Bridge Commander Central
BC Forums => BC Modding => Topic started by: MarkyD on January 09, 2009, 04:08:02 PM
-
What is it? I have heard 15k is ok, but i have read through old pages here all night and I have read different opinions on what is what...
question:-
Using atleast 4x 2048 x 2048 texture sheets.... (current base standard) whats the poly restiction fellas.
-
BC has been able to play 160K models but most like to keep it at, near, or under 15K.
-
textures, polys, all add-up , thats why 15k poly,
-
I limit my models to 8000 polys or less. Not sure if that is the limit, but it is one I havent had troubles with.
-
Texture consumption is more the issue than poly count. 15k is reasonable.
-
Generally (and that means you're using no crazy resolution for textures, so sticking to what, 1024x1024 or maybe 2048x2048 tops) I think the golden rule for any current game (may differ per engine) is somewhat around:
< 4000: Plays smooth on all PC's
4000-8000: Plays smooth on most PC's (bar the very old PC's still using GeForce3/4 etc)
8000-15000: Plays smooth on modern PC's
15000 > : Plays smooth on modern high-end PC's.
That is not taking into consideration the largest fleet battle which can quickly lag up ofcourse.
-
with all the dual core crap comming out the 512meg vid cards being reasonably priced now don't worry about 15k. Alot of games like silent hunter 4 and such are far more intensive than BC and most people build their pc's for those type of games.
-
I wouldn't be too terribly concerned about the poly counts... I've been able to use DJ's Century [about 21k last time I checked] with almost totally maxed game settings on nearly any machine. I don't think it's really so much the computer's limitations as it is the archaic game engine... on a personal level, though, my goal has been to keep my stuff below 30k... [you know me, LOVE the uber-detailed stuff. :P] I mean, the models are only a couple megs each, so for the system to load the models isn't so much an issue as it is an issue for it to load a number of large super-detailed texture files. Really, it's the textures that hurt performance; 10, 20, or even 30 megs of textures per ship [depending on the ship and all that] can really hurt performance after a while, especially on an older machine with, say, a sad little 64 or 128 meg graphics card.
-
i would say go all out
-
I wouldn't be too terribly concerned about the poly counts... I've been able to use DJ's Century [about 21k last time I checked] with almost totally maxed game settings on nearly any machine. I don't think it's really so much the computer's limitations as it is the archaic game engine... on a personal level, though, my goal has been to keep my stuff below 30k... [you know me, LOVE the uber-detailed stuff. :P] I mean, the models are only a couple megs each, so for the system to load the models isn't so much an issue as it is an issue for it to load a number of large super-detailed texture files. Really, it's the textures that hurt performance; 10, 20, or even 30 megs of textures per ship [depending on the ship and all that] can really hurt performance after a while, especially on an older machine with, say, a sad little 64 or 128 meg graphics card.
If I had that ship to do over again, (which I might, lol) it would definitely not be that high. Probably closer to 15k. I really didn't need to use that many polys to get the job done, in retrospect.
You would really be amazed at how much detail you can model in with 10-15k if the mesh is clean and efficient.
-
i would say go all out
QFT and cookied!!!
I've taken to not even looking at the poly count of the stuff I'm making till I'm done, if I even look at all!
-
Not to sound old and frumpy, but...
That's modeling blasphemy.
There is some measure of pride inherent in knowing you can build a mesh that is the best you can possibly make it, and a minimum expenditure. Working on the principle that the ideal mesh was one that described exactly what you want in as few polygons as possible. It's a recogniztion of economy and reason. Not using more than you need to just because you can.
I used to operate under that principle, but I've found that as a modeler, it is far more satisfying, and better for gamers, to build a mesh that is both beautiful and economical.
For me, poly count is a matter of pride. It is a measure of my competency, and in my opinion, a measure of the competency of others, as well.
-
Ahh, that all makes sense. :]
I'm rather pleased with a current project I've been working on [that has yet to be posted, anywhere]. It's almost done, and has merely 5,000 polies so far. I suspect it won't be any more than 10k when it's entirely done with all the little details I want to incorporate... it's been a huge learning experience that I'll incorporate into the final rendition of the Mithra, which I've been planning for quite a while now. ;)
I suppose a measure of discretion with poly consumption and constructing these models is needed in order to not only be satisfied in your own work, but for other gamers to be satisfied in it as well? :]
In any case, I do look forward to your next shot at the Century Dj, whenever you get around to her again! :]
-
I try to be under 25k polys, but I learn fast. :)
-
You certainly do.
-
Not to sound old and frumpy, but...
That's modeling blasphemy.
There is some measure of pride inherent in knowing you can build a mesh that is the best you can possibly make it, and a minimum expenditure. Working on the principle that the ideal mesh was one that described exactly what you want in as few polygons as possible. It's a recogniztion of economy and reason. Not using more than you need to just because you can.
I used to operate under that principle, but I've found that as a modeler, it is far more satisfying, and better for gamers, to build a mesh that is both beautiful and economical.
For me, poly count is a matter of pride. It is a measure of my competency, and in my opinion, a measure of the competency of others, as well.
