Not able to create mask for the image

[amey_kulkarni]

Hi All,
I am developing an application where i need to create the mask of the original image.
My image is just blue(filled)square surrounded by magenta color.
Now i want to create mask of this image which should be that blue square has to be white and rest should be black.
here is my code.


RFs session;
session.Connect();
_LIT(KImageFile,"c:\\sample.bmp");

iDecoder= CImageDecoder::FileNewL(session,KImageFile,CImageDecoder::EAllowGeneratedMask,KNullUid,KNullUid,KNullUid);
session.Close();
//Get the image attributes
TFrameInfo FrameInfo = iDecoder->FrameInfo(0);

Image->Create(FrameInfo.iOverallSizeInPixels,FrameInfo.iFrameDisplayMode );
Imagemask->Create(FrameInfo.iOverallSizeInPixels,EGray256);
//TRequestStatus Status = KRequestPending;
iStatus=KRequestPending;
//Decode the bitmap from the file to symbian type bitmap
iDecoder->Convert(&iStatus,*Image,*Imagemask,0);
if(!IsActive())
SetActive();


After this i am not getting a proper mask.
so if i bitblt my image using bitbltmasked..i get same image displayed.
and if i bitblt only mask...only white background i can see.

please tell me the problem with my code.

[A.A.M.]

See this article - How to read images to Symbian bitmap

And try to replace

RFs session;
session.Connect();
iDecoder= CImageDecoder::FileNewL(session,KImageFile,CImageDecoder::EAllowGeneratedMask,KNullUid,KNullUid,KNullUid);
session.Close();

with

...
{
 TBuf8<255> ImageType;
 GetFileType(KImageFile, ImageType);

 iDecoder = CImageDecoder::FileNewL(CCoeEnv::Static()->FsSession(),KImageFile,ImageType);
 ...
 ...
 ...
}

void C***::GetFileType(const TDesC& aFileName, TDes8& aFileType)
{
	TEntry FileEntry;
 
	if(CCoeEnv::Static()->FsSession().Entry(aFileName,FileEntry) == KErrNone)
	{
		TBuf8<255> FileBuffer;
		
		if(!FileEntry.IsDir())
		{
			TInt FileSize = FileEntry.iSize;
 
			if(FileSize > 255)
			{
				FileSize = 255;
			}
			
			if(CCoeEnv::Static()->FsSession().ReadFileSection(aFileName,0,FileBuffer,FileSize) == KErrNone)
			{
				RApaLsSession RSession;
				if(RSession.Connect() == KErrNone)
				{	
					TDataRecognitionResult FileDataType;
 
					RSession.RecognizeData(aFileName,FileBuffer,*&FileDataType);
					
					if(FileDataType.iConfidence > aResult.iConfidence)
					{
						aFileType.Copy(FileDataType.iDataType.Des8());
					}
					
					RSession.Close();
				}
			}
		}
	}
}

[wizard_hu_]

.bmp file format does not support "contained" mask, that is why CImageDecoder does not decode a mask for you. If you decide that magenta color indicates transparent pixels, that is your personal decision, obviously not part of the file format itself.
So you have to create the mask for yourself, TBitmapUtil can help you in doing that, it enables you to read pixels from the image, then compare the color to your color key, and write a black/white pixel to the mask accordingly.

[amey_kulkarni]

Thanks Wizard_hu!!

But creating a mask using TBitmapUtil is a very slow process as you have to compare each pixel. so on device it is going to take lots of time. we can create the mask for .png if it contains alphs bending information.
So do i have to change all my .bmp to .png format?

[skumar_rao]

Thanks Wizard_hu!!

But creating a mask using TBitmapUtil is a very slow process as you have to compare each pixel. so on device it is going to take lots of time. we can create the mask for .png if it contains alphs bending information.
So do i have to change all my .bmp to .png format?

why not change all images to MBM format where you package both image and it's mask.

[amey_kulkarni]

I can not change images to mbm format. Because application takes dynamic calls so depending on this you need to decode the image and display it on the screen.There are number of images so not feasible to create mbm.

[deepchand86]

I can not change images to mbm format. Because application takes dynamic calls so depending on this you need to decode the image and display it on the screen.There are number of images so not feasible to create mbm.

There is no other go.. you have to convert the images to .mbm for display....
May be You cud better go for decoding of .png files instead of .bmp files...
I am using/decoding .png files at run-time along with the mask and it's not too long.
Also .png's consume less space than .bmp...

[wizard_hu_]

I do not see the problem. If you have access to the original images (so you are do not downloading them in runtime for example), it is pretty straightforward to convert them to .mbm. Also note that a .mbm file can contain a significant number of bitmaps (just search for avkon.mbm with Google for a "famous" example). And .mbm is a lossless compressed format actually.
Check how Graphics example works (part of all S60 1st and 2nd edition SDK-s as S[eries]60Ex\Graphics, and re-introduced in S60 3rd edition FP1 - somewhere in S60Ex or S60CppExamples). And there is a downloadable example too, Docs above, Symbian C++, Graphics, Scalable Screen Drawing (or something like that).

[amey_kulkarni]

Thanks, Wizard_Hu_.
I am just porting the code which was developed such that for displaying image it takes the local path of iamge on device and decodes it and displays it. So i dont have access to change the behaviour of this code so i have no alternative except using decoder and thats where it is creating the problem of mask.
So my original images should contain the information so that i can create the mask successfully.

But still thanks friends for the guidance.