These are chat archives for highfidelity/hifi

20th
Feb 2015
Boris Skegin
@sirop
Feb 20 2015 14:03
I could not find what the minimum hardware requirements are to develop on win 7?
Thijs Wenker
@thoys
Feb 20 2015 14:35
depends on which part you're developing
what are your specs?
@sirop
Jason Hawks
@nbq
Feb 20 2015 14:44
@sirop and @thoys I learned from trying to "run" in a virtual machine in windows, you need at least something capable of running opengl 3.0 - that is the simplest answer I have so far
you can develop on anything that can run visual studio community
Thijs Wenker
@thoys
Feb 20 2015 14:45
depends also if you want to develop the interface or the assignment-clients/domain-server
for interface you obviously need some graphics power
Jason Hawks
@nbq
Feb 20 2015 14:45
yes, my info covers the interface
Boris Skegin
@sirop
Feb 20 2015 15:49
ok, thanks. I'll try now to build bullet -2.8. If it works on my machine, then the rest might also work.
Jason Hawks
@nbq
Feb 20 2015 15:50
@sirop great! They are working on a way to get all of that to auto compile so you do not need it separate but its not merged into the main branch yet. Good thing is info on how to compile it is on the build guide in the repo.
Boris Skegin
@sirop
Feb 20 2015 15:58
I got these warnings when running cmake for bullet:
OPENGL FOUND
glu32opengl32
-- WARNING: you are using the obsolete 'GLU' package, please use 'OpenGL' instead
-- Could NOT find GLUT (missing: GLUT_glut_LIBRARY GLUT_INCLUDE_DIR)
Win32 using Glut/glut32.lib
How earnest is that?
Jason Hawks
@nbq
Feb 20 2015 15:59
ignore those, everyone gets them
just warnings
Boris Skegin
@sirop
Feb 20 2015 16:00
thanks.
Jason Hawks
@nbq
Feb 20 2015 16:00
welcome
If they give you any troubles you can always use these precompiled libs: https://s3.amazonaws.com/hifi-public/requirements/win/bullet-2.82-vs2013.zip
Boris Skegin
@sirop
Feb 20 2015 16:09
the bullet build manual says:
In file: src\MiniCL\cl_MiniCL_Defs.h --- comment lines 364 to 372
Jason Hawks
@nbq
Feb 20 2015 16:10
yes, to get it to compile for vs2013, but I just gave you a precompiled solution to save you the trouble
Boris Skegin
@sirop
Feb 20 2015 16:10
should it not be from 363 to 371 ? As we now let the function signature remain, but comment out its body.
Jason Hawks
@nbq
Feb 20 2015 16:12
ahh the lines being a little off, yes, actually it is off a little bit but I personally am rewriting the windows compile manual so I figured it was just easier to give you the precompiled for now
Boris Skegin
@sirop
Feb 20 2015 16:20
I'm kind of ambitious to build bullet, sorry.
C:\VS_Projects\bullet-2.82-r2704\cmakebuild\Demos\DX11ClothDemo\AppDX11ClothDemo.vcxproj.metaproj" (Standardziel) ( 62) ->
"C:\VS_Projects\bullet-2.82-r2704\cmakebuild\src\BulletMultiThreaded\GpuSoftBodySolvers\DX11\BulletSoftBodySolvers_D X11.vcxproj.metaproj" (Standardziel) (63) ->
"C:\VS_Projects\bullet-2.82-r2704\cmakebuild\src\BulletMultiThreaded\GpuSoftBodySolvers\DX11\BulletSoftBodySolvers_D X11.vcxproj" (Standardziel) (64) ->
(FxCompile Ziel) ->
C:\VS_Projects\bullet-2.82-r2704\src\BulletMultiThreaded\GpuSoftBodySolvers\DX11\HLSL\OutputToVertexArray.hlsl(1,1 -10): error X3000: unrecognized identifier 'MSTRINGIFY' [C:\VS_Projects\bullet-2.82-r2704\cmakebuild\src\BulletMulti Threaded\GpuSoftBodySolvers\DX11\BulletSoftBodySolvers_DX11.vcxproj]
does this mean that I need DX11 intallation?
Andrew Meadows
@AndrewMeadows
Feb 20 2015 16:58
No, we don't use any SoftBody stuff yet.
So you can disable it or figure out some way to bypass that error. Also you can build without GLUT since that is only required for the Demos.
Boris Skegin
@sirop
Feb 20 2015 17:00
Ok, I''ll try.
