FlexSim Community Forum

FlexSim Community Forum (https://archive.flexsim.com/index.php)
-   Gripes and Goodies (https://archive.flexsim.com/forumdisplay.php?f=115)
-   -   next step visualisation? (https://archive.flexsim.com/showthread.php?t=2170)

Carsten Seehafer 05-21-2012 08:05 AM

next step visualisation?
The flexsim engine has an incredible grafic engine. Today flexsim models look unbelievable realistic. But how would it look like with realistic shadows?

Is there a chance to get shadows in next flexsim releases?
What do you think about shadows?

Anthony Johnson 05-21-2012 11:36 AM

We (the dev team) have had several discussions over the past few months regarding more realistic scene rendering. I think we can implement basic things like stencil-based or shadow-map-based shadow rendering, as well as support for bump maps and parallax maps. This could get shadows to be about as good as this. But that would come at the significant expense of scene rendering speed, especially for big models. Game developers do lots of tricks to clip and simplify certain areas of the scene so that they only have to render shadows and complex geometry on the stuff that's important and close to the camera. If we implement shadows and other stuff, we'd have to also provide tools to modelers that let them do the same clipping tricks that game developers do. To me that seems like a lot of work we'd be pushing on the user. And it would still never get us to the point where we're rendering like this.

So the other option we're throwing around, which I'm leaning toward us implementing first, is to implement geometry exporters that allow the geometry of a scene to be exported to a commonly-used format like renderman, or perhaps to implement an in-app plugin that can use a renderman or a mental-ray to render the scene. It would not be fast at all, but that's not the goal. The goal would be to render a scene for post-simulation play-back and video creation.

I'm interested to hear others' opinions.

Jason Lightfoot 05-21-2012 02:42 PM

1 Attachment(s)
You can pre-render the static scene lighting onto the objects like the one Anthony linked to using MentalRay. That just leaves the lighting/shadows of the parts that move around and how they might project onto the scene, but I think the overall static scene lighting might be more important to create the realism effect.


Carsten Seehafer 05-21-2012 11:03 PM

For our kind of presentation it would be perfect if you can turn the shadows on and off in the model (like startanimation, stopanimation).

Another option could be to combine runspeed and shadow details. A higher runspeed means lower shadow details.

brett forbes 06-26-2012 07:16 PM

Hi Anthony,

I must admit, I am not really a fan of the graphics provided by any current sim vendor (e.g. Simio, Arena etc). In my mind it is difficult to provide credible looking graphics in the way they do, to people who have grown up with video games. No matter how hard you try, the graphics will come out looking cartoonish, without even simple anti-aliasing of straight lines, and thus they lack credibility at boardroo-type presentations.

As you know it is much easier to realistically render a man-made object, such as a machine or a forlift, than a natural object such as a rock or a tree. In graphics, man made objects are conceptually complex, but visually simple, whereas natural objects are the other way around.

My current view is that if I am to present the outstanding Flexsim capabilities at a boardroom level or to a group of Professors, I am better off avoiding animations as they are visually simple looking. Thus board room people, or Professors, would correlate the simple graphics, with simple capability of the underlying software paradigm, which is obviously false, as Flexsim has outstanding capabilities.

Given any small software company has limited resources, it is obviously unrealistic for you to develop outstanding graphics, as the resource load is too high, and these valuable resources are better focused on driving and improving the underlying logic.

In my mind, the next gen simulation company will work out how to integrate their simulation capability with outstanding, game-quality graphics. Many sophisticated games nowdays have quite simple world-building consoles, enabling users to develop their own maps, paths, scripts and animations.

The salient question seems to be why don't simulation companies build similar capabilities? Why do simulation companies believe they should develop their own graphics, instead of licensing that caability in from manufacturers such as Unreal (http://www.unrealengine.com/features), and focus on developing a simple world-building environments on someone elses graphics engine?

Surely from a business strategy perspective it makes sense to focus limited resources on the technical aspects of your core code, and not to try to replicate what other company's can build more successfully. Thus it would seem to be an obvious move to provide basic 3D capabilities as you have done, and then integrate with a game engine to provide next generation capacity for those who want it. It seems that you could deliver industry-leading capability relatively easily if you focused on developing a world-building environment on top of an existing game engine.

This seems particularly relevent to me as I am focused on customising your excellent base capability to suit mining scenarios, which obviously have a lot of natural features (i.e. rocks and hills). I do not believe any current simulation tool has a viable perspective on visualisation as they are all trying to use limited resources to deliver graphics capabilities, and thus have simplistic graphics, rather than partnering with existing capabilities in this area.

I note that if you were to achieve this partnering approach you would then open up a massive new market, the use of simulations for training purposes. Surely the value of your tool is in two areas, logic-heavy sims with some graphics (i.e. current market), and graphics-heavy sims with some logic (e.g. training market).

I would be interested in your opinion to this post.

Thanks for your time


Luciano 09-23-2012 04:06 PM

Hi Jason,

how do you produced the 3d shape (.3ds files) you used in your lighttexture.fsm example?


Jason Lightfoot 09-23-2012 05:21 PM

The basic idea is to unwrap the UVWs and then render to texture.

All times are GMT -6. The time now is 04:08 PM.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2019, vBulletin Solutions Inc.
Copyright 1993-2018 FlexSim Software Products, Inc.