[MF3D.FORUM:521] Re: stereo cards on the web

[Marco Pauck <pauck@xxxxxxxxxxxx>]

Tom Deering wrote:
> 
> >I agree with Tom's remarks and suggestions except one:
> >
> >> Then, either I would save the images as much higher quality JPEGS, or
> >> I would save them as greyscale GIFS, maybe 64 or 128 color palette.
> . . . .
> >Never cripple high quality b/w prints by quantizing them to less
> >than 8 bits (i.e. 256 grey values)!
> 
> The word "never" is too strong.  If these were full color images, I'd
> mostly agree.  But the jpeg format loses some of it's edge with
> greyscale images.  In fact, if file size were no option, then gif
> would render better greayscale stereo images than jpeg, every time.

Agreed, as a 8 bit paletted GIF has the same dynamic range as a 24 bit
true-color JPEG but without compression artefacts.

> Due to the way the two compression schemes work, gif is less
> disruptive to the stereoscopic perception than jpeg.  When comparing
> files of the same size, a "loss-less" gif with a limited color
> palette may look better than a jpeg compressed to the same file size.
> That's why I said it was a judgement call.

Well, considering that the LZW compression algorithm employed in GIF
works very badly on photographs, even the highest quality settings for
the JPEG algorithm will result in smaller files for nearly all subjects,
esp. if they are crisp with maybe a little grain visible as typical for
many high-quality b/w scans.

> The reason is, jpeg compression artifacts are positioned with regard
> to the physical dimensions of the image.  On the other hand, the
> posterization  "artifacts" that can occur in gifs are positioned with
> regard to the different elements of the image itself.

True. Now, while I still think that JPEG is overall a better choice
than GIF, the interestiing question is "In which way is 3D image quality
affected by compression (and maybe color quantization) artefacts?"
I would guess that JPEG artefacts show up (when compression is really
strong) similar to corrugated glass positioned at the window plane.

The whole topic about JPEG compression quality has been beaten to death
for 2D images but I don't know of any research for 3D. I also haven't
looked much into the new JPEG 2000 stuff.

[...]
> Really tough without pictures.  I guess I have a new topic for my web site.

Great, I would be very interested to see some results.

	Marco
--
Marco Pauck -- marco@xxxxxxxx -- http://www.pauck.de/marco/
For every complex problem, there is a solution that is simple,
neat, and wrong.  -- H. L. Mencken