Thijs Wenker
@thoys
Feb 20 2015 17:11
are there any thoughts on using a library like https://github.com/assimp/assimp for a wider range of supported 3d-model-formats ?
Andrew Meadows
@AndrewMeadows
Feb 20 2015 17:28
As far as I know we haven't heard about assimp yet. That could very well be a good way to go forward, but our situation reminds me of this xkcd cartoon: http://xkcd.com/927/
but we're in that situation even without assimp: the FBX format is quite featureful and we probably won't support all of them... it makes sense to be able to import FBX with success if the model doesn't use any unsupported features and then save to some simplified format.
I wonder what the common shared in-memory format of assimp looks like. If it supports all of those formats, each of which supports some subset of the infinite feature possibility space, then the union of those features must be larger than any supported format, which makes assimp's format even more complicated... unless they lose some information when they boil the results down into whatever they store in memory.
David Rowe
@ctrlaltdavid
Feb 20 2015 17:45
Hi ... I'm trying to set up building Interface with Faceshift on Windows ...
In my usual lib directory tree I have a "faceshift" directory.
And in that an "include" directory which contains fsbinarystream.h.
And also a "lib" directory that contains faceshift.lib and faceshiftd.lib.
Is this correct and complete?
Andrew Meadows
@AndrewMeadows
Feb 20 2015 17:51
I dunno ctrlaltdavid. I suspect that if you could decipher the .../cmake/modules/FindFaceshift.cmake script you could figure it out. Unfortunately I don't know cmake well.
David Rowe
@ctrlaltdavid
Feb 20 2015 17:54
Yes, I've had a look at FindFaceshift.cmake and on the face of it, this is what it looks like it expects ... but it's still saying "Could not find FACESHIFT (missing: FACESHIFT_LIBRARIES)"
The readme.txt for Faceshift external doesn't cover Windows.
Andrew Meadows
@AndrewMeadows
Feb 20 2015 17:55
For those who are on Windows and suffer from the horrible scratchy audio bug there is a workaround: open Edit--->Preferences and disable the "Output Starve Detection" option. And then make sure the "Output Buffer Size" is set to about 4.
The problem appears to be that some output audio devices on Windows will complain if their output buffer is set too large -- this is probably a Qt bug. My Windows laptop gets scratchy at 6, or maybe 5... it reproduces easily at 6 but I was only able to reproduced it once at 5, and not yet at 4.
Thijs Wenker
@thoys
Feb 20 2015 18:08
I heard a big scratch right after I started up a few times
thats the one?
Lexx
@Triplelexx
Feb 20 2015 18:10
nice find @thoys the post processing features off assimp are also interesting
(of
*of
like merging of redundant materials
Do not know if it was too much or too little. Got no errors now.
Lexx
@Triplelexx
Feb 20 2015 18:16
license is sweet as well, will def check that out more later, thanks @thoys
Thijs Wenker
@thoys
Feb 20 2015 18:17
cool @Triplelexx , and nice comic @AndrewMeadows
Lexx
@Triplelexx
Feb 20 2015 18:18
@AndrewMeadows I'll see if I can test that workaround now, I've had the problem consistently
typical, no issue now on loopback ;)
Lexx
@Triplelexx
Feb 20 2015 18:23
I'll keep it set like that and see how it goes
So can I run interface on my PI 2? Damn been getting distracted, might have to try as I'm spending a lot of time playing with it. ;)
there's no support for Arm Linux is there?
the gl es 2 support does seem alright
Andrew Meadows
@AndrewMeadows
Feb 20 2015 19:16
We're working on a build for the Gear-VR which runs on the Samsung Note 4 which has an ARM CPU, so supporting general linux+ARM is probably doable.
Thijs Wenker
@thoys
Feb 20 2015 19:21
@AndrewMeadows did you guys also attempt to run stack-manager on android / iphone?
Andrew Meadows
@AndrewMeadows
Feb 20 2015 19:22
No, not that I know of.
thoys @thoys looks for his old smartphone and starts tinkering
David Rowe
@ctrlaltdavid
Feb 20 2015 19:23
Have sorted building Windows with Faceshift, thanks to @murillodigital ... will update Faceshift readme.txt.
Joe Large [Omega Heron]
@OmegaHeron
Feb 20 2015 19:42
Stack is running on my Raspi B now. It's not trivial making it happen though. Not the new quad core one - the classic B model.
Thijs Wenker
@thoys
Feb 20 2015 19:43
nice!
