locked
Error on stitching large images in Version 2.03 which were ok in Version 1.44 RRS feed

  • Question

  • From the USGS Corona (Keyhole) satellite collection, under Version 1.44 I am able to stitch 4 Tiffs to get an image of 120,000 x 10,000 pixels in grayscale on Export..Save is  to Adobe PSB  100% With version 2.03, all seems to run normally (planar motion), but the result image has only part of the four parts and the rest is totally smeared or black.  

    This happens under Windows 7 and in 8.1, 64 bits on two machines with 16GB memory and nothing else loaded,  It happens with either the default maximum memory or with double that amount in 2.03.  In 1.44 all works well with the doubled amount and the default.  .

    iscollar

    Wednesday, June 10, 2015 7:09 PM

Answers

  • Thanks for the prompt reply.  Sorry about the delay in answering, but a slow DSL upload connection for the large files plus thunderstorms out here in the country required shutting everything down for hours at a time.

    Here are the links you need to reproduce the problems:

    http://dl.dropbox.com/u/36412452/ICE_TestData.zip

    The original 4 Tiffs from USGS (Corona images)have been modified to jpegs to save upload time.

    http://dl.dropbox.com/u/36412452/TestDataResults.zip

    This shows screen captures and spj for the above data with ICE 2.03 and 1.44 with the image captions containing short descriptions.

    http://dl.dropbox.com/u/36412452/Corona.exe

    This is my own program which reads, writes, modifies and corrects Corona imagery.  The exe file is an installer, and there is a help file in chm format which includes essential information on how to get more data from USGS (860,000 7.5 micron scanned 70mm film strips) averaging 200km long and 10km wide at between 4 and 2.5 meters/pixel ground resolution as well as instruction on how to use the small program Corona64 and some links to the history and bibliography in short form.  The literature is enormous, but the final topic contains some unpublished information obtained from telephone or e-mail interviews with survivors 35-40 years later.

    iscollar


    • Marked as answer by airphoto Friday, June 12, 2015 6:36 PM
    • Edited by airphoto Friday, June 12, 2015 6:39 PM
    Friday, June 12, 2015 6:35 PM
  • After writing the above, I found that I could set "Composite" as a flag in our Adobe code.  This makes behavior identical in 1.44 and 2.03.  The problem is solved. Thanks for your attention and query.

    iscollar

    • Marked as answer by airphoto Sunday, June 14, 2015 10:37 AM
    Sunday, June 14, 2015 10:37 AM

All replies

  • Sorry about this.

    Could you post your source images and ICE project (.spj) somewhere (e.g. OneDrive, Dropbox)?

    Thursday, June 11, 2015 4:59 PM
  • Thanks for the prompt reply.  Sorry about the delay in answering, but a slow DSL upload connection for the large files plus thunderstorms out here in the country required shutting everything down for hours at a time.

    Here are the links you need to reproduce the problems:

    http://dl.dropbox.com/u/36412452/ICE_TestData.zip

    The original 4 Tiffs from USGS (Corona images)have been modified to jpegs to save upload time.

    http://dl.dropbox.com/u/36412452/TestDataResults.zip

    This shows screen captures and spj for the above data with ICE 2.03 and 1.44 with the image captions containing short descriptions.

    http://dl.dropbox.com/u/36412452/Corona.exe

    This is my own program which reads, writes, modifies and corrects Corona imagery.  The exe file is an installer, and there is a help file in chm format which includes essential information on how to get more data from USGS (860,000 7.5 micron scanned 70mm film strips) averaging 200km long and 10km wide at between 4 and 2.5 meters/pixel ground resolution as well as instruction on how to use the small program Corona64 and some links to the history and bibliography in short form.  The literature is enormous, but the final topic contains some unpublished information obtained from telephone or e-mail interviews with survivors 35-40 years later.

    iscollar


    • Marked as answer by airphoto Friday, June 12, 2015 6:36 PM
    • Edited by airphoto Friday, June 12, 2015 6:39 PM
    Friday, June 12, 2015 6:35 PM
  • Just to confirm. The ICE generated PSB files don't look correct in the Corona application? I just tried your data in ICE and exported a PSB that Photoshop does display correctly.


    Sunday, June 14, 2015 1:13 AM
  • That is correct. After writing my first posting here, I tried Photoshop and observed that it read ICE output PSB from 1.44 and 2.03 identically. Our Corona application uses the published documentation from Adobe for reading and writing PSB files.

    https://www.adobe.com/devnet-apps/photoshop/fileformatashtml/

    One of the screen shots in Dropbox (TestDataResults.zip), TestData203_144_203Composite.jpg shows the Corona output from three ICE runs, 2.03 with "Composite Only" (OK) ,  2.03 with defaults (Error), 1.44 with defaults (OK).

    Photoshop reads the layer option correctly regardless of the ICE version.   

    Our Adobe code reads the layer option default correctly when used with output from ICE Version 1.44, but not from 2.03 default. If the user selects "Composite Only", all is well in 2.03.  In 1.44, selecting the equivalent "Flat" wasn't needed, but it seems to reduce the output file size.  

    In our Corona program, Adobe PSB's are used only once for reading ICE output although they can be written for reading in Photoshop if the user needs it.  8bit grayscale Tiffs are read and written by default, since there is no color in Corona imagery.

    I will try to duplicate the behavior of Photoshop in Corona code if I can using the flags in the Adobe specification when Layers are selected by default in ICE 2.03. However, if you can make ICE code retain user option choices of Planar Motion and Composite Only from session to session using the spj file, that would be fine. 

    Sunday, June 14, 2015 9:13 AM
  • After writing the above, I found that I could set "Composite" as a flag in our Adobe code.  This makes behavior identical in 1.44 and 2.03.  The problem is solved. Thanks for your attention and query.

    iscollar

    • Marked as answer by airphoto Sunday, June 14, 2015 10:37 AM
    Sunday, June 14, 2015 10:37 AM