2007.01.17 02:25 "[Tiff] Elevation Data", by Craig Bruce

2007.01.17 19:00 "Re: [Tiff] Elevation Data", by Joris Van Damme


Libtiff uses the names PHOTOMETRIC_MINISWHITE and PHOTOMETRIC_MINISBLACK. This implies something different than BlackIsZero. Does that help?

It's confusing. I usually use the names MinIsBlack and MinIsWhite... What's in a name?

Craig Bruce referred to these as ZeroIsBlack which is quite distinct from MinIsBlack if you are thinking about signed integer values as he was.

I'm starting to see where you guys are coming from...

But still, ZeroIsBlack and MinIsBlack are different names for the same grayscale we *all know* encodes black as zero. It is unsigned. I don't care if Craig feels it's elevation data. He's writing grayscale.

Well, I do care if Crag feels it's elevation data, of course, because I feel we need to come to a solution to this problem that satisfies us all. But if it's grayscale, it *is* grayscale, and that comes with all known properties. So either

  1. we need to find something that sits good with renderers seeing the grayscale, and still sits good with Craig re-interpreting it as elevation data.
  2. or we need to simply say it's impossible, and refer him to an Undefined ExtraSample, which is really intended for this sort of purpose in the first place. The possible trouble here may be that there is no NULL Photometric, with 0 image data samples, leaving all other samples as ExtraSamples and making it clear for image renderers there's nothing there to their liking. So if he hasn't got an actual image to write along with his elevation data, and doesn't much care to write a dummy image channel, we're stuck with option a).

In another message, you wrote
> There is a certain inherent tension between those who think of TIFFs
> images in general in photometric terms, and those who think of them as
> container for transporting all sorts of gridded scientific data. But

> need to recognize they are both legitimate applications.

Of course they are both legitimate applications. But I don't know about there being room for tension. There certainly is no Purpose WORD tag with values Imaging = 1 and Scientific = 2. So the same data, hits both family of readers.

There is the ExtraSamples, where you can put anything, and it's clear to the photographic imagers this isn't intended for them to render. But then there's the base samples, and the rules that apply to the Photometric and all chosen for these, do apply.

I keep repeating myself. Though there may be no need to do so. I came up with a recommendation that I think agrees with the spec, agrees with imaging needs, and allows Craig to redefine ranges to match his view of metres metric and sea level = 0 elevation data. Is there anything in that recomendation (http://www.asmail.be/msg0055261416.html) that you see conflicts with any needs or with de facto usage (as far as the latters is consistent)?

Best regards,

Joris Van Damme
Download your free TIFF tag viewer for windows here: