Nature Photographers Network
This is the preserved archive of the original Nature Photographers Network. The community continues at naturephotographers.network.

Photoshop Elements question

Discussion

6 replies
Posted by Gary Gulash on May 25, 2010 at 4:44 PM
This has been concerning me for a while and thought I would finally ask the question. When using Photoshop Elements 6 on my iMac there are some option boxes when saving as a Tiff where I have no idea what to click. The second box shown here asks Pixel Order - Interleaved or Per Channel. The third box asks Byte Order - IBM PC or Macintosh. Do I need to be concerned about these options, for viewing here at NPN or for printing, etc. Thanks for your help.
Posted by David Bostedo on May 25, 2010 at 4:58 PM
Gary - I don't believe you need to be concerned about those. They are part of general TIFF specification, so unless you have some strange file use or specific software that doesn't fully support TIFF, you should be fine. (For instance, I wrote some processing software that could only handle Interleaved.) Here's some explanation of what they are though :

Pixel Order - Interleaved or Per Channel - This has to do with how each of the RGB numbers for each pixel are stored. "Interleaved" means that all three number, the R, the G, and the B, are stored together in the file for each pixel. "Per Channel" means that all the R numbers are stored first for all pixels, followed by all the G, then all the B.

Byte Order - In a "word" of data (typically 4 or 8 bytes), it determines whether the most significant byte or least significant byte comes first. This can be a bit confusing, but in different systems, you could store number in two different ways. Pretend that the number "1693" is a "word" of data made of 4 bytes. And that each digit in the number "1693" is represented by a byte in a unique computer memory address. In computer memory that could be stored as :

Address0 1
Address1 6
Address2 9
Address3 3

OR

Address0 3
Address1 9
Address2 6
Address3 1

PCs tend to use one way, and Macs tend to use the other. It's largely irrelevant to most software as long as it knows which is being used. And TIFF supports both, so it shouldn't matter. Where it does matter is low level programming, where you might need to know exact storage locations of all the data.
Posted by Gary Gulash on May 25, 2010 at 5:42 PM
Thanks for the explanation David. I can sleep well now knowing it really isn't an issue. :smile:
Posted by Derek Yow on May 28, 2010 at 1:20 PM
David's explanation is perfect.

Only thing I could think of that the pixel order might affect would be rate of compression. If you had a long run of RGB values for the same exact color, in the per channel mode all the R values, G values, and B values would be in separate bands. A longer run of identical values might compress down a little better than if all the different RGB values were sandwiched together (interleaved).

Not sure if that makes sense or not..
I do software development and work with TIFF and JPEG decompression algorithms, from time-to-time and I've never tested this, but I might have to now!!
Posted by David Bostedo on May 28, 2010 at 3:22 PM
You're assuming we have compressed TIFFs Derek...which isn't always the case! That's another section of the save dialog. Since ZIP and LZW compressions are lossless, I always compress mine with those. JPG compression will of course be lossy, which will negate some of the advantage of TIFF.
Posted by Derek Yow on May 28, 2010 at 4:05 PM
Yes, that's true David. Thank you for pointing that out.

I was just trying to brainstorm if there would be any advantage to using one pixel order selection over the other. That was about the only possible advantage I could think of.
Posted by David Bostedo on May 28, 2010 at 5:22 PM
For certain image processing steps (not necessarily what an average user would want) like line or color identification, object tracking, etc. it might be slightly easier/faster to use one method over another. But for most things, I agree that it probably makes no real difference.