Hi
Thanks for your answers.
The message indeed comes from my code, but I should have mentionned that
it was here for debug purpose only. Not being able to read a tile is not
a problem for Clarisse (it can reuse a file it saved), it is for Nuke or
Photoshop: they cannot open a partial file.
I previously made a modification to write all tiles of the image region,
I'll leave it there then. :)
On 11/04/2013 11:24 PM, Larry Gritz wrote:
In that case, then it sounds to me like the right thing is
happening... it's letting him close the file without writing all the
tiles, and it's letting him open it later, but he's getting error
messages about the missing tiles when he tries to read.
I don't believe that "Unable to read tile: x/y" comes from OIIO, so
presumably Michael's app is noticing the failure return code from
ImageInput::read_tile() and issuing that error. Instead, perhaps it
should just accept that failure and infer that the tile was never
written to the file (however it wants to handle that case).
-- lg
On Nov 4, 2013, at 2:08 PM, Piotr Stanczyk <[email protected]
<mailto:[email protected]>> wrote:
Hi,
libIlmImf requires that you provide the domain of definition
information a priori; failure to to fill all of the specified region
will generate what is considered to be an ill-formed image.
An application responsible for reading the image file may handle the
case gracefully, for example, by informing the user of the missing
tiles/scanlines, or catastrophically or somewhere in between.
Piotr
On 3 November 2013 21:06, Larry Gritz <[email protected]
<mailto:[email protected]>> wrote:
Sorry for the delay in answering.
Is this intended? I can't speak for the underlying OpenEXR
library, libIlmImf, but OIIO doesn't really make any promises one
way or the other about what should happen if you open a file for
output but fail to write all the scanlines or tiles.
OIIO is pretty straightforward in its operation... each call to
ImageOuput->write_tile will write one tile. If you fail to write
all the tiles in the image region, then I guess it's up to IlmImf
about whether that produces a valid file or not. Perhaps it
makes a valid file, but one that will fail to read the tiles that
were never written?
-- lg
On Oct 22, 2013, at 4:05 AM, Michel Lerenard
<[email protected] <mailto:[email protected]>> wrote:
> Hi
>
> i've stumbled across a problem i don't understand. I cannot
save a tiled image in EXR or TIFF if I don't fill all the area
defined in the ImageSpec.
> If I try to open the file I get lots of messages "Unable to
read tile: $x/$count".
>
> I would have thought that all tiles would be created for the
canvas set in the ImageSpec, and that I would need to call
"write_tiles" for the meaningful parts of the image only. ( for
example if you only render a cube in the corner of the image, you
would need to write only the tiles that contains the images of
the cube)
>
>
> Is this the intended behavior ?
>
> Thanks
>
> Michel
>
--
Larry Gritz
[email protected] <mailto:[email protected]>
--
Larry Gritz
[email protected] <mailto:[email protected]>
_______________________________________________
Oiio-dev mailing list
[email protected]
http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org
_______________________________________________
Oiio-dev mailing list
[email protected]
http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org
