by

Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
    Pavel Rojtberg
    @paroj
    besides auto_params are in some way derived by ogre - otherwise they are ordinary params
    SNiLD
    @SNiLD
    yeah i was kind of hoping there was a way to have "provider" and "consumer" kind of system for providing parameters to shaders
    but with ACT_CUSTOM i still need to know who is going to use the values so it's no different from me keeping a list of materials that need the parameter i'm providing
    Pavel Rojtberg
    @paroj
    what exactly do you want to achieve? SH is just a different cubemap encoding and therefore can be stored & used as a texture.
    SNiLD
    @SNiLD
    i want to pass them as vec3 array of size 9, there is no need to waste a texture
    Pavel Rojtberg
    @paroj
    are you looking for the k-nearest lightprobes? Ogre could handle that similarly how it currently gives you the k-nearest lights - except that it has no concept of light probes at the moment
    then you can use shared_params
    SNiLD
    @SNiLD
    oh that sounds cool! how do i set shared_params globally on cpp side?
    i'm generating the IBL from the scene and instead of passing diffuse map and specular map with sh i only need to pass specular map and 9 vec3 of sh parameters
    SNiLD
    @SNiLD
    thanks! that's exactly what i need
    except was there an issue setting shared_param_ref value as struct parameter?
    Pavel Rojtberg
    @paroj
    if your shaders do not contain an UBO, ogre will just copy the shared_params contents to the normal params and they will behave as such
    SNiLD
    @SNiLD
    ok nice then it's exactly what i needed
    Pavel Rojtberg
    @paroj
    besides there was an important bugfix, which should also fix UBO usage: OGRECave/ogre#1476
    SNiLD
    @SNiLD
    this was really good read regarding tangent space btw. https://medium.com/@bgolus/generating-perfect-normal-maps-for-unity-f929e673fc57
    steelfax
    @steelfax_gitlab
    Does anyone know a good Q1 or Q3 map loader for Ogre?
    Pavel Rojtberg
    @paroj
    SNiLD
    @SNiLD
    why don't some material trigger Ogre::MaterialManager::Listener::handleSchemeNotFound even though they don't have technique with that scheme?
    for example hydrax and skyx do this
    (i'm trying to implement SSAO according to the SSAO samples and generate the GBuffer via RTSS like done in the samples)
    Pavel Rojtberg
    @paroj
    the current scheme may be changed by the viewport or by the compositor, so probably one of these is done
    SNiLD
    @SNiLD
    well the same function is called for my own materials for example, don't know what's special about hydrax or skyx
    looking at the code it would mean that the shader is not compiled (ie. not in the supported techniques list)
    Pavel Rojtberg
    @paroj
    the RTSS ignores materials that have shaders attached by default
    so probably handleSchemeNotFound is called, but ignored by the RTSS
    SNiLD
    @SNiLD
    i put debug point in that function and it's not called for hydrax and skyx
    but it's called for my own materials, which use custom shaders as well
    anyway i added the techniques for GBuffer scheme manually to hydrax and skyx
    currently after long battle i finally get the skyx rendered, but hydrax still refuses to render
    the SSAO sample doesn't show how the compositors have to be put into a chain so that the final screen is rendered
    SNiLD
    @SNiLD
    but the deferred shading sample shows that to some degree
    currently the chain must be in this order: SSAO/GBuffer -> SSAO/CreaseShading -> CustomCompositor that references SSAO/GBuffer mrt and has new pass clear {} and pass render_scene {}
    the ultimate goal is to get SSAO in texture that can be passed to filament and used there
    SNiLD
    @SNiLD
    so i guess it's the output from SSAO/CreaseShading that would be used there
    SNiLD
    @SNiLD
    yeah, that alone produces only black screen (except for overlays, they can be enabled)
    SNiLD
    @SNiLD
    the way scene is rendered in the SSAO example is not translatable to the way SSAO is used in filament because in the sample SSAO is "post process effect" but in filament the SSAO must be passed as texture and the scene must be rendered normally
    so as i understand what should be done is first render the scene once with GBuffer scheme, then run the SSAO compositor to get SSAO, then clear the depth buffer (possible others as well?) and render the scene normally with the default scheme
    however it's still unclear to me how to do this in ogre
    SNiLD
    @SNiLD
    currently enabling the compositor in correct order and having custom compositor as last one that renders the scene again almost works (skyx is rendered, but hydrax is not)
    oh! but it was because i was clearing to depth_value 0.0
    i though i needed to do that when using reverse z-buffer
    sbuckner50
    @sbuckner50
    Hello y'all, I am working on a simulation development project using Ogre 1.9.0 for graphics. Many capabilities currently work, however I am running into issues trying to set up textures on geometries. Specifically, using Ogre::Image::load("image.png" , "valid_path...") is yielding an error for me: "SystemError: OGRE EXCEPTION(7:InternalErrorException): Error decoding image in FreeImageCodec::decode at .../OgreFreeImageCodec.cpp". Any help figuring out this problem and what could be causing it would be greatly appreciated!
    Also, I tried sending an email from the website however it seems "webmaster@REMOVESPAMogre3d.org" is not a valid address.
    If it helps, I am trying to use Ogre 1.9.0 on RHEL 6, which is not listed as a supported OS for Ogre (although Fedora is). I am not sure if this could be causing issues or not, because the issue I stated above is only one of many I am having.
    SNiLD
    @SNiLD
    you need freeimage codec and with that old version of ogre you probably need sufficiently old version of freeimage as well
    sbuckner50
    @sbuckner50
    Hmm, is FreeImage a separate FOSS?
    SNiLD
    @SNiLD
    it's a separate library