one for assigning them. make sure you are in assign mode, select one of the collision blocks, and hit a number. remember, a different number for each block. don't use the same number twice.

if you have more than 20 or so, i don't remember exactly, different blocks in your collision model, and you might have as this is a set of stairs, then you'll probably get a compile message about it being too complex, and the compile will still truncate your hitboxes. there's a line for the qc file that allows complex collision models. something like $notruncate, but i can't remember that either. meh:/

---

i eat paint

ive done exactly that.

ive made the box and assigned it to its own group

and when i am finnished assigning them i click on select and then click each number to make sure that all the blocks are alone and that they all have a smoothing group

but yet the error still comes up

---

the only problem with steam updating things every week is it doesent give the developers an incentive to playtest thier work, so we do it for them

ok i did a little test

i remade the collision model from scratch and i just randomly placed blocks throughout the grid and compiled and it works, so while i had a working collision i vertex manipulated the blocks to fit the shape and then all of a sudden it doesent work anymore.

i cant figure out what this thing wants from me

---

the only problem with steam updating things every week is it doesent give the developers an incentive to playtest thier work, so we do it for them

make sure that the shape of each of the blocks is a shape that hammer would allow.

---

i eat paint

i can assure that they are legal shapes(im drawing them with the box tool). and i can assure that each box is in its own smoothing group.

and yet that pesky error still comes up ( and i only have 3 boxes there)

is there a way to manually check that each is correctly assigned to a group

ive attempted this thing 10 times over and i double checked my steps when assiging the groups.

maybe there is a limit on how far a box can go. the dimentions of the model in its entirety are 256(left)192(front)192(height)

i wish there was more documentaion from valve on the subject

<div class="abouttext">Message submitted 32 minutes after original post:</b></div>
ok just for the hell of it ive decided to make a really simple collision model with just ONE BLOCK. ive made it 10 units tall, 80 units wide, and 64 units across.

ive textured it (cause you have to in order to export)

ive assigned it to the joint at the 0,0,0 dimentions

and ive assigned it to smoothing group "1"

then i exported it

i comple it and it still has the nerv to give me a bad smoothing groups error and it said that it truncated it even tho it is a box, i look at it in the model viewer and it looks just as i created it but the error still persisted

what the hell does studio_model want with me

---

the only problem with steam updating things every week is it doesent give the developers an incentive to playtest thier work, so we do it for them

I remember a while ago I read that when modeling for HL2 in Milkshape, everything had to be one smoothing group. I follow this rule and all (well, most) of my models compile correctly. If you're wondering, the only models that didn't compile correctly when using that rule were ragdolls that had problems with the QC... (me and my coding problems)

P.S. The physics model needs a texture to compile aswell... don't worry, you can just use the same one for the reference model. The physmodel's texture is never displayed so it doesn't matter.

i tried the single smoothing group and that too did not work?

im gonna try a few more things with dimentions and sizes through the day

<div class="abouttext">Message submitted 46 minutes after original post:</b></div>
heres the qc, altho it is a copy of a qc that i have used before and works i just changed the file names.

$modelname backalley/staircase01.mdl // this will name the compiled model, and place it in the models/warlord folder. change as desired, but make sure the folder you name exists in your game directory.

$cdmaterials models/backalley/ // looks for the material(vmt, not vtf) in the materials/fact folder (of your game dir.)

$scale 1.0

$body studio "staircase01.smd" // the .smd here should be your highest poly reference smd. it should also be the same as the 1st smd to be replaced in your lod commands, so i've changed it for now, and used your lod1 smd. the 'collide' name from the original line that was here suggests that it may have been your physics model that was referenced, so i've used it as a collision smd

$surfaceprop "metal"
$staticprop
$noforcedfade

$sequence idle "idle" fps 1 // every model needs at least one sequence. export another smd as a sequence named idle, and put it next to the ref.smd's.

$collisionmodel "staircase01collide.smd" {
$Mass 50 // Mass in kilograms
$concave
}

also i am using a material from cs_assault map's model staircase so i know that the material is correct.

and the smoothing groups are one block to the group

---

the only problem with steam updating things every week is it doesent give the developers an incentive to playtest thier work, so we do it for them

how many seperate blocks/smoothing groups are there in your phys model?

---

i eat paint

nevermind i found out what the problem was

it was using the block tool that was causing the problem. when you use the block tool it makes like 20 verticees instead of just 8 that it takes to make a box

so building the boxes by scratch solved it

thanks for all your help tho

oh! and one more thing if there is a command that will allow more than 20 groups in a collision model could you point it twards me please?

---

the only problem with steam updating things every week is it doesent give the developers an incentive to playtest thier work, so we do it for them

thats strange, because any shape with any amount of vertices can be used to form seperate parts of a phys models, just as long as they are all convex shapes, like shapes that hammer would allow.

---

i eat paint

thats studiomdl.exe for ya

---

the only problem with steam updating things every week is it doesent give the developers an incentive to playtest thier work, so we do it for them