Re: Endless gamma in eXIf 2017-0119

Here I thought eXIf was about preserving "original metadata", or more 
specifically I thought it was "userdata" - data for the user (I'm 
definitely using this term wrong, I know).

Shouldn't eXIf be considered "sensor/camera data" and thus only useful 
for ppl who know what they're doing, and not automatically mangled into 
the image? Did I miss the messages where it was said otherwise?

On 06/02/17 09:36 AM, GLENN PEHRSON wrote:
>> On February 6, 2017 at 5:05 AM Pavel Zlatovratskii <[email protected]> wrote:
>> [...]
>> At least colourspace data have special way to handle: absence of chunks
>> is 'output as-is'. So 'Where conflicting information is present in other
>> chunks in the stream that data shall be assumed to be correct unless it
>> can be determined to be incorrect.' is not enough.
>>
>> My previous comment about 'endless gamma' was missed so I try to provide
>> more detailed explanation:
>>
>> 1) Alice and Bob have same Jpeg/EXIF file with same gamma = 2.2
>>
>> 2.1) Alice recode jpeg to PNG trying to achieve best compatibility so
>> she use both eXIf and gAMA chunks.
>>
>> 2.2) Bob use Willem's pngexif tool, which write only eXIf (that's not
>> Willem's fault! proposal have no word about this!)
>>
>> 3) Both PNG goes to Charles which publish them to web. Charles knows
>> nothing about eXIf, but he knows that some browsers and image viewers
>> doesn't process gAMA. So he runs 'truepng /g1'(apply and remove gamma)
>> on all his files.
> Would making the chunk copy-unsafe ("exIF") prevent this error?
> Charles changes the image data and therefore knows that he must not
> copy exIF.
>
>> 4) Now if we give Alice's file to Bob and ask him to recode it back to
>> Jpeg/EXIF he ignore absence of gAMA chunk(he doesn't write it) and write
>> gamma-applied Jpeg with original meta.(gAMA is absent so there is no
>> conflict mentioned in proposal)
>>
>> 5) Repeat this scenario again and again and gamma will be applied again
>> and again.
>>
>> In fact after step 3 we already have different images with same
>> metadata. So we need some 'meta-metadata' to find out if eXIf could be
>> used as source of colourspace data.
> I'm interested in testing to find out if this scenario happens without
> using PNG or exIf.  What happens if I use ImageMagick to read a JPEG,
> change gamma, and write a new JPEG?  I don't think the Exif profile
> gets modified, but I could be wrong about that, and will do some
> experimenting later (at the moment I don't have access to ImageMagick
> or GraphicsMagick).
>
> And I expect that the scenario you described can happen with existing
> ImageMagick/GraphicsMagick, which preserve Exif in a zTXt chunk and
> read it back, when converting a JPEG to PNG and back again.
>
> Glenn
>   
>> Pavel Zlatovratskii scondo-JGs/[email protected]; [email protected]
>>
>>
>>
>> ------------------------------------------------------------------------------
>> Check out the vibrant tech community on one of the world's most
>> engaging tech sites, SlashDot.org! http://sdm.link/slashdot
>> _______________________________________________
>> png-mng-misc mailing list
>> [email protected]
>> https://lists.sourceforge.net/lists/listinfo/png-mng-misc
> ------------------------------------------------------------------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, SlashDot.org! http://sdm.link/slashdot
> _______________________________________________
> png-mng-misc mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/png-mng-misc

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot

_______________________________________________
png-mng-misc mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/png-mng-misc