Re: FxPlug/Final Cut Pro X Rendering issue (32 vs. 16 bit Float textures).
Re: FxPlug/Final Cut Pro X Rendering issue (32 vs. 16 bit Float textures).
- Subject: Re: FxPlug/Final Cut Pro X Rendering issue (32 vs. 16 bit Float textures).
- From: Darrin Cardani <email@hidden>
- Date: Wed, 13 Jul 2011 09:19:20 -0700
Michael,
This definitely sounds like a bug. Can you please file a bug with as much information as you have?
Thanks,
Darrin
On Jul 13, 2011, at 12:49 AM, Michael Jonsson wrote:
> Hi list,
>
> It appears there are some inconsistencies in respect to the data that
> is fed down to plugins under various circumstances when called from
> Final Cut Pro X. This concerns only GPU rendering.
>
> I see two scenarios when my plugin is called from Final Cut:
>
> 1. Expected behavior.
>
> Happens prior to rendering, say while I change parameters in my plugin
> UI and if I export when NO pre-rendered data is available (say if
> flush optimized media and turn background rendering OFF and export to
> H.264) - In this case renderOutput:withInput:withInfo: gets called
> with the following:
>
> FxTexture input depth = 16
> FxTexture output depth = 16
> FxRenderInfo.depth = kFxDepth_FLOAT16
>
> Binding the input texture and doing a glGetTexLevelParameteriv(…
> GL_TEXTURE_INTERNAL_FORMAT…) reveals that the Internal format of the
> input texture handed down to renderOutput: is GL_RGBA32F_ARB.
>
> In my fragment shader I see the data that I expect; that is,
> super-white values over 1.0 etc.. In short, everything behaves as
> expected.
>
> Motion 5 everything behaves similar to above expected behavior both
> during playback/editing and exporting.
>
>
>
> 2. Unexpected behavior.
>
> This is where things gets strange… This happens (only?) when the
> plugin is called during background rendering. Here
> renderOutput:withInput:withInfo: gets called with the following:
>
> FxTexture input depth = 32
> FxTexture output depth = 32
> FxRenderInfo.depth = kFxDepth_FLOAT32
>
> However, glGetTexLevelParameteriv(… GL_TEXTURE_INTERNAL_FORMAT…)
> reveals that the internal format of the texture is GL_RGBA8_EXT...
> (8bit pr channel...)
>
> In addition, in my fragment shader I see mostly the same data as in
> the 1st case but the over-range > 1.0 (i.e. super-white etc.) data is
> being truncated/clamped to 1.0… which is not the way its supposed to
> work.. I hope! :-)
>
> Any insights or workaround for this problem is highly appreciated. If
> not, I hope its something that can be addressed in an upcoming update.
>
> Cheers,
> Michael
> Rig: MacBookPro8,2 / Radeon HD 6750M / MacOSX 10.6.8.
> _______________________________________________
> Do not post admin requests to the list. They will be ignored.
> Pro-apps-dev mailing list (email@hidden)
> Help/Unsubscribe/Update your Subscription:
>
> This email sent to email@hidden
--
Darrin Cardani
email@hidden
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Pro-apps-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden