New Vp8 coder. v-0.9.5

[Smelter]

Hey guys, the camjo was re-released with a couple small updates and most importantly, a new build of the vp8 codec.

more info here.

http://www.thebitstream.com/?p=250

Also, added the first round of icecast server mount support. If you use nsv output , name the mount '*.nsv' or winamp will not play it as nsv. The icecast module also outputs plain audio. No title change action yet.

[streamer45]

Nice job Smelter

[Smelter]

Hey, thx

btw streamer45, Your red5 app was rebuilt and works mucho better.

hit me on skype.

[sigma_sigma]

Well, vp8 codec can be used with nsvcap thanks to nsv_coder_vp8.dll. I haven't tested yet but I think old nsvcap (from NSVtools_repack.exe) can broadcast via shoutcast without problems. I don't know what Vidcast.exe can do, looks to be a similar application like nsvcap. Unfortunately I don't have any camera in this PC to make some tests right now. The old vp62 encoder did his job several years, we hope that vp8 has a better image quality at the same video bitrate comparing with vp62. I'm using it since 2004 after I have discovered the solution here on winamp forums.
We are waiting also a brand new version of NSVTOOLS, that one i'm using is the last build ever made, pls correct me if I'm wrong.
Also in these years I couldn't find a proper solution to put the streaming in a web page.
As far as I know, the only one player is Rava's made by AOL who can connect to a shoutcast server which is broadcasting vp62 content but is working with IE only without support for firefox or other browsers. Somebody told me that there is also a flash solution, never found it, never saw a live stream example in a web page (swf) somewhere.
I've read also about red5 but i don't understand exactly what it does and seems to run under linux only http://red5.org/wiki/Install
Anyway I was curios about this new vp8 codec and for a solution to mount a shoutcast stream into a web page power by a flash solution (even for vp62 if it works will be outstanding).
My previous post here was several years ago so if I'll make this post into a wrong section please accept my apologies.

[Smelter]

Originally Posted by sigma_sigma

Well, vp8 codec can be used with nsvcap thanks to nsv_coder_vp8.dll.

Dont just thank me, Thank winamp for adding native support so no special dll are needed to watch.

I haven't tested yet but I think old nsvcap (from NSVtools_repack.exe) can broadcast via shoutcast without problems.

Yes and is quite stable if you are satisfied with that solution and limited resolution and DV support.

We are waiting also a brand new version of NSVTOOLS, that one i'm using is the last build ever made, pls correct me if I'm wrong.

There are no NSVTools forthcoming from AOL/Winamp. Winamp will point to my encoder for NSV video encoding and broadcasting when I feel it is stable enough.

Also in these years I couldn't find a proper solution to put the streaming in a web page.
As far as I know, the only one player is Rava's made by AOL who can connect to a shoutcast server which is broadcasting vp62 content but is working with IE only without support for firefox or other browsers. Somebody told me that there is also a flash solution, never found it, never saw a live stream example in a web page (swf) somewhere.
I've read also about red5 but i don't understand exactly what it does and seems to run under linux only http://red5.org/wiki/Install

Red5 runs on windows and linux, and provides flash playback of NSV.

Anyway I was curios about this new vp8 codec and for a solution to mount a shoutcast stream into a web page power by a flash solution (even for vp62 if it works will be outstanding).
My previous post here was several years ago so if I'll make this post into a wrong section please accept my apologies.

Here is flash player that connects to shoutcast server and even supports AACp and H264.

http://thundersnow.thebitstream.com/player/

[sigma_sigma]

Thanks so much. I've tried vp8 under nsvtools and is working very nice via shoutcast. Very interesting latest verion of camjo but is working only with icecast. After the transmission is started camjo can connect to icecast but unfortunately winamp client not. I wasn't able to connect to the stream with winamp using the standard url like http://server_ip:8000;stream.nsv, nothing happens.....is trying to connect for ~one second and suddenly gives up.
Any solutions? Thanks in advance!

[Smelter]

sorry, My fault. I broke it.

Ill fix it soon,

[sigma_sigma]

We are waiting the new version, nice work Smelter!

[oLRi]

Will your red5 app oder thundersnow support vp8?

[Smelter]

no until I get the alchemy version or create the alchemy version myself. cygwin anyone?

[oLRi]

Originally Posted by Smelter

cygwin anyone?

I do.

