Posts made by pitrak

I think developing sketchup in different directions might be interesting!
The fact that sketchup is still the 3D modeller of choice for many experienced modellers probably has more to do with all the great plugins around here than with the last upgrades we've seen.
The evolution of sketchup in the last years has been slow and not always successful (dynamic components anyone?)

My enthousiasm for this acquisition might have to do with Trimble's expertise in laserscanning as well. The holy grail of surveying in architecture would be an automated process from laserscan to a textured 3D model. If this would be possible with sketchup, I'ld be delighted!

So good luck guys, I hope you get all the freedom and support you need to take sketchup in new directions!

Thanks a lot for all the advice, I'll stick with nvidia then!

Solo, this is the laptop card, it's a dedicated one: NVIDIA GeForce GTX560M 1.5Gb GDDR5, PhysX + CUDA ready

It's a BTO laptop (dutch, sorry). I was doubting between this and the Radeon 6990m. But when looking again at their site, it seems they have removed the GTX560m option! Now I can only choose between the Radeon and the GTX580m which is really expensive..

Or take the cheaper P-book with the GT555 (no cuda, does have optimus, which is good for battery life.

Buggers, I thought the choice was made...

Solo,
I don't want to hijack this thread, but can you elaborate a bit on the nvidia vs AMD idea?

I'm in the market for a laptop based on a clevo barebone and I have the choice between the Geforce 560M (3Dmark06 15000) and AMD 6990M (23000). Nvidia was my obvious choice at first, but when I started looking for the advantages of CUDA I started doubting. Some minor accelerations in adobe CS5 for supported CPUs (slower or way more expensive quadros), CPU rendering in Octane (which I don't use).

For the rest I have the impression most software developers support both AMD & Nvidia. OpenCL seems the way to go, although it's not clear how well it will work and how widely it will be adopted. I hope CS6 will be OpenCL and not CUDA accelerated, but that's not clear yet.

So any other good reasons to go for the slower and cheaper 560m?

The % reduction seems to work great. File size went from 18mb to 10mb and quality is still quite ok.

But when I tried the 20pix/meter option I got different results. I posted a cropped screenshot of before and after the reduction. As you can see the right facade remains unaltered, while the new one and the left neighbour are reduced a lot.

I've used 'make unique texture' on some of these, maybe that's the problem?

BTW, this could work very well with "Component edit window". You select a component, open in another window, change texture size and bring it back. Like that you have control over the different parts of your project.

---

---

Simply amazing! I was just going to look up a way to reduce texture sizes on a 18MB SU project. I open Sketchucation and this is the first thread of the bunch

Thanks for this, I'll give it a go and let you know how it turns out!

Another question re: SU and Win 7 64bit:

32bit apps can only use 2GB at a time. But if I have multiple instances of SU open, will they all have their own 2GB of RAM allocated?
I'm planning on buying an i7 quad laptop with 8 or 16 GB of RAM, so would be interested to know this before I buy.

Found this, but is says that the principal does not apply to older software. And as SU is not always state-of-the-art in terms of resource use, I'ld just like to know for sure..
http://www.sevenforums.com/general-discussion/145160-whatis-mem-use-multiple-instances-same-32bit-program-win7-64bit.html

Thanks!

I've just tested the plugin on SU8 (win7 32bit) and it only works for groups, not components - the new SU window just doesn't load.
Any suggestions?

Nice and easy!

There's a good way to automate the process if you want to make a lot of those images. And if you have photoshop & bridge. Make three scenes with style settings as described. Include these three in the animation. For every camera scene, export those three style scenes (in an animation without transitions).

In bridge, select the three images, go to Tools>photoshop>Open files into photoshop layers. Make an action for the blurring and blending of the layers and save. Now you just have to repeat this action on the next series of images.

@unknownuser said:

okay . .. what's the catch?

