1  ...  5  6  7  8  9  10  ...  42  |   Order by default

Jul 23, 2002, 10:12am

http://tnlc.com/rw/files.html#exporters to get Maya/Max DFF exporters with source code. ZModeler (http://www.zmodeler.com/) can now import (and soon export) DFF format, due to the overwhelming demand by Grand Theft Auto 3 modders. I'm also trying to get someone to create a MilkShape 3D (http://www.swissquake.ch/chumbalum-soft/) filter for DFF format but no takers yet. :/

[View Quote] > There's nearly nothing about it. The only way I found out of using DFF files
> was to get ProGate Suite (no price given) and convert RWX DFF to compact DFF
> files. I certainly wouldn't pay for it unless i knew it worked well and I
> can't even find a price. Gah!
>
[View Quote]

Jul 23, 2002, 5:02pm

[View Quote] > Looking around, I can't find any but are there any programs that can convert
> RWX to DFF (open in RWX, export to DFF).

Not directly, but you can convert RWXes to 3DS or Maya, for example, then export to DFF in 3DS Max 3.x or Maya 2-3 with the DFF exporters.

> And does using DFF actually make it noticably faster?

I haven't noticed any speed increase but I haven't really tested it. The only speed increase, if any, would come at rendering, but to test that you would probably need to take a complex RWX that freezes AW while it renders, convert it to 3DS or Maya format, then export to DFF and compare render/freeze times, if any.

[View Quote]

Jul 23, 2002, 9:09pm

Not particularly, since the loading process is different in AW than in games, for the most part. Games tend to load all or parts of a level completely into memory, while AW loads cells and sectors. Granted, Morrowind also loads "cells" but they're much bigger and have relatively long load times all at once vs. AW's incremental smaller cell loading.

[View Quote] > so comparing FPS wouldn't be a reliable way of testing the rendering speeds?
>
[View Quote]

Jul 23, 2002, 9:13pm

Hmm, so nested clumps cause more pausing than a single clump? I've never noticed a difference, but I guess I haven't paid close enough attention to it. If this is true, I'll definitely want to mention it on my RW site. :/

Although, since AW3, nested clumps aren't really necessary for creating non-solid parts anyway since there's the "collision" RWX command. Regardless, for full backwards compatibility with 2.2 (since it's still allowed), nested clumps should also be used.

[View Quote] > ... but don't forget to optimize the RWX (consolidate clumps) before
> you start. A tree with one clump for each tiny branch will freeze
> AW quite "reliable" when it's added to the scene. It isn't fault
> of the RWX format if the design is unnecessarily complex.
>
[View Quote]

Jul 24, 2002, 4:38am

But there would still be a pause while AW loaded the objects from the disk into memory. AW needs to load more cells (perhaps at least 400m³) into memory initially (when it first starts and when entering a world) to reduce the jerkiness when moving through a world--perhaps even acting like Morrowind (optionally, of course) by loading chunks of the world at a time, with a pause and "Loading..." progress indicator to consolidate all the pausing at once instead of incrementally as AW currently does.

[View Quote] > Maybe a "second level cache" would be the solution, preloading
> one row of cells more than needed. This could happen with low
> priority, but the stuff would already be in memory when it's
> needed.
>
> I doubt that this is a program change that can be done quick
> though :-/
>
[View Quote]

Jul 24, 2002, 5:02am

Thanks. After a test, it seems you're right: nested clumps cause AW to pause slightly while rendering. :/ That would explain the jerkiness in Hole since all my shadows are in nested clumps. Guess I'll have to remove them and screw AW 2.2ers (which are most likely getting scarcer and scarcer anyway).

I'll be adding that nested clump note to my RW site, Ananas, giving you full credit for discovering it, of course. ;) Thanks again!

Anyone else have optimization tips not already covered on my site: http://tnlc.com/rw/ ?

[View Quote] > sample sent ...
>
[View Quote]

Jul 24, 2002, 5:58am

Seems I was too quick to agree. I've noticed pausing on both tree models Ananas sent me, ShanKahtus' original nested-clump gm_turtree13 and Ananas' consolidated-clump version, and it doesn't seem to matter. What DOES seem to matter is which tree is rendered first. I tested at 30m visibility (the lowest AW will go), with only flat modular 20m² ground pieces and a cloud plane being rendered, and pausing only occurred on whichever tree was closest to the camera; the tree farther away NEVER paused.

