Reflections that large and intense are never going to look good with 
cubemaps. The best solution is setting up real-time reflection, then 
plugging the resulting rendertexture into the cubemap's distortion code 
in so that it wraps across the surface correctly.

That's a lot of work, needless to say. I'd probably choose instead to 
make a high-res custom cubemap with indistinct, generic reflections that 
can be applied anywhere without looking too dodgy. That or give up on 
the idea of reflections at all!

(One final possibility: convert the windows to models and apply 
high-exponent phong.)

Minh wrote:
> http://www3.telus.net/public/a6a91102/1.jpg
>
> I think I'm gonna try splitting the glass up and using more cubemaps...
>
> Matt Hoffman wrote:
>   
>> Interesting idea but it sounds like you could better accomplish this with
>> multiple cubemaps on smaller surfaces than single cubemaps on large  glass
>> surfaces.
>>
>> There's also the concern of large BSP's from these. Each cubemap taking a
>> 1024^2 or 2048^2 screenshot, six times PER CUBEMAP is going to add quite a
>> bit of filesize if your map has anywhere near the proper amount of cubemaps.
>> Not saying that every cubemap is going to be big however...
>>
>> Could you possibly show us an example of what your trying to cubemap? Only
>> thing I can think of is large glass buildings like in mirrors edge. (In
>> which case, I'm interested!)
>> http://cloroshare.files.wordpress.com/2009/06/mirrors_edge_wallpaper_by_fuzzy13.jpg
>>
>> I haven't looked, but I imagine it's buried deep in engine code we don't
>> have.
>>
>>
>> On Fri, Jul 31, 2009 at 11:24 PM, Minh <[email protected]> wrote:
>>
>>   
>>     
>>> Has anyone tried to increase the max cubemap resolution of 256x256?
>>> We're running into some ugly ass large glass windows that look terrible
>>> because of the low cubemap resolution.
>>>
>>> _______________________________________________
>>> To unsubscribe, edit your list preferences, or view the list archives,
>>> please visit:
>>> http://list.valvesoftware.com/mailman/listinfo/hlcoders
>>>
>>>
>>>     
>>>       
>> _______________________________________________
>> To unsubscribe, edit your list preferences, or view the list archives, 
>> please visit:
>> http://list.valvesoftware.com/mailman/listinfo/hlcoders
>>
>>
>>   
>>     
>
> _______________________________________________
> To unsubscribe, edit your list preferences, or view the list archives, please 
> visit:
> http://list.valvesoftware.com/mailman/listinfo/hlcoders
>
>
>   


_______________________________________________
To unsubscribe, edit your list preferences, or view the list archives, please 
visit:
http://list.valvesoftware.com/mailman/listinfo/hlcoders

Reply via email to