[CODE] Camera settings

DavidBoulder

I have a script I'm working on where I'm trying to iterate through scenes/pages and look for faces parallel to the camera for each scene. I have the code to iterate through faces and test to see if parallel to the camera, but I could find a way to grab the camera for other scenes to iterate my tests across all scenes.

I suppose I can change to each scene and then get current camera view, but I would think I should be able to grab the camera from a scene without having to change to that scene. Keep in mind, I don't know the scene names.

edited content below

OK, I solved my problem, here is a piece of the code.


# draft script - David Goldwasser
# find faces parallel to camera and draw lines back to camera

model = Sketchup.active_model
entities = model.active_entities
pages = model.pages

faces = []
entities.each do |n|
 faces.push n if n.is_a? Sketchup;;Face
end

pages.each do |page|
camera = page.camera
eye = camera.eye
 faces.each do |face|
  status = camera.direction.parallel? face.normal
  if status then
  edges = face.edges
  edges.each do |edge|
   pts = []
   pts [0] = eye
   pts[1] = edge.start.position
   pts[2] = edge.end.position
   face = entities.add_face pts
   end
  else
  end
 end
end

The pages.each do is where I was having problems before. I was actually switching to that page and then getting current view, camera etc. But the animation was causing problems, so I had to have a dummy UI.messagebox to get it to behave. Now I don't and it is much cleaner.

So I have an edges.each nested in a faces.each, nested in a pages.each. My indentation for isn't so clear when previewed here.

chrisglasier

When I regenerate a scene without zoom extents the scale is very small. Zoom extents would be OK if I didn't want to capture the original zoom that produced close ups for example. But I think it important, so if anyone can help I would be grateful.

Thanks

Chris Fullmer

I'm not quite sure what you mean. Zoom extents is a separate matter, not really related to scenes, except that it can change the camera position.

So do you want to use zoom extents? Or not use it? Or do you want to match the scene camera settings precisely? Or do you want to get the scene camera, zoom extents, then update the scene camera to reflect the new camera? Or probably something entirely different that I'm missing?

Chris

chrisglasier

@chris fullmer said:

... Or do you want to match the scene camera settings precisely?

Yes I want to match them precisely. I just used zoom extents because the display was ridiculously tiny. Do you have have an idea why?

When zooming you change the camera position or something else? I didn't use new camera but active camera, perhaps that is what is wrong. All the scenes are reconstructed in the one Nameset tab by showing/hiding components (only layers for dims and notes).

Thanks

Chris Fullmer

So to further understand now: You have an existing model with scenes in it. Then you iterate otver the scenes and get all the cameras and save the camera settings. Then you want to use your script to re-create the cameras from the scenes. Is that the idea?

If so, then what your script does should not differ from the cameras in the scenes. so if the camera is zoomed way out in the scene, then that is what your script should make too. IS this not currently what you are achieving?

If that is the case, then it sounds like somehow you are missing something in the camera settings when you extract them from the scenes. And the field of view might be my guess. Are you checking or setting the Field of View at any point in your script? I think ideall you would check the FOV of the scene camera and then st your camera to match when you create your camera object.

Chris

chrisglasier

@chris fullmer said:

So to further understand now: You have an existing model with scenes in it. Then you iterate otver the scenes and get all the cameras and save the camera settings. Then you want to use your script to re-create the cameras from the scenes. Is that the idea?

Yes - maybe recreate views of the model is a better term.

@unknownuser said:

If so, then what your script does should not differ from the cameras in the scenes. so if the camera is zoomed way out in the scene, then that is what your script should make too. IS this not currently what you are achieving?

Please see pixs and code.

@unknownuser said:

If that is the case, then it sounds like somehow you are missing something in the camera settings when you extract them from the scenes. And the field of view might be my guess. Are you checking or setting the Field of View at any point in your script? I think ideall you would check the FOV of the scene camera and then st your camera to match when you create your camera object.

Please have a look at the code. From the console and js output it seems the fov does not change. Perhaps I have made the error there.

