#Future of 3DS and GLiNT
16 messages in this thread
Hi all,
I have been talking with Osman Kent, President of 3DLabs, about the future of
GLiNT. I am very excited with the launch of GLiNT boards and the effects that
they could have on the future of 3D in any platform and, specially, any
platform running 3DS.
But here we have the present. Right now, little could be done in R3. People
like Pavel are writing IPAS software to allow realtime 3D previews on 3DS,
because it seems that, simply, the inline drivers structure and the 3DS
graphics engine don't allow 3D acceleration (if I am wrong, please tell me).
I think that we all need/have/must push this technology as much as we can. The
boards will be really cheap, specially compared to the SGI beasts and,
althought not comparable with a Reality Engine, they have an impressive
performance, comparable to the Elan engines. IMHO, support for this technology
will lead to:
1. Lower prices caused by the general market acceptance and support.
2. Since GLiNT is an scalable chip, we will see in a not too far away future
more (faster) chips and developements. Future boards that will guide us into a
new age in the desktop imaging arena.
How we can support this kind of developments? For sure, an hypothetical port of
3DS to NT will allow direct OpenGL support. But right now, we are working with
R3, and R4 is still far in the horizon. If I buy a GLiNT board now, how can I
use it in 3DS? I made Osman. Nevertheless, I think it could be _great_ to hear
how the 3DS community thinks about this subject. We all have been talking about
realtime 3D in 3DS for years, but now it is time that we decide what this
support is going to look and feel. We have the key to make realtime 3D in 3D
Studio a reality.
Here are my suggested IPAS programs for GLiNT:
a-Move across a scene freely and in realtime, just like in a 3D game,
controlling speed/orientation/FOV/zoom/etc. with a mouse and the keyboard (with
an option to save that the moves as camera path/data for the keyframer for
editing/use in the final render).
b-See a selected object(s) in a realtime window.
c-Save high quality previews of final keyframer scenes using the GLiNT
processor. This doesn't have to be realtime (but near realtime). Posibility of
save to different formats such as TGA, TIF, FLx.
d-The posibility of applying textures to an object in realtime and adjusting
texture data on the fly. Could this be done without getting into the inners of
3DS?
e-Applying lighting in realtime/near realtime like in Painter's "Apply
Lighting" filter, using some kind of Gouraud preview window of the scene.
What do you think about this ideas? What are your ideas? Isn't 3D wonderful?
<g>
I will be grateful to recieve all your suggestions since I am planning to
archive them and send them to 3DLabs as a formal wishlist.
Thanks in advance,
ana /krazy images
I hope somebody jumps on a GLiNT IPAS. Your suggestions are right on! Most
of this stuff is being hacked out by Pavel right now for the MGA, but the
texture mapping, of course, is not. I'd be curious to know about that as well.
I'd love to see a full screen, hot-key callable preview!!! with real-time
interaction and recording.
John Tissavary (LUNA cie)
If you have any idea/suggestion/wish/order <g> to pass to 3DLabs, I
would be glad to hear you. They are open to any kind of GLiNT
impletantions in 3DS.
Come on, I am sure that everyone has "fantasies" about realtime 3D in
3DS. This GLiNT 300SX thing could make anything (up to 300k polygons
per second in 24bits, gouraud shaded, clipped, stencilled,
antialiased and alpha-blended. Still doesn't have data for the 300TX
(the highend chip). Shout them out! <g>
Your first idea is at the top of my list also. At the '89 NAB I saw Caligari's
3D program for the Amiga (I believe it was called Broadcast?) and I was
impressed by their ability to manipulate the wireframe view in realtime w/ the
mouse. I tried to find something similar for the Mac w/ no luck; I believe this
is true still today (am I correct?).
Add to your list the ability to preview IPAS effects in realtime. I know disks
6&7 make great strides toward this but all the old disks would benefit w/ this
new capability.
