High Definition Props and Textures - Discussion thread

[z]

Thanks for clarifying!  Your second post makes much more sense.  It's true that we can't ever be 100% sure that any new features we add are free of bugs.  But by sufficient testing, we can reduce that possibility to almost zero, and I think that has been done here.  I'm happy to hear that you agree!

[callagrafx]

My god this battle between the pair of you is getting old.  Give it a rest already.

Don't look at me mate, I'm just ignoring him....

[Orange_o_]

I so agree, that I launched the production of HD prop 

[SimFox]

My god this battle between the pair of you is getting old.  Give it a rest already.

While Z's test was useful, I do think that there is a point that it needs to be tested trying to load multiple files, not just one instanced copy, that is most likely to be a larger drain on the processor.

Additionally, it needs to be tested in regular usage.  IE with the computer working on a transit network, animation and additional props and automata running.  Given that only Zooms 5 and 6 are affected this may not have a large effect, but we do need to test this.

Another thing to consider is load times, this appears to make a file 4 times larger.

There is still a lot of misconceptions going on about the whole thing.
We call them HD, but that's just a name that isn't technically correct even... There is nothing HD about these. Proper name would be Zoom6 Props as they are presented in native (eg 1FSH pixel for 1 screen pixel) resolution at zoom6 unlike the regular ones that at zoom6 have 1 pixel of actual FSH for every 4 pixel of screen. Definition is always same and is determined by your screen. So, 1 or 100 different HD-Props the load on CPU wouldn't be different as it has very little to none to do with any of it.
The switch in Game option between software and hardware mode is an interesting, but somewhat unclear one. I wonder if anyone has run across any data about it's functionality.
When it refers to the renders to the renderer what renderer is that? 3D or 3D/2D?
I think we can safely assume that at anyway 3D is there for sure. Given the age of the game option of giving such task to CPU is almost certain. But what about 2D? This has some relevance to the issue we are discussing and are trying to test here. If 2d (and that wold be scaling and tinting mostly) is done by the GPU then, given the power of modern GPU even the simplest ones there is SO much power that there is not point to be bothered at all, Particularly taking in account that ANY of the manipulation is done ONLY to pixel that is to be displayed and hence at maximum equals to the resolution of your screen. Even if you have total area of FSH needing resizing 20x that number. And that 2D Engine of GPU has very little else to do. So there isn't any drain on it's resources.
If on the other hand it is done by CPU in, say software mode (although I would really be surprised) same rules would apply, but of course the CPU cycles are used for many otehr things so some competicion for resources may ensue. But as I said this highly unlikely scenario.
What on the other hand would constitute clear drain on CPU resources/cycles is to complex LODs and excess props (for the reason of each and every bush, phone booth, bench, trash can etc) being a separate S3D model that has to be constructed (in software mode definitely by CPU) textured, and even if not textured on the basis of not being visible on screen (again I'm pretty sure that such an optimization given fixed view had been written in the 3d engine) such a prop will still be (at leas if it is a ground model) taken into account for collision calculations. This last function, I believe is always processed by CPU. And here complex LODs will make A LOT more work.

So, what are implication of it all.
Size of the FSH means nothing for the CPU. More complex 3d geometry may mean much more work for GPU and alway for CPU.

This said it should be clear that there is no relation to the running of simulators, networks sims or what ever. This are simply different function performed by different parts of your computer. And you GPU in this game is clearly underemployed. Take into account that increase in 3d power will likely be mostly wasted on game like this - these thing should be coded in 3d engine, while 2D hasn't really changed in last 10 years.

So the only part of game that works directly with HD prop - GPU has been tested (and tested scientifically and fully) by Z. Other systems are irrelevant in this case.

The issue with memory getting full because of this props... Again there is nothing specific to these props that should or can be taken into consideration here. For memory it's just bytes... doesn't matter what they are. HD prop or another skyscraper, tram skin or day/night mod. It doesn't matter. So there NOTHING specific to test.

[Jonathan]

So there NOTHING specific to test.

The thing to test is, when using HD props, how does performance compare to using the same props but SD.
And when using HD props is the performance overall affected and is it too much.

