CMP to SUR Conversion Tests

No rush, Mirkha. This is a long-term project.

But I appreciate your efforts.

Guys.

OK, no more trolling -

I will always react strongly to being offended time after time. Nobody likes to be offended, and I also realise I have done that to BW too.

BW: I apologise, my response was out of frustration and being very offended by your offhanded and repeated dismissals.

For the last time, I am not trying to “fool the tool”. What I am trying so hard to do is to give you proof that the problem exists and can be reproduced, so that you can have a starting point to look for the problem.

Note: It did not matter whether the boxes are welded to each other, or overlapping each other or separate. It made absolutely no difference in my first test file.

I separated them to see if there was any difference - and once I had seen that it made no difference (the same boxes still had no surs) I left them like that, concentrating instead on the shapes, making some with obtuse angles and some with acute angles as you can see from the jpg, and you can see the results.

Try to read my previous posts to see the symptoms while I cool off and try to be logical in my next test file.

Some sad news - I just found a couple of hours ago that in a new test model with 13 boxes and two spheres, ONE of the SPHERES did not generate either. I had only made one sphere in my previous test and that had worked.

This problem is also my goal to verify before I present it, like I did with the previous file.

So give me some time and I will bring you more evidence of the problem, and I will show you welded and separated parts too.

In the meantime, those of you who have MilkShape and the .cmp exporter and the sur builder and the sur importer can duplicate my tests yourselves and report back.

W0dk4:

Why do you also say this tool is for beginners?

I understand that you may enjoy doing the occasional model by hand, it does give a lot of satisfaction when it works (finally).

But for the tool I am surprised by your comment and I do not agree with you at all.

I have had to spend those long hours on many models, and redo and redo. I have also had to struggle my way through various “tutorials” that caused more misinformation too. When you take on a project to make several models it’s no longer fun when they fail initially.

If this problem can be identified properly and fixed, it will be the best sur tool at our disposal and will cover so many requirements including large ships, bases, scenery and whatever else modders would like to try.

It’s already great with single-part surs for small ships, although personally I would far prefer to have surs closer to vanilla types, with all groups and HpMount instead of just a single wrapper. But that’s another issue.

And it will be by far better than all other methods. So what’s wrong with that?

It really is so nearly there.

w0dk4 wrote:
@ST: The bottom line is, the sur builder is a simple tool for newbie modders.

Any serious modder should invest the time to do custom surs by hand rather than put something in a tool and wait for its outcome.
Therefore its perfectly fine if BW doesnt want to spend hours trying to cater to special requests as regards the input model. A model that has parts that are not connected in any way are not a “normal” model, they are a special case.

It is unfortunate that there are issues with other methods. A reliable manual technique would solve most of this discussion.

Thats why Im eagerly anticipating Schmackbolzens attempt to build a obj->sur converter from scratch by using adoxa’s and other’s findings.

Ok, first off w0dk4, your wrong about the sur builder being a tool for newbies. Can’t believe you said that. It’s a tool to help people with their projects, just like any other tool made for FL. Next thing you’re going to tell me is i should make my own 3D model making program so i know how to do that as well :-x

Most of us, me included don’t have time to waste on things we either don’t understand or can’t be bothered to understand. If a tool comes along that hastens the overall process, i’m all for it, doesn’t matter what tool it is.

The sur builder is not for newbies that’s for sure, as i’m sure as anything that you use tools for your modelling efforts, this is no different.

Its definately a tool for newbies. Im not saying that experienced modders shouldnt use it, Im saying that if you want a good, quality sur, you better do it by hand.

And what really bothers me is that rather for someone to write a working sur-exporter, people wish for an automated solution that can only give mediocre results at best.

If hitbox generation could be automated to the extent some people in this thread would wish it to be, why are there still custom made hitboxes in todays games?

w0dk4 wrote:
And what really bothers me is that rather for someone to write a working sur-exporter, people wish for an automated solution that can only give mediocre results at best.

Exactly.

The SUR Builder is fast, sure, but wouldn’t it also be pretty quick to build a custom SUR if the Exporter worked reliably?

Then a modder could make their SUR look exactly the way he/she wants it to look.

The SUR Builder makes pretty good compromises, as far as I can tell. It is terrific for newbies and good for some experienced modders who like it. I can envision a modder of any skill level using it for the majority of models in a mod, while saving a few complex models for hand-crafted techniques.

But it will never, ever, have the fine level of control that a 3D modeling program has.