Here's what they say on their site:

How could this be? How can StudioGPU offer such a valuable piece of software at no cost? Let us briefly explain how we can do it. We firmly believe that MachStudio Pro is a revolutionary piece of software that will completely transform the way that you work in 3D by saving you both time and money. Based on our confidence in this notion, we think we can create a viable business model from generating revenue from Support, Training and Consulting Services. Please help us change the 3D production world for the better. Take a spin on MachStudio Pro today. You won't be sorry.

Sounds reasonable. A good free product will probably get hundreds of times more users than a 4000$ one in a market flooded with visualisation solutions. So if they get 20 or 30$ on average through selling extra services, textures, training, ... they might make more money out of it.

Anyone tried it yet?

@matte said:

I didn't notice anyone mention shaderlight? I played with it only breifly and thought it might have promise, but I do not yet know much about the rendering game yet. What do you more experienced rendering folk think of it?

I've played with both Shaderlight and Twilight. IMO Shaderlight was too limited. The realtime preview is a plus. But on all other areas, Twilight wins I think. The fact that you cannot use your own bumpmaps is a big downside to shaderlight. That might change in the future, but for now it is too limited to grow into I think.
With Twilight you can make more progress over time. And if you've reached the limits, you can add kerkythea or thea.
I have bought Thea just before the price went up, and I'm planning to buy twilight soon. I thinks I'll use twilight most of the time. But with the quality materials over there on the Thea forum, I should really expand my (yet to be explored...) rendering skills to thea asap.

That's the default behaviour of the free Windowizer indeed.
4 looks amazing, I especially like that you can input proportions for the different panes. Albeit unfortunately only in the horizontal division if I understand correctly.
It does cost 10 bucks though. I think I'll order it, seems well worth the cost.

This is just what I needed for a project where almost nothing is orthogonal.

Just like anyone else, I am wondering if this plugin will be developed any further? I am also convinced that this could become the standard pushpull for plugin people, but it still needs a lot of work.

1. It leaves lines at the original position, which you have to delete manually.
2. I would prefer the inferencing should to be the same as SU default PP behaviour
3. integration with JPP would be great (and be perfectly in line with the concept of this plugin)

a4chitect has not been active here for half a year.. Maybe Fredo6 can incorporate SmartPushpull in JPP? Or are there other options for collaboration?
This plugin to me is one of the most promising and SU like plugins around. It makes a common problem a lot simpler to solve without adding complexity on another level. It's just a shame that the implementation is not up to the promises...

I'm having a very similar issue here. I've modeled a house (rectangular, everything based on the same axes) and Geo-located it to have accurate sun and shadows. Now I want top view, elevations, ... all based on the original axes, but it does not seem to work.

I've tried several things:

1. Clearing the geo-location (Windows > Model Info > Geo Location > Clear Location). Does not change the World axes.
2. Resetting the axes does not help as these views apparently only work on the World axes..
3. Copy paste model to a new SU file on world axes > set north with Solar North toolbar. That does the trick.

That means I actually need two models: one geo-located for sun and surroundings and one for making plans and views. Seems quite silly.. Is there a way around this? A plugin maybe that allows you to set the world axes while keeping the north and geo-location? Or a way to make the views relate to the current axes rather than the world axes?

Thanks!

Found the best practice in this thread. Seems like I'll have to use layers after all... Would still be nice though to have a fix for the scenes/hidden geometry bug.

http://forums.sketchucation.com/viewtopic.php?f=15&t=28486&p=247297&hilit=+layers+outliner+scenes#p247297

@wo3dan said:

(*)One (big) disadvantage of this workflow is that not only axes orientation will be inherited from the higher level when exploding > grouping. Also will default colored entities and default layer0 entities inherite their groups properties. So it depends on whether you have used component coloring and layers. Which is almost always the case.
So manually is the way to go.

Thanks Wo3Dan! The biggest advantage of the 'exploding workflow' would be that I have to rename all my components after exploding and regrouping (recomponenting actually). I heavily rely on component names in outliner to navigate my model.
I did everything manually, it's actually not too much work.

