CMP to SUR Conversion Tests
-
Thanks Mirkha.
Guys, I feel I need to put down my outlook here for you all to understand what I am looking for in this project…
I am familiar with making surs by hand. Probably more than many of the greater modders. I have spent many months making bad ones and eventually succeeding, and even when I succeeded, not even being able to see why a certain one succeeded when its predecessors failed, since the model and method was identical in many cases.
I have manually made convex shrouds for sur-splicer, and later when we found they existed, I have used 3DS Max with Havoc tools and got very good results. Yes some missed welds did cause problems, but other times I needed to reshape a ship part entirely, or turn some face edges, but it’s obvious when a shape is not right and it needs fixing, so not problem. Now we have the Convex tool for MilkShape and that is pretty good although not 100%. Note there are two of them, one is not quite as reliable as the other.
Then Sur Builder comes along and it is capable of making convex shrouds for individual cmp groups. WOW!!! FAB1!
I can split my own models very well and very easily.
So I figure, if Sur Builder can make a convex shroud for each cmp group in a cmp file, then I can make all the cmp groups that I need in the right place,the right size and the right shape, including the weapon hardpoints, equipment hardpoints, weapon+wings, and so on.
And then, if Sur Builder does its bit, all it has to do is locate each cmp group, shroud it, and save it, move on to the next, until all the cmp groups have a shroud (sur part). FAB2!
And last of all, wouldn’t it be FAB3 if Sur Builder had a button that said: “Shield Bubble”!? All that has to do is take a sphere, and very simply squash it to the size of the whole model’s maximum diameters+1metre in each axis X, Y, Z. Or +2 metres so we can have 1m above & below, sideways & fore-and-aft!
Done!
We wouldn’t need any magic, only these functions.
Then we won’t need more work on an exporter or other, since the tool can be as beginner or as complex as the user wants it. For me, I will use it to add HpWeapon hardpoints and so on to make as close to a vanilla sur as I can. Just because I can, and because it is so quick!
Others will make a simpler 3-group model or a 1-group model with shield bubble and will be so happy with it too!
And for my dreadnoughts of 2,000 to 3,000 metres with claws, booms and beams and extended hulls, it will make hull-following surs without blinking - all I will have to do is break my model into sensible parts and the sur shapes will follow them!
My Crab will have sur parts following its booms, not filling in the gaps.
So the tool is not only for beginners, is it. All of us need it, perhaps more now that I have explained how I will use it.
Unless you want to call me a beginner too - I’ve been called worse!
So - I’m trying to find out which shapes the builder doesn’t like. But sadly, so far it’s not only my model’s right wing tip (actually it is not the model since the identical left tip generated fine!) but also boxes, and spheres, and fat triangles (like Pizza slices, triangles with thickness).
The more I investigate the more basic shapes are failing.
That is what I am trying to isolate but since it’s more than one shape it must be a fundamental problem.
Complex, nook-and-cranny ones like the hull and the wing roots generated fine, even when they were NOT WELDED, and even with gaps between vertices that were supposed to be one!
So - the weld theory is blown out of the water too, Sur Builder deals with that exceedingly well and no fix is needed for that at all!
All we need it to do is to stop skipping some cmp groups.
It MUST be easy to find the cause once we know what to look for and what model will reproduce it - now we do know!
From LS’s comment it could be the Duplicate Radius, but I doubt it, I tried values of 0.01, 0.1, 1, 2, 5, and 10. The bigger the number the more obvious was the effect of combining vertices, below 1 it didn’t show much difference so I left it at 1 for the remaining tests.
By the way, is sur builder’s “1” the same as a MilkShape “1” unit, approximately 1 metre scale for FL ships?
I’ll be back!
BW: you need to get MilkShape and the importers and exporters please, so you can make some simple shapes and help us to help you to troubleshoot. It’s really not hard and is compulsory to fix this problem.
-
StarTrader wrote:
BW: you need to get MilkShape and the importers and exporters please, so you can make some simple shapes and help us to help you to troubleshoot. It’s really not hard and is compulsory to fix this problem.As I read through your last two posts, ST, I began to think that you and I should sit down over a pint and have you show me how to do some of these things so that I could run my own tests.
Great minds think alike, eh?
Now that is a scary thought!
-
adoxa wrote:
@Bejaymac: Unless you have the source to 1.1, I have little choice (I think it’d still be better to use 1.2 than start from scratch).I found two posts from Louva-Deus that suggest that the 1.2 code is a failed experiment, but the eoa.seriouszone.com main page is not working right now, so I could not investigate further:
-
I’ve made a request to Louva for the v1.1 source, hopefully he’ll see it soon and say yes.
-
-
i think i have the 1.1 sources
here : msSurimporter1.1
-
Well done Mirkha.
Gisteron: Are you sure you want them here? Don’t forget they are all older and grumpier too…!
-
That is the **Im**porter. But nice try.
ST: Any chance you can get the FLMM source while you are at it?
-
oh ?
sorry -__-/me go out
edit :
@ ST :
all here
i’ve done several tests
so, 2 sphere and a box, no way to generate a entire sur
the more i can do is a sphere + the box, or 2 spheresand it’s weird because the sur builder seems to generate all the 3 shapes but it’s not show in hardcmp …
for example, Type1 general allways the box and not the spheres
here the best result with Type2 :
as you can see the second sphere is generated in the sur builder but not seen in hardcmp
-
Bullwinkle wrote:
ST: Any chance you can get the FLMM source while you are at it?