I tried also to create a vp8 coder dll by myself. Tried it with the samples from the nsv sdk, can you post the visual studio source from your coder dll?

Played arround with your vp8 coder dll...found some "bugs":
- VLC plays VP8 Stream but the picture is 180° inverted
- ffmpeg can't decode an vp8 nsv stream, error messge: Invalid (unknown) pixel format

[Smelter]

http://comserver.googlecode.com/svn/.../comDemo/coder

Extract the the codec source from http://www.webmproject.org/tools/vp8-sdk/ to the 'includes' directory.

Configure the project settings to the correct includes.

On your flipping.... So then, you'll have to flip/unflip the image output, which is unflipped I think, and set/unset the coder_dll flip flag. Winamp is not flipped at this time, so flip the output and toggle the flip flag. Winamp will still be right side up then.

Also check versions. 'Flip' shouldn't change between versions, but output can look funny. This was 9.5 from january.

[oLRi]

Sooooooo.....

I changed:

code:
RGBtoYUV420PSameSize((const unsigned char *)in,raw.img_data,4,0,cfg.g_w,cfg.g_h);

to:

code:
RGBtoYUV420PSameSize((const unsigned char *)in,raw.img_data,4,1,cfg.g_w,cfg.g_h);

the 1/0 is the toggle for flipping.

When i use a coder.dll with this settings my VLC Stream is correct, but now my winamp stream is fliped.

But i think i am missing this:

so flip the output

how can i do this?

[Smelter]

Oh shit, you're right! the flip flag is in the decoder, and that is now built in to winamp core!


We can ask winamp team to flip nsv video for vp8 and It would probably be in the next release

[oLRi]

So there is no possibility to compile a coder.dll with vp8 which has a "normal" picture in vlc (and other clients) AND winamp at the same time?

So the only thing i can do is build a dll for "winamp viewers" and "other viewers"?

[Smelter]

No, the winamp team is very cool about responding to this kind of thing.

It will just be till the next release likely.

[oLRi]

There is something other i'm playing with at this time:
watermarking.

I thought i can't be that hard to implement a watermark with a custom dll.

in this function
RGBtoYUV420PSameSize
you have 'rgb' which is the rgbpicture of each frame, right?
So i must be easily possible to modify each frame...any ideas about that?

[Smelter]

The easy way is to load the rgb bits of another and blend them.

code:


	void AlphaBlender2(char *pBits, char* m_pCheckers)

	{   // the incoming source

		//

		RGBQUAD * pSource = (RGBQUAD*) pBits;



		// the checkerboard buffer

		//

		RGBQUAD * pCheckers = (RGBQUAD*) m_pCheckers;

		// blend them

		//

		for( int x = 0 ; x < 320 * 240 ; x++ )

		{

			if( pCheckers->rgbReserved)

			{

				RGBQUAD d;

				d.rgbRed   = (BYTE) (pSource->rgbRed * pSource->rgbReserved / 256 + 

									pCheckers->rgbRed  * ( 256 - pSource->rgbReserved ) / 256);



				d.rgbGreen = (BYTE) (pSource->rgbGreen * pSource->rgbReserved / 256 + 

									pCheckers->rgbGreen * ( 256 - pSource->rgbReserved ) / 256);



				d.rgbBlue  = (BYTE) (pSource->rgbBlue * pSource->rgbReserved / 256 + 

									pCheckers->rgbBlue * ( 256 - pSource->rgbReserved ) / 256);

				

				d.rgbReserved=  pCheckers->rgbReserved;

				*pCheckers = d;

			}

			pSource++;

			pCheckers++;

		}

	}

The harder but easier way( go figure...) is with device context.

code:




#include "stdafx.h"

#pragma once

#include "streams.h"

#include "stdio.h"

#include <tchar.h>

#include "CAutoDC.h"

#include<iostream>

using namespace std;



char *CAutoDC::Translate(char * Main ,char *pBuffer)

{

	//Clear DC paintbuffer

	PatBlt(m_dcPaint,0,0,320,240,BLACKNESS);

	

	//Copy Sample to DC paintbuffer

	CopyMemory(m_pPaintBuffer,pBuffer,307200);

	//Set up Blend function

	StretchBlt(m_dcPaint2,0,239,320,-240,m_dcPaint,0,0,320,240,SRCCOPY);