I mentioned this to Ananas and he thinks it might be the collision detection for each separate clump, with more clumps equaling more pause, but I don't think so since even the nested clump version didn't pause when it was rendered into the scene after the single-clump version. However, it could be the collision detection calculation radius, which was increased in AW3, but then why wouldn't the farther tree pause when it enters the scene at the same radius as that of the first tree? It's quite odd and more testing clearly needs to be done.

Perhaps Shamus and/or Grimm can offer some insight, assuming they ever read this...

[View Quote] > Thanks. After a test, it seems you're right: nested clumps cause AW to pause slightly while rendering. :/ That would explain the jerkiness in Hole since all my shadows are in nested clumps. Guess I'll have to remove them and screw AW 2.2ers (which are most likely getting scarcer and scarcer anyway).
>
> I'll be adding that nested clump note to my RW site, Ananas, giving you full credit for discovering it, of course. ;) Thanks again!
>
> Anyone else have optimization tips not already covered on my site: http://tnlc.com/rw/ ?
>
[View Quote]

Jul 24, 2002, 8:25pm

Yea, they sure did, stupidly. For them to not at LEAST provide an RWX-to-DFF converter is quite silly. They shot themselves in the foot with that blunder. It's bad enough they didn't provide an RWX importer for RW3...they basically dumped all their RW 2.x users and told them to go screw themselves. Not very smart...

[View Quote] > Or you could just parse the text in an RWX file by yourself. Didn't Criterion drop support for RWX?
>
[View Quote]

Jul 25, 2002, 3:58am

Er, you don't have to pay anything for it considering the RW 2.x API documentation is freely available (http://tnlc.com/rw/links.html )...but it's not even needed for parsing RWX files since it's RenderWare-specific, not specific to whatever 3D viewer will be rendering the RWXes (and then converting them).

[View Quote] > Exactly, and who in their right mind would pay $40,000 for that?
>
[View Quote]

Jul 25, 2002, 9:38am

Er, the API (not just the documentation) to create games/apps using the RenderWare engine. Most game companies license an existing 3D engine (Quake, Unreal, LithTech, RenderWare, etc) because it means less development time (in theory) because they don't have to reinvent the wheel.

[View Quote] > then what in the blazes does renderware want $40k for?
>
[View Quote]

Jul 22, 2002, 12:10pm

[View Quote] > Interesting name, isn't it? Anyway, it's a new idea I came up with while
> posting a reply to Strike Rapier's "Extended Viewing Range" topic. What it
> would be would be a mode in which it only searches for objects within
> viewing range once. So any change in building would not be downloaded by
> your client and actually make the Active Worlds Browser much faster since
> it's not searching for new objects that may have been placed by people while
> you are exploring. It would be very useful for when you're looking at
> historical builds which won't be changed (or at least not while you're
> looking at them) or just any area that you may feel like exploring and don't
> need to see any objects being placed down.
>
> Also, the Active Worlds Browser currently (I think) compares what's in it's
> cache with what the world server has so the speed of your cache really isn't
> as fast as it could be. It is only the cache of models and textures but not
> the landscape. It would be a good feature if you could force the browser to
> use what's in it's cache without comparing it to the world server. And if it
> doesn't cache landscape now, it most certainly should!

AW already DOES cache everything and only "compares" it with the object server if the object refresh limit has been reached.

> Another way of making it faster for some worlds would be to download the
> objects, textures, terrain, and placement of them all before going into the
> world. When you're playing Quake or Unreal Tournament online and you go to a
> game that is using a map you don't have, it downloads all of the information
> before letting you in. It would be helpful in Gaming worlds where speed is a
> must and the world is small (so that the download doesn't last for weeks!
> :-P).

Been on the list for years: http://tnlc.com/eep/aw/improve.html

Jul 23, 2002, 9:57am

No, the constant disk access is what slows AW down the most. Sure, a complex object can completely freeze AW while it renders but all the jerking and twitching AW does while simply moving through a world, I've found, is due to disk access because AW doesn't load everything into memory, and constantly has to access the disk for the cache and cell updates. I've tried to come up with some workarounds to this on the AW improvements page at http://tnlc.com/eep/aw/improve.html#scene

[View Quote] > server if the object refresh limit has been reached.
> So I assume it's the RWX Rendering that is slowing everything. AW needs to
> support a new and faster 3D model format.

