@# Quotes DB     useful, funny, interesting





Google
 
Web www.quotesdb.info
Undernet  |  EFnet  |  Quakenet  |  Freenode  |  Dalnet  |  Ircnet  |  Galaxynet
Page: 1 2



Comments:

<0> Does anyone have any tips or links to help me understand when to use multiple techniques or p***es in an effect file?
<1> that's entirely up to you
<1> but techniques are collections of p***es
<1> and you can validate a certain technique on your device
<1> making sure it's compatible
<1> (and each p*** is a state configuration, with a certain vs, ps, certain renderstates, samplers etc)
<0> Yes I've read that p***es help you manage state, which I presume is a performance optimization
<1> hm
<1> that sounds confused
<1> they're just certain setups of the pipeline
<1> look at some of the fx files that come with the sdk
<0> I will, but...
<0> Can you give me an example of what I might do in multiple p***es?
<0> Can a subsequent p*** receive data or pixel color information from the previous p***?
<1> they're independent unless your c++ code does something special
<1> but e.g.



<1> maybe you have a p*** for normal rendering
<1> and one for rendering into the shadowmap
<1> and the .fx file might be a "material"
<1> the techniques could be fallbacks for different hw support
<1> like one technique for shader model 2.0, one for 3.0 etc
<0> I don't really care about multiple techniques right now
<1> ok :)
<0> I'm more interested in what I can do with multiple p***es.
<1> well for instance
<1> i draw my terrain in multiple p***es
<1> i might have:
<1> p*** FirstP*** (etc)
<1> p*** NextP*** (etc)
<1> and FirstP*** has alpha blending turned off
<1> NextP*** has it on
<1> with the same shaders
<1> so i dont have to set up those renderstates manually from the c++ code
<1> and then i draw e.g. four p***es... using FirstP***, NextP***, NextP***, NextP***
<1> (i don't want the first one to blend with what's in the frame buffer)
<0> Does every p*** have to have a vs and ps explicitly ***igned to it?
<1> i'll say yes because i can't remember
<1> :)
<1> you can set them to null though
<1> but that turns off shading just like it would in the D3D API
<1> no you dont need it actually
<1> hell i have a world.fx file with the "global state" i need
<1> so i do worldEffect->Begin()/BeginP***() ... render everything with material .fx'es ... worldEffect->EndP***()/End()
<1> nested(!)
<1> and the world.fx p*** is just a bunch of defaults
<0> hmmm I see
<0> I guess I need to study the different state options that can be set
<0> I'm just learning
<1> ok :)
<0> I want to do some stuff with XNA but this HLSL stuff is such a sidetrack
<0> I hate HLSL :(
<0> Hopefully a simplified architecture will be forthcoming soon
<1> huh
<0> Next I have to figure out how to build shaders from fragments
<1> don't use fragments
<1> that's going away
<1> afaik there's no fragment linking in d3d10 either
<1> http://rafb.net/p/0nVyEa36.html
<1> some examples of p***es here... my default material
<0> cool I'll check them out
<0> I'm also confused about the lighting and texture stage stuff
<0> I've been doing some reading
<1> texture stage states are deprecated when you use shaders
<1> and lighting states too
<0> that's all fixed function pipeline stuff?
<1> yup
<0> So the best practice is to manually write lighting and specular equations into your shaders?
<1> yes definitely
<0> I don't understand why fragments would be going away. I thought fragments are the mechanism that enable you to build custom shaders at runtime... such that I could create a user interface for example which allows a user to select which shader components are desired (such as specular for example) and then build a shader that only uses the required code snippets
<1> yeah i'm doing that with literal parameters instead
<1> see those Enable_Foo's?
<1> i mark them literal with ID3DXEffectCompiler
<1> which means they "compile out"
<1> and i can do
<1> if( Enable_Specular ) { color += CalculateExpensiveSpecular(blah); }
<1> for instance
<1> and that just compiles out if it's false



