View Full Version : lw8 request

10-07-2003, 02:00 PM
I have been using LW for six years. I prepare architectural presentations. I also use Autodesk VIZ, but I have always found LW more user friendly and working with it enjoyable.
I always prefer to work with LW, but only in small projects. In bigger projects, I have to work with VIZ. Here I'd like to declare some suggestions that will bring LW to top in architectural visualisation.
-Better dwg, dxf and 3ds import and export.
-Instances / Blocks. (Think of a skyscraper including hundreds of windows of same type. You modeled it and the client wanted bigger windows. The only choice is to change every window one by one.(In some cases like that i prefer to define each construction element as a different object and combine all elements in Layout, but of course it would be very nice to be able to have an option in Modeler).
-Xrefs (External References) Think that you can insert/merge many object files in your main object file and if you make a change in an inserted object file, the changes will also be seen in your main object file.
-Better boolean tools.
-Rendering engine is the best in a package i have seen, but it must be developed also.
-Integrated LW. Then it would be possible to work faster I think.
-Array lights. Again, think of a skyscraper having hundreds of lights in a grid mode and you have to move and place them one by one in LW.
-Rail clone lights and some other cloning commands that can be used in multiplying lights would be helpful in some specific big projects.
-All modelling tools should be real time adjustable. I would like to model all objects like the ease of modelling a box. Many users would like to have a modelling history, but it isn't a must i think, in many cases, it only enlarges the file sizes.
-Orthographic view rendering. In architectural visualisation, sometimes the buildings shoud be wanted to be rendered in an orthographic view. I don't think that it is hard for LW to have this option. Because it is avaiable even in some freewares.
These are the main things that i can remember now. And these will make LW one of the most preferable 3d package in arcitectural visualisation, as it is in the other industries. I wait Newtek to add these features at once, maybe in LW 8.

10-07-2003, 03:28 PM
Xrefs for modeler... never thought of that one before - that could be very nice, how would you see it working in LW?

btw, if it was developed I'd like to see it fashioned after Bentley's Referencing rather that Autodesk's, which allows true self-referencing, cos that would solve the instancing problem aswell.

Agree with all the rest, and will add my usual "I need a bigger universe" :)

10-07-2003, 08:14 PM
Originally posted by ali
-Array lights. Again, think of a skyscraper having hundreds of lights in a grid mode and you have to move and place them one by one in LW.
Remember you can use Luxigons for exactly this... ;)

10-07-2003, 10:29 PM
I thought Xrefs were added to MAX because they don't have external object files?

10-08-2003, 03:00 AM
There are many users wich use Lw for arch visualization still now.surely this field would be a great market place for Lw. Nt have to take this in a serious consideration. many, many users of Lw use it for arch Viz.
from my POV Lw is currently a good package for movie/ viseo production. modeler rocks and rendering is quite accurate if not fast, indeed. also, right now the Nt development orientation seem to be CA wich was the real weak point in Lw.

I think that the implementation of arch related tools would be a good choice for the future. surely in Lw [8] we will not see many things in this direction IMO. actually I believe that similar tools as you describe woul be not only good for arch viz, but would increase the quality of the app, allowing the user to work faster
and better.

if I was in Nt I'd develop following these guidelines
(ok some of them are not in time order)

1) CA enhance it with dedicated and good tools for animation, expressions, keys management, pose management distortions corrections, etc. this is quite done . MB and Messiah, are a good reference for that dev. though.

2) rigging.possibility to rig and setup the char in layout quite done here too so let's see how is this implemented. with weights in layout too would be even better

3) rendering engine. faster and more accurate render engine for radiosity.

4) lights. they have to be quite reworked and enhanced they have some lacks and strange behaviour as showed in these forums.in that occasion a good introduction of precise arch measurements for lights would be required.

5) interactive tools for modelling. enhanced interactive tools with snaps for axes inside to allow a precise work without using numeric, and with previews, using a common window(maybe a beefed VIPER if there is an integration?) if needed and the possibility to change the effect till the job done, and not undo and redo all. working also wit simmetry.
introduction of precise measurement systems in them, for precise modelling. sort like measurement indications in the views, wich show how long is a side, height, etc.

6) snaps sort like in CAD systems

7) edge managing and creasing.

8) introduction of NURBS. this seem banal, but what about prototyping? a car is modeled better with Nurbs as well as mechanical and design stuff. a good conversion system from NURBS to Subpatches and vice-versa if possible,would be great.

9) istances.

10 construction history with the possibility to toggle on off.so no prob if you don't want to use it.

11) a good correction of plugins implementation.allowing them to communicate better with the core and not interfering each other.

12) finally, a good integration between Layout and modeler would be incredibly good, giving to the app huge possibilities in many cases and a smoother way to work. common tools for modeler and layout will give also to animation a great boost

10-08-2003, 02:07 PM

It would be cool to see the following features in LW8 in the near future. These I guess don't need really big work and rewrite the rendering core.

1. More scalable anti aliasing settings

Let's say High AA (17) setting is not enough, but I don't want to use Extreme(33), because that would be almost the twice of the rendering time. If I could write 20 into a field and deffine 20 samples with that number would be just very useful! Keep the default settings, but also let the user to define the number of the samples.

2. Implement the DITHERED function what motion blur already has, into the DOF settings. I think this would boost up the quality.

3. A RENDER STOP button

This could be on the Render Status window somewhere next to the Abort button. If the user push this button it would stop the render what ever state it is in, and composite the already finished layers (render-passes) together and send it out to the render display/image viewer.

I made a cketch about the AA settings panel to show what I mean.

Thank you!


10-09-2003, 06:13 AM
Uhm it's no coincidence that the samples are 1, 5, 9, 17 and 33...
they are 1+ 2^x...

You need to create a somewhat regular pattern around the pixel you are rendering or the AA filter will look weird. Sure, you can create one with 20 samples, but there's not really one logical formular for that like for those settings LW offers.
And 20 samples won't really give you any visble difference to 17 when using the same type of algorithm...
it'd be better to implement different AA algorithms to choose from, like one for very crisp images, one slightly more blurry for animation (the soft filter just looks poor, sorry...and don't have to be that much types as in 3ds max where no one know what they are good for, max 3 already had 12 types :D
) etc or other adjustments like sample area or even variable sample number like splutterfish's Brazil...that's great.

But to the original post:
Orthographic would really be usefull for anyone sometimes i guess, be it for games with (somewhat) isometric perspective, any kind of technical vizualization, or just a true front/seide/top whatever rendering og your model. I don't even expect volumetric effects etc to work.

And some tools really need better feedback, every time i try using taper i wish i hadn't pressed that button :D
Already thought of writing my own one, but darn i can't rewrite the whole modeler i want to work with it not for it
(besides that mine will always be buggy... ;) )
Some instances inside modeler would be cool too, like you make an array and when you tweak the original shape the others follow, it's tedious deleting the array and making a new one etc.

10-09-2003, 09:22 AM
Okay I see...I missunderstood first. So that is simply the limitation of the AA algorithm.

Thanks for that!


10-10-2003, 09:12 AM
SNAPS are greatly needed, but without edge manipulation, would edge snapping be possible? instead of Modeler's snap tool, there should exist snap filter icons down where the modes/copy&paste buttons are in modeler, with a pull-down menu of snap options. the grid snap options should also be located down there as I know I change my grid options constantly throughout a modeling session.

sketchup's component feature is a good example of xrefs. you can group any number of objects together as a component, and edit the component to reflect all of its instances. components in sketchup could also be separate files imported as groups. there's also an edit 'single component' (vs. 'all components') to make unique changes to 1 instance.

10-13-2003, 12:19 PM
it depends from the kinda snaps you want.
for interactive tools i intended snaps to grid with their axes, for precise modeling. just draw your axe near to central origi axe and it snaps right on that, so no worry about precision.

other kinda snaps can be great at center of faces, like in Maya and at points as well.

edge manipulation & creasing implementation is a good thing, though :)