T3D = end of DIF?

#1 And what I've gleaned from the forums/blogs is that DIF will be supported, but not actively, in favour of DTS and Collada. So I'd expect it to phased it eventually.

#2 Collada is just a raw import that is automatically converted within T3D to dts. (doesnt make since to me, but,...I'm not writing the engine)
dif support from what I've gleaned is going to remain.

#3

Quote:Collada is just a raw import that is automatically converted within T3D to dts

Yeah ... it confused me a bit. I think it might be useful because it means that different modelling programs can share source files (max, Maya, even Blender has Collada support - though I haven't tried it) - I guess it would make content packs easier for use and that sort of thing, multiple artists working on one project remotely and using different modelling suites.... etc. Maybe a get-round for most DTS exporters poly limit (I'm completely guessing on that).

I suppose the future of games design really isn't brush based modelling anyhow.

I might make a second DTS version of the DIF level I'm currently working on so I can see the performance difference if/when I get a T3Dbeta.

#4 I have been wondering about stuff like this as well and from my investigations it seems DIF still has advantages over DTS for interiors, with things like zoning and collisions. e.g. Polysoup can get expensive with more complex models. As Mike says Collada support doesn't add anything new to the DTS format but just makes it easier to export models from ones modeling app of choice. If this is wrong or if there is more that needs to be pointed out I look forward to hearing more about this. Thanks for starting this discussion.

#5 @Gerry: It does convert them to DTS, but DTS has in fact been improved for it. If you take advantage of the new DTS features then your DTS won't be backwards compatible with older versions of TGE/TGEA but there are numerous new features.

From my experince with DIF and DTS, there are a few things that DIF is good at, though unless your using a lot of portals performance wasn't really a major one. In some cases the DTS models ran faster for us. But they couldn't do these things:

- Portals allow you to have certain objects inside of a building not be rendered unless they are within site. If you have a city with objects or pepole in the buildings, that is a big win.
- Portals could result in big savings in an indoor environment.
- Lightmap lighting looks a lot better on DIF than it does on DTS.
- By portalising interiors if it rains the rain won't rain inside of your building, where it would in DTS.

Now it should be noted, this is all with existing versions of TGE/TGEA, some of these things may be addressed with T3D. The advantage of DTS is that its a lot easier to work with and get better looking graphics out of.

#6 Thanks J.C.

Do you know enough about those new features to provide more information on them... or else point us to where they have been discussed already... or is this one of those new things that GG has yet to introduce to us?

#7 Portals:
Matt Fairfax told me that portalling (say individual buildings) wasn't as useful with modern tech as it was with older, and after a few tests I noticed that there is a bit a jump when the processor has to think about changing what to display, so I've scrapped portalling of individual interiors in my DIFs (my current DIF maps are huge outdoor areas so zoning really isn't going to work).

He also mentioned that they had a way to fake zoning, but didn't elaborate.

I noticed in the T3D advanced lighting vid that polysouped DTS' now seem to cast shadows on themselves, something lacking in TGEA, and a reason why I've so far stuck with DIF (lightmaps).

#8 I've no idea why anyone would say portalizing isn't as useful anymore. Even on a modern graphics card the improvement is substantial (ie, 30 fps standing outside, 300fps standing inside on a 9800GTX). If you have a complex city with many DTS shapes inside your buildings, DIFs may be your only recourse.

In the IRC chat I've counted about a dozen or so issues with polysoup and collision raised, so perhaps some work still needs to be done on that end. But hey, if DTS evolves to encompass everything DIFs do and then some - I'm all in favor of it.

Really, DIFs are but a tool. Sometimes it makes sense, sometimes it does not. I just hope they hang around for use in those projects where they make sense.

#9 The portals in T3D (including the editor placing) were definitely not added because they have no use :)

Its just the DIF for the purpose of portalling thats no longer required, because you can now get it without DIFs
And yupp as soon as the remaining issues on the collision end are fixed (or people end in the position where they need more complex things than DIF can reproduce) then it will definitely be the end of DIF.

#10 Could you explain how this really works? How can you render a complex scene without a portal system?

When I create a large interior, then it will sum up to lots of millions of polygons. Each room could have a couple of hundreds of thousands. Clipping, occlusion and other techniques are needed to just render the visible geometry, to render only visible lights and shadows to animate only visible models to execute only scripts of objects close to you.
It is even harder than that. Some invisible lights need to be rendered, if their shadows are visible through a portal.

I cannot imagine how we create big levels without this. But maybe there are good scene management alternatives I don't know of. I am not a a game engine developer myself. But I am really interested in new information in this case.

A well thought combination of DIF and DTS with unified lighting might do the trick.