No zoom_extents

original scene

chrisglasier

... and the non perspective views are the same ...

Chris Fullmer

I can't parse your code very quickly Chris. Especially once it goes out to js, I'm lost. But it looks like you are doing the right thing by getting the fov for the perspective cameras. But somewhere in there, something is getting messed up. If you get the values correctly and apply them back correctly, the cameras will match 100%. There should be no need to do a zoom extents. So it is clear that somewhere the values are not coming through correctly. I would do some solid value testing. Output all the values before they go to js, then output them again as you recreate the camera. See if they differe at all.

Also, if you could make a model with a scene, then run your script on it and recreate that camera, then create a new scene that uses your camera info it might be easier to compare the settings of the 2 cameras - the original and your recreation.

Chris

chrisglasier

Thanks for all your effort Chris. Just by posting those pix **.mm** came to mind ... errgh ... but somehow oddly satisfying. First go seems to work but still not quite it. You know when the French gave you the Statue of Liberty, it would have been nice if you had taken up their metric system.

Tally ho!

Chris Fullmer

Hehe, yeah sorry about that imperial vs. metric thing. I hope you're able to pin down the problems!

Chris

driven

@chrisglasier said:

Yes guess what - fov is in mm by default!
Here's a kind of victory pic.
I amended the previous code in case anyone is thinking of doing something similarly lunatic.

Cheers.

hi Chris,

I can't get the feb016.avi to run on any player, I'm on a Mac but don't normally have problems with .avi
can you out-put a quicktime.mov as well? Is it the same video that's up on youtube?
I'm very interested in your approach and although I'm only just starting to learn coding, I'd like to integrate namespace conventions into my efforts right from the start.

john

chrisglasier

@unknownuser said:

Is it the same video that's up on youtube?

Yes. You could also look at my old website (button on left).

Now I'm off to the real movies for a while.

Interested in your interest.

chrisglasier

Yes guess what - fov is in mm by default!
Here's a kind of victory pic.
I amended the previous code in case anyone is thinking of doing something similarly lunatic.

Cheers.

Metric!

chrisglasier

A couple of odd things I just discovered when doing something else is that whereas the perspectives are spot on the standard views are unstable. The top/plan views for example get some 164% bigger when imported. See pixs:
original recreated
The other is that I .mm 'd the vector3ds for up. I changed them but it didn't make any difference. (Obviously I don't understand this stuff well enough.)

chrisglasier

Thanks for your help.

[flash=600,432:21eq7727]http://www.youtube.com/v/6h6m_xr5IYA&hl=en_US&fs=1&[/flash:21eq7727]

For anyone interested here are extracts from the rather clunky code I ended up to support this v quick video hinting at the potential.

Ruby extractor:


pages	= model.pages
		layers	= model.layers
		styles	= model.styles
		@parray	=[]
		pages.each do |page|
			model.pages.selected_page = page
			cam  = page.camera
			name = page.name
			pl = []
				page.layers.each do |layer|
					pl.push(layer.name)
				end
			pl = pl.join(",")
			st = page.style
			st = st.name
			pts = [cam.eye, cam.target, cam.up]
			per = cam.perspective?
			if(cam.perspective?)
				fov = cam.fov
				sub = [pts,per,fov]
			else
				sub = [pts,per]
			end
			sub = sub.join("yy")
			entry = [name,pl,st,sub]
			@parray.push(entry.join("xx"))

JS converter


p = ret[3].split(";");
	pages = [];
	for (a = 0; a < p.length; a += 1){
		p[a]	= p[a].split("xx");
		lgth	= fLength(nset);
		lab		= lgth.toString(); 
		o		= {};
		o.Type	= "Scene"; 
		o.Genus	= "New";
		o.Name	= p[a][0];
		o.Backlink	= [];
		o.Backlink.push(sRoot);
		o.Hidden	= [];
		nset[lab]	= o;
		nset[sRoot].Forelink.push(lab);
		pages.push(lab);
		nset[lab].Forelink = [];
		//mark hidden layers
		//alert(p[a])
		pp 		= p[a][1].split(",");
		
