Hi Martin, yeah NURBS models have a higher overhead for things like edge structures, basically for every edge in a NURBS model there is a lot of various pieces of data including both a UV-space version of the edge as well as a 3D edge curve.
So when you get to a situation where you've got something similar to a polygon structure with just a lot of little planar faces and lots and lots of little edge curves, that is not very efficient data-size wise in a NURBS model as compared to a polygon model structure.
NURBS are more efficient in other kinds of ways, especially when your model has large sections of it made of curved surfaces, since a big curved thing can be exactly represented as one NURBS spline surface while with polygons you have to use a whole lot of little polygons to represent it.
For a case like you've got there, you'll probably be fine with those heavy pieces if you just break them out into a separate file (select them and use File > Export to put them in their own separate file and then delete them from your main working model) so that you won't bog things down so much with a huge file for your main working model.
A little more progress, the lower porting of the back-end of this Lamborghini is a lot more tricky than i first thought, also it's the first time modelling this area, no excuse i know, but sometimes i need a few goes to get things looking right! But i am happy with my artistic licence result ... What say you ?
Some more updates, this is getting more tricky and I am finding myself having to try and work from photos. If you are reading this Michael, any chance of some proper camera matching in v3 ?
@ Burrman - Glad you like it, and as you asked so nicely, and because we all like to sho off from time to time, here are two more screen grabs ...
> If you are reading this Michael, any chance of some proper camera matching in v3 ?
Usually camera matching is more associated with animation type functions, so it's not really on my radar for implementing in v3.
Are you trying to use a perspective angled reference image? Usually that type of image reference is problematic to use as a reference, you need to find things like top/front/side blueprint like reference images instead if you're trying to trace over stuff, since it's primarily intended for reference images in MoI to go on the Top/Front/Right views.
Hi Michael - Camera matching is probably the wrong term to use here, but yes trying to line up a reference photo rather than the normal top/left/right views. Hopefully this description may help explain what i think might work.
To be able to create an image plane in a perspective view only, which can then be rotated so as to be able to match up to straight lines and angles, enabling the end-user to be able to better approximate how far certain details protrude that are obscured from the normal flat 2d views
This will only be an external view type of thing, my skills are not up to doing interiors yet.
> I can imagine after tweaking on it for a month, it will be fantastic!
I hope it wont take that long, but you never know. This is what I love about NURBS, once those splines are in place the speed you can create surfaces is far superior to SubD - Well in my eyes at least. I hate to sound like a sales person for MoI, but given it's tool-set at present i am amazed at how well it is coping with this project, hopefully when it's completed i will have more time to play with trying to get an external renderer to play with MoI scenes directly.
Glad you are enjoying my suffering as much as i am - hehehehehe :)
> To be able to create an image plane in a perspective view only, which can then
> be rotated so as to be able to match up to straight lines and angles, enabling
> the end-user to be able to better approximate how far certain details protrude
> that are obscured from the normal flat 2d views
I don't really follow the rotation part of what you're describing here.
Normally that kind of perspective matching mechanism is a function in rendering or animation programs to make the virtual 3D camera line up with the position that the actual camera was at when taking a particular photo or video - that's so that newly rendered content can be added to the image.
It's a pretty involved process - there's a solver mechanism involved which may involve you drawing reference lines or picking reference points within the model and then the result of that is a camera position for where to put the camera for the 3D view which should then be locked in place and not free to rotate around anymore.
In order to implement that in MoI would require quite a lot of new systems to be implemented - a new system for applying an image onto the viewport that stayed as a kind of billboard always flat to the view, some UI mechanisms for managing those bitmaps, some UI for managing the perspective match process, some mechanism for locking the 3D view in place and UI for that, etc...
Really the biggest problem with all of that stuff is all the various chunks of UI that would be required - UI development tends to be very time consuming for me because I try to manage it in such a way to avoid things getting bloated or too complex. So usually I'm quite conservative about things that would require a lot of new UI to control it.
So anyway because of all that various stuff it's not too likely to be something added to MoI soon anyway.
Forgive my ignorance I seem to have over complicated things again ...
I thought a simple plane with a decal image that could be tilted/rotated on all axis combined with the zoom and pan options might have been enough to do a crude camera match. Being able to scale one side of the plane as required so as to create rhomboidal shapes to help with perspective issues.
It was just a thought, but as is often the case, a not very well thought out one - Lol
Hi Martin, yeah I don't think it's really feasible to do a camera match calculation just purely from image adjustments like that, from what I have seen (I have not implemented one before myself actually though) it usually involves things more like analyzing stuff in the image itself possibly including things like user guided annotations or selecting matching points between 2 photos or things of that nature.
In any case since I'm not actually 100% familiar with all the details it would involve a lot of research as a first step before even actually digging into it.
Then like I mentioned before there would likely be a lot of UI to control the whole process and in general UI design for me tends to be a highly time consuming area of work.
So yeah I would expect for it to be a quite significant undertaking in order to do that.
For rendering programs it makes sense to have it for a focus area since it's frequently needed there to merge the result of the rendering into an existing photo. So if you do want to do some kind of camera matching work you may want to look at a rendering program which has that feature in it already. If you are able to use a different program to calculate the camera position that matches a particular photo, it is possible to then set that camera position in MoI by using a script on a keyboard shortcut, see here for an example: http://moi3d.com/forum/index.php?webtag=MOI&msg=2396.11
@ Michael - Thanks for given it some consideration Michael ... It was just an idea. As you say the more heavily bloated and expensive rendering suits do tend to be where the camera matching is, other than Blender, which to be honest i cant use for modelling ... Too cumbersome. I can live without it but it would have been nice to have something like it in a pure modelling package.
Otherwise here's a quick update on how things are going. For rendering purposes those fans in the back-end are good enough, but they are a mish mash of parts that wont marry up so will need to be re-done at some stage.
Not much done again today, but i am busy doing other stuff ... Little updates, Front grill added,(but then exported out), side mirrors done, rear light clusters modelled ... File size is now approaching 20mb in size and the screen is beginning to stutter on rotations even without the grills - But most of the model is done now, not many surfaces to add but lots of fillets and blends to come yet.
My modelling skills are not up to that kind of accuracy yet - This project is to get a good body shape only and for me to test MoI on file sizes and ability to cope with medium detail level - I am dreading the wheels but at least i know where to come to for help if i get stuck. Once all the modelling is complete then it's export and import time with Blender Cycles as the target rendering engine to see how good that is too!
Reveton-moi-v3-03.jpg