	CopyMemory(Main,m_pPaintBuffer2,abs(m_bmpInfo.bmiHeader.biWidth*m_bmpInfo.bmiHeader.biHeight*m_bmpInfo.bmiHeader.biBitCount/8));

	return Main;

}

void CAutoDC::GetColor1(RGBQUAD *pQuad1)

{

	CAutoLock lock(&cSharedState);

	*pQuad1=tColor;

	return ;

}

void CAutoDC::GetColor2(RGBQUAD *pQuad2)

{

		CAutoLock lock(&cSharedState);

	*pQuad2=t2Color;

	return ;

}

void CAutoDC::GetAlpha1( BYTE * Alpha1)

{

	CAutoLock lock(&cSharedState);

	*Alpha1=alpha;

	return;

}

void CAutoDC::GetAlpha2( BYTE * Alpha2)

{

		CAutoLock lock(&cSharedState);

	*Alpha2=alpha2;

	return;

}

void CAutoDC::SetColor2 (RGBQUAD *pQuad2)

{//State changing method

	CAutoLock lock(&cSharedState);

	t2Color=*pQuad2;

}

void CAutoDC::SetColor1 (RGBQUAD *pQuad1)

{//State changing method

	CAutoLock lock(&cSharedState);

	tColor=*pQuad1;

}

void CAutoDC::SetAlpha2 (BYTE * pAlpha2)

{//State changing method

	CAutoLock lock(&cSharedState);

	alpha2=*pAlpha2;

}

void CAutoDC::SetAlpha1 (BYTE * pAlpha1)

{//State changing method

	CAutoLock lock(&cSharedState);

	alpha=*pAlpha1;

}

void *CAutoDC::Logo (void* Main, void* Pip)

{//Streaming method

	CAutoLock lock(&cSharedState);

//Picture in Picture

	//Clear DC paintbuffer

	PatBlt(m_dcPaint,0,0,m_bmpInfo.bmiHeader.biWidth,m_bmpInfo.bmiHeader.biHeight,BLACKNESS);

	PatBlt(m_dcPaint2,0,0,m_bmpInfo.bmiHeader.biWidth,m_bmpInfo.bmiHeader.biHeight,BLACKNESS);

	//Copy Sample to DC paintbuffer

	CopyMemory(m_pPaintBuffer,Main,CSAMP);

	CopyMemory(m_pPaintBuffer2,Pip,CSAMP);

	//Set up Blend function

	BLENDFUNCTION blend;

	blend.BlendOp =AC_SRC_OVER;

	blend.BlendFlags =0;

	blend.SourceConstantAlpha =alpha;

	

	AlphaBlend(m_dcPaint,140,110,160,120,m_dcPaint2,0,0,320,240,blend);

	CopyMemory(Main,m_pPaintBuffer,CSAMP);

	return Main;

}



void *CAutoDC::Pip (void* Main, void* Pip)

{//Streaming method

	

	//Picture in Picture

	//Clear DC paintbuffer

	PatBlt(m_dcPaint,0,0,m_bmpInfo.bmiHeader.biWidth,m_bmpInfo.bmiHeader.biHeight,BLACKNESS);

	PatBlt(m_dcPaint2,0,0,m_bmpInfo.bmiHeader.biWidth,m_bmpInfo.bmiHeader.biHeight,BLACKNESS);

	//Copy Sample to DC paintbuffer

	CopyMemory(m_pPaintBuffer,Main,CSAMP);

	CopyMemory(m_pPaintBuffer2,Pip,CSAMP);

	//Set up Blend function

	BLENDFUNCTION blend;

	blend.BlendOp =AC_SRC_OVER;

	blend.BlendFlags =0;

	blend.SourceConstantAlpha =alpha;

	

	AlphaBlend(m_dcPaint,140,110,160,120,m_dcPaint2,0,0,320,240,blend);

	CopyMemory(Main,m_pPaintBuffer,CSAMP);

	return Main;

}

void *CAutoDC::PipRect (void* Main, void* Pip, BLTRECT *Rect, int Alpha)

{//Streaming method

	

	//Picture in Picture

	//Clear DC paintbuffer

	PatBlt(m_dcPaint,0,0,m_bmpInfo.bmiHeader.biWidth,m_bmpInfo.bmiHeader.biHeight,BLACKNESS);