And testing those in real cities, so not test cities, taking a typical built up city with over 100,000/200,000 sims I 'd say.
I would test myself except I don't have any cities big enough.

btw,

We call them HD, but that's just a name that isn't technically correct even...

HD could stand for Higher Definition , in which case it is correct.

Jonathan

[choco]

how about, the arguments over semantics cease, so as to salvage this otherwise important thread and make something meaningful out of it....the continued drivel and useless bickering about what the meaning of "is" is has pretty much ruined any hopes i had of salvaging information from this thread.

[wouanagaine]

actually a gf8 card can handle around 8000 drawcall per frame at 60fps. What I name a "drawcall" is  sending to the gpu some vertices , 1 shader and 1 to 8 textures. Given the age of the game, I suspect there is no shader and the game use a fixedpipeline with 1 texture per drawcall ( ie 1 fsh ), the vertices send will be the one from the S3D that participate in the FSH sent
What I mean is that I can stall my gf8 by asking it to render 8000 3D cubes with 1 texture if I ask it to draw each cube one by one. Of course I can also draw more than 8000 cubes if I prepare my vertices so I can do only 1 drawcall for 8000 cubes ( of course all cubes will have the same texture )
So I hope SC4 engine actually one draw call for all props sharing the same FSH ( this is only a hope, I'll do some investigation ) because texture switching was the biggest hit for any 3D engine at that time.
As for size of the 'HD' FSH, it will impact the GPU only if all needed FSH for a particular view can't be stored on the video card, the driver will need to throw a texture out of the video card to store the needed one. Given the GPU we have now and the size of the city we see on zoom 6 I don't think this really matter anyway. If the game doesn't store the texture on GPU but relies on system memory to video memory each time, then it can have a bigger impact, however for the same reason ( a gf8 card can handle that ) it might not matter.
As for software mode, this is a totally different story, a bigger FSH means more CPU ( or more exactly more memory transfert involved which usually translate to more CPU )
Anyway someone playing in software mode shouldn't bother using HD props, his game is certainly already awfully slow

[z]

Wouanagaine, most of what you say makes sense (at least the parts I know enough about to understand).  However, I have to take issue with your last line:

Anyway someone playing in software mode shouldn't bother using HD props, his game is certainly already awfully slow

This implies that HD props display slower than SD props in software mode, when all my experiments have shown that this is not the case.  And so far, no one has come up with experiments to the contrary, nor do I expect they will.  This point actually has significant implications:

• For those people who run in software mode, if displaying HD props takes no longer than displaying SD props, they  might as well use HD props and get their benefits.
• If displaying HD props takes no longer than displaying SD props, there is no need to maintain duplicate sets of prop files at SC4D, one HD and one SD.  HD props should work for everyone.

I understand that people haven't had time to develop elaborate experiments, although I don't think these will shed any more light on the issue.  But for those who think there is still some question here, I'd like to see a hypothesis consistent with the data I've presented that explains how HD props could slow down the game.

[callagrafx]

But for those who think there is still some question here, I'd like to see a hypothesis consistent with the data I've presented that explains how HD props could slow down the game.

One Swallow does not make a summer....

More testing with more than one prop (not the same prop repeated over and over again) in view at zooms 5 and 6 in actual developed cities rather than sandboxes needs to be done.  Isolation is one method, live testing is quite another.  Why do you think we beta test everything in fully functional cities?  We are just guessing as to the structure of the game engine as we don't have the source code, so any new development requires exhaustive testing to make sure it won't screw people's cities up....And remember, while Simcity is technically a game, it's first and foremost a simulator without an end and people spend years developing their cities...we would be wholly irresponsible to shout Eureka!, create a bunch of HD props and unleash them on people who are attracted by the higher level of detail without knowing for sure that they are safe to use and don't cause slowdowns, CTD's etc.  The Prop Pox taught us that...

[buddybud]

Well i've seen some people testing and some people just complaining. How can we test if we don't DO first. Lets keep the testing going and see where it goes....the arguing is pointless. Nothing is released and there is presently no danger. You fear nothing except your own beliefs that are unfounded either way you look at it. The only way we can see the path is to clear the forest. Let's start chopping.......ie lets all shut up and get to work and report results........not opinion. Then we can decide what to do.


Cheers.