StackBerryPi
Thijs Wenker
@thoys
Feb 20 2015 19:49
@murillodigital the new functional test video's are awesome :)
Leonardo Murillo
@murillodigital
Feb 20 2015 19:52
@thoys thanks, still early prototype work in progress
Thijs Wenker
@thoys
Feb 20 2015 19:57
yeh also seeing it trigger at #4321 , which is just an edit in the readme.txt
Bradley Austin Davis
@jherico
Feb 20 2015 20:50
I wish there was a nice powerful Android set-top device that you could use with the DK1 or DK2 to prototype Gear VR stuff.
I tried with the Kindle Fire TV, but it did not go.
Boris Skegin
@sirop
Feb 20 2015 22:45
I seem to have set all paths properly , also the Qt paths, but when building hifi.sln mtc.exe complains about missing Qt5Core.dll. Any hint?
Bradley Austin Davis
@jherico
Feb 20 2015 22:50
when building? It shouldn't care about the DLL while building, only the lib file. However, if there's some external tool that's being built and then run as part of the build process (like scribe) that itself depends on Qt, it might be failing because it can't find the DLL.
Brad Hefta-Gaub
@ZappoMan
Feb 20 2015 22:50
there is a tool that runs during build that is based on qt
Bradley Austin Davis
@jherico
Feb 20 2015 22:50
there you go.
Brad Hefta-Gaub
@ZappoMan
Feb 20 2015 22:51
I think the build instructions make that clear. But it might have been lost in recent reworking on the build instructions.
Bradley Austin Davis
@jherico
Feb 20 2015 22:52
@sirop perhaps your PATH environment variable doesn't contain the Qt binaries folder, or contains the wrong version of Qt.
Jason Hawks
@nbq
Feb 20 2015 23:00
@sirop here is my path settings, if it helps any, this is off windows 8.1 but you can pick the needed parts: %HIFI_LIB_DIR%\tbb\bin\ia32\vc12;%HIFI_LIB_DIR%\sdl2\bin;%HIFI_LIB_DIR%\soxr\bin;%HIFI_LIB_DIR%\zlib;%SystemRoot%\system32;%SystemRoot%;%SystemRoot%\System32\Wbem;%SYSTEMROOT%\System32\WindowsPowerShell\v1.0\;C:\Program Files (x86)\Windows Kits\8.1\Windows Performance Toolkit\;C:\Program Files\Microsoft SQL Server\110\Tools\Binn\;C:\Program Files (x86)\Microsoft SDKs\TypeScript\1.0\;C:\Program Files (x86)\CMake 2.8\bin;C:\Qt\Qt5.3.2\5.3\msvc2013_opengl\bin;C:\Qt\Qt5.3.2\5.3\msvc2013_opengl\lib\cmake;C:\Program Files (x86)\Git\cmd
Bradley Austin Davis
@jherico
Feb 20 2015 23:04
@ZappoMan I tried out the new build stuff yesterday with the external_project stuff in it. Kind of neat. Trying to integrate the Oculus SDK in the same way. going to want to chat with somebody about how to refactor the rendering output to abstract away the distinction between a screen and an HMD, as well as between various HMDs of different capabilities.
@sirop Does C:\Qt\Qt5.3.2\5.3\msvc2013_opengl\bin contain Qt5Core.dll? If so, look in your build directory for the CMakeCache.txt file and see if the Qt directory listed there matches the PATH environment variable.
Brad Hefta-Gaub
@ZappoMan
Feb 20 2015 23:08
@jherico - wrt your comment about HMD vs screen - sounds interesting - I’m not sure what you’re thinking, would love to hear more. We definitely intentionally made the HMD “overlay” work slightly differently from the screen overlay - to handle the smaller viewable area of the HMD in that case… is that what you’re talking about?
Bradley Austin Davis
@jherico
Feb 20 2015 23:08
@sirop ... unrelated but... putting custom PATH entries in front of the %SystemRoot% ones is generally a bad idea in my experience.
@ZappoMan I don't think so. More about basic architecture.... I'm basically trying to ponder how to minimize the touch points between the core application code and all the potential output mechanisms for rendering (along with the potential input mechanisms that HMDs are likely to expose).
Bradley Austin Davis
@jherico
Feb 20 2015 23:16
as for the overlay, I think that for screen based experiences, just making it an overlay is fine, but that for a VR experience, something like that should be rendered to a avatar relative in-world positionable 'window'.
