Posts made by dsarchs

@ notareal

I hadn't considered integrated rendering -- thanks for the info. That still wouldn't affect Thea or other stand-alone engines, though, would it?

I don't understand why everyone is so down on the new version. Before I go on, though, let me say that I haven't yet used it, just watched the new features video.

Solid tools! I don't know how well bool tools works (congrats to Whaat for another awesome plugin either way) but integrated boolean operations and automatically created solids is a VERY good thing. I hope that 3rd party rendering software (here's looking at you, Thea) finds a way to import the solid property for calculating density (fog, SSS, dispersion, etc.)

Google Earth integration: we know that Google is focused on this and we would be getting improvements here no matter what, but these look actually useful. Normally I would have to remap a color image over the imported topo -- so this will save time and might be a little more accurate. Increased triangulation will be good as it used to get kind of dicey for closeups. As for making buildings (I know this already existed at some level) might be useful but I'd have to try it before really commenting on it. Still, overall this is a nice improvement.

Layout. I hated layout when it was first introduced and honestly haven't used it since. The new version does look nice, though -- specifically the examples they show. If sketchup can produce decent detail drawings (depends on the work-flow, though) then I'd be excited, especially as it can export (cleanly?) to autocad.

As for the "missing" items, how many of these really matter?
The shadow bug would have been nice to have fixed, but I never use SU output -- it's just so easy to get an unaltered "clay model" type image from Thea or whatever.

64bit support? Would have been nice, I suppose, but how much would it have improved anything? Being able to load more into memory doesn't help because it's the inference engine that slows it down (and that makes SU so good -- a pharmakon, I suppose) and more loaded into the model would make it worse, not better.

Multi-core support? I would rather them spend time on openCL (screw CUDA and it's required hardware) since almost the entire SU user base probably has a decent graphics card not to mention that a proper GPU can be vastly more powerful than an extra core or two (try smallluxgpu to see an impressive, real-world comparison).

Animation? At some point it's better to have multiple programs than one that does everything -- especially when blender is free, easy to import to, and already has powerful animation (3ds for you rich people, I suppose

Basic other light sources would have been nice, as would proper UV mapping but as with the previous comment, they're are other programs that doe that -- I would like them to get integrated sometime soon, though.

One thing I haven't heard people mention that I am disappointed with is that they didn't clean up dynamic blocks. That seemed like it had potential but had been awkwardly implemented so that it made it easier (most of the time) to continue doing without it.

I had pretty low expectations and am VERY glad they didn't waste their time improving styles (does ANYONE use styles? With one or two exceptions they look unprofessional -- and those exceptions are people gifted enough to come up with awesome product with whatever tool -- NOT because they had 'styles'). This looks like a quality release -- although it might turn out to be unfinished when I actually get to try it. I have to say, though, at this point I'm pleasantly surprised. I'm done with my rant.

Is there a way to align the scale tool to the plane you're working on? I love the tool, in general, but hate that as soon as the object being scaled is no longer perfectly aligned with XYZ axis then the tool distorts. I don't think it does what I want, but I'm hoping someone will tell me a way to make it work like the rotate tool, where you can hover over the axis you want, hold shift, and it becomes locked on that new axis. Ideally this would work with both single planes and complex groups.
I suppose this would probably be a plugin request -- I'm just hoping it already exists and I've just never figured out how to do it.

Thom- Import to Blender, decimate, export back to sketchup.

You have to give us something to start with. Can you post a plan (or better a SU model) of your kitchen?
What are you trying to accomplish? Who uses the kitchen? Lifestyle?

Without any of that I'll say that uppers are kind of lame. Go for under counter storage with a paperstone top.

See how lame advice is without the context?

nvidia scales much better than ATI, so if you're thinking about multiple cards, definitely dual 480's in SLI. Having said that, they run very hot, noisy and draw a huge amount of power, so make sure you have a hefty (probably 1200 watt) PSU and a water cooling system.

I haven't had a chance to try it out yet, but that sounds really helpful. Thanks for the advice. Any good tips on setting the scale of the photo?

Can anyone explain how to photomatch the same building multiple times?
I want to rough-out the shape of the building from an overall photomatch but then rotate to different views and even zoom in to model detail work with subsequent p.m.'ing.
I'm not sure if this is even possible. Does anyone have any experience with this?

Solo,

That last rendering is pretty great. My only comment is that the desktop is so clean -- it hardly fits with a mad scientist. I would throw in some dust/cobwebs, scorch/burn marks, and/or cracks in the tiles. Make it look like he his done crazy (gone wrong) experiments in the past, not just that he collects odd desktop trinkets.

Did you mention what program you're using to do the renderings?

I agree with others that some of your images have too wide of a FOV. Also, A lot of your interiors show small light leaks that make the walls look like they're not completely touching the floor. Also, and this is the most noticeable critique I have, is that your bump maps are MUCH to strong. I would turn the value down to about 10% of what you are currently using.

I'm impressed, though, with how much detail you put into adjacent properties -- almost as much as the project building.