@startrader
go test havoc tools and try it on the simple test
test to see if havoc sur can do it
i talk with lance about it
a while back to make it like that tool
but he never looked into it

but i did find with havoc some times it did miss bits out
but just a simple weld fixed that problem

but looking at the test its like it aint counting the parts
or is it and shrinking them to 1 point ??
if it is i would say its a weld fail
simple to fix then but then it would be down to the model failing not the tool

like i say now and again when i sur i get a total collapse
on some parts but it seems it might be the same fail the tool is having
but test that first on it if it fails its model if not its a simple fact it aint counting the amout of parts

well, as far i can test
the object was always in solar
i haven’t test as a ship yet, and alone it’s a little difficult

tie_0 : tie without wings
tie_1 : tie_1.ms3d, is a tie with each group in each other
tie_1b : tie with welded parts
tie_2 : tie_2_separated_groups.ms3d

tie_multi, and so on : obvious

the tie_1 works ingame without a sur
and a single group sur

the tie_0 works ingame without a sur
and a single group sur

using the sur builder :

single part :
Min MAx sort
No sort : ok
Type 1 : ok
Type 2: no

Secondary sort :
no sort : ok

multi part :

no sort : all seems to be good
type 1 : partial hitbox
type 2 : partial hitbox

no sort secondary : only the sphere

• Make sure that the wings have some thickness (not 2-D).
  yes, as you can see in the ms3d

• Build the main body, then make a SUR for it. Then add the wings. Does the resulting SUR work in-game? That would be very close to what you seek, and could probably be done as a single-part SUR.
  no differences

• Try the wings 1) connected to the body and 2) disconnected (not touching). Is there a difference in the resulting SURs?
  no differences

in the zip all the cmp/mat/sur/ms3d
and screen shots to see what i’m talking about

the only one who works is the tie_multi_no_sort.sur
the cmp is 5 groups,
how i built the cmp :

export it with milkshape choosing 5 groups at the export, back to front and scale 1
then, rename all the materials in the cmp with utf editor to match what i’m used to
and make by hand a .dat and import it in the cmp with utf editor for the cons/fix
the cmp/sur/mat are fine ingame

http://soler.sebastien.free.fr/test_sur_builder/
all is here

i hope it helps

ps : the tie_multi_no_sort.sur with the groups all separate (tie_multi2.sur) seems to also works, in hardcmp
i haven’t tested ingame because here is 2:10 AM

all is in the zip file

i cannot resist
so a test with a real tie, the model is more complex 1022 Vertices, but not so complex event so

i build the model the same as above
it works ingame
the best sur i can do is this one :

Uploaded with ImageShack.us

Type 2 also works, this time No sort doesn’t work

not perfect but good

Mirkha wrote:
i cannot resist

Holy Smoke!

Both the model and the SUR look gorgeous!

Is this some mod that I can play? Forget the programming… I want to blast some Tie Fighters!

(Also, if I figure out what you said, I will put your process into the documentation for the SUR Builder. )

@Lonestar: I feel as though I should respond to your message. But, honestly, I have no idea what to say. So consider this an “ACK” (in communications, an ACK is an Acknowledgment – it means that I heard you).

@ST: Thank you; no hard feelings. I lump you and String Theory in the same category – you both say everything that could possibly be said; in all universes and all dimensions.

PS: New Rule: From now on, ST, I am only going to read the first ten lines of your messages.

I spent most of this afternoon trying to get it to make a usable multi part SUR without any luck. I used several of my own CMPs as well as a couple of vanilla ones, I made several SURs for each CMP trying to find a setting that would work, every single one of them was a fail of some sort, the Root was the only part to be done properly, of the parts that it actually made a hitbox for most didn’t cover all of the part, and depending on the setting it does miss out parts.

After reading the first couple of pages of this thread it’s become clear what LS’s intentions were when he made this, it was to save him having to make SURs for all of his single part CMPs, an automated tool to make a simple “wrap” SUR for his CMPs, and IMHO that is exactly what is should be used for as it does that beautifully.

Now what’s this about the SUR exporter being unreliable, the “shield bubble” never put a foot wrong for me when I was making ships, and the “shrinkwrap” did what it needed to do when I needed to use the splicer, it just took me 2 years to figure out how to get it to work
People have a pre conceived idea that the exporter is broken, which is based on thier & other peoples crappy no-brainer attempts, believe it or not but Colin came within a fraction of creating the perfect “shield bubble” SUR with v1.1, it can create a SUR just like one of DA’s, only the “wraps” covering parts with equipment hitboxes stop it from working like one of DA’s, it’s the “shrinkwrap” that has the problems.
Gibbon also hit the nail on the head with his “time & inclination” comment, for most the idea of spending the best part of a working day to do one simple ship is to much, and that’s when you know what your doing, so the idea of possibly spending weeks learning how to get a CMP & SUR to work right is a non starter for most.

Which is why ppl like ST are desperate for a one click automated app that can do it all for them, never likely to happen, but that doesn’t stop them wanting one.

Bejaymac wrote:
so the idea of possibly spending weeks learning how to get a CMP & SUR to work right is a non starter for most.

Oooh. That explains my experience.

So, now that you have completed your two years of learning how to make good SURs, Bejay… is it possible to explain in a tutorial? Or is it like asking a hacker how to hack (impossible to teach)?

