Last night I dreamt I watched vray render my scene in the viewport. The white buckets were cubes that rendered over the wireframe. Then I moved, rotated and zoomed the rendered model around in the viewport to pick a view. Similar to the way you can move around in the Imap viewer utility.
So I have two questions, Vlado. Do you think one day this will be possible?
Secondly, does anyone else have obsessive dreams about the software they use? It worries me sometimes…
its a mental strenght you have to develop, all tho often I question if some things i see are real or not…and mix up dreams with reality. Like once I was working soo much that when I went to sleep, my dad woke me up (i was 18 maybe back then) my dad woke me up and said:
get up and get your passport we got to go. I answered like yeah hang on. Then 5 min later he came in and said get up already, so I did. I came out to the kitchen and asked why did you need me to take my passport? he said: i never asked you for a passport…turns out first time he came it it was all in my dream…was so real…
Sounds like what your asking for is the Render Region tool in XSI.
A little OT but yeah, I’ve had dreams that were so real that I was convinced that it was for like 4 days, then I was in my car and reached behind me to the back seat and was like, hey, where’s my turkey? I know it’s back there. Then it dawns on you and your like… waaaait a minute, that muust have beeen a dream! haha
no. he isnt talking about that. he is refering to the white buckets becomming cubes…3d cubes which render volumetrically rather than 2d. so once its rendered you can rotate and its still rendered.
Doesn’t renderman’s pipeline work something like that? It doesn’t actually render cubes, but it breaks stuff up into visibility buckets for large farms and large renders with umpteen million passes?
At least thats what I remember from seeing some of the SIGGRAPH talks this year.
sounds kinda like “F-prime” for lightwave. Not exactly, but it’s pretty nifty how F-prime works, you move a light or change a property and the render updates automaticly, and pretty dang fast.
Da_elf understands what I mean and describes it better- volumetric rendering. Imagine…
I’m certain that in ten years we’ll laugh about the past when we actually had to wait hours for a render, and when it was finished it was in 2D.
Morbid Angel, I guess your right. Imagine when you hit render though, and it authors a vray quality 3D realtime environment, no messing about, it just does it.
Actually I would say that Vray may not be that far from being able to do something like that. In its root engine. Did you see the last one with the volumetrics? That was 130 cpus. Those bastards over at Pixar have more CPUs than Nasa… seriously.
what i ment by not comparing is how the scene renders on the fly on 140 cpus…so really vray dr system is by far not that…because to send one scene open it in vraydummy and etc takes several seconds etc etc..
cpnichols, well, that’s not quite the same but not too far off is it? Perhaps with 139 less computers as well! We won’t need them in ten years time though, probably even less I would guess. We’ll all be riding high on Cell Mk3s or some such thing.
I enjoyed your Gnomon DVD, particularly the baking for animation by the way. Fresh perspective and all that… Thanks.
Yeah I agree… what RAT has is something called Irma which does the interactive stuff, just keep a bunch of data floating around in g-buffer. It looks to me like this is simply irma on steroids.
RAT also has the ability to do netrender which is similar to DR and mental rays thingy.
the thing is tho I dont doubdt that render man can do the job well… but how many people and resorces you need to do it. Vray on the other hand can do the job well and have much fewer people involved. I guess everything has its balance.
Totally agree with you on that one. Renderman has some VERY smart people working there. Catmull is president of Pixar after all. But renderman is early 80’s technology that was designed to run on a vax. It has so much legacy to it is like a supertanker trying to make a left turn. They do very smart things in renderman, but it is still a scanline system at heart, and it can’t do GI and reaytracing the way others can. But lets see what happens after “Cars.” I think that was the big push to get renderman’s raytracer up to snuff.
The other problem is re-educating old fuddy duddy renderman artists and shader writers that have been doing it for 12 years to think raytracing rather than fake everything.
I still would love that kind of interaction inside my viewports though
In all honesty though, it’s not to far from what we have with XSI. All I have to do is drag a render region inside of a viewport and it’s basically the same… minus the massive cpu power they obviously have. If I had that kind of power and if MR supported unlimited satellite nodes, we would be sitting pretty.