	PatBlt(m_dcPaint2,0,0,m_bmpInfo.bmiHeader.biWidth,m_bmpInfo.bmiHeader.biHeight,BLACKNESS);

	//Copy Sample to DC paintbuffer

	CopyMemory(m_pPaintBuffer,Main,CSAMP);

	CopyMemory(m_pPaintBuffer2,Pip,CSAMP);

	//Set up Blend function

	BLENDFUNCTION blend;

	blend.BlendOp =AC_SRC_OVER;

	blend.BlendFlags =0;

	blend.SourceConstantAlpha =Alpha;

	

	AlphaBlend(m_dcPaint,Rect->Left ,Rect->Top ,Rect->Width ,Rect->Height ,m_dcPaint2,0,0,320,240,blend);

	CopyMemory(Main,m_pPaintBuffer,CSAMP);

	return Main;

}





void CAutoDC::TextSample (void * pBuffer,char * text)

{//Streaming method

	CAutoLock lock(&cSharedState);

	//Clear DC paintbuffer

	PatBlt(m_dcPaint,0,0,m_bmpInfo.bmiHeader.biWidth,m_bmpInfo.bmiHeader.biHeight,BLACKNESS);

	//Copy Sample to DC paintbuffer

	CopyMemory(m_pPaintBuffer,pBuffer,CSAMP);

	CHAR szText[256];

	sprintf( szText,text);

	//Text background color

	SetBkMode(m_dcPaint,TRANSPARENT);

	SetTextColor(m_dcPaint,RGB(0 ,0 ,0 ));

	//Do the text

	if( !TextOut( m_dcPaint, (m_bmpInfo.bmiHeader.biWidth/2-((strlen(text)/2)*8))+1, m_bmpInfo.bmiHeader.biHeight-19, szText,_tcslen( szText ) ) )

		return ;

	// Text Color



	//SetTextColor(m_dcPaint,(COLORREF)tColor);

	SetTextColor(m_dcPaint,RGB(t2Color.rgbRed ,t2Color.rgbGreen ,t2Color.rgbBlue ));

	

	//Do the text

	if( !TextOut( m_dcPaint, (m_bmpInfo.bmiHeader.biWidth/2-((strlen(text)/2)*8)), m_bmpInfo.bmiHeader.biHeight-20, szText,_tcslen( szText ) ) )

		return ;

	//Copy sample back to original buffer

	CopyMemory(pBuffer,m_pPaintBuffer,CSAMP);

}

void CAutoDC::TextSample (void * pBuffer,char * text, int Linenum)

{//Streaming method

	CAutoLock lock(&cSharedState);

	Linenum++;

	//Clear DC paintbuffer

	PatBlt(m_dcPaint,0,0,m_bmpInfo.bmiHeader.biWidth,m_bmpInfo.bmiHeader.biHeight,BLACKNESS);

	//Copy Sample to DC paintbuffer

	CopyMemory(m_pPaintBuffer,pBuffer,CSAMP);

	CHAR szText[256];

	sprintf( szText,text);

	//Text background color

	SetBkMode(m_dcPaint,TRANSPARENT);

	// Text Color

	SetTextColor(m_dcPaint,RGB(0 ,0 ,0 ));

	//Do the text

	if( !TextOut( m_dcPaint, 4, (((m_bmpInfo.bmiHeader.biHeight-18)/12) * Linenum)-19, szText,_tcslen( szText ) ) )

		return ;

	//SetTextColor(m_dcPaint,(COLORREF)tColor);

	SetTextColor(m_dcPaint,RGB(tColor.rgbRed ,tColor.rgbGreen ,tColor.rgbBlue));

	//Do the text

	if( !TextOut( m_dcPaint, 3, (((m_bmpInfo.bmiHeader.biHeight-20)/12) * Linenum)-20, szText,_tcslen( szText ) ) )

		return ;

	//Copy sample back to original buffer

	CopyMemory(pBuffer,m_pPaintBuffer,CSAMP);

}

void CAutoDC::MakeMediaType()

{

	ZeroMemory(pMediaType, sizeof(CMediaType));

		VIDEOINFO *pvi = (VIDEOINFO *)pMediaType->AllocFormatBuffer(sizeof(VIDEOINFO));

		if (NULL == pvi) 

			return ;//E_OUTOFMEMORY;



		ZeroMemory(pvi, sizeof(VIDEOINFO));



		pvi->bmiHeader.biCompression	= BI_RGB;

		pvi->bmiHeader.biBitCount		= 32;

		pvi->bmiHeader.biSize			= sizeof(BITMAPINFOHEADER);

		pvi->bmiHeader.biWidth			= 320;

		pvi->bmiHeader.biHeight			= 240;

		pvi->bmiHeader.biPlanes			= 1;

		pvi->bmiHeader.biSizeImage		= GetBitmapSize(&pvi->bmiHeader);

		pvi->bmiHeader.biClrImportant	= 0;



		SetRectEmpty(&(pvi->rcSource));	// we want the whole image area rendered.

