You can also use the tools in the Windows Automated Installation Kit (AIK) to create a custom boot image. For more information, see Windows PE Walkthroughs (http://go.microsoft.com/fwlink/?LinkId=120691). When creating boot images, ensure that the image is stored in .wim format, and it is marked as bootable from RAMDISK using the /boot option with ImageX. To create custom boot image, you must manually copy the Windows Vista or Windows Server 2008 Setup.exe binary files into the custom image. The process for doing this is as follows:

1. Create a top-level folder named Sources in the custom boot image.
   
   
2. Mount the boot image that is RAMDISK-bootable in the Boot.wim file (Boot.wim contains two Windows PE images, and the bootable image is the second image).
   
   
3. Copy all of the Setup.exe files from the \Sources folder in the mounted image to the \Sources folder in the custom boot image.
   
   

In addition, you must ensure that the Windows Deployment Services client is started by Windows PE. To ensure that this occurs, create an entry in the WinPESHL.ini file to start Setup.exe in Windows Deployment Services mode. For more information, see “When Setup Is Started in Windows Deployment Services Mode” in How the Windows Deployment Services Client Works (http://go.microsoft.com/fwlink/?LinkID=147067). For more information about editing WinPESHL.ini, see Include a Custom Script in a Windows PE Image (http://go.microsoft.com/fwlink/?LinkId=120692). For information about managing and modifying the boot menu, see Managing the Boot Menu.

When creating custom boot images, the version of Windows PE that you use must match or be newer than the install image. For example, you can use Windows PE 2.1 to deploy install images of all versions of Windows (including Windows Vista with SP1, Windows Server 2008, and all earlier versions of Windows). You cannot, however, use Windows PE 2.0 to deploy Windows Vista with SP1 or Windows Server 2008.

If you are deploying Windows Server 2003 and your boot image does not contain the Windows Deployment Services client (for example, if you are booting a Windows PE 2005 boot image into the command prompt instead of into the user interface screens of the Windows Deployment Services client), we recommend that you use the latest version of Windows PE. You can use Windows PE 2004, Windows PE 2005, Windows PE 2.0, or Windows PE 2.1, although the following caveats apply:

• If you are applying a .wim image of Windows Server 2003 using ImageX, you can use either the x86 or x64 version of Windows PE.
  
  
• In the past, if you were running Winnt32.exe for Setup.exe, you could only use x64 versions of Windows PE, but that has changed. For details, see Knowledge Base article 931761 (http://go.microsoft.com/fwlink/?LinkID=110354).
  
  
• If you are using Windows PE 2.0, you might run into the issue documented at http://go.microsoft.com/fwlink/?LinkID=110354. In this scenario, run the command Bootsect.exe /nt52 c: to set up the correct NTFS file system boot sector.
  
  

For advanced scenarios if you want to create a custom deployment, you can create a discover image by using the tools provided in the Windows AIK. For more information, see Windows PE Walkthroughs (http://go.microsoft.com/fwlink/?LinkId=120691).

To create a discover image manually

1. In the boot image, create a temporary folder in the path pointed to by the environment variable, %TEMP%.

2. Copy the appropriate Setup.exe files from the \Sources folder in the boot image located on the Windows Server 2008 DVD (it is the second image in the Boot.wim file) to the temporary folder.

3. Create a Winpeshl.ini file in the Windows\System32 folder of the applied image with the following section. For more information about editing WinPESHL.ini, see Include a Custom Script in a Windows PE Image (http://go.microsoft.com/fwlink/?LinkId=120692).

   [LaunchApps]
   %SYSTEMROOT%\sources\setup.exe, "/wds /wdsdiscover"
   

   or

   [LaunchApps]
   %SYSTEMROOT%\sources\setup.exe, "/wds /wdsdiscover /wdsserver:<server>"
   

   See the following table for more information about these options.

4. Capture the modified image into a new .wim file.

5. Update the image metadata to reflect any changes to the image name or description.

There are two command-line options for Setup.exe that control the discovery behavior of the Windows Deployment Services client. These options are described in the following table.

Capture images provide a subset of the functionality included in the ImageX /capture command. For information, see ImageX Technical Reference (http://go.microsoft.com/fwlink/?LinkId=120693). The following table compares these two tools.

In advanced scenarios, you can create a custom capture image.

To create a capture image manually

1. Create a temporary folder in the path that is pointed to by the environment variable, %TEMP%.

2. Apply the contents of the source boot image from the Windows Deployment Services server’s image store to the \Temp folder.

3. Create a Winpeshl.ini file in the Windows\System32 folder of the applied image with the following section:

   [LaunchApps]
   %SYSTEMROOT%\system32\wdscapture.exe
   

4. Capture the modified image into a new .wim file.

5. Update the image metadata to reflect any changes to the image name or description.

To convert your files, right-click the RIPREP image you want to convert in the Legacy Images node, and then click Convert to WIM. The default conversion process copies the updated version of a file to another location. There were two main factors that influenced this design decision:

• The original image remains unmodified, in case the conversion process fails or you want to continue to use the original RIPREP image after the conversion process is run.
  
  
• In RIS, you could associate multiple unattended setup installation files (.sif) with a particular image. If an image is so configured, a conversion process that is run for one .sif file would alter the backing files used by the other .sif file.
  
  

The conversion process requires at least twice as much free disk space as the size of the image. This space is needed for a copy of the RIPREP image placed in the %TEMP% folder and the .wim file that was created by using the contents of the converted image in the %TEMP% folder. Note that the data in the %TEMP% folder can be removed only after the new image has been captured.

You can force an in-place conversion of a RIPREP image, which will save time and the amount of disk space that you use during the conversion process. You can do this by using the /InPlace option with the WDSUTIL /Convert-RiprepImage command. It is common for multiple variations of a single RIPREP image (differing only by HAL type) to exist on a server. You can save time during the conversion process by using the /Overwrite:Append option of the WDSUTIL /Convert-RiprepImage command to take advantage of single-instancing technology within the .wim format. For instructions, see the "Install Images" section in How to Manage Images.

The append operation is much faster than a traditional capture because it does not need to compress and insert files that already exist in the .wim file. Files that are identical between images and that already exist within the .wim file will simply have their reference count incremented to indicate that the single file belongs to multiple images within the .wim file. The general conversion process entails first converting the first RIPREP image in the set by creating a new .wim file, and then converting the remaining RIPREP images (for the other HAL types) by appending them to the .wim file you created previously.