		for (c = 0; c < pp.length; c += 1){
			for (d = 0; d < combo.length; d += 1){
				if(pp[c] === nset[combo[d]].Name){
					nset[combo[d]].Filter = 1;
				}
			}
		}
		for (d = 0; d < combo.length; d += 1){
			nset[combo[d]].Backlink = [];
			if(!nset[combo[d]].Filter){
				nset[lab].Forelink.push(combo[d].toString());
				nset[combo[d]].Backlink.push(lab);
			}
			else{
				delete nset[combo[d]].Filter;
			}
		}
		//name,pl,st,camry
	nset[lab].Style = p[a][2];
	nset[lab].Camera = p[a][3];
	}

JS scene call


function fScenes(){
	var cell,nt,node,name,kids,dn,cm,genus,cmps;
	cell = document.getElementById("ScenesOn")
	nt		= config.nametrail;
	node	= nt[nt.length-1][0];
	type	= nset[node].Type;
	if(type === "Scene"){
		name	= nset[node].Name;
		kids	= nset[node].Forelink;
		dn		= [];
		cm		= [];
		for(a = 0; a <kids.length; a += 1){
			genus = nset[kids[a]].Genus
			if(genus === "Dimension" || genus === "Notes"){
				dn.push(nset[kids[a]].Name);
				continue;
			}
			dn.push(nset[kids[a]].Name);
			cmps = nset[kids[a]].Forelink;
			for(b = 0; b <cmps.length; b += 1){
				cm.push(nset[cmps[b]].Index); 
			}		
		}
		name	= nset[node].Name;
		style	= nset[node].Style;
		rec		= nset[node].Camera.split("yy");
		cam 	= [];		//eye,target,up
		for(a=0; a < 3; a += 1){
			cam.push(rec[a].slice(1,-1));
		}
		cam.push(rec[3])
		if(rec[3]){
			cam.push(rec[4]); 
		}
		data	= [name,dn.join(","),cm.join(","),style,cam]; 
		document.getElementById("trans").value = data.join("|");
		window.location = "skp;scene";
	}
}

Ruby callback


@dlg.add_action_callback("scene") {|d, p|
		a = @dlg.get_element_value("trans")
		a = a.split("|")
		name	= a[0].split(",")
		lyrs	= a[1].split(",")
		cmps	= a[2].split(",")
		sty		= a[3].to_s
		cam		= a[4].split(",")
		model	= Sketchup.active_model
		entities	= model.entities
		entities.each do |entity|
			if  entity.class == Sketchup;;ComponentInstance || entity.class == Sketchup;;Group
				entity.visible = false
			end
		end
		cmps.each do |c|
			a = Integer(c)
			#p entities[a].name
			entities[a].visible = true
		end
		layers = model.layers
		layers.each do |layer|
		layer.visible = false 
		end	
		lyrs.each do |lyr|
			layers[lyr].visible = true;
			#p layers[lyr].name
		end	
		
		pages	= model.pages
		styles	= model.styles
		page	= pages["Nameset"]
		styles.selected_style= styles[sty]
		page.use_style = styles.selected_style
		eye		= Geom;;Point3d.new [Float(cam[0]).mm, Float(cam[1]).mm, Float(cam[2]).mm]
		target	= Geom;;Point3d.new [Float(cam[3]).mm, Float(cam[4]).mm, Float(cam[5]).mm]
		up	= Geom;;Vector3d.new [Float(cam[6]), Float(cam[7]), Float(cam[8])]
		camera = model.active_view.camera
		camera.set(eye,target,up)
		if cam[9] == "true"
			camera.perspective = true
		else
			camera.perspective = false
		end
		if camera.perspective?
			camera.fov = Float(cam[10])
		end
	}

It seems to work OK but I don't like the zoom extents (I'm working on that).

As they say in the gallery ... C&C welcome.

Edit 15 Feb: Changed code in callback to match post post revelations

Scene extraction and view generation

chrisglasier

And here's the matrix returns for one component in the original and recreated view.