Google is Listening!
-
@jbacus said:
I don't know what application you're comparing SketchUp against in performance terms
.I've Been a Maya user for about 6 Years until I enrolled into Architecture Studies.
For the past 5 Years, I've been getting away with mostly SU, but also a mix of Rhino (Not relevant - Apples & Oranges), C4D and Maya. I've been hearing great things about Modo, but haven't got a chance to actually use it more than a few minutes at a time.
I want to make clear that I'm not looking forward to Mudbox / ZBrush Billions of Polys.
But when a big % of the Component in 3DWarehouse (a big feature of SU) are like kryptonite to SU, or when trying to load Evermotion props...@jbacus said:
I don't know what you would consider a "medium project", and I don't know what your hardware configuration is either.
.It's holiday here, so I won't be in the office until monday, So I'll get back to you with numbers as to poly count.
Since it's an ongoing competition, i can't yet publish any photos, but i'll see what i can do.
My Workstation is an Win7 / i7/ 12Gb / GTX470 Machine. Build to crunch number and to render fast.
at the Office we also have an older (Yet Faster) 8-Core MacPro.@jbacus said:
In game-theoretic terms, SketchUp's modeling performance is a classic "arms race". Every time we make SketchUp faster, you start making bigger models. Then we have to make SketchUp faster again, and then you start making even bigger models. There is no logical conclusion to this game where performance is infinite and you can make an infinitely large model. Eventually performance will plateau, and you'll have to learn how to work with the system as it stands.
@jbacus said:
You have to take some responsibility as well, though, and manage your model to work within the constraints of the system.
john
.Those two comments really amuses me.
I Had a University project which we ported to Unity3D for real-time walk-through for an exhibition.
it was a whole Blvd, with over 30 Historical Building and surroundings.
SU couldn't handle the Building themselves, yet in Unity we got REAL-TIME geometry, Trees (Not Low Poly/2D) and Improved Textures (Bump/Spec which SU can't handle) at high resolution and good FPS (60+).
any modern Game Engine can Handle a Huge amount of data today, even the free ones (UDK, Unity3D).
Most 3D Packages can handle much more geometry than SU, and that's a fact.Yes, improvements have been made on version 7 & 8 - but if you take into account the increase in computer power from 2007 (SU6) until today you'd see that SU doesn't begin to utilize what is a fairly reasonable workstation TODAY.
I Realize that Google doesn't want to do heavy work on SU core.
I want to emphasize that in my humble opinion - Doing so would take SU from a "Google Earth Plugin" to worthy "Pro Arh Viz Tool", even a competitor to 3DSMax and others, thanks to it's ease of use and wide spread popularity among Architects. -
@unknownuser said:
There is no logical conclusion to this game where performance is infinite and you can make an infinitely large model.
I might argue John... Google Earth for instance is one large model right? I mean of course not really but from the perspective of me flying through it and using it, it is one large model, so perhaps one day SketchUp could use that same theory in it's modeling that you only work or see the area you are in... loading each consecutive region as you came to it.
perhaps also if we are to be responsible for our own model efficency, maybe you could add more tools to help along that line, such as the purge tool that is already available, a tool that could locate dramatically complex geometry within a simple model. or things such as ThomThom's new plugin Similar Objects http://forums.sketchucation.com/viewtopic.php?f=323&t=30143&hilit=Similar+Objects this would allow us to analyze a model for all groups or geometry that was similar and replace them with one component.
I guess what I'm saying is we expect the impossible from you... you're Google...
with each respective release of SketchUp I have noticed a great improvement in speed. and I will continue to expect that... the first thing I do when a new version comes out is open an older, clumsy model and spin it. with every release I've noticed a difference.
on the plugin conversation. I love the simplicity of SketchUp... I love being able to add the plugins I want to use. and adding them one by one I learn them as I go.
I personally have been using this software for years and have even been an admin on this site and I will still find plugins that I didn't know about... as others were saying a database of plugins with a plugin manager in SU... that would be awesome. if Google were to run a repository of plugins and evaluate each one as it is added. they could purchase plugins from the creators and sell bundles for Architecture, Wood Working, Mechanical, Engineering, Civil etc etc... this would allow another avenue of monetization for Google. it would also spur even more Ruby Writers... if they could make a residual income from writing rubies? if the million people who have downloaded Google SketchUp in the last week had instant access to rubies like they do components... it would be incredible.
