Related
Hi
I'm trying to write an app to capture video through integrate cameras using directshow. Now, the prob is though the app is doing this now, but after capture is stopped, encoding takes a loooooong time to complete. The encoder being used is WMV9. Now, the inbuilt camera apps must be doing this differently because they dont take any extra time after capture is stopped. So, any suggestion how to optimize the process (how to use a different encoder, or otherwise) to speed up the capture?
TIA
Just found out that HTC devices ship with a filter named "HTC Source Filter". Somewhere on directx.audio group I found out that someone had delved into it a bit and found that it was a capture filter from HTC (both audio and video). I checked in registry and that is true. So, it must be some optimized filter that we can use for capture.
Now, I tried to enumerate the pins on this capture filter, for which I have to first query for an IPersistPropertyBag interface on this filter (used the clsid of the filter from the registry and CoCreateInstance for it returns S_OK). But the QueryInterface returns E_NOINTERFACE. Any opinions/suggestions here? Because the registry clearyly shows that this is a capture filter. Is there any other way I can use this filter?
---My Code---
HRESULT hResult = S_OK;
IGraphBuilder *pFilterGraph;
ICaptureGraphBuilder2 *pCaptureGraphBuilder;
htcDDR.Data1 = 0xA06DF275;
htcDDR.Data2 = 0xC4F3;
htcDDR.Data3 = 0x46C7;
htcDDR.Data4[0] = 0xB5;
htcDDR.Data4[1] = 0x20;
htcDDR.Data4[2] = 0x3A;
htcDDR.Data4[3] = 0x16;
htcDDR.Data4[4] = 0xB3;
htcDDR.Data4[5] = 0xD1;
htcDDR.Data4[6] = 0xB8;
htcDDR.Data4[7] = 0xC2;
hResult = CoCreateInstance(CLSID_FilterGraph, NULL, CLSCTX_INPROC,
IID_IGraphBuilder,(void**)&pFilterGraph);
hResult = CoCreateInstance(CLSID_CaptureGraphBuilder, NULL,
CLSCTX_INPROC, IID_ICaptureGraphBuilder2,
(void**)& pCaptureGraphBuilder);
hResult = pCaptureGraphBuilder->SetFiltergraph( pFilterGraph );
IBaseFilter *pVideoCaptureFilter;
hResult=CoCreateInstance(htcDDR, NULL, CLSCTX_INPROC,
IID_IBaseFilter, (void**)&pVideoCaptureFilter);
IPersistPropertyBag *pPropertyBag;
hResult=pVideoCaptureFilter->QueryInterface( IID_IPersistPropertyBag, (void**) &pPropertyBag );
/* This is where I get the E_NOINTERFACE*/
---------------
Capture Performance on WM 5
Hello shantzg001,
I'm quite new in C++ mobile developpement and have the same problem with the encoding performance of the WM9 encoder. Have you find something new to improve that perfomance or change the default encoder (HTC built in) ?
Thanks
hey thecrets, Im afraid I couldn't...
Hello guys,
I've got an HTC Athena and I find encoded videos are crappy. I'm not a coder, at least I'm not a windows mobile coder, Id just like to know if that would be possible to record videos directly to the microdrive or a SD card in raw mode, so I would encode the video later on my desktop PC.
Do you think a machine like the Athena is powerfull enough to handle video recording in raw mode ? Would the microdrive fast enough ?
Another thing, I can't access more than CIF video mode, is it possible to reach higher resolutions with directshow ?
I know there is a software called coolcamera, but it was not updated since 1 year, and my device is not compatible.
Very very last question, if raw mode is possible, could one of you upload here a release of his proggy ?
Thanks dudes ! I don't expect n95 videos on our devices,but I'm sure it can be really better than what we have.
Hi~
have you solve your problem now? I came across the same problem, and have no idea.
I could be wrong ( or just a moron) but isn't there a webcam program for ppc out there that does exactly what you are trying to create?
@arigold1: If that was directed me, then Yes, there is a webcam program but that is not what I "was" trying to create..I needed to do the recording but not as a webcam but smthing else as part of another project that I was doing at that time..
I'm currently doing some programs myself with the free MS embedded VC++.. and I'm finding it comfortable to do a simple dialog-based programs for PPC. I think I can have most of the background code going, and I've just got the GUI .. alright.
Now the question, how do I do a copy/paste to/from clipboard? I had most of the stuff done using the included MFC Wizard. I can get and send data to/from an EditBox (TextBox, whatever you call it). However, the click-hold thing on PPC doesn't seems to work on my EditBox, and hence I'm thinking what's needed to enable a simple Copy/Paste on an EditBox.
