Hi Saurabh, Your question at first disturbed me, because I thought you were saying that the -int_region option was not always applying the correct region...
Thanks Costa, You are quite right. This bug was introduced in version 5 with the upgrade to arbitrary DWT structures (beyond the simple Mallat structure). I...
HI Yas, I'm just going through all past bug reports to see if there is anything else I need to fix for the new release. Your report mystified me a bit. It...
When opening some potentially non-conforming JPEG2000 files like http://memory.loc.gov/gmd/gmd433/g4332/g4332g/np000131.jp2, kakadu's kdu_show seems to hang up...
HI Mike, Sorry for taking such a long time to respond to this report. I have been going through Kakadu in an attempt to remove any of these (usually quite...
Thanks for the quick response Dr. Taubman. I apologize for the confusion regarding the pixel location vs pixel values. Your reply was really useful. -saurabh...
Does anyone know of or have a utility to do fast validation on a JP2 or similar JPEG2000 file? Ideally this would step through the codestream without actually...
Nice idea, Mike. It would be hard to make it 100% rigorous of course. One thing you could do is to pass the codestream through `kdu_transcode', perhaps...
Hi David, Thank you for your response. Yes, I am trying to execute "kdu_compress" from my own multiple threads. Currently, I want to individually compress...
Hi Scott Can u please clear one of my doubt. What is the use of sample gap and row gap used in the kdu_stripe_compressor file. Please give me an idea about the...
Since reporting this problem I've had a chance to look at the DSTO port of the Kakadu server code. A crude fix for the Kakadu server is attached. The fix...
I am no longer working on this project, but I have forwarded this to the new project manager. I suspected that the "struct pollfd pollData[2];" line was the...
I am not using shared buffering as far as I can tell. I never call the share_buffering() function, nor do I use restart(). The problem does seem to be...
Some additional information for threading crash that may or may not be related. I am getting a crash in kd_precinct_ref::active_deref() when processing...
HI Mike, This is very useful. I did not want to lock the general mutex to access precincts which already exist, since the logic prevents a precinct from going...
Hello, I would like to further investigate JPEG2000 files more by generating a type of index file from either .jp2 or .jpc files. It doesn't appear any of the...
I am using kdu 5.2.6 to compress a raw R8G8B8 image of 7500x7500 pixels(168,750,000 byte in size). I can't get the command line options right for kdu_compress,...
Not quite sure understand your question. But I guess kdu_maketlm is your friend. Here's what I did. I have bunch of images(15000x15000, RGB) used for texture...
... My use case is encoding with the RLCP progression order. With that I am wanting to see how the data is laid out in the file, thus an index file. My...
Hi Michael, It depends on how detailed you want the index file to be. It is not true that each resolution gets its own tile. Nor does it automatically get its...
Hi Dr. Taubman The region decompressor actually decompresses 2 lines more than it has been asked to. Although this does not affect the region that was...
Hi Lokesh, The SIMD processing is not a huge deal inside `kdu_region_decompressor'. I would not focus on that right now. Needless to say, though, 24-bit/pixel...
... Dr. Taubman, Thank you very much for your response. I slipped up in my original post and did mean tile-part not tile. I have been reading your JPEG2000...
Hi! I've been pulling my hair off the last couple of days trying to get the colour space conversion from BT.601 YCbCr to work with kdu_show, without success....
Hi Dr. Taubman, Â I did get the tile decompression working. The reason why I was having problems was because the kdu_multi_synthesis returns a line that has a...
Hi Lokesh, You cannot get 8-bit data directly from the core synthesis engines; Kakadu at its core works with 4 types of numerical representation, depending on...
Hi Dr. Taubman, Â Â Â Â Thanks for the confirmation. I had concluded that from my experiments. Â This transformation causes an overhead of about 20% in the...
Hi Dr. Taubman, Â Â Â Â The "open_tile" function takes up about 20-25% of the total decompression time. Does this point to slow disk I/O? Â Regards, Lokesh ...
Hi Lokesh. Yes, this means that portion of time is single-threaded and represents the incremental parsing of tile-parts, walking through from the start of the...