How do I save an 8 bit grayscale jpeg with CImage

When I remove the SetColorTable call the image turns out solid black.

This is a grey scale image, so why are you setting three consecutive bytes inside the inner loop? I should think this would mess up the bmp and png files. You should just set one at a time. Or are you copying from a color image? (What is imgBits?) If copying from a color image, you need to do some conversion from color to grey scale -- I'm not sure what the conversion should be.

Also, you seem to be using unsigned char *, char *, and LPBYTE interchangeably. Be careful about using char * in case it is doing some conversion.

The RGBQUAD that I pass to the color table makes the grayscale.

You might try something like the following code. I didn't test this, but think it should be close to workable.

for ( int y=0; y<pdib->biHeight; y++ )
{
  for ( int x=0; x<pdib->biWidth; x++ )
  {
    BYTE red = imgBits [(pdib->biHeight-y-1)*pdib->biWidth*iMultBitCountFactor+x*iMultBitCountFactor+0];
    BYTE grn = imgBits [(pdib->biHeight-y-1)*pdib->biWidth*iMultBitCountFactor+x*iMultBitCountFactor+1];
    BYTE blu = imgBits [(pdib->biHeight-y-1)*pdib->biWidth*iMultBitCountFactor+x*iMultBitCountFactor+2];
 
    BYTE grey = (BYTE)(((int)red + (int)grn + (int)blu) / (int)3);
 
    pbitThat = (char *) (LPBYTE(pImgBits) + (y * iPitch ) + x*iMultBitCountFactor8+0);
    *pbitThat = grey;
  }
}
 
                                              

1:
2:
3:
4:
5:
6:
7:
8:
9:
10:
11:
12:
13:
14:
15:

Your code sample compiled perfectly, however without the imagePtr->SetColorTable(0,256,clrs); call the image was solid black and still 24 bit, and with the call it was grayscale but still 24 bit.  I attached the jpg that was produced in case it may help reveal a solution.

The only application I found so far to actually produce an 8-bit jpeg is gimp, if you open the image then grayscale it and save it again in gimp it produces an 8-bit grayscale jpeg, according to its windows properties.  This is the format I need my jpegs to be, the gimp version of the image is attached.

It appears that CImage does not provide a means to specify the type of encoding, but if you start with a 8-bpp indexed grayscale image, it will output a jpg that is in that format.    See:
http://www.dotnetmonster.com/Uwe/Forum.aspx/dotnet-drawing/3644/how-to-save-grayscale-image
Incidently, JPG is a "lossy" format.  This shows how to control the quality of of the JPG output:
http://www.bobpowell.net/jpeg_compression.htm

"It appears that CImage does not provide a means to specify the type of encoding, but if you start with a 8-bpp indexed grayscale image, it will output a jpg that is in that format.    See:
http://www.dotnetmonster.com/Uwe/Forum.aspx/dotnet-drawing/3644/how-to-save-grayscale-image "
-This currently isn't a problem for me, in this post he wants to atleast make an 8-bit bitmap, my code when saving in BMP format produces an 8-bit BMP fine, or PNG, or GIF, just not JPEG.

"Incidently, JPG is a "lossy" format.  This shows how to control the quality of of the JPG output:
http://www.bobpowell.net/jpeg_compression.htm "
-JPEG compression for me is fine, only bit-depth is a problem

Also, I use the following calls for CImage:

imagePtr->Create(pdib->biWidth,pdib->biHeight, 8); //8 specifies the bit-depth

result = imgPtr->Save("C:\\test_JPEG.jpg", Gdiplus::ImageFormatJPEG);
 
result = imgPtr->Save("C:\\test_BMP.BMP", Gdiplus::ImageFormatBMP);
 
result = imgPtr->Save("C:\\test_PNG.PNG", Gdiplus::ImageFormatPNG);

...so CImage DOES provide a means to specify the type of encoding

I meant to say that ("does not provide a means to specify that kind of encoding"); that is, 256-color jpeg-compressed format.   It seems to me that if you use 256 colors (or fewer) even when starting with a 24-bit jpeg, then the result will be the same.

I suppose that one thing you could try would be to blit it to a 256-color BMP, then try to sav it as .JPG.

Where did you get the idea that you needed 8bpp JPEG?  It's not a normal format.

See: http://www.tech-archive.net/Archive/Development/microsoft.public.win32.programmer.gdi/2007-01/msg00087.html

"Keep in mind that gdi+'s .jpeg encoder will not create a 8bpp .jpeg. It will create a 24bpp
..jpeg image. "

"Keep in mind that
gdi+'s .jpeg encoder will not create a 8bpp .jpeg. It will create a 24bpp
..jpeg image."

-So who's jpeg encoder does create an 8bpp .jpeg?

I've never tryied it, but you could look into setting the Encoder.ColorDepth  as described (with C# example code) here:
  http://msdn.microsoft.com/en-us/library/system.drawing.imaging.encoder.colordepth.aspx

They show its use with TIFF format, but the JPEG encoder might also accept that parameter.

Found an open source encoder/decoder, multi-platform, with code examples which can decode/encode 8-Bit Grayscale JPEGS, called TinyJpeg:
http://www.stud.fit.vutbr.cz/~xpolok00/proj_tinyjpeg.htm.en.iso-8859-2