Jul 23, 2002, 10:02am

You can compare AW to games easily enough. When moving through game level=
s ("worlds") you don't usually encounter pausing since most, if not all, =
of the level is completely loaded into RAM and it's just a matter of the =
far clipping plane, portal rendering, and/or LOD keeping the frame rate m=
anageable (if the game engine is designed sufficiently, of course). AW's =
frame rate is mostly quite crappy BECAUSE it lacks such optimization feat=
ures, but I have managed to get decent performance by setting visibitilit=
y to the max (now 200m) and just using the fog far clipping plane to hand=
le actual object visibility (vs. scene refresh). Unfortunately, since the=
fog far clipping plane is a per-world effect, and can't be changed in th=
e browser, it's quite useless this way (and it's annoying how the plane i=
s always facing the camera, clipping things at odd angles--a better idea =
would be to have a clipping box or sphere).

[View Quote] > RWX is fast. As I don't know any real competitor of
> AW yet, with all features that AW allows me to use,
> I cannot say if AW is slow or fast, I just have
> nothing to compare it to.
>
[View Quote]

Jul 24, 2002, 8:28pm

That has to do with physics (or AW's lack thereof) and collision detection. Perhaps some eon AW will actually get competent physics...

[View Quote] > Why can't there be Create KeepSolid? Instead of when you go in a lift that
> moves you going through the floor, cant you stay on the floor level even if
> it is going up.

Jul 25, 2002, 3:07am

One would think, but considering AW's gravity isn't even correct--and there's no momentum (which is why avs fall straight down if not moving forward or backward)...

[View Quote] > I think at least simple vertical and horizontal collision detection with
> moving objects should be pretty simple.
>
[View Quote]

Jul 25, 2002, 7:13pm

But avatars can't "ride" each other, which is what John is referring to in terms of lifts/elevators--hell, avatars can't even walk up each other anymore.

[View Quote] > It would be easy to implement as far as I can tell: Avatars have "avatar
> collision" which could be easily added to objects, as avatars are objects
> too, it would just mean that the avatar has to move in the direction of the
> object.
>
[View Quote]

Jul 25, 2002, 8:10pm

OK, so feel free to email Shamus and Grimm with all the technical details and/or apply as a programmer at AWI. ;)

[View Quote] > Sorry... didn't finish putting all I meant to, lol!
>
> All it would involve is setting the Y position of the avatar (for lifts) at
> the height of the object that the person was on. Maybe make it an
> auto-enabled function - would stop running down the stairs & flying into the
> air. :p
>
[View Quote]

Jul 29, 2002, 5:24pm

Yes, terrain should be surface .5 1 0--of course, they should really support action commands and even be RWX-editable via the world options (changing per-terrain-cell textureaddressmode, mipmapstate, lightsampling, surface, etc, etc). Changing all object surface settings to match the terrain's screwy ones is just silly. The terrain simply needs to be more configurable since it's essentially useless in its currently designed state.

Increase line length to reduce/eliminate quoted text breakup. I had to fix your broken quoted lines. Compare how much better it looks than with your previous quoted text.

[View Quote] > Don't get me wrong, I'm only hilighting the correct issue here. 0.2 Ambient
> and 1.0 Diffuse are still really bad settings.
>
[View Quote]

Jul 29, 2002, 9:04pm

How about at least using a configuration text file with bump noises for the terrain textures? That would at least make terrain actually useful for me anyway since, as you know, having a terrain cell vertically near a bumpable object causes the bump trigger to fail. A mipmap state command would be nice too but it's not as important.

In the meantime, don't you think the terrain's surface settings should at least be the standard .5 .3 0 setting common in AW? I recommend .5 1 0 because I've found it to be the most realistic for outdoors (and the most reactive to directional lighting). .2 ambient (if that's what it is now) is just too dark.

[View Quote] > Yes, ambient and diffuse settings should be a world option. This is on the
> list of 3.4 things I want to do, but have yet to schedule.
>
> As for adding rwx commands and such, that is a long ways off, due to the
> required complexity.
>
[View Quote]

Jul 26, 2002, 2:19am

I'd like an option for the terrain to disappear (turn into holes) when it is covered by another object. Terrain is nice for making a world have more visual continuity but since it can't accept action commands and have its mipmapping state altered (among other things), it's fairly useless to me. But instead of having holes everywhere where I have ground objects, it would be nice if the terrain disappeared when something covering it was rendered. Since terrain can be fairly complex and unnecessarily add to the total polygon count (which, of course, lowers frame rate), if it disappeared it would give the best of both worlds in terms of visual continuity (more visible ground) without the abilities it lacks (no action commands, etc).

Jul 26, 2002, 4:25am

Well, a per-terrain-cell could be implemented then but I would be happy with just an overall all-or-nothing option.