		SetRectEmpty(&(pvi->rcTarget));	// no particular destination rectangle



		pMediaType->SetType(&MEDIATYPE_Video);

		pMediaType->SetFormatType(&FORMAT_VideoInfo);

		pMediaType->SetTemporalCompression(FALSE);



		const GUID SubTypeGUID = GetBitmapSubtype(&pvi->bmiHeader);

		pMediaType->SetSubtype(&MEDIASUBTYPE_RGB32);

		pMediaType->SetSampleSize(pvi->bmiHeader.biSizeImage);



		m_bmpInfo.bmiHeader = pvi->bmiHeader;

		m_bmpInfo2.bmiHeader = pvi->bmiHeader;



	return ;

}

void CAutoDC::MakeDC()

{

	HBITMAP hDibSection = CreateDIBSection(NULL, (BITMAPINFO *) &m_bmpInfo, DIB_RGB_COLORS,&m_pPaintBuffer, NULL, 0);

	HBITMAP hDibSection2 = CreateDIBSection(NULL, (BITMAPINFO *) &m_bmpInfo2, DIB_RGB_COLORS,&m_pPaintBuffer2, NULL, 0);

	

	HDC hDC = GetDC(NULL);

	

	m_dcPaint = CreateCompatibleDC(hDC);

	m_dcPaint2 = CreateCompatibleDC(hDC);



	SetMapMode(m_dcPaint, GetMapMode(hDC));	

	SetMapMode(m_dcPaint2, GetMapMode(hDC));	

	

	HGDIOBJ OldObject = SelectObject(m_dcPaint,hDibSection);

	HGDIOBJ OldObject2 = SelectObject(m_dcPaint2,hDibSection2);

}

CAutoDC::CAutoDC()

{

CAutoLock lock(&cSharedState);

alpha=0;

alpha2=0;

pMediaType= new CMediaType;

this->MakeMediaType ();

this->MakeDC ();

t2Color.rgbBlue =t2Color.rgbGreen =t2Color.rgbRed =255;

tColor.rgbBlue =255;

tColor.rgbGreen =255;

tColor.rgbRed=255;

tColor.rgbReserved =0;

t2Color.rgbReserved =0;

}

CAutoDC::~CAutoDC()

{

DeleteDC(m_dcPaint);

DeleteDC(m_dcPaint2);



}

[oLRi]

Hmm..

And how do i parse a bmp file from filesystem to a rgbquad array?

[Smelter]

code:


#pragma once

#include"stdafx.h"





class CBitmapHandler

{

private:

	

	char * buffer;

public:

	int width,height,depth;

	CBitmapHandler(void)

	{

		buffer=NULL;

		depth=32;

		

	}



	int getSize()

	{

		return width*height*(depth/8);

	}

	char * getBits()

	{

		return buffer;

	}

	int readBitmap(char * path)

	{

		 HANDLE   m_hFile = CreateFile(path, GENERIC_READ, 0, NULL, OPEN_EXISTING, 

                         FILE_ATTRIBUTE_NORMAL, NULL);

	

	    if (m_hFile == INVALID_HANDLE_VALUE)

			return -1;



		DWORD dwFileSize = GetFileSize(m_hFile, NULL);

		if (dwFileSize == INVALID_FILE_SIZE)

			return -1;

		BYTE* m_pFile = new BYTE[dwFileSize];

		 DWORD nBytesRead = 0;

		if(!ReadFile(m_hFile, m_pFile, dwFileSize, &nBytesRead, NULL))

		{

			return -1;

		}

		int cbFileHeader = sizeof(BITMAPFILEHEADER);

		BITMAPFILEHEADER *pBm = (BITMAPFILEHEADER*)m_pFile;



		int m_cbBitmapInfo = pBm->bfOffBits - cbFileHeader;



