[ome-users] bfconvert of .r3d file

Tue Mar 15 18:53:51 GMT 2016

Hi Sebastien,

Thanks for taking a look at this. I verified the output of bfconvert with the r3d file solo in a directory and it seems to be correct. So I went ahead and repeated the same bfconvert call on a crowded directory and the output is the same - I used a diff to compare files. Then I compared these results with ones I created earlier and they do differ. So, while there was a problem in the earlier conversion it seems that bfconvert 5.1.8 is generating the right results.

Given that the generated tiffs will be used by other programs, I wonder if the problem reported by tiff lib (e.g. "Failed to read directory at offset 0") is actually preventing 'gmic' to work properly. I will need to investigate this further, including a fresh installation of gmic to iron out any possible outdated behavior due to the new OS kernel update.

Best,
- Alex

On 3/15/16 8:30 AM, Sebastien Besson (Staff) wrote:
> Hi Alex,
>
> I have been able to reproduce the tiffinfo error message with some r3d files both on
> CentOS and OSX. However the message seems to be unrelated to your original issue
> as the converted TIFF files do not have any satured pixel value.
>
> Is there any other content in the same directory as the r3d file (output of `ls`?).
> If so, does moving the file to a separate directory modify the outcome of bfconvert?
> Also, could you upload the r3d file to http://qa.openmicroscopy.org.uk/qa/upload/
> in order to allow us to test independently the conversion on our systems?
>
> Best,
> Sebastien
>
>> On 14 Mar 2016, at 17:50, Alexandre Cunha <cunha at caltech.edu <mailto:cunha at caltech.edu>> wrote:
>>
>> Hi Sebastien,
>>
>> We have tried bfconvert on the .r3d file onRHEL5 and RHEL6 distrosand with three different versions of java (1.6, 1.7, 1.8) and the problem persists. One thing we noted is that when we try to read the generated tiffs, we get an error message - see below tiffinfo output. Would this be any indication where the problem might be?
>>
>> Best,
>> - Alex
>>
>>
>> [56] : tiffinfo vala.0.10.tiff
>> TIFF Directory at offset 0x8 (8)
>>   Image Width: 1025 Image Length: 1024
>>   Resolution: 153846, 153846 pixels/cm
>>   Bits/Sample: 32
>>   Sample Format: IEEE floating point
>>   Compression Scheme: None
>>   Photometric Interpretation: min-is-black
>>   Samples/Pixel: 1
>>   Rows/Strip: 1
>>   Planar Configuration: single image plane
>>   ImageDescription:
>>   Software: OME Bio-Formats
>> TIFFFetchDirectory: vala.0.10.tiff: Can not read TIFF directory count.
>> TIFFReadDirectory: vala.0.10.tiff: Failed to read directory at offset 0.
>>
>>
>>
>> On 03/11/2016 01:17 AM, Sebastien Besson (Staff) wrote:
>>> Hi Alex,
>>>
>>> thanks for letting us know. Keep us posted about the outcome of your
>>> investigation.
>>>
>>> Best,
>>> Sebastien
>>>
>>>> On 10 Mar 2016, at 00:22, Alexandre Cunha<cunha at caltech.edu>  wrote:
>>>>
>>>> Hi Sebastien,
>>>>
>>>> Many thanks for taking the time to check on this and reporting here.
>>>>
>>>> I have done an investigation on the redhat cluster that I am having the problem and, although not conclusive yet, I found out just a few minutes ago that it seems the problem might be with the system itself. A simple run of "pnmpsnr a.pgm b.pgm" is reporting that the "images don't differ" while in fact they strongly do. There has been a major kernel update on the system and this might have screwed things up. I apologize for the possible false alarm. Again, this is not conclusive yet and the investigation continues but please put on hold any work with the .r3d problem I reported. The culprit might be on our end after all. I will report back.