[View Quote] > I guess that's not so easy, as it depends very much on the
> type of object covering it. If it's a masked, translucent,
> invisible or non-solid model, the terrain needs to be there.
> Even if a model has gaps where you can look through the
> terrain needs to be there. I guess it would need more time
> to find out wether a terrain has to be rendered or not than
> actually rendering it.
>
[View Quote]

Jul 26, 2002, 6:31pm

Too bad the gradient sky doesn't do it too. I have it going from light blue horizon to black bottom but the fading isn't the same as that of the top half so it's mostly blue on the bottom half. :/

[View Quote] > If you have a dark skybox, activating a black fog helps to hide the holes that could be seen far away.
>
[View Quote]

Jul 27, 2002, 7:51am

The problem with this is that it just isn't realistic. I agree that the avatar's FEET and body (when crawling, laying, etc)--depending on the type of avatar, of course--should angle to the surface, just not when walking/running.

[View Quote] > I think, though this would just be a tiny add-on, that an avatar's rotation
> should depend on the terrain or object they are walking on. For example, if
> it was slanted like this / the avatar would be slanted as if it were walking
> up that. According to what I hear about the way AW works, this and many
> other 3D ideas such as mine would not be easy.

Jul 28, 2002, 1:48am

Right, hence, "depending on the type of avatar".

[View Quote] > Yes, true, but for certain avatars such as the skunk (my avatar so don't use
> it), it would be appropriate for their bodies to slant as they walk/run up
> hills, etc. as well as when they are idle.
>
[View Quote]

Jul 28, 2002, 1:47am

Could be the overlapping masked texture and transparent polygon bug, but AWGate has some coronas inside masked-textured polygons (the pyramid lightbeam things), and I've seen coronas visible from inside objects, so it IS possible to have coronas show from inside something, but I don't know the command.

[View Quote] > why can't you see coronas through glass objects, they're bursts of light,
> light can be seen through glass yes? couldn't there be a way to program it
> to go through clear objects? something in the rwx code that they would
> respond to or something...

Jul 28, 2002, 7:52am

As usual Roland didn't know what he was talking about because it's most indeedly possible since there are coronas inside objects in AWGate at 1s 12w inside a spinning sphere (which, oddly enough, still disappears when the spinning bars cover it) and there used to be one inside a tree (which is no longer there) in Corona world, where I first saw a corona inside an object and even telegrammed Grimm or Shamus about. Unfortunately, since I can't select objects in AWGate, I can't see how the corona is positioned, but it looks to be right on (in) the sphere.

[View Quote] > Roland said it was "too complex" to do make the coronas visible from behind
> transparent/masked objects. I don't think it is, lots of games do this.
> Then again I haven't been developing programs for the past 6+ years.
>
[View Quote]

Jul 28, 2002, 7:55am

Oh and they're even inside some of AWGate's walkway lamps too.

[View Quote] > As usual Roland didn't know what he was talking about because it's most indeedly possible since there are coronas inside objects in AWGate at 1s 12w inside a spinning sphere (which, oddly enough, still disappears when the spinning bars cover it) and there used to be one inside a tree (which is no longer there) in Corona world, where I first saw a corona inside an object and even telegrammed Grimm or Shamus about. Unfortunately, since I can't select objects in AWGate, I can't see how the corona is positioned, but it looks to be right on (in) the sphere.

Jul 28, 2002, 9:04am

When I put coronas on objects they become immediately blocked by the object. I have to end up making the object "create visible no" in order to see the corona. <shrug>

[View Quote] > I don't understand this concept of "inside an object", to me the coronas in
> AWGate you mention appear to be working as expected. For the "spinning
> sphere" example, the corona is on the center sphere so its always drawn
> infront of the centre of that sphere no matter which direction its viewed
> from, but its not inside the sphere. When one of the rings passes on front
> of the centre of the sphere, the corona disappears momentarily. The same for
> the walkway lamps, the lamp is in two parts (lantern and pole) and the
> corona is on the lantern part so the corona is seen in front of the latern
> all the time the origin of that object is visible.
>
> What are you expecting to happen? This is what I understood coronas to be.
>
[View Quote]

Jul 28, 2002, 9:39am

Hmm, they don't seem to disappear anymore...I just tried it again and they show through the object fine. Odd...ah well. I put one on a transparent sphere and it showed through it fine too.

[View Quote] > When I put coronas on objects they become immediately blocked by the object. I have to end up making the object "create visible no" in order to see the corona. <shrug>
>
[View Quote]

Jul 28, 2002, 10:15am

Just found the world corona inside the tree is: Terrain at 2n 0w

[View Quote] > Oh and they're even inside some of AWGate's walkway lamps too.
>
[View Quote]

1  ...  5  6  7  8  9  10  ...  42  |   Order by default