1. unbuffered_camera_software_release_4download.zip Driver
2. TwainSpec.txt File Contents

TwainSpec.txt Driver File Contents (unbuffered_camera_software_release_4download.zip)

One of my applications needed some scanner support.  I thought it might be a good idea
to get into TWAIN and try it out. Well, here are the results.


Firstly, this is NOT a complete implementation of the TWAIN specification. I have followed 
version  1.8 and have used just the basic  functionality to get an image from a scanner, 
actually , multiple images too can be acquired one after another. By the way - it will work 
with didigtal cameras too - anything which exports a TWAIN interface is supported.


Well, lets delve a little further into it. Before we do that though, I would have to mention
that I will not be able to tell you all about TWAIN here. You can download the specification
at http://www.twain.org. They also have a sample Twain source, if you do not have a scanner 
and wish to try out TWAIN.

For our purposes here, the important things are the Data Source Manager (DSM) and the Data 
Source (DS) itself. The DS is the actual scanner or digital camera or any other source 
which implements TWAIN. The DSM is the module that provides us an interface to the DSM.
And yes, this is a very simplistic view of TWAIN ( I shudder to think of what the TWAIN
designers would make of my description ).

Now for the class itself. First, the header Twain.h has to be included. I've called my
class, rather unimaginatively CTwain. Now this class requires the TWAIN_32.DLL module
to be loaded. The way I have handled it is by keeping a static Module handle and incrementing
and decrementing the reference count. When the interface is released for the last time,
the module is unloaded.

Mostly, the class is quite independant of everything else. What it does require though,
is a handle to window to which the DSM can send messages. These messages are not meant
for the application itself but for Twain implementor, in this case CTwain. This window
handle can be passed either to the constructor, or if as is more likely, if the window isn`t
ready yet, in a call to InitTwain.

The CTwain class is an abstract class as it has one pure virtual method. So to use this
class you will have to derive from CTwain.

I will now explain the important methods - the ones which will let you start scanning .

<bold>
CTwain(HWND hWnd= NULL) -- Constructor
</bold>
	The hwnd is optional. If a non Null value is given, the Twain interface is initialized
	using this handle.

<bold>
~CTwain() -- Destructor
</bold>

<bold>
InitTwain(HWND hWnd) -- Initializes TWAIN
</bold>
	This is called by the constructor if the handle passed to the constructor is not null.
	Else, it can be called later . Loads the Twain Dll and initializes it.

<bold>
ReleaseTwain() --  Releases Twain interface
</bold>
	NOTE : If TWAIN has been initialized, this must be called before the window handle passed
	to it is destroyed. Not doing so will result in resource and memory leaks.

<bold>
GetIdentity()  --  Identitifies application
</bold>
	This is called by InitTwain to initialize the TW_IDENTITY structure. Please refer to 
	twain.h for the structure memnbers and to CTwain::GetIdentity for an example as to 
	how to fill these members. As for now, you need not implement this as long as the
	default behaviour suits you. 
 
<bold>
IsValidDriver()  -- Returns true if the Driver was loaded successfully
</bold>

<bold>
SourceSelected() -- Returns true if a source was selected successfully
</bold>

<bold>
SourceEnabled()  -- Returns true if a source is enabled
</bold>
	In TWAIN parlance, a source being enabled means a scan is in progress.

<bold>
SelectDefaultSource() -- Selects the default source for the current machine
</bold>
	
<bold>
SelectSource() -- Shows the Select Source Dialog box and allows the user to select one
</bold>


<bold>
Acquire(int numImages=1) -- Starts image acquisition
</bold>
	This does not necessarily mean the scanning process is started. All it actually 
	means is that the Source has been enabled and typically a Scanning parameters 
	dialog box has been opened. Scanning typically starts from there.
	numImages is the number of images that the application can handle or
	TWCPP_ANYCOUNT for any number.

<bold>
ProcessMessage(MSG msg) -- Processes messages from Twain
</bold>
	This should be called from the message loop of the window which is intially
	passed to the class. 
	NOTE : All messages can be passed to this routine. It ignores all Non-Twain messages
	and will not act on them unless the source is enabled - so it is not a performance
	botte-neck either.

<bold>
ShouldTransfer(TW_IMAGEINFO& info)
</bold>
	This is called every time an image is to be scanned. It should return one of
	the following values :
		TWCPP_CANCELTHIS	: Cancel this image transfer
		TWCPP_CANCELALL		: Abort all transfers
		TWCPP_DOTRANSFER	: Continue with transfer

	The default implementation returns TWCPP_DOTRANSFER.

<bold>
CopyImage(HANDLE hBitmap,TW_IMAGEINFO& info)
</bold>
	This is a pure virtual method which will get called everytime an image is
	transferred from TWAIN. How the image is to be used is upto the 
	application.


Well - these are the routines you would typically use . You can probably do a lot more
too, but the TWAIN specification can probably help you a lot more there than I can.

Now for the demo application. 

What I have done is used mutiple inheritance with regard to CMainFrame. I figured that would 
be the simplest way to handle things. So CMainFrame is derived from CTwain. InitTwain is 
called from the OnCreate member of CMainFrame. Though the destructor would automatically
be called when the window closes, the window handle would not be valid at that time. So
I call ReleaseTwain from the OnDestroy member function of CMainFrame.

The two additions to the File Menu are 
	Select Source
	Acquire

Select Source shows the default dialog which lists the TWAIN sources available.
Acquire starts the actual acquisition process.

CTwain::ProcessMessage is called from the PreTranslateMessage member of CMainFrame.

Also CMainFrame implements CopyImage. This in turn calls CMainFrame::SetImage which
creates a new document along with a frame and assigns the bitmap scanned to that document.
So as new images are scanned new documents are created.

I guess thats it as far as explanations go. I will now very briefly just go through the
steps required to start scanning .

First , create a class derived from CTwain. Implement CopyImage to handle bitmaps.
Bitmaps are sent as a handle to a Device Independant Bitmap . In the demo app ,
this is handled with a class -CDIB .

Now in the pretranslate member of the window`s class,
insert this line :
		ProcessMessage(*pMsg);

Add the two menu items - Select Source and Acquire as done in CMainFrame in the demo app.

And well - you`re done. Thats it - the only thing you really have to work on is CopyImage.