>>>>
>>>> Best,
>>>> - Alex
>>>>
>>>>
>>>> On 03/09/2016 12:54 PM, Sebastien Besson (Staff) wrote:
>>>>> Hi Alex,
>>>>>
>>>>> thanks for the additional information. I tried to reproduce your issue on CentOS 6.7
>>>>> with one of the r3d files in our data repository. However I was unsuccessful at
>>>>> reproducing it.  After converting the image using the same command as you
>>>>> posted  and importing both the original and the converted one into OMERO, pixels
>>>>> have the same intensities.
>>>>>
>>>>> Would it be possible for you to upload one of your failing images at
>>>>> http://qa.openmicroscopy.org.uk/qa/upload/
>>>>>
>>>>> Best,
>>>>> Sebastien
>>>>>
>>>>>> On 8 Mar 2016, at 19:13, Alexandre Cunha <cunha at caltech.edu  <mailto:cunha at caltech.edu>> wrote:
>>>>>>
>>>>>> Hi Melissa,
>>>>>>
>>>>>> The output of md5sum is the same for both systems. The tiff validation is nothing special. I did a visual check of a few slices and they are totally off, the .r3d is a FISH image with a few bright spots on a dark background, all float values, and the generated tiff is a saturated bright all over for some slices and a mix of saturated and dark pixels for others. I also checked a few pixel values using "gmic -o -.asc". Though the tiffinfo reporst 32-bit floats on the Redhat system, the actual values seem to be integers at 16-bit quantization, it doesn't make sense. Some unsolicited conversion is being done.
>>>>>>
>>>>>> I tried different versions of java, installed an up-to-date libtiff, but the problem persists on the Redhat system - not on Ubuntu. I didn't build bfconvert from source but used bftools.zip to deploy it in both systems. I can't tell for sure if there is something wrong with the Redhat cluster - it is, after all, a possibility - but this is the only application I am aware is behaving oddly.
>>>>>>
>>>>>> Let me know if you would like me to send directly to you the .r3d I am using for tests and a few generated tiff slices.
>>>>>>
>>>>>> Best,
>>>>>> - Alex
>>>>>>
>>>>>>
>>>>>>
>>>>>> On 03/08/2016 06:32 AM, Melissa Linkert wrote:
>>>>>>> Hi Alex,
>>>>>>>
>>>>>>> Thank you for reporting this.
>>>>>>>
>>>>>>>> We run bfconvert to convert .r3d files, (Deltavision, 32-bit floats) into tiff slices, as in
>>>>>>>>
>>>>>>>>     bfconvert -separate file.r3d slice.%c.%z.tiff
>>>>>>>>
>>>>>>>> for further processing. It works well under Ubuntu 14.04 but it chokes under Red Hat Enterprise Linux Server release 6.7 producing faulty tiff slices (pixel values are incorrect, e.g. all pixels have 65535). In my tests I use bfconvert versions 5.0.2 and 5.1.8 on both systems.
>>>>>>> We're trying a few things to duplicate the problem now, but in the
>>>>>>> meantime could you please confirm that the output of 'md5sum file.r3d'
>>>>>>> is the same on both systems?  It would also be useful to know which
>>>>>>> software is being used to validate the converted TIFF files, as that
>>>>>>> may help to point us in the right direction.
>>>>>>>
>>>>>>> Regards,
>>>>>>> -Melissa
>>>>>>>
