Question

I’ve seen a few questions regarding multipage tiffs and a few questions about compression, but none (that I’ve seen) linking the two. This question is as close as I’ve seen and puts me incredibly close, so I hope. I went into the Oracle forum thread mentioned (it is talking about a multipage PDF to TIFF with compression) and I think I’m closish to having the code finished to do this. Can anyone assist? I’m going to remove the try/catches to try and shorten this down a hair (basically all they did was output a message in console and return false).

 public static boolean CompressedTiff(List<BufferedImage> images, File path)
 {
    if (!path.getParentFile().exists())
         path.getParentFile().mkdirs();
    path.createNewFile();
    ImageOutputStream ios;
         ios = ImageIO.createImageOutputStream(path);

    Iterator<ImageWriter> imageWriters = ImageIO.getImageWritersByFormatName("TIFF");
    ImageWriter writer = (ImageWriter)imageWriters.next();
    writer.setOutput(ios);
    TIFFImageWriteParam writeParam = (TIFFImageWriteParam)writer.getDefaultWriteParam();
    writeParam.setCompressionMode(2);
    writeParam.setCompressionType("LZW"); 
    writer.prepareWriteSequence(null);

    for(int i = 0; i < images.size(); i++)
    {
        ImageTypeSpecifier spec = ImageTypeSpecifier.createFromRenderedImage(images.get(i));
        javax.imageio.metadata.IIOMetadata metadata = writer.getDefaultImageMetadata(spec, writeParam);
        IIOImage iioImage = new IIOImage(images.get(i), null, metadata);
        writer.writeToSequence(iioImage, writeParam);
        images.get(i).flush();//modified after release.

        images.get(i).flush();
        writer.endWriteSequence();
        ios.flush();
        writer.dispose();
        ios.close();
    }
    return true;

}

It failed out on the next pass at writer.writeToSequence saying I needed to call prepareWriteSequence. I changed it to

 writer.prepareWriteSequence(metadata);
 writer.writeToSequence(iioImage, writeParam);

also removed the earlier writer.prepareWriteSequence(null);

and it appears to be navigating the files properly, however, the output isn’t of any type of renderable tif. Multipage or otherwise.

I have JAI installed so if it is possible to use that in some way to achieve the compressed image, that’d be fantastic. The code I am using that generates a TIFF is using this, but I haven’t seen anything that worked as far as adding compression to the pages.

edit: I added a bunch of ios.flush(); ios.close(); calls in the catch blocks and it prevents the non-renderable TIFF issue. However, it isn’t adding any page beyond the first.

Answer 1

If it helps, this is code that I use for modifying a TiffImageWriteParam to set the compression:

try {
    jWriteParam.setCompressionMode(_compression != TiffCompression.NO_COMPRESSION 
                  ? ImageWriteParam.MODE_EXPLICIT : ImageWriteParam.MODE_DISABLED);

    if (_compression != TiffCompression.NO_COMPRESSION) {
        // this code corrects the compression if, say, the client code asked for
        // CCITT but the actual image pixel format was CMYK or some other non-1 bit
        // image type.
        TiffCompression mode = recastToValidCompression(_compression, pf);
        jWriteParam.setCompressionType(getCompressionType(mode));
        TIFFCompressor compressor = getTiffCompressor(mode, jWriteParam, shouldUsePredictor(pf));
        jWriteParam.setTIFFCompressor(compressor);
        if (_compression == TiffCompression.JPEG_COMPRESSION) {
            // Java supports setting to 1.0 (ie 100), but it will not actually do lossless (maybe)
            if (_jpegQuality == 100 && !jWriteParam.isCompressionLossless())
                continue;
            jWriteParam.setCompressionQuality(toJavaJpegQuality());
        }
    }
}
catch (UnsupportedOperationException e)
{
    // this shouldn't get here, but you should consider what to do if it does.
    // set a default? throw?
}

here is getTiffCompressor():

private TIFFCompressor getTiffCompressor (TiffCompression compression, TIFFImageWriteParam writeParam, boolean usePredictor)
{
    int predictor = usePredictor 
            ? BaselineTIFFTagSet.PREDICTOR_HORIZONTAL_DIFFERENCING 
            : BaselineTIFFTagSet.PREDICTOR_NONE;

    switch (compression) 
    {
    case GROUP_3_FAX_ENCODING:
        return new TIFFT4Compressor();
    case GROUP_4_FAX_ENCODING:
        return new TIFFT6Compressor();
    case JPEG_COMPRESSION:
        return new TIFFJPEGCompressor(writeParam);
    case MACINTOSH_PACKBITS:
        return new TIFFPackBitsCompressor();
    case DEFLATE:
        return new TIFFDeflateCompressor(writeParam, predictor);
    case LZW:
        return new TIFFLZWCompressor(predictor);
    case MODIFIED_HUFFMAN:
        return new TIFFRLECompressor();
    case NO_COMPRESSION:
    case DEFAULT:
    default:
        return null;
    }
}

TiffCompression is my own enum that models the compressions that I offer for TIFF files. Finally, here is getCompressionType():

private String getCompressionType (TiffCompression compression)
{
    switch (compression)
    {
    case GROUP_3_FAX_ENCODING:
        return "CCITT T.4";
    case GROUP_4_FAX_ENCODING:
        return "CCITT T.6";
    case JPEG_COMPRESSION:
        return "JPEG";
    case MACINTOSH_PACKBITS:
        return "PackBits";
    case DEFLATE:
        return "Deflate";
    case LZW:
        return "LZW";
    case MODIFIED_HUFFMAN:
        return "CCITT RLE";
    case NO_COMPRESSION:
    case DEFAULT:
    default:
        return null;
    }
}

Now, I can’t show you everything because my code is built to encode an arbitrary number of images and yours is not, so our code structures differ wildly. In my case, I set up the encoder for using a sequence writer with a much more open architecture. I pull in an image, fire an event to optionally change the default compression, create the writer and write param, set metadata/image tags, fire progress events then write the sequence. Then I have to go in and patch the last written ifd because the tiff encoder writes them broken, so they need to be patched.