To clarify, I'm not condoning wasting polys. For example, using any more than 2 for a flat square is pointless and wasteful. However, in my opinion, quality should NEVER be compromised for poly count, which is why I don't have a limit for my Voyager model. I don't want to get tied down with a limit that I'll either have to break, or compromise on quality.
-
Quality and low poly counts are not mutually exclusive.
-
Low polycount is relative.
-
I'm sorry Prime, but you're incorrect. In Star Trek: Bridge Commander, a low poly count is not relative. In fact, the numbers are quite well established. It happens that I know this kind of thing fairly thoroughly, and you can trust that I hold a certain degree of intellectual authority in this matter.
-
I agree whole-heartedly with DJ. I couldn't have put it better myself.
-
Perhaps I'll elaborate, I was speaking generally. In BC? It's a 4 year old game, what do you expect?
-
It happens that I know this kind of thing fairly thoroughly, and you can trust that I hold a certain degree of intellectual authority in this matter.
Pop quiz: What the highest number of polys you can get one one ship in BC on a mid end machine with no lag?
-
Like already said, the polycount of a ship depends on the textures. BC can handle high polycounts much better than high res textures. When you want to do a ship with high res textures (multiple 2048x2048 maps) it's best when you hear on DJ.
Though when you go with low res textures, you can go much further than 15k...
Personally I try to stick below 15k and go for high res textures.
@limey: when i remember right, the deadalus of the upcoming ultimate stargate pack have around 200k, Is that right?
-
@limey: when i remember right, the deadalus of the upcoming ultimate stargate pack have around 200k, Is that right?
Yup, 220k. Works with no lag on a mid end machine. And thats just as high as its gone so far!
-
how well does it damage?
IIRC the higher the poly count the harder BC has to work to damage it.
-
how well does it damage?
IIRC the higher the poly count the harder BC has to work to damage it.
I don't think Dave ever said if it damages ok.
-
Fine, so you can plug in hundreds of thousands of polygons into BC's engine, even on a mid-range PC. (Which is what, these days, exactly?)
I had a conversation with Cord, who has managed to get as many as 800k into the game... a highly unstable game.
I did not claim to know everything, I claimed to have some intellectual authority. There are plenty of reasons why any good modeler in BC is economical about polygon use...
1. High numbers of polygons have been shown to make the game unstable.
2. Ships do not damage properly when they start having many tens of thousands, and then hundreds of thousands of polygons.
3. A lot of people out there still use older PC's, probably many more than we think. They should be able to enjoy the game too.
4. Due to the nature of BC's lighting rig, objects with too many edges look strange under direct light and weapons glow because the brightness goes through the mesh. There are no true shadows. An object in BC will almost always look better flat and smooth with a convincing set of textures.
5. Due to the nature of BC's camera, you can not get very close to a ship without the mesh starting to disappear. This means that using an excessive number of polygons, particularly in small areas on on small pieces of the mesh, is wasteful of system resources.
6. Making one ship tens of thousands of polygons quickly begins to add up if people want to play large battles. All of a sudden, all those mid range and older rigs out there start becoming exponentially more unstable.
7. Conversion can become vastly more complex and bug-ridden.
8. There is a much higher possibility that mesh errors such as broken edges and double-sided faces will escape the builder's attention and affect the mesh negatively. (In qualitative and technical respects, as well as in-game visual quality and stability.) To illustrate, if the average model is 97% error free, a mesh of 100k polys is going to have significantly more errors than a mesh of 10k.
9. High poly meshes are almost never a single-unified mesh, which BC prefers for damaging purposes. Also, a clean and unitary mesh is, without recourse, the mark of a skilled artist.
Now, here is why I am arguing this point so fervently against you limey. You've got an eye for ships, no doubt, but you could build your ships using a quarter to a half of the polygons that you do and achieve the same visual quality that you strive for. I know this because I've looked at your meshes. Wiley sent me his Excelsior a couple of weeks ago, which he was proudly (*and rightly so) touting as high poly. It's a beautiful piece of work, and many times, the community will get excited if you simply say something is high-poly. However, Wiley, who is growing and learning very quickly, was able to remove 20,000 polygons from his 40,000 polygon mesh with, and I stress this, zero degradation in the the visual quality of his mesh. So I see it this way: A mesh that is twice as economical with no visual degradation is a good trade off. Your Nova mesh, for example, could easily have the same thing done to it, and lose none of it's obvious visual quality.
I'm not saying all this to piss you or anyone off. I say all this because I've been around here for a while, I do know what I'm doing, and I see it as a responsibility of mine to ensure that new and young modders who come into this community are given some support and education when and where it is appropriate. The fact is that, despite some of the great things going on in this community, it has never been slower, and it is slowing down. If we allow ourselves to simply overlook a cornerstone of good mesh production, the overall quality of the products that come from this community will have diminished which has in other communities most certainly been a death knell.
In conclusion, I am not 100% opposed to a 50k mesh. Honestly, I am not. However, if we are going to be pushing boundaries in this community (which I am for), we must be sure that we are pushing the boundary for the right reason. We must be sure that a mesh that's 50k polygons NEEDS every single last one of its triangles, and that we are not making a mesh "high-poly" just for the sake of making it so, or "exciting" the community at large, who in my experience, knows virtually nothing about anything that we do here.