Thanks for your suggestions. I am sure that the realtime flythrought
will be in the final 3DS support package. I have my doubts about how
this could be done. Not because of the chip but because of 3DS.
Maybe Gary could jump in and make a few comments about these ideas.
Or the Vibrant guys. Or Pavel. 🙂
I am not sure about your suggestions about IPAS effects realtime
preview. If the maker of the IPAS take into account GLiNT, maybe this
could be done. Even then, only a some IPAS programs will be able to
use the GLiNT features. Gary? <g>
Ana, you're jumping the gun with this GLint stuff. At least wait till
it exists in product form!
After the release it will be too late. Please note that they want to
give direct support to 3DS with the first shipping boards, which are
due in Fall Comdex. Certainly, we CAN'T say that about ANY board
maker arround. I think that all the support we can give to the chip
and this kind of technologies will be always be too small.
We need to push the envelope. Moreover, this kind of support will only
benefit 3DS and the 3DS community… why do you find it excessive?
1) The boards won't be out till this Fall.
2) Without a board, an IPAS developer can't do a thing.
3) These boards will require PCI-bus machines, which are
still a very small percentage of the installed base.
While it is true that the boards from our customers will not be out
until this fall last week we released a GLiNT simulator to our Early
Access Customers to kick start the third-party software development.
This simulator is curretly limited to Windows and is implemented as a
VxD mimicking GLiNT 100% (except the speed of course). If there was a
genuine need we could release a dos version too.
We have been using this tool (and its Sun equivalent) in house
ourselves for the past 12 months and it has been very useful for us
to port a compliant OpenGL, X-Windows and Windows drivers. So in
theory at least porting of 3DS could start now. Our close developers
are also porting their 3D APIs using the simulator right now.
However, our view is that we need to understand the needs of the 3DS
community before we embark on a major development effort. Hence, we
are watching the dialogue here with interest.
The comment about the PCI is right but again our view here is that 3DS
users demand top machines with Pentia where PCI would fit the bill.
IDC predicts that there will be around 4.5M PCI machines by the end
of 1995.
I would be happy to answer any other questions anybody has about GLiNT
and related matters.
Having a texture map capable hardware accelerator would be very
helpful to me. I'm a very visceral kind of person, and have a hard
time dealing with reduced "bandwidth" previews. You products will
certainly influence the type of computer I purchase in the future,
and who's grapics boards I'll be buying. There is no question that
the 3DS community will rally behind an affordable 3d hardware
accelerator as long as it is supported by 3DS applications (including
3ds itself at some point, I would hope).
There ya go… IPAS is a DOS platform, and as you said, your simulator
is Windows-only.
Regardless, I'm not here to talk about development issues. Ana asked
me a question and I just gave her my spin on it.
You already have read Osman's reply about the DOS simulator. Seems to me that
the ball is in your side now.
As for the PCI need… well, who is going to buy an ISA/EISA/VLB machine in the
next couple of months in this forum? <g> Do you mean that it won't be
profitable in the short run because there are not enough PCI based computers
arround? PCI is here to stay. I love the idea of be able to put that GLiNT
board in my DEC now, and then put it in the PowerMac or whatever it comes next.
As for the IPAS, if someone gets smart enough to associate their IPAS with the
GLiNT, they will have an advantage over the next guy in the block. These guys
at 3DLabs are willing to give support to everybody supporting GLiNT. This means
that Vibrant could call them and say, ok, give us a DOS simulator and we will
write something for the GLiNT boards. And who says Vibrant says Pavel or Joe
"the IPASMAN" Smith. If there is enough demand from IPAS developers (and that
could mean you pushing the technology, since the Yost IPAS are models of what
an IPAS should be), the DOS simulator WILL be a reality.
I see people building systems arround this class of hardware, not only arround
a CPU, only to run a killing combo (3DS/GLiNT=SGI class graphics engine) at an
unbeatable price.