Edit complaining is the wrong word....rather i should have said speculating

[callagrafx]

Who's complaining?  No one is complaining, merely urging CAUTION when dealing with things that we don't fully understand....by the same token lets not run off at the mouth, half cocked and proclaim that this method is sound and will not cause problems. 

[RippleJet]

I also think it's time to stop arguing!

So there NOTHING specific to test.

We know that having a lot of custom content slows down the game.
We know that this is more true on zoom 1 than on zoom 6 though.

Having 5 GB of custom content is not ideal if you're having an old computer.
But we cannot yet say for sure what happens if those 5 GB went up to 10 GB due to all props becoming HD props.
At least loading time should increase...

We do know that between 1/2 and 2/3 of the savegame consists of the prop subfile.
A large city can have over 200,000 props.
Nobody can today say if and how much the game would slow down at zoom 6,
if the lion part of all custom props were made into HD props.

As long as we do not have thousands of HD props, we cannot give a final verdict.
And we certainly won't have thousands of HD props created over night.

I think we all can agree that one or a few HD props won't make any difference.
I also think we can agree that nothing can be said about it if
e.g. all BSC mega prop packs were modified to include a zoom 6 render.

The result will most likely take years to find out.
Till then, I'd suggest that we slowly start making HD props as optional add-ons.

That way they can always be removed, if the strain would increase after all.
We do not want a situation where we'd find that all props have been made into HD props,
and suddenly nobody can play their games without lagging...

[z]

I have to agree with just about everything you say, Tage.  And yes, if all props were HD, loading time would definitely increase, all other things being equal.  Yet it wouldn't double; among other things, custom buildings make up a significant portion of that load time.  (Although it's probably only a matter of time before we get our first HD skyscraper...)  The only other thing I'd like to add is that since it will likely take years for a total conversion to HD props (assuming the game is still being played then  ), then computing power will almost definitely have increased enough to counteract factors such as load time differences, and HD props in a few years will probably be a lot faster in general for the average user than SD props are today.

[dedgren]

I can never resist throwing my 2c worth in- this one might take a nickel.

Just about any ground-breaking thing truly worthwhile that has been developed for this game doesn't arrive fully-formed and ready to start using.  What's been going on in this thread is part of the process.  It's really important for everyone to keep in mind that there's no right or wrong approach or concept here.  At worst, HD textures and props (and, yeah, yeah, I've read why those aren't really precise terms) have the potential to make a big visual impact on the game- at best they could go a long way toward refreshing a game whose graphics challenged computers in 2003, but which increasingly suffer by comparison to state-of-the-art games today.  I'm not talking about 2-D versus 3-D, either.  Don't get me started- 3-D is, imho, little more than a trap (albeit a really, really attractive trap)- but that's a subject for another day.

I'm presuming that the folks who want to do serious work with this stuff, whatever is said in a thread like this, will.  Some will collaborate in groups of two or three, others will work alone.  To that end, it might be worthwhile to separate the public discussion part from the R&D part, and create some quiet workspace for those who will be doing the heavy lifting in taking this development wherever it goes.  It also might be worthwhile, to the extent the fears about memory overload are borne out, to think about setting up a NAM Team-like entity to assist the community in prioritizing what might be converted and what should be left for the days when memory and graphics are too cheap to meter (that's what our electric companies told us, some might remember, would happen when all power plants were nuclear).

I'm always amazed by how many great minds there are working on SC4 stuff at any given moment.  I'll cut things off at that- what more is there to say?


David

[Ennedi]

Dear friends, I think any emotions are completely unnecessary here. We have a really interesting problem and we should enjoy researching it.

1. A scientific experiment can be done both in undeveloped and in developed city. Both experiments can be useful. The only thing we absoulutely need is to create the same conditions for various experiments. If I have a developed city, I must make its backup first, then make an experiment without HD props (measure loading times and whatever you want), save the city if it is necessary for my experiment, but then return to the previous state (using backup file) and repeat an experiment with HD props. Even in the real life, for example in physical experiment, we can't eliminate all secondary conditions. We only must be sure that all these conditions will be the same if we will repeat our experiment.

2. If we want to make an useful experiment quickly, I would suggest to create some (not too many) most popular props and make a replacement mod for them. Even if an argument that 100 various props will affect the game more than 100 props of one type is true, we will at least know an influence of a big amount of props.

