Realtime HDR Rendering with SM3.0 question

Started by
4 comments, last by Tree Penguin 19 years, 7 months ago
Hi, i read a few articles/papers on it and i wondered if there's a way to retrieve the final 32bit per component image to store or modify it later on. I mean the 32bit per component color values before tone mapping and clamping occurs (eg. range 0.0 to MAX_FLOAT). The answer might be obvious for those who are experienced with PS coding and therefor my question might seem stupid but please keep in mind that i am not yet experienced with PS coding at all because of my crappy old hardware. Any help would be appreciated. EDIT: or even better, is it possible to pass multiple float4 variables back to the app or some 0.0-MAX_FLOAT limited buffer which i could read in the app? Thanks. [Edited by - Tree Penguin on September 14, 2004 12:58:39 PM]
Advertisement
Could you render it to a texture instead of the framebuffer?

It'd have to be a 32bit floating point format texture which i know nothing about [smile] but that seems like a logical way of doing it.

Andy

"Ars longa, vita brevis, occasio praeceps, experimentum periculosum, iudicium difficile"

"Life is short, [the] craft long, opportunity fleeting, experiment treacherous, judgement difficult."

Yeah i thought of that but i don't know if all texture formats are supported as render target (16 and 32 bit floating point per component textures are supported). And if they are they will probably get clamped to a 0.0-1.0 range.
No one? Ok, does anybody know if the 32 bit texture values are clamped when rendering to a texture?
apprently, no they arent clamped (although, i've not tested this myself I admit).
iirc with NV cards you have to be explicate about wanting it to be floating point but with ATI cards it just works once you setup a floating point context and bind a floating point texture to it.
Ok, thanks. I am targeting for NVidia CG (which is afaik not compatible with ATI cards (am i right?)) so i think i will have to test it at an NVidia card. You gave me some hope anyway ;).

The reason i am hoping to use 16 or 32 bit float per component is because i am working at an animation renderer (an old project of mine which i resumed working on) and i at least want to make it possible to do such high quality rendering.

This topic is closed to new replies.

Advertisement