<1> uniform bool Enable_Specular;
<1> can do the same thing with lightsource count etc, and just use a for loop
<0> Hmmmm
<0> Is that going to be the new recommended way of doing this instead of fragments?
<0> It's odd because the MDX SDK has examples for fragments but not the other method as far as I know.
<1> MDX is a big hack
<1> and yeah
<0> lol
<1> these if's also enable you to make use of hardware branching in the shaders, if you want to
<1> very simply
<0> Well the SDK seems to advocate fragments for native code as well
<0> Would you say that p***es also help you avoid exceeding instruction count limits?
<1> hm they can
<1> you could break up stuff into multiple p***es
<0> ok
<0> I've also had no luck in trying to find information about all the possible data that can be included in X-files or FBX files.
<0> I know X-files can contain info for textures, materials, and animation
<1> documentation is pretty nonexistent
<1> .x files are very old but can be extended with "templates"
<0> I've tried exporting x-files with a couple of different tools, but then when I load my mesh into the mesh viewer, the textures don't show up
<1> thats probably just a path problem
<1> have you seen www.ziggyware.com for xna stuff btw?
<1> ziggy in #gamedev runs it
<0> no I'll check it out
<0> Do I have to manually edit the x-file to fix the path problem?
<0> x-files can be text or binary format right?
<1> mmm well that's one way
<1> yes
<0> I feel it's also possible that my textures are not showing up possibly because I'm not setting them up correctly prior to export
<0> Have you devised a workflow for yourself that generates good x-files with properly pathed textures?
<1> well yeah
<1> we use this btw, http://www.andytather.co.uk/Panda/directxmax.aspx
<0> yay another exporter to play with
<0> is it free?
<1> sure
<1> there's also the kilowatt thing
<1> haven't tried it though
<0> Are you a professional game developer?
<1> not yet :P
<0> :)
<0> So you use Max to make all your models?
<1> yeah
<1> but i'm not an artist ;)
<0> looks like that exporter is only for Max which I don't have :(
<1> yup
<0> question
<0> if your textures are referenced properly in the x-file, are you supposed to be able to render them out properly with alpha blending, etc, using a simple for loop?
<0> The only way I've been able to get my textures rendered is by hardcoding the texture file names into the app and p***ing the textures to the shader and having the shader explicitly alpha blend them
<0> well I guess you are busy... thanks for the info aav
<1> Dazed, ah sorry
<1> dunno what you mean
<1> if there are no specific paths in the x file, they ought to work if they are in the same dir as the x file
<0> ok
<0> I'll figure it out eventually I suppose
<2> Hello
<2> I am thinking about buying a new system and this is a silly question but - I cannot run dx10 code on something like an ATI 1950XT right? I need at least an NVIDIA 8800 ?
<3> okay... i'm little stuck...
<3> i have vertexdata, vertex buffer and all... but i need to fill the buffer with my vertexdata
<3> and that's where my knowledge ends
<3> and i can't find any advice from tutorials that could help me... :(
<4> lock the buffer, and fill the references array with your vertex data
<4> -s+d
<4> at least thats how im doing it with dx7 :(
<3> is the locking a must? one my friend said that it's very slow :P
<4> well, yes
<4> you CAN just throw an array of vertices at the card, but if you want to use buffers you have to lock it before you use it, afaik
<3> /lock buffer (NEW)
<3> pQuadVB->Lock(0,sizeof(pData),(void**)&pData,0);
<3> /copy data to buffer (NEW)
<3> memcpy(pData,aQuad,sizeof(aQuad));
<3> /unlock buffer (NEW)
<3> pQuadVB->Unlock();
<3> ups sorry :/
<3> i hate screen...


Name:

Comments:

Please enter the result of the sum 63 + 46 (to avoid spam):






Return to #directx
or
Go to some related logs:

unable to boot slackware on vmware
ben athlek
#politics
ubuntu xarcade
#fedora
#stocks
#nhl
SHIT JUICE
#politics
#stocks



Home  |  disclaimer  |  contact  |  submit quotes