Hello Flo, as far as I know, QC works the following way: It first examine which render patches are active, then walking through all the inputs of that rendering patch looking for the values that are connected to them (walking recursively upwards the patch connections). Each of the inputs has certain input types (e.g. an integer value, a structure or RGBA-8 bit-Texture). Now QC is asking the patch thats provides the value for that input for its value. If the value type doesn't match then the value gets converted if a conversion is available: e.g. an Integer number gets converted into a string if the input required to be of type string. Also a texture gets converted (e.g. from a non-power-of-two-sized texture 640x480 to a power-of-two-sized texture 1024x1024 if you specify a "sampler2D" in an GLSL shader.) My guess is that the QC patches are all requiring a RGBA-8bit-texture as their inputs because their internal code is optimized for this and therefor QC converts the textures automatically. I would try to create a GLSL shader with the correct texture definition (which I din't know for a floating point texture) and see, if the GLSL shader indeed gets a floating point texture if you connect your exposed input directly to it.
best, Achim Breidenbach Boinx Software On 25.01.2013, at 15:30, Flo Loferer wrote: > Hey there guys, > > we are trying to implement some sort of pipeline with custom QCPlugins and > CIFilters within QC. > It all worked out well at first but we ran into problems lately. > > The main issue is that we don't know how the CGLContext operates in > conjunction with pixel formats. > We want that all our components are working on floating point buffers to get > the maximum quality possible. > > However, we don't understand how Quartz Composer handles the input and output > image buffers/textures. > It seems that regardless of how we set the output image, it gets scaled down > to RGBA 8 bit per component. > > So, to phrase a question: is there any way to tell QC that it should use RGBA > floating point buffers? > Are the input and output images converted to fit the pixel format QC expects? > > Maybe someone can shed some light on this! > It's really confusing. Especially as Apple is using the word "pixel format" > in CGL and QCPlugIn with totally different meanings. > > Thanks in advance!! > > Flo > _______________________________________________ > Do not post admin requests to the list. They will be ignored. > Quartzcomposer-dev mailing list (Quartzcomposer-dev@lists.apple.com) > Help/Unsubscribe/Update your Subscription: > https://lists.apple.com/mailman/options/quartzcomposer-dev/achim%40boinx.com > > This email sent to ac...@boinx.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Quartzcomposer-dev mailing list (Quartzcomposer-dev@lists.apple.com) Help/Unsubscribe/Update your Subscription: https://lists.apple.com/mailman/options/quartzcomposer-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