SCF just celebrated 50,000 users, so of that Million users alone, a great majority of SketchUp users are missing out on a lot of great plugins.
-
i don't really know what causes the slowdown in higher poly models so bear with me in the suggestion.
i'm assuming much of the performance loss is due to giving the model a certain look at any given time (shading all faces etc.).. if that's in fact the case, i'm wondering how people would feel about a mode where the look is plain jane.. for instance, if i have a group or component (say 3d trees) that i feel is bogging down performance, i could click something such as 'optimize performance' for that group and it would dumb itself down look wise.. ie, i wouldn't have to hide it in order to speed things up.. i could see it plain as day albeit the color/shading doesn't change as i orbit the model... any style options (profiles etc which may cause slowdowns) are disengaged from that component..
or maybe, if the above is even sensible to begin with, there could also be a mode in which the entire model is dumbed down and only when you open a group or component does the look become the sketchup look.is this making any sense or am i heading down the wrong path? how would some of you high poly modelers feel about this type of option?
-
I feel like that would be great. My limited insight in how all this works makes me wonder if there would be a big wait while the suspended refresh completes after you finish navigating.
In the meantime my attitude(modeling development) seems to be coming around to trying to simplify my stuff by reducing poly's, or doing a lot more thinking about my approach. -
@unknownuser said:
After reading the last posts i'm feeling like the bad guy in here...
i dunno daca, i don't think you're coming across as the bad guy or whatever.. honestly, i want what you want and i think a lot of other people do as well.. thing is, i really think that the best way to accomplish such a thing is to start from scratch or nearly from scratch..
for a lot of the power user wishes, i think more programming has to occur than "add this to sketchup".. i would think the core app needs reprogrammed in order to really fix some of the issues.. anything else is just adding even more duct tape to your hammer analogy..
but hey, why dig in and rewrite the core just to get better performance? let's see some root level nurbs in there.. 64bit.. etc,etc,etc.
you're talking about a whole new app and while i think the sketchup programmers are capable of building such an app, i don't think they're going to do it under google.. why would google want to make one he11 of an architectural modeler that sizes up well against the heavy hitters? i can't think of any google software that does this..the thing with sketchup (that i feel not enough people either realize or admit to) is that it handles 70-80% of my drawing needs very well.. that's with just the standard tools or minor modifications of the standard tools.. i design/build some relatively complex pieces/shapes in my work but i imagine if i did say houses, sketchup could handle even more of my workload..
most of the stuff the power users want is the polish, they (and i) want to be able to handle the other 25% percent of needed task but those things come at a much higher cost (system, sophistication of the app, money of course,user knowledge/experience) than what it takes to do most of your drawing.. [analogy- i can buy a car for $10,000 that will go 100 mph but i can't spend $20g and get a car going 200 mph]
in this thread, my stance boils down to enjoying and preferring sketchup for most of my needs.. when i get to the point of needing something that sketchup can't do, i don't cuss at sketchup, i use something else that can accomplish what needs to be done.. but realize, that other something can take care of my more pressing needs but it can't handle the majority of my work in a similar fashion as sketchup..
i think it's a bit shortsighted to expect sketchup8 or 9 to be some kind of amazing, all-in-one, arch/viz package giving us everything we want.. if you're going to continue using sketchup, i think it's best to get in line with or at least try to understand the developers wishes or expectations of the app.. ultimately, that's the direction the software is going to go in so for me, it's a matter of recognizing it's strengths (and there are many of them) and utilizing those aspects but when it comes time to do something that sketchup can't handle, i switch to another method.. it's that easy (ok, well, sort of "it's that easy"
)that's not to say sketchup can't be improved on because it can and it is.. i think it's highly more productive and less stressful to think of ideas that can be realistically integrated into sketchup and further refining it's use while staying true to it's key values..
and honestly, if the suteam 'added 64bit' then i might infact start showing some anger.. if they're going to 'add 64bit' then i want a 64bit app built from the ground up.. you don't just add 64bit to sketchup or add multicore support and hope to see any real gains.. you can however add some more simplistic things fairly easily which fully enhance the strengths of sketchup
-
@mitcorb said:
I feel like that would be great. My limited insight in how all this works makes me wonder if there would be a big wait while the suspended refresh completes after you finish navigating.