I am not an advocate for low polygon meshes. I am an advocate for good ones.
-
Here's my thoughts on this...
Poly counts are easier for the computer to handle, but this also depends on the game engine, gfx card and the power of your cpu. The more you put in, the more each of these has to work. Especially if they are a moving object. And I would believe that the BC game engine would have to work that much harder in order for the damaging effect to work properly. Thus the reason for some very high _vox files, or very low ones. (In terms of Megabytes) Plus loading times for when you load those objects for the first time.
I would prefer to set a limit for myself and create a model that does not exceed that limit. Not to challenge anyone else, but to challenge myself. I am the only person that I need to beat. If I can create a model with such high detail, yet have a fairly low poly count...then I have done what I set out to do. I look up to the best, not to find out how they create those masterpieces...but to find out why? when? where? what? There is a reason for everything, even if you don't agree with it.
As for ultra high poly counts, there is a new geometry card (Like a graphics card) that can help take the load off the geometry engines of your pc. It's like it's own cpu that specificaly targets polys and their positions, movement, etc.
The poly count of 1 ship may not take up a lot of power. But remember, there are usually more ships in the game so it all adds up.
Yup, over 800k polys in one game. Game crashed on the fifth time taking a screenshot.
-
I would like to present this to all takers in this thread, to get opinions.
I like to keep my models around 6000 to 8000 polys, one single object with no errors, 2 sets of 3 2048 maps (diff, spec and glow).
The question I am posing is is there anything in there I should change? Are the maps too big? etc.
Thanks in advance.
-
Well, with regards to the maps, I typically use 4 2048x2048 maps. It seems to have become the standard. However, I try to include directions on how to resize the textures to a lower level for anyone who's running a lower end rig. I don't include 1024's or 512 in the Medium and low folders because it quickly ramps up the download size.
It sounds like you're running in about the right areas. 6k-8k for polys is definitely lower end for BC these days, no doubt, so i wouldn't worry a lot about going over too much if you feel the mesh will benefit from it. That being said, you do a great job of beveling textures and making things pop, so you can get away with a lower poly mesh in areas of detail.
-
@ everyone
Lots of info, makes a good read. Thanks.
-
I agree, Moed. Me being a noob modeler, this discussion really helped. Just in case anyone thinks of it, don't delete any part of these posts - they are very helpful.
As for the question about poly count, I heartily agree with DJ: if it requires it, do so, but no more. If a 40k mesh looks as good as a 20k, go with 20k. As stated above, some of us cannot afford kick-a equipment whenever we want it. My rigs are as such:
-Laptop with GeForce 9300m GS 256 MB
-Desktop with Integrated ATI Radeon xpress 1350 (I think)
-
More rarely equals better. Paying more for an item certainly doesn't mean it's better quality, and in mathematics we know, courtesy of Pythagoren, that taking the shortest route is the most efficient.
Quite frankly, you can't see any differences in game between two similar meshes, but one with double the poly count - there is no point.
The husband speaks truth. :arms:
-
More rarely equals better.? Paying more for an item certainly doesn't mean it's better quality, and in mathematics we know, courtesy of Pythagoren, that taking the shortest route is the most efficient.?
Quite frankly, you can't see any differences in game between two similar meshes, but one with double the poly count - there is no point.?
The husband speaks truth.? ?:arms:
Would that be him in your display pic? ?LOL
Seriously, I side with DJ on this one also. ?I can easily run 2 or 3 CG Sovereign's about 3 of DJ's Galaxy's and a few LC Intrepids with hardly any lag at all, and this is a piece of crap PC :D. ?They all look great in game. ?It should be fun for all of us in this community :). ?That's why when I actually learn how to model properly, my models will be somewhere between 8 to 12k, and not anymore than that. And I'll probably run 1024x1024 textures with those models. You can still get a great ammount of detail in those sizes.
-
What I have to add on stuff DJ said:
1) Make a strategy of how you're going to build the model as easily as possible
2) Make your model and then optimize it (remove unneeded polies (vert weld))
3) Think of a strategy of how you are going to map the model using less texture maps vs. texture crispness (I usually use a single map on ships below 200m's - my Oberth is a nice example)
4) map the model
Now, a few things I want to point out (no pun intended for anyone, just using recent examples):
1) Use the symmetry modifier both in building the mesh and mapping
2) make the mesh unified (single object with single element)
3) use up as much of the map area, use the tiles also and areas not used on textures should be used for changed structure texturing (Novi Sad's saucer and secondary hull textures are nice examples, also the Akyazi packs textures)
4) I always make an ID map, where I put the areas where the registries and penants, stripes, etc. deflector, impulse engines and shuttlebay doors go. Gives more room for variety (again, the Novi Sad pack is a nice example)
5) when kitbashing a ship, make sure that you build upon the existing mesh, and keep the results unified meshes, textures interchangeable (the Luxor fleet so far has 24 ships using (not counting the extra ship ID maps) a total of 8 textures. Just imagine how much texture resources and disk space I have saved by doing that.
