I have not been successful using the context 'stretch' facility for textures in sketchup, i.e. the UV info doesn't seem to get transfered to Vray.

Therefore I always adjust the scale of the texture using the SU Material editor. As an example, of you have an image of brickwork, which is say 10 bricks high, then I set the height of the texture in the SU Material editor to be 750mm (75mm per brick!).

The only time I use context texture editor is to set up a 'projected' texture or to position it.

David

OMG!! 10 hours! You must be very patient. (I don't use displacement)
Having said that, it reminds you that comparable tests need to be done on:
a) the same scene
b) the same machine
so time differences are actually all relative.

David

Adam

I tried your external (medium) setting, and found it slower than my current external setting.(Quality seemed comparable to me)

My setting, for a 1800pix wide render (with bumps etc) took 7m 19sec, using your medium setting took 24m 28sec. I haven't really got the time to test the 'higher' setting.

I will give your internal setting a go later today)

David

I had the same problem and now its fixed - it was the vray plugin.
I purged my sketchup materials, then I opened the vray materials and right clicked on 'scene materials' and you get a 'purge unused materials' option - purge these. Now, very important, open up the vray 'options' window and then close it. Save the sketchup file as your default, and close it. When you re-open sketchup all the additional materials have gone!
(I'll post this on the vray site too)

David
(vray 01.05.30 and SU7)

honoluludesktop - thats a neat solution. It didn't occur to me to do that, I like your thinking! In effect you are re-creating the other part of the image which has been cropped. Very good.

David

You see how the verticals are vertical. Thats because the image has had them corrected, either in the camera or later in software. A regular camera lens tilted at this angle would result in converging verticals. Therefore . . . you will not be able to match this photo.
Having said that, if the model you wish to insert is a low rise, then you will probably be able to eyeball it to get it close.

David

Excellent explanation.

The other thing with the photo, in this case, is that it doesn't allow the whole of the landscape works to be shown (the height of the monolith) - which somewhat defeats the object.
A decent photo has to be a priority, and if that is not possible then Lapx will just have to work with the material he has, and accept that it will be a compromise.
My own view is that as it doesn't have to be 100% accurate (after all, you are not exactly trying to line up brick courses!) then just getting the kerb lines to match, with the background building in their approximate positions, should still give an acceptable solution.

David

Lapx

I had a go at this one . . . and I see what you mean - its a real b*****

The photo may be cropped. Photomatch doesn't like this and it does make it difficult - this may be the case here (but I'm not sure)

The elevation of the buildings is a factor. What we don't know is what level they are relative to each other.

Having said all that I think you can use photomatch to get a reasonably close match.

I started with a new model. Go into photomatch and use it to draw a line down the centreline of the side road. So now in your drawing you just have one line. Import your main model and position and rotate it and scale it so that the centre line of the road in the imported model sits on the line you have just created. Go back to photomatch and you should see that it is at least in frame.
You can now use the various handles to adjust the model to match the photo.
I turned on the edges, made it wireframe and fiddled.
However it is such an awful photo, I agree, it ain't easy.
You also have to remember, that photomatch is often not 100% accurate anyway, due to lens distortion etc in the original photo, so adjustments usually need to be made to the final composite in Photoshop (etc) to get the best match.
I've sent you a PM.

Regards

David

lapx

As you have a cad model, I would use it to model the key features, road kerb and distant buildings (if you can). Turn wire frame view on, and use photomatch to adjust the different parameters. As you adjust the handles on the green and red perspective controls you should be able to get the kerbs to line up with the photo, and having it in wireframe helps you to see what you are doing more easily.

David

lapx
You have some references - the centre line of the road, the road markings (stop line) and the street lamps (for verticals -if you are lucky!). If you have a scale plan on which you can plot these it would be a good start.
However, the SU model you've produce so far looks very close, so eyeballing the model might do the trick.

David

This 'component not rendering' thing was a known problem in earlier releases.
The latest release (01.05.03) has fixed this.
It is important to apply textures within a component, not to the component as a whole.

Good Luck

David

You could also try this excellent tutorial on the ASGVIS forum

http://asgvis.com/index.php?option=com_smf&Itemid=90&topic=5225.0

David

Well, at 30Mb, your model isn't very large. Problems can occur especially with nested components. If you have a lot of these then vray may just be giving up on them. So I don't think that making the whole model a group is going to help. The other issue is large textures. Sometimes if these are very big then it can cause crashes.
Saving the model and re-opening it just before rendering is worth a try - gives a clean model.
It is worth checking the 'render progress' in the seperate window which opens, to see if it crashes at a particular point, which may help you to diagnose the problem.

I would make a copy of the model and explode any nested components. then save it and re-open it and then try rendering it.

Hope this helps

David

Linea
I had a look at sti spirit, and it looks interesting. Can't seem to find out a cost for it though - I suspect it is expensive, as things usually are when they won't tell you how much it costs!

David

Stefanq

This looks great. At least doing your own you can get the appropriate level of detail. Nice one.

David

I am currently trying to develop a workflow link from SU to CAD drawings for architectural work (small scale stuff - houses).
So far I have worked out that the plans and sections can be exported from SU with a thickness applied to the section cut, thus eliminating the need for linetype adjustment later. I am currently using CAD to add dims and some shading to the SU import. That just leaves the actual detailed drawings and as I import these as 'standard' details (to be adjusted per project) this is quite simple.
With discipline in the creation of SU model, so far this is working quite well.

David

. . . and if you make a mistake, do you use correction fluid on the screen? I'd love to see you working on a sketch!

I think SU actually creates a 3 point perspective, but because most photos are taken at eye level they look like 2 point.

David

G

Here is the rendered model. It is then merged in with the photograph in PS.
The mirror ball in the foreground was to check out the HDR sky, to see if it was positioned correctly.

Cheers David

Modelhead
Well I admire your skill with the eyeballing!
Before photomatch I used to eyeball the perspective too, but it often used to drive me bonkers. You know the thing, "is it the angle" or "is it the degree of zoom" that needs adjusting, so when photomatch came along it was a gift for me. I must say, I don't have to fiddle with it - just sort it out once and its done.

David

Jeff
All that a SU photomatch does is to set a viewpoint and zoom, such that the view will match the given photograph. It creates a scene tab and you can go back to this tab any time and render that scene. Otherwise you work on the model in the normal way.
If you render the model (e.g. with vray etc) then the finished render can be composited in Photoshop with the photo you chose to match.
If you just export a jpeg of the view, then I believe you can also composite that in the same way.
I think the latter is what Utiler has done.

Modelhead

@unknownuser said:

Used photo-match once upon a time....fun (great for a bird house)...but not for production and not for revisionist approaches.

Not sure why you should think this. Photomatch is essentially a way to put your object (building) in a context. Some greate renders have been done this way, and there are lots of excellent examples around.

David