		BITMAPINFO* m_pBmi = (BITMAPINFO*)(m_pFile + cbFileHeader);

		height=m_pBmi->bmiHeader.biHeight;

		width=m_pBmi->bmiHeader.biWidth;

		depth=m_pBmi->bmiHeader.biBitCount;

		



		if(buffer)

			delete[] buffer;



		buffer=new char[height * width * 4];

		

		char * dataSrc=(char *) m_pFile + cbFileHeader + m_cbBitmapInfo;

		char * dataDst=(char *) buffer;





		for(int kk=0;kk< width * height ; kk++)

		{

		

			

			if(depth==32)

			{

				*dataDst++=*dataSrc++;

				*dataDst++=*dataSrc++;

				*dataDst++=*dataSrc++;

				*dataDst++=*dataSrc++;

			}

			else if(depth==24)

			{

				*dataDst++=*dataSrc++;

				*dataDst++=*dataSrc++;

				*dataDst++=*dataSrc++;

				*dataDst++=0xFF;

			}else

			{

				return -1;

			}







		}



		CloseHandle(m_hFile);



		m_hFile = INVALID_HANDLE_VALUE;





		return 1;



	}



	void makeBitmap(char * name, int Width,int Height,int bitCount,void * pBuffer,int BufferSize)

	{

		TCHAR szFilename[MAX_PATH];

		wsprintf(szFilename, TEXT("%s\0"),name);



		HANDLE hf = CreateFile(szFilename, GENERIC_WRITE, FILE_SHARE_READ, 

                               NULL, CREATE_ALWAYS, NULL, NULL );



		BITMAPFILEHEADER bfh;

		memset( &bfh, 0, sizeof( bfh ) );

        bfh.bfType = 'MB';

        bfh.bfSize = sizeof( bfh ) + BufferSize + sizeof( BITMAPINFOHEADER );

        bfh.bfOffBits = sizeof( BITMAPINFOHEADER ) + sizeof( BITMAPFILEHEADER );



        DWORD Written = 0;

        WriteFile( hf, &bfh, sizeof( bfh ), &Written, NULL );



        BITMAPINFOHEADER bih;

        memset( &bih, 0, sizeof( bih ) );

        bih.biSize = sizeof( bih );

        bih.biWidth = Width;

        bih.biHeight = Height;

        bih.biPlanes = 1;

        bih.biBitCount = bitCount;

		bih.biSizeImage=Width*Height*bitCount/8;

		bih.biCompression=0;

		bih.biClrImportant=0;

		

		Written = 0;

        WriteFile( hf, &bih, sizeof( bih ), &Written, NULL );



        // Write the bitmap bits

        //

        Written = 0;

        WriteFile( hf, pBuffer, BufferSize, &Written, NULL );



        CloseHandle( hf );











	}



	~CBitmapHandler(void)

	{

		if(buffer)

			delete[] buffer;

	}

};

[oLRi]

Hmm

I've made a test file, i opens 2 bmp's and gets the bits. this works, but if i use your alphablend function my application crashes...

here's the code:
http://pastebin.com/VZrLhVm0

i think it has something todo with the pointers, i'm very new to c

[Smelter]

probably not the same size as the movie.

[oLRi]

Originally Posted by Smelter

The easy way is to load the rgb bits of another and blend them.

code:


d.rgbRed   = (BYTE) (pSource->rgbRed * pSource->rgbReserved / 256 + 

			pCheckers->rgbRed  * ( 256 - pSource->rgbReserved ) / 256);

According to:
http://msdn.microsoft.com/en-us/libr...=vs.85%29.aspx

"rgbReserved" can only be 0? But this makes no sense because it would ALWAYS mean:

(pSource->rgbRed * 0 / 256 + pCheckers->rgbRed * ( 256 - 0 ) / 256);
which would be:
(0 / 256 + pCheckers->rgbRed * 256 / 256);
which is:
(0 + pCheckers->rgbRed);
which is:
(pCheckers->rgbRed);

[Smelter]

yes it is zero typically where the media type is 24 bits per pixel, and the whole frame work supports systems that can only access a full word. At the pixel level it is the 'quad'. at the pixel row level it is the 'image stride'.