6) the Skinning and Damaging tool is a really underrated script
7) the scripts/ships plugins have some really useful commands in them that can help with performance of a ship
8) NEVER make a ships damage radius in the HP bigger than 1 for a ship that's shorter than 2km's. It's plain stupid, and can result in ships which have all their hull gone and still alive...baaaad idea
9) NEVER make phaser damage radiuses greater than 0.1 and torpedo damage radiuses greater than 0.25, same reason as under 8)
10) Don't make fanboy ships, and fanboy retextures of existing ones, it's boring and a waste of time. We have too many armoured <insert random ship class name here> and similar pointless mods
11) don't reinvent the wheel, do "TNG Mirandas" with Sovereign parts only as a private test or practice, do something constructive for a change
12) don't rush your mods, take your time with them, finetune them, redo them a few times if you're new to modelling and texturing. Do it right, and not slapping some random ship just so you can have a release per week
13) Think your ship through. Count your decks, imagine how those decks would be set up on a ship, where goes what, how big should it be
14) (among the most important rules) learn how star trek ships work, what parts they have, where are they going, what subsystems do what, how much there is of each, which ones are visible on the exterior, etc. A well thought out ship deserves more praise than "a 100 kitbash releases of the week".
15) when you're doing a kitbash, you must remember two things: how the exterior change will affect the interior, and how hard would for an engineering team be to do those changes
16) we today know how to animate textures and how to make transparent textures, etc. use them on your ships (I have a ship in the works which has the interior of the nacelles visible - buzzard collectors and the warp coils)
-
More rarely equals better. Paying more for an item certainly doesn't mean it's better quality, and in mathematics we know, courtesy of Pythagoren, that taking the shortest route is the most efficient.
Quite frankly, you can't see any differences in game between two similar meshes, but one with double the poly count - there is no point.
The husband speaks truth. :arms:
As I have said before, I'm against wasting polys. What I really don't like though, is imposing absolute, unbreakable upper limits on poly counts.
-
I come from SFC, and the polylimit on that tends to be a bit harsher than on BC.
When I model, I tried to keep a certain rule of thumb.
Fighters and shuttles shouldn't break 1k
Frigates and Destroyers shouldn't break 4k
Light cruiser < 6k
anything else <8k
Stations <10k
However, I've recently learned that my textures aren't getting the level of detail I want, so I've now doubled each one.
Fighters and Shuttles <2k
FFs and DD <8k
Anything else (I've moved Light Cruisers up) <15k
Stations <20k
I do the same thing with my textures. When I build Textures I keep certain a certain scale.
Fighters, Shuttles, and Frigates = 1024
Destroyers, CLs, and CAs = 2048
DNs and above = 4096.
I imagine, though, when it gets converted to games, the cap will be 2048.
-
My God, i'm posting here again..
Well, DJ forced me into this, because the poor guy is arguing alone on a discussion people had here for years.
(Thanks on god points too Adonis).
Also, good quote on Exe's sig DJ ;)
Fact:
BC engine can handle millions of polies if using a high end system, but the engine was not designed to do so.
In fact it was not designed to deal with any object that has more than 10k (stock ingame models had 3k on high lod).
When it's forced to do so, it straves and you'll lose the right ability to damage the ship or the chance to use good textures.
The hard limit people list here (15k) was stabilished by SNS and CG when they did both the galaxy (13k) and sov (15k) and by me when i did the excal 0.9 (12k), before those 3 ships we used to count the bc limits around 9k.
But even with that limit, i still consider 8-10k a very hard limit and i challenge any of you to make a ship with more than that which i cannot duplicate under 10k, with no visible quality loss.
Just to post an example, my intrepid stays confortable under 8,5k and it's smooth and detailed as a good ship must be.
So, to answer the basic question at the start.
IF you cannot keep it under 15k, then you need to go back and refine your modelling skills and learn more. No single trek ship should go over it.
If you can do it under 10k, you're starting to get into the "game modeller" feeling
If you can do it under 8k and it does look good, i can learn to respect you.
LC
-
Here here :P
It's always harder to make a ship look good with less polies, than it is using more polies.
-
Legacy
My models might be under 8k and look decent, but I think I have a LONG way to go before you think my work deserves the respect of a modeler as skilled as yourself ;)
10k is the absolute max for me. I wont go over it no matter how big I make the concept...ill delete modeled details and put them in in textures if I have to.
The one thing I have always been good at, even when my modeling skills were poor (wait...they still are pretty poor lol) was greebles in textures. Ive been a concept artist since I was a child, and learned how to busy up an area to make it look mechanical/industrial.
So, simply put, put more into the textures and less into the model. The right texture details can make all the difference.
-
Well Raven, it seems like you've made quite a progress since you started, at the point you can make your own ships alone and at the end they will look good.
Obviously there's the fact that a canon ship is a lot harder to do as the surfaces usually are put in a position that does not match where you have the polygons (usually becomes painfull pretty fast), but as you model over your own concepts, i still think it's a very good showcase.