Currently, I'm using the simple
Code:
m_editBox = _T("the message I want to show");
UpdateData(FALSE); //send it to the EditBox
Any guide from here would be appreciated. However, I'm thinking there may not be an easy way to do that, hence I've also tried adding a 'Copy' and 'Paste' button to do the job, but I've tried things like
Code:
SetClipboardData(x, x)
GetClipboardData(x)
None works.
I have also tried
Code:
COleDataObject DataObject;
and with the handle etc etc .. but I can't seems to find this COleDataObject , is that in some other environment (e.g. not PPC env)?
Help
Fast solution:
http://www.pocketpcdn.com/articles/sip.html
(this shows/hides sip on get/lost focus in edit controls and add the context menu too)
and this is a simple example how to copy datas into clipboard
if(OpenClipboard(NULL))
{
EmptyClipboard();
HLOCAL clipbuffer = LocalAlloc(0, 100);
wcscpy((WCHAR*) clipbuffer, (WCHAR*) (vtNumber.bstrVal));
SetClipboardData(CF_UNICODETEXT, clipbuffer);
CloseClipboard();
free(szMsg);
LocalFree(clipbuffer);
}
I hope this help u
bye
Thanks for your respond.. things work.. a bit
Code:
//put a test char
char *test;
test = (char*) malloc(100);
strcpy(test, "blah blah blah");
//codes you've given
if(OpenClipboard()) //OpenClipboard(NULL) gives me error
{
EmptyClipboard();
HLOCAL clipbuffer = LocalAlloc(0, 100);
wcscpy((WCHAR*) clipbuffer, (WCHAR*) test);
SetClipboardData(CF_UNICODETEXT, clipbuffer);
CloseClipboard();
//free(szMsg); //not sure what 'szMsg' is
LocalFree(clipbuffer);
}
Things somewhat works. I'm not really sure which part I've got wrong. I'm suspecting some memory allocation is giving me problems. The thing is, if I were to use 'CF_UNICODETEXT' on the 'SetClipboardData(x,x)' line, I get something to paste on other programs (PPC Notes). BUT, the thing pasted is some funny stuff (e.g. letters that cannot be rendered, hence I get the little squares). If I were to use 'CF_TEXT', I don't seems to able to send my stuff to the clipboard or it made it invalid for (PPC Notes) pasting (e.g. I'm not able to paste it in PPC Notes).
Thanks.
BTW, if you are in the mood, can you give me a Paste function as well. Thanks a bunch.
Hi hanmin.
Odd I didn't notice this thread sooner.
Any way if you still having problems with this code here is the solution:
You are working with char and strcpy so your text is in ASCII (each letter one byte).
But you are calling SetClipboardData with CF_UNICODETEXT so it expects WCHAR (UNICODE) where each letter is two bytes.
The strange letters are the result of two consecutive bytes being interpreted as a single letter (probably lends you in Chinese or Japanese region of the Unicode table)
Windows mobile doesn't usually work with ASCII so the text you get from the edit box will already be in Unicode and won't give you any trouble.
The code should look like this:
Code:
//put a test char
CString test; //since you are working with MFC save your self the trouble of memory allocation
test = L"The text I want on clipboard"; //The L makes the string Unicode
//codes you've given
if(OpenClipboard()) //OpenClipboard(NULL) gives me error
{
EmptyClipboard();
//not sure why you need to copy it again, but here goes:
HLOCAL clipbuffer = LocalAlloc(0, test.GetLength() * 2); //remember: every letter 2 bytes long!
wcscpy((WCHAR*) clipbuffer, (WCHAR*)(LPCTSTR)test); //LPCTSTR is an overloaded operator for CString
SetClipboardData(CF_UNICODETEXT, clipbuffer);
CloseClipboard();
//szMsg probably belongs to some larger application and is irrelevant
LocalFree(clipbuffer);
}
I never used the clipboard APIs my self so I can't guide you farther but this code should work.
Hope this helps.
Wooo hooo.. Thanks levenum. I'm back on business!
You code works wonderfully.. just the final code "LocalFree(clipbuffer);" seems to cause problems. Without that, it works. I'm not sure if it will cause a memory leak.. but that's not much of my concern now
Now my Paste also works, and it seems that the magic code is the "LPCTSTR", which I have NO idea what it is (I'm more of a pure C person and.. a Java person ) Thanks again.
Glad I could help.
