Slion.net>Dev>ChatAboutReflection

(12:54:38 PM) StephaneLenclud: hibread bigger bitmap for you http://img375.imageshack.us/img375/9313/transxo6.jpg
(12:55:34 PM) huperniketes [n=huperni@67.232.249.72] entered the room.
(12:55:34 PM) hibread: StephaneLenclud: so which transparent bits are you talking about?
(12:57:34 PM) StephaneLenclud: Well I'm having this transparent model underneath that's supposed to be the reflection of the one above
(12:57:53 PM) StephaneLenclud: but u can see the back face are blended too
(12:58:30 PM) StephaneLenclud: maybe there is a trick in glBlendFunc
(1:00:21 PM) hibread: hmmm, im not sure exactly what you're talking about. I can see the reflected object... but it looks right to me
(1:01:10 PM) StephaneLenclud: let my show you
(1:02:33 PM) StephaneLenclud: http://img440.imageshack.us/img440/1014/transxh4.jpg
(1:02:51 PM) StephaneLenclud: You can see for instance the little wing in transparency
(1:03:57 PM) StephaneLenclud: maybe it's not that obvious in static picture but when the whole thing is rotating it's not looking good
(1:04:57 PM) hibread: Since you're performing blending, you're probably rendering triangles further away first which is affecting the resultant image
(1:05:39 PM) hibread: what you could do first is render the objects to update the depth buffer
(1:06:15 PM) hibread: then do a secondary pass to actually do your blending which would only pass the depth buffer if its equal (or lequal)
(1:06:17 PM) Plagman left the room (quit: Remote closed the connection).
(1:07:03 PM) StephaneLenclud: I see
(1:07:24 PM) hibread: why are you performing blending with the reflected render?
(1:07:35 PM) hibread: to "darken" so to speak?
(1:07:53 PM) StephaneLenclud: Well I guess I could just change it's color and not use blending
(1:08:18 PM) StephaneLenclud: that would probably be the easiest way right
(1:08:19 PM) hibread: the reflection is due to what? glass/water or something?
(1:08:31 PM) StephaneLenclud: yeah something ;)
(1:09:18 PM) sohail left the room (quit: Read error: 113 (No route to host)).
(1:09:32 PM) hibread: so your sole reason for changing the "appearance" of the reflected object is to make it look like there is an imperfect reflecting medium?
(1:10:41 PM) huperniketes left the room (quit: "Leaving").
(1:10:52 PM) hibread: if thats the case, you could draw both objects (normal and the reflected version) as per normal (no blending with the destination pixels), then render the reflecting medium which has the colour distortion properties
(1:10:57 PM) hibread: ie, using blending
(1:11:44 PM) StephaneLenclud: I see
(1:11:51 PM) mollitz [n=mollitz@e179108195.adsl.alicedsl.de] entered the room.
(1:12:10 PM) StephaneLenclud: So like render a blending transparent plan between the 2 models
(1:12:26 PM) StephaneLenclud: That sounds fun
(1:12:57 PM) hibread: yep that'd be how i'd tackle the problem.. you could also use that plan to etch out the stencil buffer before rendering the reflected object
(1:13:14 PM) hibread: so that it doesn't get reflected where there is no reflection medium...
(1:13:18 PM) hibread: if that makes any sense
(1:13:29 PM) hibread: *plane not plan
(1:13:48 PM) StephaneLenclud: Well it does not make sense cause I'm lacking vocabulary there ;)
(1:14:13 PM) hibread: say you've got a quad acting as the "mirror"
(1:14:28 PM) StephaneLenclud: yup
(1:15:21 PM) huperniketes [n=huperni@67.232.249.72] entered the room.
(1:15:41 PM) hibread: the parts of the object you see reflected from the mirror only lie on that plane....
(1:15:46 PM) hibread: sorry, hard to explain
(1:16:47 PM) hibread: you obviously dont see any parts of the object outside the "mirror".... so you can use that quad to etch out the stencil buffer so that subsequent renders only draw to that portion (the mirror in this case) of the window
(1:21:55 PM) gotan666 left the room (quit: Read error: 60 (Operation timed out)).
(1:26:44 PM) ins\away is now known as ins\
(1:26:56 PM) StephaneLenclud: hibread Thanks for those explanations