Lemme know if you have any info or help, but i don't think you actually need those.
-
Is it me or are most good quality BC ships over 15k these days? Shoot me if I'm wrong but thats the impression I got.
-
Two schools of thought here....
Texture mechanics/purists/founders/conservatives....people like BBarr, Moonraker, Myself, LC, Adonis, Mindblip, Wicked Zombie....we are talking people that were here at the beginning of modding Star Trek games (I myself started with modding SFC in 1999....Newships project anyone?) or at the very beginning of modding BC. People that, for one reason or another, keep the polys low and take pride it.
We are from the days when you had to limit ships to under 2k polys (500 was best) and 256 textures. Although I am a novice with respect to modeling, I was around at the very beginning of modding star trek games...from that point I think I might have something to say about modding ships with merit. Try to remember that lots of these guys are veterans, and should receive some respect for that.
Except for me...throw bricks, just make sure they are white ones....I hate red brick. Tacky.
Then you have what I call progressives....lots of examples here, incredible work where the ships push 25k polys and knock your socks off. There seems to be more put into the model and less in textures...by no means an insult to the fine texture work, but rather an example of priority. These are ships that look lovely even skinny (sans textures) and dress to impress.
Which is the right way? Depends on a few things.
1) Do you want lots of ships onscreen?
2) Do you want a single ship that burns with beauty?
3) How many platforms do you want to support?
It takes ALOT of skill to keep the polys low. This is beyond debate. The reason is you have to trick the eyes with textures...that means alot of weathering, drop shadows, sub contrast lines, washouts, etc. Making a ship that has an impressive presence AND is under 10k is something to be very proud of.
However, you cannot ignore the obvious skill, beauty and craftsmanship that goes into a high poly model.
I guess the point is this...making a high poly model (modeling in details) is easier, at least in the eyes of veteran modelers perhaps, compared to low poly work. Perhaps it really depends on perspective.
Just please keep in mind that LC has been around since the beginning and is a marquis modeler. I doubt many would dispute this.
I also think that people like Chronowerx and Limey98 have made huge contributions to this community, and should not be dismissed or ignored.
An opinion of my own to add to this wall of text (not like ive ever been opinionated or anything :b). I feel that doing canon work is MUCH harder than original work. I just wanted to add that tidbit.
Edited by El.
-
On more thing to add. I have to give some props.
If anyone here thinks my work is halfway worth anything thank LC and DJ for it. LC built quite a few of the initial BC2 designs (and if I hadnt been such a tightwad dweeb he probably would have done alot more)...it was those meshes, specifically studying them, like the settings, how they were made, proper lighting, etc that got me really started.
DJ is my only hope for making reasonable feds. My feds suck, period lol. However, DJ has a wonderful eye when it comes to small visual cues...like how the graduated decks are on a saucer, or the proper proportions on a saucer edge or nacelle...good profiles, good clash on paneling, etc. that gave me a million ideas that were worlds better than what I was doing.
When I do post some shots Ill make sure to specifically point out where I ripped him off and therefore bow to his greatness lol.
-
Raven put it perfectly. Neither way is right or wrong generally, but rather based on what your/your targets PC can handle. There are those of us like myself who would rather the typical trek conflict of two or three ships, that look spectacular and can almost trick the eyes into not seeing the edges and add to the immersion. There are also those of us who either don't have the computer or just generally prefer to have massive battles, dodging and weaving between phaser fire, debris and other vessels that whiz by so fast you don't notice the lower polycounts and the corners that ensue with such. (Not that I'm saying it looks terrible, just that you can't deny a lower polycount generally means less smooth edges.)
Edited by El.
-
Ok my personal views.
Don't sacrifice smoothness for polies, but don't make it unnessessarily high. From what I see 20k looks nice to me. But i'd worry more about textures making an impact on a PC. Along the lines of what DJ does seems the optimal polycount for me.
-
Ok, I've cleaned the general spam, antagonistic posts, and most of the sarcasm from this thread.
This means some of you have had your posts edited so they still make as much sense as possible.
To avoid me losing another 15 minutes of my life which I'll never get back, can we keep this thread civil, spam free and respectful of the opinions of others.
Thanks
El
-
I dunno some of the higher poly models kill my rig, A64 3500 and a Geforce 6800 256mb
-
I dunno some of the higher poly models kill my rig, A64 3500 and a Geforce 6800 256mb
Might also be the texture size. I had a Geforce 6800 with 256 MB and it couldn't handle anything over 1024x1024 but I could run models with ~50k polys.
-
well it might be, but some of my SFC models use 2048x2048 on a few places, doesnt lag me, but it might on BC dunno
-
In some of the 3d programs i have used, Light Maps were a huge cause of concern if you had to many light objects PLUS the Polygon count,Plus textures=stacked up to kill if rendered in a non high end system..does BC converted meshes worry on light maps?
-
I have s ship that is 58000 and another that is 40000 and they can run smoothly on ATI 3400 series VGA
-
The last ship I modelled came in at 64k polies I think. I honestly don't know how I'd reduce that number without compromising quality, but that's why these guys are the mod gods and I'm not. ;)
-
Is it the gfx card of cpu that does the work when it comes to polies?
