Opened 10 years ago

Closed 10 years ago

#5500 closed defect (invalid)

gdaltranslate: polynomial transformation can be buggy

Reported by: georg Owned by: warmerdam
Priority: normal Milestone:
Component: default Version: unspecified
Severity: normal Keywords:
Cc:

Description

Here is an example:

gdaltranslate \ -gcp 5502100.94896 7035747.41292 5504355.8413 7032651.71328 \ -gcp 5505158.4301 7035365.22778 5506801.8262 7032231.30962 \ -gcp 5507451.54094 7036014.94252 5508636.31488 7033148.55396 \ -gcp 5505081.99307 7038346.27189 5505081.99307 7038346.27189 \ -gcp 5507031.13729 7038155.17931 5507031.13729 7038155.17931 \ -gcp 5508636.31488 7038575.58297 5508636.31488 7038575.58297 \ -gcp 5511235.17384 7037505.46457 5511235.17384 7037505.46457

Input: 5511235.17384 7037505.46457

Output: 322909212.073118 695072513.273818 0

There is only 7 gcps so it's polynom order of 2. Problem can't be detected where there are 6 or 10 points (i. e. when polynomial transformation is exact)

Change History (10)

comment:1 by Even Rouault, 10 years ago

Please provide a sample dataset or the output of the gdalinfo utility on the source dataset

comment:2 by georg, 10 years ago

There is no dataset. Just gdaltranslate output and bug.

comment:3 by Even Rouault, 10 years ago

Perhaps you meant gdaltransform ?

comment:4 by georg, 10 years ago

Yes, misprinted. Sorry.

comment:5 by Even Rouault, 10 years ago

Here's what I get with GDAL 1.11 : 

$ gdaltransform -gcp 5502100.94896 7035747.41292 5504355.8413 7032651.71328 -gcp 5505158.4301 7035365.22778 5506801.8262 7032231.30962 -gcp 5507451.54094 7036014.94252 5508636.31488 7033148.55396 -gcp 5505081.99307 7038346.27189 5505081.99307 7038346.27189 -gcp 5507031.13729 7038155.17931 5507031.13729 7038155.17931 -gcp 5508636.31488 7038575.58297 5508636.31488 7038575.58297 -gcp 5511235.17384 7037505.46457 5511235.17384 7037505.46457 

5511235.17384 7037505.46457 

5511324.11318636 7037342.62464452 0

comment:6 by georg, 10 years ago

Could you try 1.8.0 (which is mine)? Were there any changes in the code of GCP transformer?

When?

comment:7 by Even Rouault, 10 years ago

Yes there must certainly have been some changes. 1.8.0 dates back to 3 years ago. If you have some funding, I could do some archeology for sure, and backport in a branch if needed.

comment:8 by georg, 10 years ago

As far as I see, the only change in the source between 1.8.0 and 1.11.0 is related to outliers removal.

As far as I understand, there is no refinement turned on for above example. May be there is something wrong with my build.

comment:9 by Even Rouault, 10 years ago

Hum, you are right. There were fixes in TPS, but they must be older than 1.8.0, so it is not obvious why you get this error.

comment:10 by georg, 10 years ago

Resolution: → invalid
Status: new → closed

Hmm... Strange. I can't repeat the problem with gdaltransform. Seems, that the bug is on my site. Sorry for inconvenience.

Note: See TracTickets for help on using tickets.