Roger Bullwinkle

I hope you can put into words what you want to say. And i hope that my post didn’t rub someone the worng way, but I see Startrader, NOT Startracker trying hard to address a problem and doing it well. And then getting told he is fooling the program kinda boils my blood. It took me quite some time to make that post, and many times starting over as it was too harsh, not saying what i wanted to say HOW i wanted it too.

But meh, Its over, and well not much I can say except sorry if that post i did offended you. Its a full moon and my ADHD goes crazy durning fullmoons, Also must be the wolf inside me too.

Hooooooooowwwwwllllll !

Lonestar out

now, problem is, there is no way to make a hand-made sur working properly in all cases. even if you follow any advices about convexity, part numbers, yes, me even used Sushi’s video tut on custom hitboxes - even than the sur you get is just rubbish, not working properly. me understand ST, who wants a tool that makes it well, because there is none such. we all use tools for surs, whether we model them ourselves or use the builder, we do not write the sur file by hand and if we won’t get a tool or plugin or whatever program, that does it job as it is supposed to, we won’t get a proper sur, regardless, how accurate it is. that single part surs are possible now, is great but a modder who is happy with single part surs only is yet to find.

lol Bullwinkle, it’s your lucky day
you have two server to satisfy you :p, an Us and a French ^^
and if you like this tie, you will be mad at the Actis XD

anyway it seems that preparing the cmp is very important, i mean the structure of the file, not the 3D apparently …
therefore include the way to make cmp is a good thing

on a side note, keep in mind that it’s very very very very very very very difficult to express on a forum, it lacks tone, facial expressions, gestures, ect ect. everything that informs the caller about the true meaning of what they say and especially the intensity …
Here all the information are raw, if i can say, and something that we had not meant nastily, immediately seems very aggressive

@ Gisteron,
i’m very surprise you didn’t succeed,
i follow the same tutorial and don’t have find any model who doesn’t work
i recognize that the tutorial is not very … didactic
i have allready make another, but in french,
is someone know how to build subtittle ^^ …

I will be updating the SUR exporter, but I’m finishing off the Freelancer Resource Compiler now and I’ll be updating JFLP next (had a critical bug report, otherwise it would’ve waited).

mirkha, it depends on what you call success. it lacks on collision detection with both, OP’s method and splicing method (while splicing method is far better in terms of shot hit detection) and it is definately not a question of ability rather than of knowledge of the sur. all those custom sur methods except for the resizing are outdated, old, messy. thank you, adoxa, that you promise to get on it, it likely will help a lot of us, but so long, with what we have now, making custom surs that work always right is almost impossible. oh and, no, modern tutorials on splicing method and OP’s method seem to be just fine and work in many cases but as long as the exporter codec is messy, you can put your files together as much as you want and in any order you want, it just won’t work perfectly. that’s a fact and that’s the problem and reason why were discussing it. we won’t talk or work deeper into it, if all was perfect.

well, i call it success when each surface register hits and collision properly
so for me the sur splicing method is very good
if i take a look on the sur builder tool is because he can save a huge amount of time

anyway a new/better exporter will be fantastic ! ^^

Bullwinkle wrote:

Bejaymac wrote:
so the idea of possibly spending weeks learning how to get a CMP & SUR to work right is a non starter for most.

Oooh. That explains my experience.

So, now that you have completed your two years of learning how to make good SURs, Bejay… is it possible to explain in a tutorial? Or is it like asking a hacker how to hack (impossible to teach)?

It was on the old Lancers Reactor, so it should be in the archive, it might even be on here as I can’t remember if I posted it here after the site went live (that gives some idea just how long I’ve been making SURs ;)). You are correct in that I can teach how it’s done, just not how to apply it to each mesh, that’s trial & error even for me. Wouldn’t be the first time I’ve regrouped a mesh then exported it into a CMP, only to find when building the mesh for the SUR that the grouping wouldn’t work, and end up scrapping the CMP and redoing it with the new grouping.

@ adoxa, please don’t use the source code for V1.2 as even Colin will tell you it’s crap, it has a lot of bugs that just weren’t there in the v1.1, biggest one being it doesn’t make hitboxes properly.

Well…

I moved on from testing boxes to testing spheres since in the last test one of two spheres was missed. (The second was a Geosphere and was sur’d OK though).

This is a TOTAL failure:-

In a 2-sphere, 1-cylinder model only the cylinder is sur’d.

Somehow it seems the builder has either missed both spheres completely, or it has decided the shape is already done but has not saved it…

Here’s the screen shot below.

Please: someone else do this test and give us your findings:-

Make the two spheres and one cylinder, add a material, export the cmp, and make the surs with sur builder, and import them back into Milkshape, select the surs only and print your screen.

Then please remove the surs, add two Geospheres, export and gen the surs again with sur builder, mport, select the surs only, print your screen.

It only takes 10 minutes to do all of this.

Let us see your results please.

Thanks.