-
It is safe to say that the poly limit for BC is over 9000.
-
Wow my Yorktown Refit is about 50 000 polys
-
Wow my Yorktown Refit is about 50 000 polys
Note: High amount of polygons doesn't automatically mean "high quality". It all depends on how you use the polygons. I've seen ships with 5-6k polygons that looked better than some of the 10-15k models.
-
Absolutely true DT.
-
definitely...
-
The Yorktown has a around 350 polys per phaser. First it was 6500 polys per phaser but Milkshape crashed and I tried to reduce polygons to around 300-400. In my christmas fix pack I will fix everything buggy.
-
In my more recent works [namely bridge and interior design project for the Mithra, oh how long I have worked in it... lol] I've often found that putting a crap ton of polies just to smooth the whole thing out is pretty wasteful; you don't always notice it anyway. best thing to do, is to put those wasted polies into the edges. A grand example could be saucer sections. pisses me off seeing saucers all jagged-edged or sharp-edged when they should be [well, generally anyways] smooth and clean, not like someone could cut their foot walking on the darn thing. Spend the polies where "jaggedness" is going to be noticed and criticized, not on the project as a whole wherein a lot is going to be wasted.
Aeries, slowly learning not to be wasteful. xD
-
A thought occurs... Could a couple of modders post a few wire mesh views (or solid views with the poly edges highlighted or something), in order to demonstrate good mesh quality? Perhaps a few comparisons with less advanced meshes in order to highlight some common mistakes?
Could help the rest of us learn where we're going wrong.
-
You got it, RC. :]
Pic one:
Base shape created from a spline outline then lathed. Pretty raw, untouched and not modified at all. Obviously, very disgusting and not usable for any model these days, even low-standards.
(http://img.photobucket.com/albums/v97/Zake_1/A3r13s/Icky.jpg)
Pic two:
Using the loop tool and highlight functions, remove the un-needed edges then the un-needed verts they connect to, in areas they aren't required, to keep the mesh at a nicer and significantly lower poly count. Then, using the Chamfer tool, chamfering the vertical edges of the saucer to created a nicer, smoother and far less jagged outline.
(http://img.photobucket.com/albums/v97/Zake_1/A3r13s/Alittlebetter.jpg)
Pic Three:
Be sure to weld/target-weld the un-needed verts!!
(http://img.photobucket.com/albums/v97/Zake_1/A3r13s/Better.jpg)
Pic Four:
Chamfer the horizontal edges to dull the saucer edge, so as to not look like you can cut your finger on it. ;)
(http://img.photobucket.com/albums/v97/Zake_1/A3r13s/Evenbetter.jpg)
Pic Five:
Not WLAYS needed for modeling in CGI settings, but in BC and many/most games, triangulating is a must. a VERY simple way to do this is to convert the mesh into an editable patch, then back to editable poly/mesh/whatever you wanna use. [I like poly, much more versatile.]
(http://img.photobucket.com/albums/v97/Zake_1/A3r13s/Andkindasotradone.jpg)
NOTE: If you need to keep the mesh perfectly symmetrical, make sure the pivot point is centered to the mesh and the axis you want to be symmetrical, and use the Modifiers -> Mesh Editing -> Symmetry tool
Like I said, this is really rough. there's still a bunch of things you can do to improve upon the mesh and optimize it a bit, but this should get the point across. Happy modeling. :D
-
Not WLAYS needed for modeling in CGI settings, but in BC and many/most games, triangulating is a must. a VERY simple way to do this is to convert the mesh into an editable patch, then back to editable poly/mesh/whatever you wanna use. [I like poly, much more versatile.]
The nif exporter triangulates the mesh itself. I think others would too, so that's not much of a concern.
NOTE: If you need to keep the mesh perfectly symmetrical, make sure the pivot point is centered to the mesh and the axis you want to be symmetrical, and use the Modifiers -> Mesh Editing -> Symmetry tool
I personally find its easier to model the thing as a whole, making sure at least one half of it is perfect, then splitting it and using a symmetry modifier right at the end. Though if you do, remember to remove any unneeded faces making the seam will create.
-
Thanks Aeries. Getting a good edge on something without wasting polies in the middle is something I've always found difficult, and you've solved it straight off. Cookied!
I've always modeled by symmetry, and it is important to keep the polys centred exactly on the axis. In lightwave there's a 'set value' tool. If I screw up and end up with a few off-axis, I can select all the affected polys and set the offending coordinate back to zero.
-
Agreed, very handy, Aeries, mate! Cookie for you!
Also, RC, seeing as we both have Lightwave (Albeit different versions), Is Chamfer listed under a different name? I can't find the dang thing!
-
Hmmm, turns out we don't have a chamfer tool in lightwave. That's going to make things trickier.
-
Found something handy...
There is an alternative in Lightwave. Not as great, and very basic, but using Smooth Shift gives a similar result.
-
It doesn't actually. I use smooth shift a lot for box modelling, and it doesn't increase the number of polys around the edge, it just extrudes the existing polies.