I am working from Ubuntu right know (Linux distro in case you didn't know) so I do not have access to my off-line MSDN files, but I recommend you check out the documentation for SetClipboardData.
It is possible it releases the memory it self so when you call LocalFree the handle is no longer valid.
That could also be the reason why you need to allocate memory instead of passing it the string directly.
As for LPCTSTR it is simple and not C++ related:
#define const* WCHAR LPCTSTR
Its M$ way of saying Long Pointer to Constant STRing
T changes meaning based on what you are working with:
If you work with ASCII TCHAR is char
If you work with Unicode TCHAR is WCHAR
Basically these are just all redefinitions of variable types so you can distinguish what they are used for.
In C++ you can overload operators. For example you can have a function which changes the way ++ works with certain types of variables.
In our case CString class has a function which determines what happens when you try to cast (convert) it to a pointer to string.
Thats all the "magi" code.
Good luck with your app.
Small update:
Since I had to go in to XP anyway (to change PDAMobiz ROM which kept hanging at random and didn't let me use BT to latest PDAViet which for now seem very good) I took a quick peek at the help files.
Here is why you should not release the memory:
After SetClipboardData is called, the system owns the object identified by the hMem parameter. The application can read the data, but must not free the handle or leave it locked. If the hMem parameter identifies a memory object, the object must have been allocated using the LocalAlloc function
Click to expand...
Click to collapse
levenum, thanks. You've got me almost there. There are several stuff I need to polish up though. Attach is a pre-mature version of what I wanted to do. There are several issues (including the fact that, only the 4 characters of the password are effectively used, which can be easily fix, I think. Just need to find the bug and squash it) that I like to polish up. They are sorted in order of importance (to me):
[1] Keyboard (SIP) pop up.
For this, I digged around and got to know that the function
"SHSipPreference( HWND, SIP_UP)" is the one to used. However, it never did what it suppose to do. I have had it put inside the "OnSetfocusConfirmPasswordEdit()" of the edit box, which should be called when it is set focus. I suspect that is I haven't set the HWND correctly. I have tried "NULL" and also tried using the "CWnd* pParent" from my dialog constructor (generated code my MFC Wizard). None of them worked.
[2] Editbox focusing.
For some reason, the focus on my main-dialog is correct on the editbox of the 'message'. However, on the dialog which is to confirm the password (which I called using
Code:
CConfirmPasswordDlg confirmPasswordDlg;
int nResponse = confirmPasswordDlg.DoModal();
is focusing on the 'Ok' button. What I like it to do is to focus on the 'confirmPasswordEdit' box, and it ought to automatically pop up the keyboard (SIP).
[3]Reduced size pop up dialog
I was trying to make the 2nd confirm password dialog smaller, something like a pop up in the PPC rather than something that take up the whole screen without much contents in it. How would you go about doing that? Is it not possible in PPC? E.g, if you were to use Total Commander, and start copying files around, they do have a pop up that does take up the entire screen. I'm suspecting I shouldn't do a "confirmPasswordDlg.DoModal()", and should some what do something myself. I have tried, SetVisible(1) thing, but that doesn't work. Or it shouldn't meant to work because my 1st screen is a dialog screen?
[4]Timer?
I would like to have a function of which after a certain period of idle time, it will clear off the clipboard and close itself. How would I go about doing this? Some sort of background thread thing?
Anyone can shine a light on my issues above? On MS-embedded Visual C++ (free), with Pocket PC 2003 SDK (free)
Attached the Blender-XXTea edition
Works on PPC2005 and WM5 (should work on WM6)
Does not require .NET framework
VERY small (54K)
hanmin said:
[2] Editbox focusing.
For some reason, the focus on my main-dialog is correct on the editbox of the 'message'. However, on the dialog which is to confirm the password (which I called using
Code:
CConfirmPasswordDlg confirmPasswordDlg;
int nResponse = confirmPasswordDlg.DoModal();
is focusing on the 'Ok' button. What I like it to do is to focus on the 'confirmPasswordEdit' box, and it ought to automatically pop up the keyboard (SIP).
Click to expand...
Click to collapse
In your CConfirmPasswordDlg::OnInitDialog handler, call GetDlgItem(confirmPasswordEdit).SetFocus() and return FALSE. That should handle the focus and possibly the SIP popup.
3waygeek said:
In your CConfirmPasswordDlg::OnInitDialog handler, call GetDlgItem(confirmPasswordEdit).SetFocus() and return FALSE. That should handle the focus and possibly the SIP popup.
Click to expand...
