[gdal-dev] -allow_projection_difference equivalent in GDAL CLI?

Scott public at postholer.com
Fri Dec 19 16:36:04 PST 2025


Yes, the raster being created from the.vrt is total garbage.

I was completely unaware of gdal driver gti create. I'm gonna go look 
into that.

Apologies for all of the noise on the list!
Scott

On 12/19/25 16:19, Even Rouault wrote:
> It all depends what "works" mean for you. For sure - 
> allow_projection_difference will result in *something*, but that will be 
> pure garbage. It will completely ignore the CRS information. The - 
> allow_projection_difference flag is meant for tiles that are 
> fundamentally in the CRS but coming from different formats/providers 
> where the CRS don't exactly compare to equal due to mundane 
> differences.  If you want on-the-fly reprojection, you should rather use 
> "gdal driver git create" to create a GTI compatible index, since the GTI 
> driver can perform on-the-fly reprojection.
> 
> Le 20/12/2025 à 01:15, Scott a écrit :
>> That's kinda why I was asking. This works:
>>
>> gdalbuildvrt -allow_projection_difference \
>>    result.vrt \
>>
>> STACIT:"/vsis3/sentinel-s2-l2a-stac/2025/12/18/ 
>> S2C_OPER_MSI_L2A_TL_2CPS_20251218T212314_A006718_T11SMA.json":asset=scl \
>>
>> STACIT:"/vsis3/sentinel-s2-l2a-stac/2025/12/17/ 
>> S2B_OPER_MSI_L2A_TL_2BPS_20251217T200148_A045870_T10JBQ.json":asset=scl
>>
>> This does not:
>>
>> gdal raster stack \
>>
>> STACIT:"/vsis3/sentinel-s2-l2a-stac/2025/12/18/ 
>> S2C_OPER_MSI_L2A_TL_2CPS_20251218T212314_A006718_T11SMA.json":asset=scl \
>>
>> STACIT:"/vsis3/sentinel-s2-l2a-stac/2025/12/17/ 
>> S2B_OPER_MSI_L2A_TL_2BPS_20251217T200148_A045870_T10JBQ.json":asset=scl \
>> result.vrt --overwrite
>>
>> Is there a way to create .vrt with different projections in GDAL cli, 
>> like gdalbuildvrt does?
>>
>> Thanks!
>> Scott
>>
>>
>> On 12/19/25 15:56, Even Rouault wrote:
>>>
>>> Le 20/12/2025 à 00:46, Scott a écrit :
>>>> Hey Even,
>>>>
>>>> Will do.
>>>>
>>>> In this case I was working with Sentinel L2A. One raster was zone 10 
>>>> and the other was zone 11, which would cause stack to complain.
>>>
>>> huh, and what result do you expect from stacking tiles in different 
>>> UTM zones, when no on-the-fly reprojection is done ... That will 
>>> result in non sense.
>>>
>>>
>>



More information about the gdal-dev mailing list