P3D JPG destruction

[Ole Hansen <oha@xxxxxxxxxx>]

Dan Shelley <dshelley@xxxxxxxxxxxx> wrote:
> For MOST anaglyphs, it is Red/Blue - not Red/Green that is correct.
> And, you can get the glasses from my web page....
Thanks Dan, I know that you give away red/blue glasses (I have
tresspassed
your website quite a few times ;-) but for color anaglyphs they are
useless,
for black and white they work, but the color tint is anoying. The
problem is
to get filters with the right passband in red and cyan, and with a
density,
that allows light to pass so that you can see the anaglyphs.

Tom Deering <tmd@xxxxxxxxxxx> wrote:
> There are jpeg artifacts in these images...
I agree, and that is the problem.

> Different programs use different methods. Perhaps PhotoImpact
> cannot make good JPEGs from your anaglyphs....
It seems so. 

> Certainly there would be more room for error if you were compressing
> color stereo pairs....
As I mentioned in the first mail, there is less distortion in JPS, but
there is distortion, and you have to use Depth Charge.

> Any artifacts, no matter how minor, will be visible when you view
> the image in stereo. Settings that make medium and low quality
> images, while perfectly adequate for 2D, will simply not work in 3D.
I agree, anaglyphs are sensitive to just one (or two) misplaced or
miscolored pixel.

> If you send me the PNG original of your worst looking JPEG, I will
> recompress it for you with a different program....
It is on its way.


John W Roberts <roberts@xxxxxxxxxxxxxxxxx> wrote:
>>> ....The effect of JPEG coding scheme on the perceived quality of 3-D
>>> images was investigated by using the CCIR subjective assessment
>>> method. Lossy compression affected the quality of 3-D image pairs
>>> less than that of 2-D images of the same scene....
>> So much for chineese science....
> I don't think it's appropriate to instantly dismiss something like
this
> without a fair hearing. This apparently instant dismissal of a claim
has
> the appearance of itself being *bad science*.

As i wrote in the reply to Boris: I just commented on the scientific
level of this particular mail. I have not read the paper, and therefore
I cannot comment it. If the mail is to the point, it is correct to point
out, that here we might have another "big white whale".

> If you are implying that we don't need to read the paper, that we
> already know everything we need to know, it seems to me that you're
> the one who is promoting ignorance.....
On the contrary. I think you should read the paper and give us your
opinion. I only made comments to the mail as stated above.

> You are able to provide examples which you say indicate that there
> is a problem with JPEG compression. Even supposing that's correct...
I do not suppose. Have a look for yourself.

> Tom Deering noted possible causes for the problems you had....
That is right, and I intend to take grave advantege of Toms knowledge,
if he permits me to do so.

> In the case of the paper by the researchers from China, I would be
> interested in knowing the methods used, particularly the subjective
> evaluation technique that was mentioned in the abstract....
So would I.

> I saw the images, but didn't have anaglyph glasses available...
That is just to bad ;-)

> and I did not see the PNG versions for comparison.
They are not there due to their volume, and due to the fact, that no
browser (as far as I know) can show PNG.

> if JPEG affects color, perhaps this is a particular problem for
> anaglyphs, and perhaps side-by-side JPEG images would not be as bad.
It is a problem with all graphic files, but with anaglyps the faults
destroyes to some extend the space in the stereogram by putting ghosts
all over the place.

> examples show that there *can be* a problem...
Correction - that there *is* a problem....

> but they don't show that there *must always be* a problem...
Agreed - In some of the anaglyphs the problem is small, but in some
of the good ones (to my taste), the distortion ruins the anaglyph.

>> Is there a browser where you can import PNG ?
> You might be able to get a "plug-in" program for the regular browsers.
>From where ?

best regards - Ole Hansen - olehansen@xxxxxxxxx



------------------------------