I guess applying your original methode like you wanted plus the single colored texture will work, though, you would be cheating agains the rendering engine in one way
Btw, I said the unorthadox and the contradictorical stuff with a smile such as it would be like a joke

I'll run some more tests with the stencilzfail and etc. considering your request but I'll doubt if it's even possible. Within the SDK itself, you might be able to change it temporarily.

Cheers

Frazzle

Keep me posted on your results.

I know you were joking, and if cheating will get me the results when it comes to this 2nd issue than so be it.

The reason I wanted to use my method was because All the geometry is already there. It's as if I paused the rendering, and held the result without rerendering. On the release version, it won't use the shaders and such, just the generated images. The shader should be possible with FFP, so I may not even use any shaders in the release, but we'll see where I end up...

I'm writing a storyline now, I'm going to have my friend Mike help me make it more creative/interesting. I'll begin work once I get home...

Now that we can use concave geometry in WED, I had thought about doing it this way some time ago. I'd just build a super complicated/detailed level complete with shaders or whatever else I need and take the appropriate screenshots. I'd take care of the collision hull by reducing the geometry and leaving in the more detailed sections for the shadows. If I want to get a render of specific sections for my layers then I just turn off the parts I don't want showing and do this for each individual layer.

After that it would be a process of using panels. Everything should line up nicely. However, using panels would cut out the ability to move a realtime camera into or out of a level. I suppose you could fake it by scaling panels in realtime to make it look like you're getting closer or farther away from certain sections, but that would probably look terrible. That's why I'd want the layers in the actual level itself and not use panels for background/foreground art. This is where the pain comes in because you have to manually line everything up and hope it's to scale. The background plate would be easy but the foreground elements or things you can walk in front of or behind are the tricky ones.

"I'm writing a storyline now, I'm going to have my friend Mike help me make it more creative/interesting. I'll begin work once I get home... " - Good Luck With This Project !

I've run all of the test I could think of but none gave a perfect result.
I guess you'll have to keep the player visible, apply stencil shadows and cheat agains the rendering engine
Maybe Guitar can get better results using his idea.

Cheers

Frazzle

Yep, that's what I've decided to do. I don't even know if it's possible to render a shadows at all if it's not completely visible. It might be a directx thing. Thanks for trying and for the effort. I've discussed this new use of realtime models with Dan, but I haven't heard from him, so I hope he's still on board.

To reiterate: I'm building a high poly representation of what I want in Poser 7 and saving out a large number of animation frames and attaching this to the lower poly model. This model will serve as my shadow caster. I'll have to place it a bit in front of the model to avoid clipping, and I just hope this doesn't look strange in regards to the shadow lining up wrong. I'll combine this with the results of the subject of this thread (I guess we got that working?).

Of course, if the model can cast an alternate shadow from a LOD model, why can't it cast a shadow from an invisible model?

Can someone who knows about these things post if it's possible for an invisible model to cast a stencil shadow? Is there some flag that can be set in a material?

While testing, I discovered that z-buffer shadows can be rendered when a model is invisible. Stencil buffer shadows are rendered as well when a model is invisble with render state settings. However, the result isn't very pleasant and the reason is quite obvious.
I've ran some tests with alphablending, the model is invisible due to alphablending and the shadows are casted but while the shadows are situated at a range where the model is visible/invisible rendered , this leads up to some kind of distortion. Alphablending is a pain in the as* when it comes to stencil shadows because when the model itself is invisible due to alphablending, part of the shadow is not visible. Alhpablending doesn't allow stencil shadows to be completly visible when they are situated behing a model's mesh. Strange, because z-buffer shadows work just fine. Here are two pictures who prove it:




The pictures show a model with a standard alphablending render state and a stencil shadow.
The sun rotates to make the shadow move and mark the difference.
Picture one: The shadow is being covered up for +-20 percent by the model's mesh.
Picture two: The shadow is situated underneed the player but as I've mentioned, the shadow is being coverd up for +- 95 percent due to alphablending.
I guess you get the problem But the sad thing is that I can't get this solved, not even with the stencil render states. I think you're probably right that it's an DirectX (D3D) issue is.
Good luck on finding a potential solution and completing your game OB

Cheers

Frazzle

Well, if we can make a model invisible for rendering, but keep it in the z-buffer, can we not do the opposite? Let it render, but keep it out of the z-buffer so that shadows are not cast onto it?

...of course this would probably not render correctly, and possibly on top of everything else

@xXx:

Take a look at the game Secret Files: Tunguska. I'm not recommending it because of how it may or may not relate to the technology discussed in this thread, but instead as a heads up to a modern adventure game that gets the UI right. The mouse management and space bar hint system is 2nd to none. Just about every review of this game, whether it was positive or negative, was highly positive about these two particularly aspects.

Myself, I loved the hint system. All it did was when you pressed and held down the space bar, all of the available interactive spots and exits would be highlighted. In fact, if you check out the official site, it is a Flash site and uses one of the backgrounds, and if you press and hold down the space bar, it highlights everything (that was an awkward sentence ). The mouse management is probably the best implemented "smart cursor" I've ever seen in the the genre.

Anyway, there's a demo if you want to try it out first hand.

http://www.secretfiles-game.com/