Time to dig up an old thread..

Same issue here - and while googling around for a solution, it seems like there is none. I went through the effort of making a really well structured model. The nested and named scenes make for an easy to navigate model in outliner. But not being able to have visibility states in the scenes is really frustrating. It would speed up the workflow considerably and would be great for exporting jpegs afterwards.

There seems to be a ruby or a workaround for most of those problems, but I can't find any in this case. I'm not too eager to start using layers as it will make the model more complicated and confusing. Will eventually use layers for plants, furniture, ... but don't like the logic of using it for visibility.

Any suggestions? Is this like the shadow bug? Some sort of annoyance that might suddenly get fixed but you just have to live with until then?

Similar problem here. I've made a fairly complex model for a house renovation. There are plenty of axes and components in the model - so now I'm restructuring the whole file - both the component names and structure and the component axes.
Let's say I have a facade component with 5 other components: windows, brick, stone, gutter, ... I would like all the components inside facade to have the same axes as the main component. I know I can do it manually, but that's quite time consuming. Is there a way to do that? A plugin maybe?

thanks! p

@aerilius said:

@pitrak said:

questionable implementation

SketchUp has already very flexible tools (-> modifier keys) that allow a basic interface with only very few but powerful toolbar buttons. Similar to your request, there are ideas to "complete" or polish existing tools so that they have less limits:

I was only referring to DC and now Back edges. I agree that the other tools are simple, flexible and powerful. But there is indeed a lot of room for improvement.

Personally I am a big fan of MoI3D, I just use it for different purposes than Sketchup. Developed by one single, brilliant guy, who gives amazing user feedback and constantly works to get the program better. Big advantage for him is his users are mainly pros and give very meaningful suggestions.
If you talk about simple buttons with a lot of power, for me MoI beats Sketchup. You can trim a line with a solid, operations are dependent on your viewport, infinite construction lines can be drawn while you're inside your command, it all works flawlessly and powerfully with very simple functions. If you're a newbie, you'll get up to speed fast enough. If you know all the tricks and hidden shortcuts and functions, you have lots of control and drawing power.

But MoI is 'just' a nurbs modeler, and for some tasks I like to model with textures applied. What I want to say: we all know what sketchup is good at, but we also know that it could be improved upon without giving up on simplicity.

I'll get back to my model now.. Just have to push the right buttons and she does everything I want!

@jbacus said:

As a matter of policy (both Google's and before that @Last's), we won't comment on unreleased features or future directions for the product in any but the broadest possible terms. I recognize how much you'd like to know what's coming next, but we need to preserve a quiet space in which the team is free to experiment with projects that may not work out. Software development is more like painting than manufacturing, and you're better off in the end if we're free to be creative.

john
.

Thanks for your answer. Exactly what I'ld like to do: paint and be more creative. Still haven't got a clue why Back Edges were not implemented in depth, but now at least I know why I don't know.

I was initially quite happy with the improvements - maybe because my expectations were rather moderate and because I was happy there was something new at last..

I was especially excited about the new 'back edges' because I like using SU for NPR. I had tried to get something like back edges by overlaying normal and X-Ray imagery in photoshop. Back edges could take the process to another level. BUT... it seems like you can only use the back edges in the Default Styles. In all other styles it's disabled. And nothing about back edges in Style Builder, so you cannot design your own back edge style.

If something in the styles department is changed one would at least expect it to be fully implemented. This looks like a first simple tryout of a new function. Since it's all part of the same render engine it should not be that difficult to implement it in all styles and design some extra hidden line styles for the existing styles? After all these years, even the 'competition' styles haven't changed, and no new styles were added. So I'm not disappointed about the new functionality - that's how SU works - but rather about it's shallow implementation.

Is there a roadmap for Back Edges mr. Bacus?

For me it's not so much about the lack of new functions (we have rubies for that), but about the poor implementation. Dynamic Components in 7 - great idea, questionable implementation - and now back edges in 8.