What SketchUp could/should have been like...
-
... I didn’t dare to came talk again on this topic after the last (failed) attempt, but it’s seems that it raised some interest, so, let’s try again...
I could talk about the software in question, as I’m testing it since some days, but I think it’s maybe more interesting, at least at first, to make a point on the general subject behind this discussion... It’s a personal opinion, but I hope interesting to share...
It’s all about a technological issue :
-
On one side they’re is a need for representation and simulation, and it leads to modelling software linked to rendering engines. The goal is to create efficient images (most of the time as realistic as possible), by simulating accurately a lot of physical phenomenon.
-
On the other side, they’re is a need for immersion, interaction, and it gave birth to games engines. The goal is to create convincing environment, by accumulating a lot of different parameters (visual effects, sounds, animations..).
Both sides have been greatly limited during most of they’re existence by the calculating power of computers, and it leads to two different way of handling situations :
-
rendering engine users accepted the fact that they will have to wait to get an usable result (minutes, hours...) because accurate result is the goal.
-
gaming engine users accepted the fact that things will have to be a lot biased visually (baked texture to simulate light, low poly elements,etc...) because an average amount of frame par second is the goal.
Things could have stay that way for a long time if technology was not evolving as fast as it does, and , again, we’ve got :
-
On one side, recent evolution in GPU and CPU have gave birth to a generation of rendering engines able to deliver images not in minutes or hours, but in just a few seconds (regarding to you’re hardware configuration, still...), improving greatly the sensation of interaction between the user and it’s creation.
-
On the other side, pushed by the same technical evolution, an new kind of rendering engine have emerged, able to deliver in real time what their predecessor have to precalculate ( direct lighting, ambient occlusion, complex shaders...), getting closer to an illusion of simulation.
... Those parallel evolutions leads to a rather confusing situation where two really different things seems to be really similar, where they’re not... As a matter of fact, even if GPU solutions are really efficient, they’re far from being really interactive ( lets say, at the best, something like 1 picture in 5 seconds, where you need something like, at least, 15 to 20 in 1 second to get a decent result in real-time), and even if Gaming engine visual results are now really convincing, they’re still not able to get close to simulation of physical phenomenon, and still need some "tailoring" to work smoothly.
So, it’s really important to take a step back and realised that this two ways of doing things have both something to bring to creation, and should not be opposed like they tend to be this days, because it hide the real interest of them both... ( To be really honest, I think that the name “rendering killer” is not, on that matter, a good choice, because it tend to promote this opposition, where the interest of such a software reside elsewhere...).
What should be seen is the fact that a “gaming engine” allow a completely new experience of a built environment. The fact that you’re able to move around freely, add sounds, interactive animations, or dynamic reaction of the place (day / night cycle, lamps, moving people, etc...), is really not the same thing than just capturing a picture...
Those interactions capabilities opens a wide new range of possibility and reflection potential when it come to architectural conception, without removing the need for true simulation at some point.
... To end this long post, I just want to add that the way Pixero named its topic (“What SketchUp could/should have been like...”) is, to me, interesting, because it rises more a question like “how can conception and representation of architecture be seen with a new angle ?” than “ Is this tool able to deliver exactly the same thing than the one we’ve already got ? ”, and it lead to better opportunities to discuss further the subject...
-
-
wow.. this thing looks really nice... hope its affordable!!!
a lil OT...
@ earthmover/ Adam,
from what i have seen in the demo/ intro videos Iray for Max is more like an unbiased Mental ray. Is there an interactive response like in thea or Octane in this yet? if not.. do u know if they will introduce it in 3dsMax2012 or something?
thanks!
JJ -
Sketchup was created to look like a vector illustration (eg Adobe Illustrator)-- this is not even remotely near "Realism"... any realistic effects you can achieve in Sketchup are diversions. It is designed to be graphic in its visual effect on purpose.
To say this is what Sketchup should have been is to ignore the very specific intentions of the creators. Sketchup will never be this and if this is what you want then you are able to use this product to get it.
For my industry there is a great deal of value in the graphic vector style of Sketchup and it would not be nearly as useful to us if it were "realistic".
Best,
Jason. -
tx Pixero,
I am sure all 3D modeling applications will work like this sooner or later. SU probably later or maybe too late. I just had a discussion on the Vectorworks board were I wondered if it was possible to model in a semi rendered mode or/and perspective view. It turned out it was not possible to do that really well.
Many there answered that orthogonal modeling in wireframe was the ultimate way to work. That also seemes to be the programmers idea of Vectorworks. To me it just shows how old generations of software somtimes fade away rather quick.Francois
-
New movie.
The WHOLE video rendered in 12 minutes 27 seconds!
[flash=640,390:2vur6s6y]http://www.youtube.com/v/HoYervlEdsM?fs=1&&hl=en_US[/flash:2vur6s6y]
-
Twinmotion 2 is now live.
I've played a little with the demo and am a bit disappointed.
Rather slow performance. I have a Nvidia 9800GT so that might be in the lower range for the application.
Very cool to build something and see it lightened realtime.
Quality so, so...
No control over camera/player speed and smoothness.
My main issue is the prize v.s. the quality/performance.
€ 1920 = $ 2662 is to high in my opinion. -
You have to explode everything in SU before exporting, then triangulate every face... yikes!!!
That's gonna be some very heavy models.
-
@solo said:
You have to explode everything in SU before exporting, then triangulate every face... yikes!!
That is not a very nice workflow. Exploding a large model kills SU.
-
Plus in one tutorial video I can clearly see SketchUp as a import option but he chooses fbx.
I read something about problems with SU import so thats not working at the moment. -
My big problem with this kind of approach is that it cannot support anything but trivial light sources. ie PointLights (omni) and Direction (sun) lights. eg You cannot support Area lights with this tech.
Adam
-
Hi Sketchucationers,
I test it yesterday on an iMac running Xp with bootcamp (don't have all the specs here)
Exporting from Skp to Fbx > OK (and I didn't explode anything before)
Importing Fbx in Twinmotion > OK
Collapsing the group by material in TwinMotion > OKBUT I don't get anything in the preview window.
I mean I can see my imported model with all the groups in the file tree on the left but the main window remain all black !!!I will post this problem on their support forum as soon as they accept my account but I post it here in case someone else here got the same issue.
Nico
PS: the original sketchup file was around 30Mb (just to give you an idea)
-
@adamb said:
My big problem with this kind of approach is that it cannot support anything but trivial light sources. ie Point Lights (omni) and Direction (sun) lights. eg You cannot support Area lights with this tech.
Adam, sorry but your assertion is false, there's area lights in TM2, and it works well (you also have spots and omni). The difference with, for example, light up, which bakes the lighting, is the fact that you cannot set a random surface of your model as an emitter (by applying it to a material)...it reduce a bit the range of possibilities, but still, it's quite efficient.
@solo said:
You have to explode everything in SU before exporting, then triangulate every face... yikes!!!
The importer works by splitting the object by material (just like Su2Thea actually), so you don't have to do anything by hand (exploding everything, and so on). I've beta tested the early versions, and it was rather slow with a full sketchup hierarchy loaded in real time. But, by using this method (which is automatic with the import preset "sketchup"), it's much more efficient.
-
@mrwip said:
Adam, sorry but your assertion is false, there's area lights in TM2, and it works well (you also have spots and omni). The difference with, for example, light up, which bakes the lighting, is the fact that you cannot set a random surface of your model as an emitter (by applying it to a material)...it reduce a bit the range of possibilities, but still, it's quite efficient.
I'll take a look at the App. I think you may be mistaken wrt Area light support because of the way their engine works.
-
Adam, I've been in touch with the developers during the building of the software, and ask them the same question than yours, because usually, deferred rendering engine ( cryengine and such) never include planar lights, and are always limited to point elements. But actually, in twinmotion, they've fine a way to do so. I don't know how technically, or what trick they're using, but in the end, you're able to get a planar ceiling lamp for exemple (that you can't achieve with an omni or a spot).Isn't that what your looking for ?
-
OK, I've had a play with it. Its precisely what I knew it would be - had to be even.
Doesn't do area lights.
The way these engines work is creating a view of the world from the light position to create a "shadowmap" that can then be used in your shader to determine how much direct light falls on a surface.
For a small number of shadow casting lights this works great. For a real area light or hundreds of lights, its a none-starter.
(Mr Wip, the Area light effect is done by tweaking the shadowmap projection transform to create a orthographic projection, but ultimately its still a shadowmap. Its not really an area light the way most people think about the lighting effect of an area light.)
Summing up: Definitely an Interesting Application. The best advice I'd give is for people to try it out on your own models, if what it does is what you need, you've got yourself a winner.
Adam
-
ok, it's clearer now, thanks for the explanation Adam.
-
[quote="Pixero"]Another teaser for "the rendering killer".
I just saw the release announcement in the Daily CatchUp. looks like the rendering killer is also a budget killer at USD$2,990.00
Seems to be great fun, but I think I will stick to Chucks' new plugins for now.
-
@unknownuser said:
@pixero said:
Seems to be great fun, but I think I will stick to Chucks' new plugins for now.
Link?
-
@cadmunkey said:
@unknownuser said:
@pixero said:
Seems to be great fun, but I think I will stick to Chucks' new plugins for now.
Link?
cadmunkey - sorry for the mixed metaphor - The "killer" announcement came out about the same time Chuck Vali announced his plugins. http://www.valiarchitects.com/
I intended my comment to make a comparison of my sticker shock for the $2,900 renderer versus the very reasonable ($15 to $39 each for pro, free basic) and usefull Instant roof-road and grader plug-ins.
Since then, there was the announced release of "Luminon" which seems to be a very cool, feature laden and similar rendering program at about 1/3 the cost of the "killer".
-
I tried the demo again and its really too slow to be usable. I might not have the newest top of the line grahipcs card but for an empty scene with a plane and a sphere I think you could ask for better framrate than it gives.
Now lets wait and see what Lumion is like...
Advertisement