>>>>>>> On Mon, Mar 7, 2016 at 7:01 PM, Alexandre Cunha <cunha at caltech.edu  <mailto:cunha at caltech.edu>> wrote:
>>>>>>>> Hi,
>>>>>>>>
>>>>>>>> We run bfconvert to convert .r3d files, (Deltavision, 32-bit floats) into tiff slices, as in
>>>>>>>>
>>>>>>>>     bfconvert -separate file.r3d slice.%c.%z.tiff
>>>>>>>>
>>>>>>>> for further processing. It works well under Ubuntu 14.04 but it chokes under Red Hat Enterprise Linux Server release 6.7 producing faulty tiff slices (pixel values are incorrect, e.g. all pixels have 65535). In my tests I use bfconvert versions 5.0.2 and 5.1.8 on both systems.
>>>>>>>>
>>>>>>>> I was wondering if this is a known problem on RedHat machines and/or .r3d files. Any advice?
>>>>>>>>
>>>>>>>> Many thanks,
>>>>>>>> - Alex
>>>>>>>>
>>>>>>>> ---
>>>>>>>>
>>>>>>>> [607] : bfconvert -version
>>>>>>>> Version: 5.1.8
>>>>>>>> VCS revision: 9eb8da31c1f01989f439963b9cc215b4398bedd8
>>>>>>>> Build date: 12 February 2016
>>>>>>>>
>>>>>>>> OR
>>>>>>>>
>>>>>>>> (bfconvert -version
>>>>>>>> Version: 5.0.2
>>>>>>>> VCS revision: 0c4215a
>>>>>>>> Build date: 27 May 2014)
>>>>>>>>
>>>>>>>>
>>>>>>>> lsb_release -a
>>>>>>>>
>>>>>>>> LSB Version:    :base-4.0-amd64:base-4.0-noarch:core-4.0-amd64:core-4.0-noarch:graphics-4.0-amd64:graphics-4.0-noarch:printing-4.0-amd64:printing-4.0-noarch
>>>>>>>> Distributor ID: RedHatEnterpriseServer
>>>>>>>> Description:    Red Hat Enterprise Linux Server release 6.7 (Santiago)
>>>>>>>> Release:        6.7
>>>>>>>> Codename:       Santiago
>>>>>>>>
>>>>>>>> Distributor ID: Ubuntu
>>>>>>>> Description:    Ubuntu 14.04.4 LTS
>>>>>>>> Release:        14.04
>>>>>>>> Codename:       trusty
>>>>>>>>
>>>>>>>>
>>>>>>>> _______________________________________________
>>>>>>>> ome-users mailing list
>>>>>>>> ome-users at lists.openmicroscopy.org.uk  <mailto:ome-users at lists.openmicroscopy.org.uk>
>>>>>>>> http://lists.openmicroscopy.org.uk/mailman/listinfo/ome-users
>>>>>>>>
>>>>>> _______________________________________________
>>>>>> ome-users mailing list
>>>>>> ome-users at lists.openmicroscopy.org.uk  <mailto:ome-users at lists.openmicroscopy.org.uk>
>>>>>> http://lists.openmicroscopy.org.uk/mailman/listinfo/ome-users
>>>>> The University of Dundee is a registered Scottish Charity, No: SC015096
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> ome-users mailing list
>>>>> ome-users at lists.openmicroscopy.org.uk
>>>>> http://lists.openmicroscopy.org.uk/mailman/listinfo/ome-users
>>>>>
>>>> _______________________________________________
>>>> ome-users mailing list
>>>> ome-users at lists.openmicroscopy.org.uk
>>>> http://lists.openmicroscopy.org.uk/mailman/listinfo/ome-users
>>> The University of Dundee is a registered Scottish Charity, No: SC015096
>>> _______________________________________________
>>> ome-users mailing list
>>> ome-users at lists.openmicroscopy.org.uk
>>> http://lists.openmicroscopy.org.uk/mailman/listinfo/ome-users
>>>
>>
>> _______________________________________________
>> ome-users mailing list
>> ome-users at lists.openmicroscopy.org.uk <mailto:ome-users at lists.openmicroscopy.org.uk>
>> http://lists.openmicroscopy.org.uk/mailman/listinfo/ome-users
>
>
> The University of Dundee is a registered Scottish Charity, No: SC015096
>
>
> _______________________________________________
> ome-users mailing list
> ome-users at lists.openmicroscopy.org.uk
> http://lists.openmicroscopy.org.uk/mailman/listinfo/ome-users
>