Click to collapse
HEY! The focus works! The working code is
Code:
((CWnd*) CConfirmPasswordDlg::GetDlgItem(IDC_CONFIRM_PASSWORD_EDIT))->SetFocus();
BTW, I'm wondering, whats the effect of a return TRUE/FALSE on a 'OnInitDialog()'?
Anyway, the keyboard pop up is still not working. I'm using the command
Code:
void CConfirmPasswordDlg::OnSetfocusConfirmPasswordEdit() {
SHSipPreference( (HWND)g_pParent, SIP_UP);//
}
which I suspect the 'g_pParent' is NULL. If it is NULL, would it work?
Ok, I haven't used MFC for a while and almost not at all on PPC but I will give this a shot:
1) MFC forces dialogs to be full-screen. Here is a detailed explanation on how to change that. Note that for some reason this will work only once if you use the same variable (object) to create the dialog several times.
If you use a local variable in say a button handler thats not a problem because the object is destroyed when you exit the function.
2) There is a simple SetTimer API. You can give it a window handle and then add an OnTimer message handler to that window. Or you could give it a separate function which will be called (say TimerProc). In that case you can give it NULL as window handle.
Note that CWnd objects have a member function with the same name (SetTimer) which sets the timer with that window handle (so that window will receive WM_TIMER message). If you want the raw API call ::SetTimer.
Also note that the timer will continue to send messages / call your special function every x milliseconds until you call KillTimer.
3) I am not sure what the problem with the SIP is. CWnd and derived classes like CDialog have a function called GetSafeHwnd (or GetSafeHandle, I don't remember exact name). Try passing that to SHSipPreference.
If that does not work here is an article with an alternate solution.
WOHO!! Everything works NOW!!.. MUAHAHHAHA.. wait til you see my release version
Non maximized windows works using the code suggested at the page. Although I still do not understand where the heck this '.m_bFullScreen' property came from. It is not anywhere obvious to be seen.
Code:
CNfsDlg dlg;
dlg.m_bFullScreen = FALSE;
dlg.DoModal();
Timer works using the
Code:
xx{
//your code...
CBlenderDlg::SetTimer(1, 5000, 0); //event 1, 5 seconds, something
//your code...
}
void CBlenderDlg::OnTimer(UINT nIDEvent){
//do something here for the timer
}
although somehow, the OnTimer() only works if I went to the MFC class wizard to add the WM_TIMER function. Doesn't work when I add in the OnTimer() myself. Must be something else that I've missed. Anyway.
Keyboard issue solved using
Code:
SHSipPreference( CBlenderDlg::GetSafeHwnd(), SIP_UP);
Glad its working out for you.
Couple of comments:
1) Somewhere at the top of the cpp file, if I am not mistaking there is something called a message map. It's a bunch of macros that lets MFC know what window messages it handles. An entry there is what was missing when you added the function manually.
2) m_bFullScreen is just another among many undocumented features. M$ likes to keep developers in the dark. For instance WM 2003 and up have an API called SHLoadImage which can load bmp, gif, jpg and some other formats and return HBITMAP which all the usual GDI functions use.
This API was undocumented until WM 5 came out and even then they said it only works for gif...
hanmin said:
BTW, I'm wondering, whats the effect of a return TRUE/FALSE on a 'OnInitDialog()'?
Click to expand...
Click to collapse
The return value indicates whether or not the standard dialog handler (which calls your OnInitDialog) should handle setting the focus. As a rule, OnInitDialog should return TRUE, unless you change the focus within the handler (or you're doing an OCX property page on big Windows).
I haven't done much WinMob/CE development -- I've been doing big Windows for 15+ years, so window message handling is pretty much second nature. I started doing Windows development back in the days when you didn't have C++ or MFC boilerplate; you had to implement your own DialogProc, crack the messages yourself, etc. It's a lot easier now.
CommandBar / MenuBar
I'm back.. with more questions
Not much of a major issue, but rather an annoying thing I've found. Probably that's what evc/mfc/m$ intended to do that.