Brad Hefta-Gaub
@ZappoMan
Feb 20 2015 23:17
@jherico - ok - wrt HMDs and SDK… I am guessing you’re thinking about getting rid of the shader we have for doing the oculus distortion and using the SDK properly? AFAIR - the current code in there today was for the very original oculus SDK and we never updated it when they provided more SDK services for the distortion… am I following you?
Bradley Austin Davis
@jherico
Feb 20 2015 23:17
speaking of.... maybe I should work on a similar VR based version of the chat window. Still need to get qxmpp working though.
Brad Hefta-Gaub
@ZappoMan
Feb 20 2015 23:18
I think your expertise in the oculus SDK would be well applied to the HMD question… and getting us up to date using the latest and greated Oculus SDK
Bradley Austin Davis
@jherico
Feb 20 2015 23:18
@ZappoMan yes, part of what I was doing last night was checking out how I could go about dumping the client side distortion in favor of SDK based distortion. But that requires Interface having the ability to surrender control of the buffer swap.
Brad Hefta-Gaub
@ZappoMan
Feb 20 2015 23:19
interesting…
Bradley Austin Davis
@jherico
Feb 20 2015 23:19
I've started the work here: https://github.com/jherico/hifi/tree/oculus
Brad Hefta-Gaub
@ZappoMan
Feb 20 2015 23:19
we did some work a while back to move to the latest SDK… and we found that we didn’t get big FPS gains out of it… but we didn’t have time to dig deeper
I had suspected that there was something about how we were managing openGL that was likely getting in our way
it sounds like you’re all over the root issue.
we should get some time to chat between you and Sam and I about this stuff...
we want to rework the rendering pipeline in general...
Bradley Austin Davis
@jherico
Feb 20 2015 23:20
well, FPS gains are only one potential benefit. Reduced code maintenance is another, but probably the biggest is support for Direct HMD mode, which you currently don't have as far as I can tell.
Brad Hefta-Gaub
@ZappoMan
Feb 20 2015 23:21
correct
my instinct is that Direct HMD mode improves the FPS as well… but I’m just speculating.
Bradley Austin Davis
@jherico
Feb 20 2015 23:21
but I'm also trying to think forward, so that the amount of incremental work done for each new kind of HMD that comes out will be minimal.
Brad Hefta-Gaub
@ZappoMan
Feb 20 2015 23:22
what are the other benefits of direct HMD mode?
yes yes - re: more HMDs
Bradley Austin Davis
@jherico
Feb 20 2015 23:22
Is this a test?
;)
Brad Hefta-Gaub
@ZappoMan
Feb 20 2015 23:22
lolz - no!
Bradley Austin Davis
@jherico
Feb 20 2015 23:22
well, for OpenGL it greatly helps reduce judder.
Brad Hefta-Gaub
@ZappoMan
Feb 20 2015 23:23
that’s what I suspected… and we should get to take advantage of oculus time warp?
Brad Hefta-Gaub
@ZappoMan
Feb 20 2015 23:23
and/or… maybe we could get time warp without direct hmd mode...
Bradley Austin Davis
@jherico
Feb 20 2015 23:23
Windows has serious issues with syncing OpenGL to anything but the primary monitor's refresh rate. Hence you end up having to do things like set the Rift as the primary monitor when you're running in extended mode. or set it to 60 hz, or set your main monitor to 75hz (if you can). It's a giant pain in the ass.
Direct mode eliminates that.
Brad Hefta-Gaub
@ZappoMan
Feb 20 2015 23:24
ah - I see
Lexx
@Triplelexx
Feb 20 2015 23:24
I only played with DK1, yeah same issues
Bradley Austin Davis
@jherico
Feb 20 2015 23:24
You can get timewarp without direct HMD mode. In fact I'm not sure you don't already have it in the current code.
@Triplelexx interesting. I'll have to take a closer look at that, but for that price I'd probably be more likely to see if I could toss an x86 build of android on one of my existing netbooks or something.
@ZappoMan Hmmm... yeah it looks like you're already using timewarp... at the very least you're calling ovrHmd_GetEyeTimewarpMatrices in the existing code and passing them to the shader.
Brad Hefta-Gaub
@ZappoMan
Feb 20 2015 23:27
yeah - my memory is a little fuzzy on this.
Lexx
@Triplelexx
Feb 20 2015 23:30
That does look like a pretty speedy box though, don't know the spec or bench scores on Gear
I want that box really. Nice features :)
prob run Win10 as well, still waiting for news on what they might offer for devices like that
Sata and ac wireless is pretty novel regardless of the chips.