-
Ah, that would explain why it won't double up... Seems only 9.5 has a chamfer called "Rounder" :(
-
9.5? Lol, I'm using 5.6. From my KA days. I think if you have a reasonably up to date LW, you should be able to find plugins for it.
-
Here is something also.
When working on a low poly mesh, that needs that EXTRA detail in places.
I will continue to add examples of this during the build of this andorian ship.
by cutting into the mesh and connecting in new lines here and there, I managed to modify my simple shape to allow me to add those detailed extrusions in a clean manner.
-
Dang guys! I tried makin a ship back in 03' for BC but I did not have the get-go, nor the understating of this kind of work. To this day, I cannot even understand basic modeling lol. Work by pure genius if you ask me. But if it weren't for you guy's expertise, my videos would not be possible. It all goes to you guys.
*Raises glass of beer*
-
Hmm. One of these days I'm gonna have to write some tutorials, methinks.... xD
-
Dang guys! I tried makin a ship back in 03' for BC but I did not have the get-go, nor the understating of this kind of work. To this day, I cannot even understand basic modeling lol. Work by pure genius if you ask me. But if it weren't for you guy's expertise, my videos would not be possible. It all goes to you guys.
*Raises glass of beer*
just grab gmax and fiddle lol thats how i got going granted im not that good yet
-
It was always drilled into me to avoid triangles and 5 sided polys like the plague. Part of my frustration has been hunting down all faces with less or more than 4 edges and fixing them. If this is not true in this type of artwork, then that will make things significantly easier.
Does anyone else use Maya?
-Joe
-
I can't think of anyone that uses Maya.
As for the number of sides, keep it either to 3 or 4. I'm not entirely sure, but I believe the conversion process breaks all polies down into triangles anyways, which is easy to do for 4-sided polies as it just bisects them.
-
Well that's the general logic for video games, but I studied animation more as it pertained to film which is why they taught Maya at school lol.
So the logic was tri's would not sub-divide correctly, but quads would. When you were rendering out to static frames you didn't have to worry about tessellation ruining your frame rate, so it was more important to avoid them for the sake of geometry appearance after smoothing. Since the ships in this game will not be sub-divided to keep poly counts low, that becomes a moot issue.
This might be why I am having trouble modeling mechanical objects like ships. I need to think less "pixar" more "CAD" haha
-Joe
-
I'd encourage you to take a look at the meshwork of DJ Curtis's and WileyCoyote's ships. They're pretty much the upper end of BC standards these days. If you're not sure how to convert BC's .nif's to something usable in Maya, read:
- download and install NifSkope
- download relevant models
- open relevant .nif's in NifSkope
- Under "File" there should be options to convert the .nif to .3ds or .obj. I'm sure Maya can import one of these.
-
I'd encourage you to take a look at the meshwork of DJ Curtis's and WileyCoyote's ships. They're pretty much the upper end of BC standards these days. If you're not sure how to convert BC's .nif's to something usable in Maya, read:
- download and install NifSkope
- download relevant models
- open relevant .nif's in NifSkope
- Under "File" there should be options to convert the .nif to .3ds or .obj. I'm sure Maya can import one of these.
thanks for that info. do people usually start from an existing model or something?
-
No, unless someone is refitting or updating a design. I'm just saying this so you get an idea of what to shoot for before you send it off for hardpointing and conversion.
-
thats a good point, thanks.
-
We're always in need of good modellers here at BCC, and from the way you talk, I have the impression you're already familiar with modelling, just not quite with BC. So it's anything I can do to help. ;)
-
hmm...he's got a point mentioning CAD (which i trained on in the early ninties as part of my Drafting teaching). the models are actually on the primitive end, as the game was originally developed in the late 90's. things have come a very long way since this game was developed, in both technology and appearance tolerance.
by the way, if you want to see scary, take a look at the bone-stock BC's models. only four or so survive in today's modding scene. most of the others have been replaced or redone with higher quality models.
the survivng stockers are
Ferengi Marauder
Cardassian Hybrid
small federation stations
small cardassian stations
a little side note is that the character models have never been tweaked. the game engine refuses to recognise character models from a higher version than the one used to create the game with. sets, such as the bridges, also have to be made using the older versions of the MAX program they were created with in order to be used. people don't want to mess with the character models, however.
a lot of Ideas and Modders have migrated over to the Excalibur project, a scratch rebuild of the game engine and so forth to use modern higher resolution and tighter polygon versions. since the original game engine and so forth has NEVER been released to the public for reapir, people are simply starting over. the Bridge Commander game engine has a horrendous memory-leak problem that can't be legally fixed.
-
a little side note is that the character models have never been tweaked. the game engine refuses to recognise character models from a higher version than the one used to create the game with.
That's a bummer. I would love to create more realistic bridge officers with different uniform options and species.
-
you'd have to figure out how to attach revised bodies to the modeling dummies, figure out how to get the head and hand textures to make sense, etc. something about the character animations make it hard (but at least someone could fix the hands!)
if you take a shot, i think, like the Bridges, they need Max 3.2
believe me, some of us would like to make a few body tweaks for personal use. best I can do is some costume tweaking.