Anyway, I'm starting my way of getting around CommandBar. I created a MFC skeleton, studied the code, and that's what I've found, after I've created a CommandBar/MenuBar on evc and putting it in
Code:
if(!m_mainMenuBar.Create(this) ||
!m_mainMenuBar.InsertMenuBar(IDR_MAINMENUBAR) ||
!m_mainMenuBar.AddAdornments() ||
!m_mainMenuBar.LoadToolBar(IDR_MAINMENUBAR))
{
TRACE0("Failed to create IDR_MAINMENUBAR\n");
return -1; // fail to create
}
where I have the variable 'CCeCommandBar m_mainMenuBar' and I have created a MenuBar on evc with the Id 'IDR_MAINMENUBAR'. The menu bar works flawlessly on my dialog based application, when I have the 1st level as a pop up. Example
MenuBar --> 'File' --> 'New', 'Save'
Where 'File' is a folder-like thing that pop-ups and show the contents (i.e. in this example, 'New', and 'Save'). However, given the SAME code to load the CommandBar/MenuBar, it will not work, if I were to put the actual command at 1st level. Example, this will not work
MenuBar -> 'New', 'Save'
where there isn't any folder-like pop-up to store the commands 'New', and 'Save'.
I know that I can have buttons for these commands, and probably works. But, what I'm trying to do is to utilize the bottom-left-right softkey in WM5/6. If I were to have the 'File'->'New','Save' structure, it works fine with WM5, showing it as a softkey. But, if I were to do just 'New','Save' it will not show up in both WM2003 emulator and WM5.
As a matter of fact, even if I have (say) File->New,Load, and I added a new command (i.e. not folder-like-pop-up), example 'Help' on the CommandBar/MenuBar, the File->New,Load will not show up too. It seems like the 1st level command (ie. without a folder-pop-up), causes some problems and stop it from loading further.
Guys, ring any bell?
two bytes more
levenum said:
Hi hanmin.
Odd I didn't notice this thread sooner.
Any way if you still having problems with this code here is the solution:
You are working with char and strcpy so your text is in ASCII (each letter one byte).
But you are calling SetClipboardData with CF_UNICODETEXT so it expects WCHAR (UNICODE) where each letter is two bytes.
The strange letters are the result of two consecutive bytes being interpreted as a single letter (probably lends you in Chinese or Japanese region of the Unicode table)
Windows mobile doesn't usually work with ASCII so the text you get from the edit box will already be in Unicode and won't give you any trouble.
The code should look like this:
Code:
//put a test char
CString test; //since you are working with MFC save your self the trouble of memory allocation
test = L"The text I want on clipboard"; //The L makes the string Unicode
//codes you've given
if(OpenClipboard()) //OpenClipboard(NULL) gives me error
{
EmptyClipboard();
//not sure why you need to copy it again, but here goes:
HLOCAL clipbuffer = LocalAlloc(0, test.GetLength() * 2); //remember: every letter 2 bytes long!
wcscpy((WCHAR*) clipbuffer, (WCHAR*)(LPCTSTR)test); //LPCTSTR is an overloaded operator for CString
SetClipboardData(CF_UNICODETEXT, clipbuffer);
CloseClipboard();
//szMsg probably belongs to some larger application and is irrelevant
LocalFree(clipbuffer);
}
I never used the clipboard APIs my self so I can't guide you farther but this code should work.
Hope this helps.
Click to expand...
Click to collapse
I know it is a bit late! But there is a mistake the code snippet:
HLOCAL clipbuffer = LocalAlloc(0, test.GetLength() * 2); //remember: every letter 2 bytes long!
needs to be
HLOCAL clipbuffer = LocalAlloc(0, test.GetLength() * 2+2);
the terminating 0 is als 2 bytes long!
Those errors are sometimes fatal, with very less chance to find them!
ms64o
Hallo!
I'm working on an application called hTorch and although it is my first vb.net app I was quite successfull so far (at least I hope so ). But one think I'm just not able to achieve: Preventing the device from going into standby/suspend while the app is running.
In C++/C# there semes to be a system call "SystemIdleTimerReset()" but either there is no equivalent for vb.net or I was not able to find it.
Another thought was to modify the according registry key (HKLM\System\CurrentControlSet\Control\Power\Timeouts\BattSuspendTimeout) and to restore the original settings on program exit. This works fine except that WM does not apply the settings change immediately when it's done via registry. Again I did some research and again I only found a solution for C++/C# only. A system call named "NWUS_MAX_IDLE_TIME_CHANGED" tells the OS that the settings have changed and should be reread. I wasn't able to find something similar for vb.net though.
Does anybody know a solution for my needs?
Maybe someone could provide me with some sample code how he achieved this within vb.net?
Maybe some C++/C# programmer has build a library to access the power functions?
I would be really thankful for any help/hint you can give me!
Thanks in advance,
DeepThought
The standby mode is in the registry. Make a timer in VB.net and set it with an interval of about 5000 ms. Then make sure that ever time the timer ticks the standby mode will be disabled in the registry
Thank you for the quick response!