Seriously, I think that making cheap 3D in 3DS a reality requires a compromise
from all the parts involved. If the Yost Group support this hardware, everyone
will follow you. That's why I find your comments so important (like always 🙂
Gary, please take this message as a friendly suggestion. I am not pushing GLiNT
because it is a great piece of technology but because I want to see 3DS at the
top. 3DS supported the Targa and the Vista because they were/are industry
standards. Now I think it is time for 3DS to make an industry standard (that
BTW, will help 3DS to win more fans and clear one of the last SGI software/3DS
frontiers).
Cheers!
ana /krazy images
>a-Move across a scene freely and in realtime, just like in a 3D game,
>b-See a selected object(s) in a realtime window.
It is possible without problem.
>c-Save high quality previews of final keyframer scenes using the
GLiNT There is the problem with morphing. But this can be solved (I
hope).
>d-The posibility of applying textures to an object in realtime and
adjusting >texture data on the fly. Could this be done without
getting into the inners of >3DS?
As far as I know, the GLiNT in the SX version has no support for
hardware assisted textures. The DX version will be able to do HW
textures, with separate texture memory.
>e-Applying lighting in realtime/near realtime like in Painter's
"Apply
We will see. I haven't enough informations about GLiNT API etc.
But definitely, I am trying to collect as much informations as
possible. I would like to make something for GLiNT/3DS.
Pavel,
This is starting to sound very good. Thanks for the reply!
ana /krazy images
>> a-Move across a scene freely and in realtime<<
I would definately put this as the priority item. The more interactive the
previewing process, the more artistic or realistic we can be. It is a MUCH
NEEDED tool.
>> b-See a selected object(s) in a realtime window. <<
Yes, this could be incorporated indirectly with item a.
>> c-Save high quality previews of final keyframer scenes using the GLiNT
processor. <<
This also dovetails nicely with item a.
>> d-The posibility of applying textures to an object in realtime <<
I think this is a great idea. I would like to have a tool where I could
stretch a texture around an object like silly putty.
>> e-Applying lighting in realtime/near realtime like in Painter's "Apply
Lighting" filter, using some kind of Gouraud preview window of the scene. <<
I will second that idea too. Lighting is another area that needs more
interactivity.
If we want 3DS to remain on the bleading edge, we need to do more than
sit around and let the Yost Group do all the work. If nothing more,
we can vocally support some hardware improvements that will close the
gaps against the major players. ("Pomp and Glory" music playing in
the background.)
I think GLiNT can do this, but timing is of the essence! In three
years my solar pocket calculator will be able to do real-time
playback. We need this stuff NOW and I am confident that we would
gladly divy-up to anyone that fills the need!
Glad you liked the ideas! (do you have any more ideas?)
As for the timing and support, you know what I think about that.
That's why 3DLabs need the ideas now: to have the 3DS support ready
for GLiNT based boards first release. I DON'T want to see GLiNT
supporting only OpenGL, 3DR, other 3D APIs, AutoCAD and Microstation
but not 3DS. And I don't want to wait for 3DS to support OpenGL or
any other 3D API in order to get a GLiNT board. Now, the only way is
IPAS. The future is on Gary's hands, though (and I am happy about
that 🙂
We have there Matrox ignoring 3DS (until Pavel comes with his IPAS),
Artist ignoring 3DS and we have GLiNT willing to support EVERYTHING,
including 3DS. Are we going to turn our backs to a company who is
offering us what nobody offered before? Are we blind? 🙂
Maybe I am taking this too personally, but I don't think so. I want
affordable realtime 3D and that's why I want feedback from everybody
in the forum. Like you say, realtime 3D is what 3DS needs NOW. 60K
polys/S could be good for optimize, but 300K/s is what we really need
for maximum productivity. Not to talk about what the 300TX chip could
do.
I am not asking support to Autodesk or the Yost Group, just to the
individuals in this forum, to add ideas and make a positive statement
about this technology.
ps.BTW, I have nothing to do with 3DLabs, except for my support to
GLiNT.