hey,
I was wondering if you could elaborate a little bit about the bmap_rendertarget function. I can'st seem to get the render targets to preserve the content even when I set the 3rd paremeter to 1. please see my thread under A7 engine for more info: http://www.opserver.de/ubb7/ubbthreads.php?ubb=showflat&Number=253136#Post253136

I would REALLY appreciate help - this is causing a complete stop in development of my Iridium plugin for GS. thanks in advance - Mike -

"Preserving the content" means the content that the bitmap previously had before it was used as a rendertarget. This is used for one-time rendering an object into an already existing scene, but is probably not what you attempt to do. What content do you want to preserve?

ok, to be more precise I will explain the actual situation:
so each frame is defined as rendering the camera and all following views (in my code).
Each frame, I first render into the camera's bmap and 2 other render targets. After I render other views. HOWEVER, I want to keep the rendered images from the camera's bmaps (both the .bmap and the rendertargets). So basically, I want the rendertarget bmaps to always store ONLY the camera's color0 through color2 data and not update with the following views (but I must be able to retrieve the data in the following view's shaders).

I want to preserve the render target bmaps from the camera rendering stage.

The rendertargets are only rendered into when you use the COLOR1..COLOR3 semantics in a shader in that particular view. When you don't use them, the render targets are supposed to preserve their content until it's rendered into them again.

when I use a view with VISIBLE set following the camera, the view overwrites the rendertargets even when I set the views to a pixel shader that only outputs to the COLOR0 semantic. Maybe this is a problem?

It seems that each view that renders (VISIBLE) renders into the rendertargets regardless of whether I create the COLOR1,2,3,etc semantics or not. This is not a problem with post processing views though. -Mike-

Hmm, the render targets should only react on the color number that you set in the call. Rendering in COLOR0 must only affect render target number 0... except when you have a hardware that does not support multiple targets. But then the engine should give you an error message. Can you upload or send a project where the problem is visible? We'll check.

hey,
Thank you so much for taking your time to look at this problem. I have fought this for a long time however it IS possible that I'm just doing something stupid. here is the link to a project I made specifically for you to examine:
http://www.megaupload.com/?d=PH9H8KMY

whats in it:
the program is inside, run it and press 0 and move the camera.
what you should see:
there should be 4 quadrants: the upper left is the light1 render target (see code for details)
the lower left is the camera color render target
the upper right is the camera scene normals render target
the lower right is the camera scene depth render target

to experience my problem, toggle between the k and l keys to switch the stages. When you press l, the right two render targets should become blank. This is because the stages are switched so that the light1 view renders AFTER the camera and seems to override/erase the render targets (the fx file does NOT write to any other semantic other than COLOR0).

I forgot to upload the source: mrt_test.c I'll do this in a sec - thank you very much!

here is the .c source code:
http://www.megaupload.com/?d=NZW5ERWU

Ok, I'll download it and look into it. I can't do it right now, but will check the project someday next week.

The problem was that the rendertargets were cleared with the sky color at the begin of a view render cycle, just like the normal screen. I found that the DirectX Clear function affects all color targets, not only COLOR0.

I'll mention this in the manual. Simple solution: Either don't clear them,

vec_set(sky_color,vector(0,0,0));

or disable the render targets with a view event when the view should not write into them.