But the change via registry doesn't work. When I change the BattSuspendTimeout to 0 that does change the setting correctly. But they become not active. Only after a soft reset. It seems, that you somehow have to notify windows, that the settings have changed. Otherwise the will not be reread.
Any Idea how to achieve that?
I would also like to know if there is a way to prevent suspend.
I think only the pocket pc winmo versions do a real suspend where wifi and applications stop processing, and the smartphone winmo devices only kind of black the screen.
The only software I know is S2U2 which successfully can prevent "real suspend" and let e.g. my led notification work correctly.
Cause i cannot really use it on ks20 due to compatibilty issues i would be very interested in a reg hack, or (if there really is no reghack, i tried alot) some code snipets with which i could build a little app.
There's a pretty good article on CodeProject covering power.
*digs out link*
Here you go - http://www.codeproject.com/KB/mobile/WiMoPower1.aspx
It covers:
* Displaying the voltage, current, and temperature of your device's battery
* Changing the power state of hardware within the device
* Enumerating the hardware in a Windows Mobile Professional device
* Enumerating the power modes that a Windows Professional device supports
* Preventing a device from sleeping
* Toggling the state of the screen's backlight
* Waking up the machine to perform work without alerting the user or turning on the screen
[solution]
Thanks to AndyZap I can now answer the question myself
It is so simple, that I'm really ashamed now.
The magic word for me was PInvoke. Since I new the function I was searching for was an available win32 systemcall the missing link was how I can make this system call within vb.net.
So the actual solution are just two lines:
Code:
Declare Sub IdleTimerReset Lib "coredll.dll" Alias "SystemIdleTimerReset" ()
where "IdleTimerReset" is the name I've given the Sub. Which than just needs to be called like:
Code:
IdleTimerReset()
NOTE: This just resets the Idle timer once. So if you want to prevent the device from going into standby you have to reset this timer every time before it reaches the threshold. (The windows default setting is 60 seconds AFAIK).
THANK YOU AndyZap for your Help via PMail!
DeepThought
Thank you for your answer Northernmost!
I really love this community!
Hi everyone,
I have a small technical problem to solve and hopefully someone knows an answer for it. Someone gave me a HTC Diamond 2 (very nice toy I think). I don’t know which hardware is exactly inside the phone but I assume it’s an ATI ImageOn 2300 which fully supports OpenGL ES 1.0 + Extension Pack (that’s what GL_EXTENSION is saying). And all of these extensions are doing fine. As everyone knows HTC is not providing a D3D driver for their phones and all other drivers I have seen are just a wrapper around OpenGL ES. So using D3D is currently not an option.
I have to write an application which requires some render-to-texture functionality in realtime and here my problems are starting to grow.
[1] glCopyTexImage2D:
I recognized that glCopyTexImage2D() is very slow. My framerate is dropping dramatically from >100 frames down to ~20 frames. Somehow this is done in software by the driver and not hardware- accelerated. glCopyTexSubImage2D() is even more slow (down to ~10 frames per second), but both functions are working. I tried to move it into a different thread, but the driver is not supporting shared contexts. Also it’s not supporting two bound contexts at the same time with two different windows and threads. So this can’t be improved or I’m doing something wrong.
[2]PBuffers:
PBuffer are working fine. But PBuffers which can be bound to a texture are not supported by OpenGL ES 1.0. And that’s what happens also on this ATI card.
[3]GL_OES_framebuffer_object:
This was my preferred choice. But framebuffers are not officially supported by the installed OpenGL driver. But the libgles_cm.dll on the HTC is exporting these functions so I tried them and recognized that they are not working correctly. Somehow the vertex pipeline is allowing only triangles in the center of the viewport. All others will be discarded. When I turn it off it renders correct. Using glDrawTexiOES() will be ignored while using framebuffers. I assume that either the current implementation of the framebuffers is just waste of some development guys or the functionality is locked in some way.
[4]D3D:
Normally D3DM is supporting to switch between different rendertargets. But we are all know the D3D problem. I tested some other D3D drivers but they are not usable. Also I can’t imagine how render-to-texture will be implemented in these drivers while it’s not working in OpenGL.
I’m wondering what TouchFlow3D is using internally, whether they are using render-to-texture or not. The only thing I know is that they are using OpenGL ES and some extensions. But Manila.exe is querying the functions of libgles_cm.dll during runtime. So I have to write a few proxy dlls first and need to hook into the system to track what they are doing. And I don’t want to spend time on this.
