I don't think the palette files are commonly used, but I do have a couple
simple examples. There were clearly wrong before, definitely right now. So I'm
fairly confident about the fix.
OK, I will merge.
-- lg
On Apr 30, 2013, at 1:23 AM, Michel Lerenard wrote:
> Thanks.
> Master branch is enough for me, we're working with it.
>
> I've already done some tests on various TGA files (RGB, RGBA, Grayscale), the
> modification I posted does not seem to have side effects. Unfortunately I
> have no palette encoded file here, I won't be unable to test the other fix
> you made.
>
> On 04/30/2013 12:13 AM, Larry Gritz wrote:
>>
>> Works for me. Here's the pull request, and while I was there I noticed
>> another problem with TGA reads and fixed it:
>>
>> https://github.com/OpenImageIO/oiio/pull/587
>>
>> Let me know if you need this backported to any release branches, or if the
>> master is sufficient for you.
>> Also if you are working with TGA files, it would be helpful for you to grab
>> this pull request and test it on your end (including my fixes) to confirm
>> that it works for you and doesn't break anything else. I have a limited
>> number of TGA files to test against.
>>
>>
>> On Apr 29, 2013, at 12:16 AM, Michel Lerenard wrote:
>>
>>> Yes of course.
>>>
>>> I've attached the file to this message.
>>>
>>>
>>> On 04/29/2013 01:01 AM, Larry Gritz wrote:
>>>>
>>>> This sounds reasonable.
>>>>
>>>> Can you send me an example TGA that has this problem?
>>>>
>>>>
>>>> On Apr 24, 2013, at 6:13 AM, Michel Lerenard wrote:
>>>>
>>>>> Resuming my previous monologue...
>>>>>
>>>>> Digging a bit in the targa plugin i found that no check was done between
>>>>> the flags of the channels and the actual channels in the file.
>>>>>
>>>>> I modified the code from
>>>>>
>>>>> m_alpha = (m_tga.attr & 0x0F) > 0 ? TGA_ALPHA_USEFUL : TGA_ALPHA_NONE;
>>>>>
>>>>> to
>>>>>
>>>>> m_alpha = TGA_ALPHA_NONE;
>>>>> if (((m_tga.type == TYPE_RGB || m_tga.type == TYPE_RGB_RLE) &&
>>>>> m_tga.bpp == 32 )
>>>>> || ((m_tga.type == TYPE_GRAY || m_tga.type == TYPE_GRAY_RLE) &&
>>>>> m_tga.bpp > 8 )) {
>>>>> m_alpha = (m_tga.attr & 0x08) > 0 ? TGA_ALPHA_USEFUL :
>>>>> TGA_ALPHA_NONE;
>>>>> }
>>>>>
>>>>> The idea is to make sure there is a channel in the file, even if the flag
>>>>> says the data is used. In the file I have, the m_tga.attr value is equal
>>>>> to 8, which is, as far as I understand, absolutely inconsistent with the
>>>>> channel in the file.
>>>>> If I understand the meaning of the flag, I should be set to 1, meaning
>>>>> the first channel has useful information, instead of 8 meaning the last
>>>>> channel (here, alpha), has useful info.
>>>>>
>>>>> I'd appreciate very much if someone could tell me if my fix is correct,
>>>>> and if i understood correctly how the tga header work ?
>>>>> Thanks.
>>>>>
>>>>> PS : I used this page http://www.paulbourke.net/dataformats/tga/ to get
>>>>> TGA specs.
>>>>>
>>>>> On 4/24/2013 1:30 PM, Lerenard Michel wrote:
>>>>>> I found why OIIO finds two channels in the file. The tga_alpha_type of
>>>>>> the file has been set to TGA_ALPHA_UNDEFINED_RETAIN.
>>>>>>
>>>>>> I do not know how TGA works, I then have a question: what is the use of
>>>>>> such a flag ? How is it set ? Can i ignore it ?
>>>>>>
>>>>>> To be able to load correctly this type of file (ie discard the alpha
>>>>>> layer), i have to parse the data buffer returned by OIIO, and it costs
>>>>>> quite some time. I there another way to detect such an empty channel
>>>>>> before calling "read_scanlines" which i use ?
>>>>>>
>>>>>>
>>>>>> 2013/4/24 Michel Lerenard <[email protected]>
>>>>>> Hi,
>>>>>>
>>>>>> I've spotted a weird behavior when opening grayscale files, i'm
>>>>>> wondering if I have to check something else in the header or if it is
>>>>>> the correct behavior: the ImageSpec structure tells me there are two
>>>>>> channels in the file, one called 'I' and another called 'A', which is
>>>>>> absolutely black. (ie empty ?)
>>>>>>
>>>>>> The file i'm opening is a simple black and white TGA used as a mask,
>>>>>> with, according to Gimp, photoshop, xnview, etc... a single color
>>>>>> channel. ( 8 bits )
>>>>>> Is there a way to check if a given channel is empty ? Apparently the
>>>>>> file does not hold any data for alpha, so i'm wondering if OIIO is
>>>>>> creating an alpha channel for compatibility and if I should check some
>>>>>> attributes to detect if there is actually any data in it ?
>>>>>>
>>>>>>
>>>>>> Thanks for your help.
>>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Oiio-dev mailing list
>>>>> [email protected]
>>>>> http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org
>>>>
>>>> --
>>>> Larry Gritz
>>>> [email protected]
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> Oiio-dev mailing list
>>>> [email protected]
>>>> http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org
>>>
>>> <monochrome.tga>_______________________________________________
>>> Oiio-dev mailing list
>>> [email protected]
>>> http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org
>>
>> --
>> Larry Gritz
>> [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
--
Larry Gritz
[email protected]
_______________________________________________
Oiio-dev mailing list
[email protected]
http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org