well, that's what it does now.. it's constantly updating itself and providing us with a rendered view.. i'm more or less suggesting the user can choose when the refresh occurs instead of it happening constantly.
for instance, if you have a material inside a group that's checked as optimized, the material will just be the base color at all times until the user decides to tell it to act normal. -
@unknownuser said:
i don't really know what causes the slowdown in higher poly models so bear with me in the suggestion.
i'm assuming much of the performance loss is due to giving the model a certain look at any given time (shading all faces etc.).. if that's in fact the case, i'm wondering how people would feel about a mode where the look is plain jane.. for instance, if i have a group or component (say 3d trees) that i feel is bogging down performance, i could click something such as 'optimize performance' for that group and it would dumb itself down look wise.. ie, i wouldn't have to hide it in order to speed things up.. i could see it plain as day albeit the color/shading doesn't change as i orbit the model... any style options (profiles etc which may cause slowdowns) are disengaged from that component..
or maybe, if the above is even sensible to begin with, there could also be a mode in which the entire model is dumbed down and only when you open a group or component does the look become the sketchup look.is this making any sense or am i heading down the wrong path? how would some of you high poly modelers feel about this type of option?
Your 'optimize performance' idea would be a possible improvement but it would then emphasize another bottleneck that hasn't been mentioned yet. The Maxwell render plugin (and probably others) has the ability to use 'proxies' where you have some simple geometry component, say a box named PalmTree_Proxy. THen, you have your high poly object named PalmTree in the same model on a hidden layer. That way it doesn't slow down the viewport because it's hidden, and at render time Maxwell will substitute the high poly object for the proxy.
It works just fine but for some reason SU file sizes get verylarge with even 1 or 2 high poly objects (as compared to 3ds Max for example which can hold quite a lot of polys without getting too large). Combine that relatively minor inconvenience (file size) with the much large inconvenience which is save times, and it really bogs you down. To me, SU's save times once you start getting into higher poly's becomes almost as burdensome as viewport issues because it's something you have no control over. And again, I will compare it to 3ds Max which saves even huge models very quickly.
While I wouldn't compare SU's modeling capability to 3ds Max's, I think this comparison is only fair because if you're claiming that SU is purposely a much less complex program than 3ds Max and so it's modeling tools shouldn't be as complex as Max's, then shouldn't that lack of complexity also translate into smaller file sizes and faster save times?
-Brodie
-
gotta admit, file size has never crossed my mind as any sort of concern.. storage is way too cheap (relatively) to even worry about it.. i mean, one or two raw frames from a 20+megabeam digi cam = the same file size as a big model and some of those people are shooting hundreds if not thousands of pictures a day.
save times too.. that's like 5 seconds at the most for me (but if you're saying it takes more than say 30 seconds to save files then yeah, i'd be concerned too but i've never had a file do that.)..
point being, if i could have a 2million poly model [exaggeration] perform the same way as a 20,000 then i'd surely accept a longer save time as a consequence.. pretty much a no brainer tradeoff
-
@Jeff:
I wonder what it is that the Crytek game engine does. A while back there was a link to a Crytek demo that showed real time smooth refresh and in-scene remodeling. Yes, I am aware that it is a game engine and you probably need a gamer type video card, but I'm just saying.. -
I'm seeing alot of great ideas in the voting area -- and also some submissions that are already covered somewhere in Sketchup.
I love Sketchup and I love the ruby plugins that are available that make Sketchup much more powerful... I'm pretty sure I don't want the Sketchup team wasting dev time on reproducing the work already done by the ruby developers. I'd rather see them focus on giving the ruby developers the most robust and stable environment they can have to design in. Really the people who should get the strongest vote as to what gets included in future releases is them...
As far as the UV editing issue goes, I guess my reasoning for wanting this as a stand alone component for Pro (like style builder) is it is just so annoying to get things into and out of Sketchup that if I can keep the project within the borders of "Sketchup land" then I feel it would be much more productive and faster to work with textures.