Does someone knows an alternative to do some render-to-texture on the ATI or knows some secrets of the libgles_cm.dll which I don’t know? There are a lot of private functions inside but can’t find some documentation about it. Also ATI and Qualcomm are not very helpful to me.
Thanks.
Maybe, glReadPixels and then glTexSubImage2D (what probably glCopyTexImage2D is doing)?
(I know that this goes two times over the graphics bus, but you never know...)
What kind of scene (number of triangles, lights, textures) are you rendering with > 100fps?
glReadPixels() + glTexImage2D() is even more slow than glCopyTexImage2D...
Also this can't be parallized. I thought using AHI2DATI.dll instead to do the same thing, but I don't know how to get a surface handle from a OpenGL texture id.
>>>What kind of scene?
A very simple scene yet. Only a few depth sorted + material sorted objects (via VBO) with some textures (backed lighting) on it. Textures are compressed. Currently no lighting, no skinning or other things. While the render thread is waiting for glFinish() to return, a second thread prepares the next frame. Also the rendering thread is not redrawing the entire viewport each time.
The OpenGL texture id is the handle. There's nothing more you can do with it. PBuffers or framebuffer objects are the only way I know for doing performant render to texture in OpenGL.
No...I mean the surface handle of the AHI2DATI library. Here you have access to the raw data of the surface. Somehow the libgles_cm.dll uses these surfaces for it own buffers and/or textures or not?!
But this is not really useful unless someone tells us whats going on inside libgles_cm.dll.
See:http://greengalaxy.wordpress.com/2009/04/18/ati-direct-access-to-hardware/
Hi jeansmsixer, I don't think there is no efficient way to do it. Have you tried eglCopyBuffers?
Aren't all textures stored in system memory? - they surely have to be because the device reports no available video mem.
Even if you get a pointer to the color buffer, it's impossible to wrap the memory in a HBITMAP to select onto a HDC and use GDI fonts for instance. (which is what I need to do).
If you do write the proxy dll like you were suggesting, can you please let me know what TF3D does for fonts? Are they textures or have they somehow mixed 3D with GDI?
Hi jeansmsixer, something else you could try is what is described here:
http://brewforums.qualcomm.com/showthread.php?t=10668
Looking on glbenchmark.com it looks like HTC devices support the extension. You should be able to get a pointer directly to the color buffer (which I presume is in normal system memory) so you could copy off pixels fairly efficiently with your own memcpy(). However, you will need to eglWaitGL() etc to ensure 3d stuff is complete before attempting to access it.
I love the irony of the only helpful information for WM opengl being found on a brew site for symbian. If WM7 is as terrible as the current mess, then I'm moving to iPhone.
I know some people have already discussed this in the noRefresh App thread.
But I got a new idea
I don't wanna create a module and embed that to applications
Instead, I created a new Input Device in the kernel
After that, I set the device ( dev/input/event3 ) permission to allow read / write for others.
Then, write an app using JNI to catch the Input Device and set the mode of refreshing.
The prototype proved that this idea works, as i worked out it already, but not very perfect, still in experiment
I am just in some problems of threading, as I am not a professional programmer
This is wt i have now:
http://www.youtube.com/watch?v=GkFyvRR6In8
NEW!! : http://www.youtube.com/watch?v=cucG03rg3tg
I know that my method is a bit dirty,
but at least it works XD
I hope that someone would like to help
Upload those code soon
By the way, why can't i change content of init.rc ?
It removes my changes after reboot..... How to solve?
wheilitjohnny said:
By the way, why can't i change content of init.rc ?
It removes my changes after reboot..... How to solve?
Click to expand...
Click to collapse
You need to modify uRamdisk to change init.rc.
Would you like to tell me how to unpack the uRamdisk? I use both Windows and Ubuntu, any methods on these two platform is ok. Thanks!
Try the script from this post http://forum.xda-developers.com/showpost.php?p=24135886&postcount=72
This is simply amazing. Since you already have it working, polishing it shouldn't take too long (I think, I'm still a beginner programmer).
I don't think /dev/input is a good place to park that.
There are any number of things that could show up there and affect order.
Why not put it in its own directory?
Maybe I'm missing something, but if we use your method you still expect applications to have to be modified to read/write from event3 to control display mode?
As it is now, you only need a single function call to switch display modes. Yes, there is a little bit of housework to do before that, but what could be simpler than a single function?
I think no place else is more suitable than /dev/input.
As /dev/input is the linux kernel's input system.
In the driver, I would only use the input report system but not use the I/O system.