I’ll ask, I think it’s Matthew Leider’s? Don’t know if he was in EOA?
Mirkha wrote:
… as you can see the second sphere is generated in the sur builder but not seen in hardcmp
Good work, Mirkha, many thanks. I import the surs into MilkShape because the HardCMP display is not always clear.
In your file the shapes that are generated are there - it’s just the red colour you chose for the material, it obscures the sur wires, but in my HardCMP v1.8 it shows as pink, see below…
Well I too have noticed that the outer shroud in the sur builder display seems to always cover the whole model, but we can’t see if the missing parts are being generated and not saved perhaps -
BW, is it fairly easy to put a line in that prints the name of each cmp group as soon as it is found? Or even just a counter that is printed, if we see the cmp groups are all being found, (or not) it will give us a good starting point.
Thanks.
-
-
StarTrader wrote:
I’ll ask, I think it’s Matthew Leider’s? Don’t know if he was in EOA?Yes. I sent Matt an email but got no response. Apparently he has left the scene. Somebody that I spoke with mentioned that Louva Deus might have a copy, and I have not gotten around to asking because I have my hands full with other stuff.
However, since you already initiated a conversation with Louva, FLMM is in seriously need of an overhaul. The SUR Builder is the kind of thing that a few modders will use a few times, but everybody uses FLMM. And most have trouble with it in one way or another.
BW, is it fairly easy to put a line in that prints the name of each cmp group as soon as it is found? Or even just a counter that is printed, if we see the cmp groups are all being found, (or not) it will give us a good starting point.
Why an immediate display? No matter… the main request is for a result report, and I will put that on the list to investigate.
-
Doesn’t need to be, could be in a log file or whatever is easiest as a tracing tool.
Actually I meant programmatically speaking, as in the next line of code! lol
Thinking about it a log may be best if we need more tracking output later to get closer to the problem.
-
Not sure if LD will have the the source for 1.1 as that was Colin’s project, but you never know.
As for FLMM LD does have the source for it, as it’s now an ongoing EOA project, Matt dropped the whole lot in his lap a couple of years ago, Crazy released FLMM 1.5 last year as part of that project.
-
yea, ST, i think it is sensible if they get up here. ofc, we are kinda new generation, many of us are either teenagers or at least under 30 but moralic standards are irrelevant for technical work. what we need is knowledge, the ability to explain and the wish to keep the game alive. it is unimportant whether we like each other as long as we keep rationalic (sounds vulcanic, don’t you think?). even to mention \v\ again: yes, we troll each other, which is most stupid time consumption ever invented. in end effect, you won’t disagree, that even such a project does its, even if tiny, but however is being a part of the initiative.
now, sorry, for offtopic…
isn’t it, that this v1.5 FLMM is even more buggy than 1.31? fortunately they call it a beta, so clean work is not to expect but once i tried out 1.4 and 1.5 me, and many others got back to 1.31, didn’t you? -
Update:
It’s taking me forever to isolate this…
OK, managed to get a 14-mixed-shape test file to generate complete sur parts as follows, but I know you won’t believe it!
What is not working?
1. Regular boxes (8 vertices, all 90 degree angles). One or more or even all always fail to be sur’d, no matter what the size. This is our problem.
2. Solid triangles (6 vertices) are frequently missed. When the length is close to 2 x width they can work but are unreliable,
3. The warning from LS & BW that minimum 5 vertices are needed for sur builder to work on a shape. So two faces with one common seam (4 vertices) will not work.
What is working?
Here are the shapes that do work correctly:
1. Sur Builder Settings: Sort: Type 2, No secondary sort, Duplicate Radius = 0.
2. We can have “boxes” made of 2-stack, 4-slice cylinders then they work fine - basically this is a double box, 12 vertices.
3. Thick triangles made using 1-stack 3-slice cylinders (8 vertices) are fine, even when one side is welded to make a knife-edge too (6 vertices).
4. Other shapes: spheres, geospheres, cylinders, “eggs” (stretched spheres) are OK.
5. Irregular-shaped boxes (looking down from above):
5a. Boxes MUST be tapered at the front or back, depending on their position and also depending on whether or not there is another shape beside it but further out from the centre line. This changes!
5b. If another shape is added the taper may need to be reversed!
I told you that you wouldn’t believe me.
One GREAT confirmation for all time: Parts do NOT need to be welded together to generate a good sur part in sur builder.
But Bejaymac is very possibly still right in that the sur may not work in-game, although his long and bad experience was not with surs made with sur builder.
On the other hand if the sur parts are all fine and present? So we need to test such a sur in-game to confirm. I don’t have time at the moment, please go ahead someone, make a ship with gaps and sur it and test it out. It WILL be fun seeing a rickety ship working in space!
How long did this proof take me? Don’t ask, my wife is so pissed with me you won’t believe. If we had a dog he’d be in the house and I’d be in his house for the next 2-3 weeks.
But this is how we avoid the box problem - now we found the cause, we have to get BW & LS to fix it.
Here’s the .ms3d file in a zip, it includes all the shapes, and the surs that were generated, so test it yourselves or make similar ones: