Thanks. Of course, I'll be keeping a local copy of the final compressed video. What I don't want to keep is the edited video project and the original source footage.

Because two 4GB USB keys are so expensive?

Snap to grid can be helpful to align objects more precisely, but I doubt it'd be faster IMO.

I don't think it would be that much slower for someone that knows what they are doing. It's approximately 3 clicks to get it exactly where you want it anyway.

Not that your work is trivial, I would just highly caution against saying it's "better".

Yeah, by now I'm starting to see it's better for me to shut up and let the demonstration speak for itself. I don't seem to make things any better when I explain the motivation behind the Slide technique and the weak points of widget ways.

I doubt epic didn't have a reason for doing it this way

I'm sure it's very much related to the fact the original UnrealEd was made more than 10 years ago, and back then there were limited advances in 3D UIs. And Epic probably never allocated enough resources to redoing something that's already done, because I'm sure there are always more important things (in their view) for them to do.
[/quote]
I'm sure Epic hasn't thought of UI/Productivity improvements in the last decade.

Out of curiosity, how does your demo handle complex/convex geometry?

I don't think it would be that much slower for someone that knows what they are doing. It's approximately 3 clicks to get it exactly where you want it anyway.

It's really hard to discuss things like this over the internet via text. So much easier, and more convincing to show a live demo.

Out of curiosity, how does your demo handle complex/convex geometry?

Equally well. It's very general, and works with any set of triangles defined to be objects. This is a common concern, and it's a direct consequence of me showing simple geometry in most of my demos. The reason for that is because my modelling efforts were the bottleneck, not the Sliding algorithm hehe.

If anyone's interested, you can glance over my thesis pdf that describes Slide in excruciatingly full detail.

Also, in appendix E you can find a very early pilot study that compared an early version of slide against an XYZ-axes method, with the associated result graphs (page 175). It's not unusual to see 2-3x difference in task completion time. However, this result was known from many previous studies going back as far as 2005~, and taken as a given. Hence it wasn't my main goal to replicate those results, but rather to work on other aspects of the system.

[quote name='way2lazy2care' timestamp='1329399762' post='4913647']
Out of curiosity, how does your demo handle complex/convex geometry?

Equally well. It's very general, and works with any set of triangles defined to be objects. This is a common concern, and it's a direct consequence of me showing simple geometry in most of my demos. The reason for that is because my modelling efforts were the bottleneck, not the Sliding algorithm hehe.
[/quote]

What did you export your geometry from to begin with? If you can import geometry from Maya or Blender, you can use the stock Teapot/MonkeyHead models that come packaged with each.

That would be much more impressive than blocks imo. Almost to the point of doing yourself a disservice by not getting some complex meshes.

I have a pretty basic loader that opens .dae files (Collada format) that I export from SketchUp. It doesn't do advanced things like materials, textures, just plain geometry. If Blender exports as .dae, then what you suggested is possible. But it's almost much easier just to get some random hi-poly model from SketchUp's 3D Warehouse.

I already have a few more complicated scenes that I used to test various things. I just used the scene with boxes because I wanted a task where you just have to move around a lot of small objects, rather than one highly detailed one. Of course, I could've used highly detailed boxes, but that didn't happen.

I have a pretty basic loader that opens .dae files (Collada format) that I export from SketchUp. It doesn't do advanced things like materials, textures, just plain geometry. If Blender exports as .dae, then what you suggested is possible. But it's almost much easier just to get some random hi-poly model from SketchUp's 3D Warehouse.

http://code.blender.org/index.php/2010/12/collada-importexport-team/
It looks like it might be native o.o