Maybe even setup a input/event searching function to solve the problem later.
The case now is that, the touch screen originally just provide event2,
but if we need to extract move and fingerup information on upper level, it may waste some time.
In order to make the whole algorithm easier and faster, I added 1 more event in the zForce driver,
to only output FingerMove and FingerUp state.
As the reporting system is starting from the kernel, this hack would need to change uImage + uRamdisk + Add an App. The project is quite huge.
My final target is to make the application remembering your choice on what mode u need for the focusing application.
AlwaysOn / OnlyWhenDragging / AlwaysOff, so on.
Of coz the click-to-call function still work.
Isn't it a better workflow and more intuitive, isn't it the thing that we would expect?
I already have an dev/input/event3 on my system and sometimes event4, 5, 6.
That's why I don't think that whatever it is you are doing belongs there.
Is the point to allow coders for user applications to interface with your driver?
Is this supposed to work without modifying user applications?
What information would be going in/out of event3?
Clearly I am missing something here.
Actually not necessary to be event3, the system will auto-ly create a new eventx, just in normal case, without any extra USB devices, a nook should only have up to event2. So, I use event3 as an example here.
We can later make the program auto-ly search back which is the needed eventx.
Yes, u r right, we no need to modify any other applications and this is exactly the point y i am creating this!
Anyone know how to call a FullScreen Refresh in a service?
always on?
Look really great. Since blinking after scrolling is incomfortable is it possible to have also "always on " mode using your new ideas?
My final target is to make the application remembering your choice on what mode u need for the focusing application.
AlwaysOn / OnlyWhenDragging / AlwaysOff, so on.
Of coz the click-to-call function still work.
Also an Over-Ride mode, for more flexible using.
But the most basic part need to be handled well now, as it is not very perfect now.
I still don't figure out how to call the e-ink driver to refresh the screen =-=
wheilitjohnny said:
I still don't figure out how to call the e-ink driver to refresh the screen =-=
Click to expand...
Click to collapse
Have you tried
Code:
echo 1 > /sys/class/graphics/fb0/epd_refresh
?
wheilitjohnny said:
Yes, u r right, we no need to modify any other applications and this is exactly the point y i am creating this!
Click to expand...
Click to collapse
Are you saying that your idea does not require modifying user applications?
If it doesn't, then there is no need to have a public interface.
It will be only your code talking to your code.
What is the point of this /dev/input/event3? You say that it will be writable. What's going in and out?
Some apps will be using gestures, some dragging. How are you going to keep track of that all?
I have one application that works perfectly fine now, one activity uses swipe gestures to page up/down while another activity uses drag with a user choice of A2 and display while dragging or else only panning at ACTION_UP.
All this requires less than 10 lines of code.
With multitouch, many applications don't even need A2. Even normal panning in Opera Mobile works much better now that Opera doesn't try to display while panning.
Maybe my english is too bad, cannot express the idea well.
I know, we can make such an application with noRefreshDrag working on its own well.
But how about other applications, it is impossible for us to change all applications.
So, my idea is making it system based.
My prototype is very good now, after several adjustment.
Not limited to only 1 application, but the whole system.
The approach is like this:
1. zForce driver provide extra information to InputEvent
2. A JNI catch the InputEvent
3. A service get the data and set the update mode
We only need to write 1 application to handle the setting of this chain.
This is what i mean, hope that u get what i mean now.
mali100 said:
Have you tried
Code:
echo 1 > /sys/class/graphics/fb0/epd_refresh
?
Click to expand...
Click to collapse
Let me try it now!
wow, it works great
wheilitjohnny said:
zForce driver provide extra information to InputEvent
Click to expand...
Click to collapse
I guess that this is the part that I don't understand.
What is this extra information?
Renate NST said:
I guess that this is the part that I don't understand.
What is this extra information?
Click to expand...
Click to collapse
It created an extra Event, l called it Event3 before.
Driver reports only move and finger_up to Event3.
Just providing a channel to pass an information from driver to user-space.
You may ask why not directly using the existing Event, need to create another one.
That is because, the original one only have touch and position information, parsing them back to move information need a bit of work. As the hardware provided the move information, then don't waste it.
Code:
public boolean dispatchTouchEvent(MotionEvent event)
{
switch(event.getAction())
{
case MotionEvent.ACTION_DOWN:
break;
case MotionEvent.ACTION_MOVE:
break;
case MotionEvent.ACTION_UP:
break;
}
return(true);
}
}
Isn't that everything that you could ask for?