3. I'm sure that Barby, Lee, Tage and some other people know much better than me which props are most popular and which ones are the best material for experiments. But I would suggest to use some props which are not only "popular" (there are a lot of them in our cities), but which are the most demanding for our computers.
- Cars and people; these are animated props - can we use HD textures for them?
- Trees and bushes; I played with them a lot and I know that they are one of the most slowing down the game stuff. The same for ploppable water.

4. I suppose that we are rather safe and we shouldn't expect slowing down the game when it is running (Edit: However we should expect increasing of the game loading time). HD textures are stored  in .dat files at the place reserved for Zoom 5 (it is called Z4 in the dat) and the game takes them to create a Zoom 6 view. So they haven't any impact for zooms 1-4. When we are in Z6, we see 25% of an area visible in zoom 5. The game doesn't render the whole city tile when we are in zoom 4, 5 or 6 (however there is probably a minimal area which is always rendered depending of the place we look at). I suppose much more CPU (or GPU) resources is saved because of decreasing the viewing area than added because of using HD props. But of course it is only a supposition   and we must check it.

[wouanagaine]

Z, I meant that someone using software mode certainly have an old computer, and even if it can benefit the HD props, I really don't see a reason why he will overload his computer. And if you really want to make a test in software, then you shouldn't use a modern PC, but an old PC that has the only option to run in software mode, and I bet you'll see the difference . However I don't really think this is something we should care about. My point is, make HD props, those who have a very old computer are already limited by other factors

As for loading time, this won't make it longer as I expect the FSH to be read the first time it is needed, ie if you never go to zoom5/6, the corresponding FSH won't never be load. The only potential increase in loading time will be when you'll go to zoom5/6, and you'll notice it only if you zoom on a portion of your city where almost all props are HD  or when you scroll around forcing new FSH to be loaded and I doubt it will be noticeable anyway as there isn't that much FSH to be loaded ( very small portion of the city )

[Orange_o_]

Yesterday evening I have create hundred of different props HD, if somebody wishes to make an experiment I am ready to supply him.

[A200]

Uhm... it is 2009. This game came out in 2003. 6 years, 6 years. I read that by 2014, XP will not be supported by Microsoft. That is 5 years.

Apparently SC4 doesn't work with SOME Vista setups, and after that, Windows 7, who knows if it will be compatible?

Seeing as HD could take years to fully develop with all props, the fantastic work you are doing with these props might be wasted on a game which, in a few years, may be outdated and incompatible.

By then, I assume many other simcity style games will be out, who knows, something better than SCS might be out.

I think we should all just enjoy and play the game as it is, it will expire someday, it can't last forever, just like SimCity Classic, SimCity2000 and SimCity3000 have all reached their fate and have been superseded, SC4 itself will be superseded.

I know by 2010 I probably won't be playing SimCity anymore, it has grown old, not new or exciting. and I am soon going to stop downloading props- slowing down the computer, etc. we have reached the apex of SC4, any more plugins I feel would be wasted and just clog up the game and make it less enjoyable.

My advice: Don't do this, cherish the game as it is, and when it comes time to say goodbye, then say goodbye!

A200

Mmmm.. .where is your contribution or constructive criticism to this development ?  -jeronij-

[jeronij]

Hello to all,

Everytime that there is a new discover, some people wants to push it to its limits, and other people does not even hear about it. This is the human nature  . But this is a development thread, and we must focus on this: development. People with old/outdated computers may not be able to run these props ( I said "may not be able") but this applies for everything in computers world. I dont think we must waste a single more word on this sort of dicussions. It is up to the final user to use the new props. In RL,if you try to load your 1.5tons car with 3tons of anything, and your car crashes because of this, who is the reponsible for that?. I'd say you and not the car manufacturer 

Enough discussion about "what if"'s and more props (thanks O) and testing is what we need 


Oh, and pointless posts like A200's one are also not necessary in this thread. We need people contributing and giving ideas and energy here.

[A200]

So what, I am not allowed to comment and say that wasting 3 years in doing HD props, just as a visual eye candy, that will push out file sizes, slow the game down and just make it worse rather than better is a bad idea? O_O

Oh, and from what I have seen, there is no real improvement with these HD props