Actually, I just found another problem. When the SketchUP scenes are imported into Kerkythea, every camera is identical. I assume this is a problem with importing axonometric views -- does anyone know if there is a work-around for this?

Thanks, that technique seems exactly what I need.

If anyone else is interested in, Tomasz's Bring My View Back plugin seems made for this, allowing you to create a grid on the screen so you know where to make each frame.

http://www.tomaszmarek.eu/sketchup_1.html

Is anyone aware of a plugin that would create multiple scenes/views that would piece together to form a much larger single image?
I need to do a larger rendering (using Kerkythea) than my computer can handle and thought I remembered hearing about something like the above a while ago. I've tried multiple images with slight camera rolling (like you would do for a panorama with a real camera) but stitching these together is always a little imperfect and it seems like a computer rendering should be able to do better.
Can anyone think of a solution or plugin that would work?

I was just responding to an earlier question about whether or not you could use different types of cards for cross-fire/sli.
As for directx, I was saying that the benefits of directx 11 are useless unless you get the top level of cards. I know that SU doesn't use cross-fire or dx, but presumably the machine will be used for programs other than su. GPGPU rendering/processing is coming soon to say nothing of games
Besides, if SU is the sole reason for investing in new hardware your money would be better spent on cpu/cooling/ram or something and just getting an 8800gt video card.

This might be a little late in answering but I'm almost positive you need two of the same card to use cross-fire (eg. (2) 5850 CARDS).
I'd be wary of spending a bunch of money on a workstation graphics card for SU. The only real difference between "gaming" cards and workstation cards are the drivers and the tech. support. For some programs -- 3d Studio Max, for example, the drivers make a huge difference. I haven't heard that they've been optimized for SU.
For ATI cards, the DirectX 11 seems to be useless for anything less than a 5850 as the cards aren't powerful enough to compensate for the performance hit from using dx11 instead of dx10. The 5000 series is MUCH more energy efficient but you can get more power/$ from using the 4000 series (not to mention more mature drivers). The 4890 is a very powerful card and can be had for ~$180 or so. That said, the 5850 and 5870 are very nice cards.
Recently there was an article (http://www.tomshardware.com/reviews/2d-windows-gdi,2539-11.html) about 2d (and 2.5d) graphics where the ATI cards perform VERY poorly (and the NVIDIA cards are not much better). Fixes are supposed to be in the works, but it's still worth knowing.

NVIDIA cards tend to be more expensive than ATI for comparable power although they are sometimes more stable and you need an NVIDIA card to use CUDA (having the graphics card help with general processing) in programs like Octane Render (which looks like it could be amazing).

Hopefully some of this is helpful.

I made a dynamic window component in an attempt to learn what goes into them. This window is based on the Milgard picture window profile. It seems to work just fine -- feel free to use it. I did have a couple of questions, though, if anyone can help answer them. (By the way, the material on the glass is for use as a bump in renderings [KT] to give the reflection just a hint of waviness).

1. I would like to be able to scale the size of the window as well as just entering sizes. I understand how to do either but not both. Entering data seems to require that you have set variables while scaling requires... grayed-out variables. I thought I might be able to do this through proxy/referenced variables but haven't given it enough thought to know if that will work. Any ideas?

2. I want to add mullions to the window could be resized and repositioned inside the window. I'm more sure that I can do this but haven't gotten around to it yet. Can you make a scalable array (of mullions) inside the window while the window proper does not use the scale tool to resize?

3. Would it be possible to (completely apart from this component) draw a plane over the window opening and replace that plane with this component already sized correctly? This seems like it would be easier for using in modeling. ...or is this something for a ruby rather than a DC?

4. Can you set limits on user input? For example, if x<12 then x=12?

LINK

For some reason attaching the SKP doesn't seem to work.

Lastly -- SU DC seem really poorly implemented and awkward. I was really hoping they would be easier and more powerful -- like a 3d version of AutoCAD dynamic blocks. At least taking advantage of stretching rather than needing to have every piece in a separate group getting moved around. It seems to add a lot of time building DCs and makes their organization pretty ungainly. I'm glad there's something now, though.

As I said earlier, this is my first DC and I'm just learning the basics, so any feedback would be greatly appreciated.

The solar panel is under the "solar panel" tab for this component

Solo,

when it asks you to vote for a similar idea, you can still choose to vote for yours. At least that worked for me.

Hi everyone, I thought I'd just tack this request onto a related previous one.

Would it be possible to change this script to create components instead of groups for the individual pieces of lumber being created for this script? I was trying to texture a wall made with this (orienting the grain) but I'd have to texture each individually or erase them and copy the 1 textured board all over (which makes this plugin worthless).

Could someone modify the script to create all the boards as instances of a single component -- ideally with each different size being a different component (stretching a single component would distort the texture). At least this would I would only have to texture 2 different pieces for a single wall.

I'm really not sure if this is a huge request or very simple, but I would be so glad if someone could help me out here.

You could also do a quick model in Blender [or similar] using the fluid simulator and import that into sketchup. Cheating, I suppose, but that's a specific thing that other programs are just much better at [and free, in the case of blender]. Use the right tool for the job.