It doesn't have to be really all that complicated -- but based on my personal experience poor control over UV's is the thing that really undermines the NPR render engine in Sketchup the most... that NPR engine is the reason why I use Sketchup for illustration. However I (like many others) use Maxwell Render to also produce realistic renders and the UV issue becomes even more of a problem in that arena. This of course really become a issue on curving/organic surfaces that are becoming more and more prominent on everything from buildings to cars to toasters.
I've got some ideas for names too -- I like MapUp, WrapUp, or TextureUp.
Best,
Jason. -
emage, if you have already used Maya, it seems to me, that you would be better off investing in a copy of 3DS Max. I did one module in ArchViz when I studied for my degree, and we used mainly 3D Studio Max. I did prefer SketchUp for its easy of use, but found that I couldn't use SketchUp with Flash (I ended up using Unity instead of Flash, but I had to use the 3DS Max/Flash combination set up to pass the module). You could also look at SpaceClaim, the version without all the engineering tools. The developers call it "SketchUp on Steroids"!
Failing that, try Bonzai 3D.
-
The thing is... I don't feel Bonzai handles more polygons than SU. Nor is it either 64bit. (and doesn't have as many render engines available.)
The key difference is that it's a nurbs and solid modeller. -
This is what I am up against often, using SU without needing photo realistic modeling.
I've got of model of a treatment plant. The footprint is large. Equipment (piping, channels, galleries, and bays) can span 2 or 3 floors. Components are used to keep file size down and reduce repetitious modeling. The compressed SU file size can easily be 25 meg. The associated Layout file (depending upon the settings) could double or triple that size.
Goal of the model has two basic needs. 1- produce technical illustrations (usually thruough Layout for highres transparent PNGs) and 2- Walk throughs.
#2 was a big a reasons for getting SU way back in version 2. However, with the complexity I mentioned above this is not fluid. Waiting to render even with the improvements since version 7 is a bummer. Perhaps there could me a mode where SU reads only the next or previous Scene or two and ignores "remembering" any other geometry of the model of scenes which are not in the immediate queue?
#1 usually Layout works good enough but depending upon the settings used with a big model or one that has a lot of textures it can seem to be slow and exporting a transparent PNG can fail and if it does you don't know it will until the end of the exporting process when you get an error message. Perhaps utilizing multicore processing would be possible for Layout rendering? This could make buying the Pro version more desireable and bolster the difference between the Free and Pro versions.
-
@jclements said:
Perhaps utilizing multicore processing would be possible for Layout rendering?
You guys want to know a funny thing? LayOut 2.0 on Windows
renders in another thread while you continue to work on the
document. What's so funny you ask? Well in LayOut 2.1 we
had to pull out the background rendering and guess
how many people noticed? ZERO! -
Perhaps if you hadn't have updated/improved the SU rendering engine we would have
. Kind of kidding you here. -
@jhauswirth said:
@jclements said:
Perhaps utilizing multicore processing would be possible for Layout rendering?
You guys want to know a funny thing? LayOut 2.0 on Windows
renders in another thread while you continue to work on the
document. What's so funny you ask? Well in LayOut 2.1 we
had to pull out the background rendering and guess
how many people noticed? ZERO!Touché!
-
@jhauswirth said:
@jclements said:
Perhaps utilizing multicore processing would be possible for Layout rendering?
You guys want to know a funny thing? LayOut 2.0 on Windows
renders in another thread while you continue to work on the
document. What's so funny you ask? Well in LayOut 2.1 we
had to pull out the background rendering and guess
how many people noticed? ZERO!Well I certainly didn't...since I don't use Layout.
-
@hieru said:
@thomthom said:
.....a Plugin Manager would let you browse and download plugins directly from SU. And of course update when updates are available. The Plugin Manager should be available as an one-click-installer.
Basically something like Dreamweaver's Extension Manager?
Now there's a good idea!
-
I would think google should keep the SU in 32bit for the less intense users and release the 64bit for more intense users. Let us (the users)test it out and you will get your response of what we think that if the 64bit version actually benefit more to the intense users. Now a day, I believe computers set their 64bit application as their standard. I have seen laptops as low as $400 with 64bit window base application. I believe 64bit is the future if not the current then why SU stays as 32bit.
-
Peep
I want a way to selectively disable inferencing -- a checkbox list for which inferences should be on or off.
Not having that option is a major bummer for me.
Advertisement