'reserved' was 'reserved' for any 32bit pixel type even where the reserved byte data does not translate to visual 'alpha'

The 'reserved' byte could hold any data related to the pixel.

The 'must be zero' key 'must be zero' where you are only sending 24bits of color to a machine that it is unknown what Endian it is. Your RGBQuads are handled in the same system, and only to shuffle the pointed values. Clearly your render pipeline requests ARGB pixels, which are 32bits packed. You can create your own class handler to handle the pointers.

You should also look at the 32bit alpha media type from the directshow samples.

[Smelter]

If a PC sends 24bit pixels to a MAC, the 32bit word must have the reserved byte as zero so the MAC can determine what 8 bits is blue, then swap.

'must be zero' would not apply to 32 bit alpha image, and really only apply in an windows function that explicitly takes a RGBQuad as a parameter, and even then, 'must be zero' I think was auto inserted any time a developer left a parameter undocumented , as in the case of future media types.

[MrX_1980]

Hello

The VP8 v0.9.7 "Cayuga" is out since about one month ago (2011-08-02).

The changelog sounds nice:
http://www.webmproject.org/tools/vp8-sdk/changelog.html
http://blog.webmproject.org/2011/08/...-released.html

I hope we will see with this release more VP8 Shoutcast TV streams if Winamp supports it.

[Jeeper One]

Your site seems to be under maintenance Smelter but this has me intrigued now.....

In fact, I just may take those videos that can't be streamed elsewhere (You know THOSE videos ) & put them on SHOUTcast TV (If only AOHell would re-add the SHOUTcast TV YP > )

Cheers

[pythy]

I have compiled a new "nsv_coder_vp8.dll" based on VP8 v0.9.7 "Cayuga". You can download it under http://bommels-technik-welt.de.
It works very fine on my NSV-Tools.

Have fun and enjoy it

cu Pythy

[sankt]

Nice Job Pithy !

[sankt]

It is a shame that NSV encoding cant be added to VLC. That would take care of any update for NSV tools.

There is actually an NSV module that some one created a few years ago, I still wonder why it could not be added to VLC.

[rockouthippie]

Shoutcast needs to stream webm. The best current tool for making webm (VP8/vorbis) files is ffmpeg. FFmpeg is now suitable for making webm in a production environment. It is about half speed vs. h264 in the encode direction. That's a lot better than it was last year. Mostly it's not really well threaded.

VLC (and I think it's almost on purpose) is a terrible encoder.

[MrX_1980]

VP8 Codec SDK v1.0.0 "Duclair" Released (Friday, January 27, 2012)

"performance improvements up to ~10,5%" :-)

Update:
libvpx 1.0.0 Duclair: VP8 Video Encoding Benchmarked

[sankt]

Good catch !

Dirac has updated their encoder as well.

http://diracvideo.org/2012/01/schroedinger-1-0-11/

[MrX_1980]

VP8 Codec SDK (libvpx) before 1.0.0 "Duclair" allows remote attackers to cause a denial of service (application crash)...
http://web.nvd.nist.gov/view/vuln/de...=CVE-2012-0823

[pumbaa2]

Indeed, I've been using VP62. I just installed VP8 and everything so far looks good. I will find out shortly about the Quality. A copy of the VP8 Codec along with Install Instructions has been posted to my files area on my website. The German Webpage was a little painful to look at

[pumbaa2]

Originally Posted by pythy

I have compiled a new "nsv_coder_vp8.dll" based on VP8 v0.9.7 "Cayuga". You can download it under http://bommels-technik-welt.de.
It works very fine on my NSV-Tools.

Have fun and enjoy it

cu Pythy

The codec settings field is a little to big and partially covers the Okay Button. Infact, on my version of NSVTools, the Okay button is invisible but right below the "Logo" in the Output Format Settings Dialog inside FlaskMPEG for NSV. Is there a fix for this?

[pumbaa2]

Ohh also, I noticed that whatever bitrate I use seems to be "halfed" when its done. I can set it to 256kb/sec but when I play it back in Winamp 5 it reports closer to 128. Ideas?

[slavas]

probably You encoding at 15 FPS, and wrapper didnt set correctly FPS for encoder.
Sample Xvid coder didnt handle FPS neither marked keyframes IIRC

[pumbaa2]

Hahhaha, okay... Yeah, your right.. My FPS was set at 15, that does explain a few things.

[slavas]

yeah just tested myself, that's the case