-
What about new characters from scratch? The you wouldn't have to worry about conforming things to the existing character. New rigs with newly painted weights eliminates the texture issues.
-
take a look at Lurok's take on the DS9 ops. it shows the base pose skeleton in shots. again, the game engine will CRASH with any bits newer than what it was created with back in 1999-2000.
if guys like Sovereign and Defiant would chime in here (they're the master scripters around here), they'd give you a better idea of how much of a pain in the butt it is to work with the characters, let alone the character models.
-
yeah about that, I'm not sure if I exported this right, since I've been doing it the Gmax way.
I tried to export my Crossfield class mesh into BC, and so far it works like a charm until
(https://cdn.discordapp.com/attachments/351879241438068736/502851817965748235/unknown.png)
This only happens with the saucer and god forbid if you lop off a nacelle and the whole thing goes explosion.
This does not make any logical sense to me as to why this is happening, because I have converted much higher resolution meshes I made without problems or worry.
-
Have you tried deleting the VOX.nif files just to see if they're the problem?
-
I have and that's not the issue.
So far the only explanation I can come up with is that the model I made is too high poly.
-
yeah about that, I'm not sure if I exported this right, since I've been doing it the Gmax way.
I tried to export my Crossfield class mesh into BC, and so far it works like a charm until
This only happens with the saucer and god forbid if you lop off a nacelle and the whole thing goes explosion.
This does not make any logical sense to me as to why this is happening, because I have converted much higher resolution meshes I made without problems or worry.
This is not necessarily a high poly issue. Unless your model is over 100k polys, BC will still load it (although keeping your model to less than 25k is better because it is far less laggy on the older game engine).
This might be the same mysterious 'missing poly' issue that I have had with a few models I've been trying to convert. Both Gmax and Nifskope show the model completely intact, but in-game the ship is missing certain parts.
Your problem seems to look similar to mine, although I'm not completely sure TBH. A more experienced 3D modeler than me might be able to shed more light on this.
-
In my case, the model runs perfect so long as not a single photon blast or something that can make a decent sized hole decides to make a dent on her, then the faces disappear.
Idk what to do here. I tried cleaning the mesh, decimating it, even removing unnecessary parts....nothing works...
-
I've asked to test this myself since I have an older build, scripts, and modeling programs. having the identical problems to Ported ships is an eyebrow raiser.
unfortunately, around here, this former texturer and tweaker is LITERALLY the 'senior modeler', now. :P
-
Well, after a bit of tweaking and experimentation, I came to the conclusion that the standard Disco version of the Crossfield class may perfom rather....fragile for Bridge commander. I have made sure the model is water tight, strength and radius mods for damage were implemented and I've done a cross experiment with ANOTHER version of the Crossfield, one that is more TOS like. Turns out the TOS Discovery performs better because the saucer has a type of grill connecting it all together, where as the Discovery version has just clamps connecting two saucers together.
All of the mesh "vox problems" I have is with the saucer section: a problem solved in the Refit but still there for Discovery. Times like this I wish Baz1701 were still around..
This is the standard (yes no textures were made yet but the models are perfectly UV'd for experimentation)
https://cdn.discordapp.com/attachments/371722416436674561/503343125339570196/unknown.png
Atm I came to one final conclusion: a final part that I have not fully looked after.
This is the Refit.
(https://cdn.discordapp.com/attachments/371722416436674561/504050981919653888/unknown.png)
-
Play around with the damage radius, the default 1.0 is waaay to big for things to look reallistic ;)
-
From testing I did several months ago here are some observations I made about how models work in BC and some of their limitations.
1: It appears that each element of a model (each section of it) has a limit of approximately 97,821 triangles
2: Elements of a model can be combined (like the nacelles on the stock galaxy class are their own element) to have a total of around approximately 166,658 triangles
3: The game appears to have an extremely high limit on the total amount of triangles it can display, a ship that has the max 166,658 triangles can have many multiples of it loaded on screen without issue (for my test I had a total of 10,666,112 triangles on screen)
4: The game has a hard time when vertices too near one another, on high detail models this will cause faces to not appear in game.
(https://cdn.discordapp.com/attachments/351879241438068736/683835971476783104/unknown.png)
Here is 27 stock galaxy models compressed into one model element, 97,821 triangles
(https://cdn.discordapp.com/attachments/351879241438068736/683840212043956303/unknown.png)
Here is 81 stock galaxy models compressed into one model element, 293,463 triangles. Notice how the nacelles have vanished off of 2/3rds of the ships, they are being culled due to limitations.
(https://cdn.discordapp.com/attachments/351879241438068736/683864159750062086/unknown.png)
Here is many hundreds of stock galaxy models that were combined into groups of around 30 (I don't remember). Total onscreen triangle count is 10,666,112.
(https://cdn.discordapp.com/attachments/351879241438068736/683818808250859520/unknown.png)
Example of a high triangle count model that has vertices too near one another.
-
Very useful and awesome info!! :thumbsup:
-
Very useful and awesome info!! :thumbsup:
Yea, thats right! :bow: :shocked: