Images that display in IE, but not in Picture-and-fax viewer (nor FireFox)

On Tue, 18 Dec 2018 18:01:36 +0100, R.Wieser wrote:
Hello all,

I've got some (old) JPG and GIF images here that display in IE, but not in
picture-and-fax viewer (nor FireFox. The FreeImage DLL refuses them too).

What is causing it, and what can I change (to the image) to fix it (other
than trying to pull them thru some image convertor) ?

Or, said otherwise: What is different to such JPG and GIF images that allows
them to display on IE, but not on/in three other programs (ok, two and a DLL
:-) )

Chances are that the images have corrupted data either unintentional or due
to bad image encoder; or contains non standard image parameter(s). MSIE must
have a persistent image handling where it serves more like an image salvager
than a simple image decoder.

If you use an image file analysis tool which breaks down the data structure
on those image files, it should show the cause of the problem.

R.Wieser wrote:


In other words, you doubt - without stating a reason for it - my findings.
I find that offensive.

Alright, you're cut off, dickhead.

Why would you post a question with no
physical evidence whatsoever to back up
your premise, then *shun* any attempt
at collaboration ?

The mind boggles.

Why not just sit in your ****ing room
and solve it yourself ?

You notice I'm finding the lack of logic... offensive ?

Well, yeah, I do.

Paul

JJ,

Chances are that the images have corrupted data either unintentional
or due to bad image encoder

I would expect that to be rather visible, especially on JPG files and on the
ammount of images I've currently got that exhibit this problem. But they
don't. From batch of 29 images *all* seem to show perfectly (just tested to
make absolutily damn sure).

or contains non standard image parameter(s)

That was my first thought. I have no idea what such non standard
parameters look like though.

MSIE must have a persistent image handling where it serves more like
an image salvager than a simple image decoder.

Yeah, I also thought about that. But as with the first paragraph, such
salvaging should leave marks. Which there seem to be none.

If you use an image file analysis tool which breaks down the data
structure on those image files, it should show the cause of the problem.

:-) Pretty-much the reason I came he Either someone recognising this
behaviour, or knowing about how to do such analysis / where to find such an
tool. Beyond walking thru the GIF/JPG files block structure I mean. That
I've already done.

Regards,
Rudy Wieser

On Wed, 19 Dec 2018 12:10:18 +0100, "R.Wieser"
wrote:

Paul,

How did your Irfanview tests go ?

With all due respect, but I already mentioned three different programs that
would not accept the images, and one that does. What more proof do you need
?

In other words what should I try IrfanView for ?

If its just about being able to view the images I can simply assign IE to
the involved extensions and be done with it (wouldn't like it though).

If you are hinting at image conversion, I've mentioned that I see such an
action as a last-ditch option (doesn't mean that I'm not already
investigating it. The build-in GdiPlus DLL shows some promiss).

Regards,
Rudy Wieser

Probably suggested it because IrfanView seems able to find incorrectly
coded images and offers to fix them. (and does)

Paul,

In other words, you doubt - without stating a reason for it - my
findings. I find that offensive.

Alright, you're cut off, dickhead.

Ok, goodbye.


What, aren't you gone yet ?


Oh wait, first some ranting - to which a rebuttal ofcourse will be ignored
(yeah, right :-) ).

Why would you post a question with no
physical evidence whatsoever to back up
your premise, then *shun* any attempt
at collaboration ?

*WHAT* collaboration ? A demand that I prove that some random other
program does(n't) exhibit the same problem ? WHAT WOULD THAT PROVE.
Nul. Nothing. Zilch. Nada.

.... to which I've not seen a single word claiming, let alone explaining
otherwise.

And I already tried to make clear why I do not like to install random crap
on my machine.

So, underbuild why you think its a good idea to do it regardless, or
otherwise good riddance. Go away. Shoo.

Why not just sit in your ****ing room
and solve it yourself ?

If you would have read my posts in this thread (just 8, this one included)
you might have noticed that I've got at least two solutions lined up: Either
set IE as my default image viewer, or re-encode the offending images.

And, if its not just a matter of flipping a few bits in the image headers, I
will most likely go with the latter one.

I've already brewn me a small program to do so.

You notice I'm finding the lack of logic... offensive ?

I guess you *haven't* noticed - although I think I was rather clear about
it - that your/other peoples the lack of underbuilding why throwing another
image viewer at it would be a good idea does not really impress me in the
least (understatement).

Do I really need to be *this* blunt to get thru to you ? And you have
problems with *me* ? Pot, meet kettle. :-)

Regards,
Rudy Wieser

Default,

Probably suggested it because IrfanView seems able to find
incorrectly coded images

:-) "incorrectly coded" ? If that where so IE would not be able to display
them either.

But I take it you ment "using a non-mainstream encoding" there. Yes, in
retrospect that seems rather likely (though I had hoped for a simple
image-header problem).

and offers to fix them. (and does)

I also mentioned that I only wanted to look at re-encoding as a last-ditch
option. As such I cannot truly believe that IrfanView was offered for
that reason alone. But it beats me what other reason they had, as it
looks they are not quite certain themselves either ...

And by the way, I can automatically find the offending images myself just
fine. I even enjoyed the process of figuring out how get the OS to do the
re-encoding for me and putting it into a small program.

Regards,
Rudy Wieser

R.Wieser wrote:

Beyond walking thru the GIF/JPG files block structure I mean. That
I've already done.

You could have pointed that out in your first post.

If you've already used a parser/validator, it already
contains the known defects information.

Paul

On Thu, 20 Dec 2018 14:40:12 +0100, "R.Wieser"
wrote:

Default,

Probably suggested it because IrfanView seems able to find
incorrectly coded images

:-) "incorrectly coded" ? If that where so IE would not be able to display
them either.

But I take it you ment "using a non-mainstream encoding" there. Yes, in
retrospect that seems rather likely (though I had hoped for a simple
image-header problem).

and offers to fix them. (and does)

I also mentioned that I only wanted to look at re-encoding as a last-ditch
option. As such I cannot truly believe that IrfanView was offered for
that reason alone. But it beats me what other reason they had, as it
looks they are not quite certain themselves either ...

And by the way, I can automatically find the offending images myself just
fine. I even enjoyed the process of figuring out how get the OS to do the
re-encoding for me and putting it into a small program.

Regards,
Rudy Wieser

Well, I'm way outside my depth here, but the error message I
frequently encounter sounds more like there's some incorrect type
information in the image compression protocol itself.

Like the image is correctly encoded but not to the standard the header
field claims it is. (if that makes any sense?)

I encounter it so infrequently that I haven't really been paying
attention to exactly what the error message is telling me.

IrfanView just tells me there's a problem and would I like them fix
it, click yes, and the image is shown.

In message , R.Wieser
writes:
John,

Apart from that it's a dedicated image viewer, and opens a LOT quicker -
and with less fuss - than IE,

Not what I was-or-am looking for.

As the thread proceeds, what you are looking for is becoming clearer
(and possibly changing, though I'm not going to claim that definitely!).

it also can tell you a lot _about_ the image: colour depth (or greyscale
equivalent), number of colours,

That I can do by simply looking at the JPEGs or GIFs header.

You hadn't revealed that you've actually done that (which you are now
implying you have).

and probably other things, which I'm pretty sure IE can't

And thats just guesswork.

Like you, I don't want to load something - especially not something as
complicated as IE (if it's different to the version I have) - just to
try it. But I'm pretty sure IE does not offer much information about
images.

I do not really like to put a program on my 'puter just to "have a look" and
be removed 15 minutes later. *Especially not* when it needs to install
itself, most always meaning it drops stuff all over my OS and registry.

Fair enough. I don't think IrfanView touches the registry much,
certainly if you don't let it become the default viewer for filetypes.
You _can_ just copy the executable - I think it's called i_view.exe -
from a running system, and it'll work on its own (though I think if you
change settings, it will make a .ini file).

Basically, IrfanView is a well-written piece of software that I think
you'd actually like; I don't know anyone who has removed it who has
tried it, though obviously there must be some. However, you definitely
don't _need_ it for what you want, and with your knowledge of image
headers (I presume you are using a hex editor/viewer of some sort).
[]
If push-comes-to-shove my current experiments seems to show that I can
re-encode the images using the GDI+ DLL. The resulting JPGs even seem to
become smaller, with no discernible visible differences.


Other than pixel-by-pixel comparison, you can't be sure, though; JPG
coding (and thus re-encoding) is very subtle. (Yes, I did see you'd used
the word "discernible".)
--
J. P. Gilliver. UMRA: 1960/1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

I'm the oldest woman on primetime not baking cakes.
- Anne Robinson, RT 2015/8/15-21

In message , R.Wieser
writes:
Default,

Probably suggested it because IrfanView seems able to find
incorrectly coded images

:-) "incorrectly coded" ? If that where so IE would not be able to display
them either.

Not necessarily; as JJ has said, IE may be repairing too.
[]
--
J. P. Gilliver. UMRA: 1960/1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

I'm the oldest woman on primetime not baking cakes.
- Anne Robinson, RT 2015/8/15-21

In message , R.Wieser
writes:
JJ,

Chances are that the images have corrupted data either unintentional
or due to bad image encoder

I would expect that to be rather visible, especially on JPG files and on the
ammount of images I've currently got that exhibit this problem. But they
don't. From batch of 29 images *all* seem to show perfectly (just tested to
make absolutily damn sure).

If the "bad" data was in the header, and IE corrects it, then there's be
nothing visible. (When JJ said "corrupted data", he I'm pretty sure was
not distinguishing between the various parts of the file, such as header
and pixel data.)

or contains non standard image parameter(s)

That was my first thought. I have no idea what such non standard
parameters look like though.

You _have_ said that you could examine the headers (-:. I took that to
mean you _do_ have knowledge - more than I do, anyway! - of the
structure of a .jpg image.

MSIE must have a persistent image handling where it serves more like
an image salvager than a simple image decoder.

Yeah, I also thought about that. But as with the first paragraph, such
salvaging should leave marks. Which there seem to be none.

See above: if the salvaging involved only header correction, it would
leave nothing visible.

If you use an image file analysis tool which breaks down the data
structure on those image files, it should show the cause of the problem.

:-) Pretty-much the reason I came he Either someone recognising this
behaviour, or knowing about how to do such analysis / where to find such an
tool. Beyond walking thru the GIF/JPG files block structure I mean. That
I've already done.

And not found anything, presumably.

Regards,
Rudy Wieser


As an aside, what sort of _size_ are these images (I mean in pixels)?
Not that I'm suggesting that has anything to do with the problem, I'm
just being curious (nosey).

As a second aside, you've I think spent enough time on this thread that
you could have recoded them by now (including installing IrfanView [and
uninstalling it]; or, using the software you've written yourself). Not
that I'm saying you should do that: I think I now do understand your
question: was it "has anyone else encountered images that view OK in IE
version ## or Fax viewer ??, but not in zzzz, and _knows why_?"
--
J. P. Gilliver. UMRA: 1960/1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

I'm the oldest woman on primetime not baking cakes.
- Anne Robinson, RT 2015/8/15-21

Default,

Well, I'm way outside my depth here, but the error message
I frequently encounter sounds more like there's some incorrect
type information in the image compression protocol itself.

Possible. But not in a way which makes the image fully unusable.

Like the image is correctly encoded but not to the standard the
header field claims it is. (if that makes any sense?)

It certainly does make sense. It was the first thing that came to my mind
too. Only later I realized that the encoding itself could also be the
problem..

IrfanView just tells me there's a problem and would I like them
fix it, click yes, and the image is shown.

I might be a bit odd in this regard, but I'd like to know what went wrong.
Maybe even fix the header data myself. But I'll keep the "fix it"
capability of IrfanView in mind. Thanks for the "this is why you could do
worse than to install it " explanation/hint.

Regards,
Rudy Wieser

John,

As the thread proceeds, what you are looking for is becoming clearer (and
possibly changing, though I'm not going to claim that definitely!).

:-) When I posted I was thinking about some image-header related problem.
Only later on I realized that the encoding itself might be it. So although
the problem (and the question) itself hasn't changed, my idea to what might
be causing it has.

That I can do by simply looking at the JPEGs or GIFs header.

You hadn't revealed that you've actually done that (which you are
now implying you have).

Does knowing it bring you closer to knowing what goes wrong ? I don't
think so.

Does me being able to look at that data, but not knowing what aspect of it
is out of the ordinary - let alone if it actually has any influence on the
mentioned programs - help in any way ? Again, no.

That, and my experience with over-feeding - causing all kinds of
discussions/arguments about all kinds stuff which only serves to muddy the
waters - made me decide to keep that to myself.

Like you, I don't want to load something - especially not something as
complicated as IE (if it's different to the version I have) - just to try
it. But I'm pretty sure IE does not offer much information about images.

Having information is one thing. But that doesn't mean sh*t if you do not
know what you should be looking at. And as I definitily do not know
gathering such info would constitute to a waste of time.

And you guys do not seem to know either. Not a single "you could take a
peek at {X}", or a "what does {Y} say" mentioned anywhere. Just a blanket
"go take look" suggestion (of sorts). :-(

I don't think IrfanView touches the registry much, certainly if you don't
let it become the default viewer for filetypes.
.....
Basically, IrfanView is a well-written piece of software that I think
you'd actually like

Is it now *my* turn to say "You hadn't revealed that ..." ? :-p

Other than pixel-by-pixel comparison, you can't be sure, though; JPG
coding (and thus re-encoding) is very subtle.

Having read about how JPEG encoding is lossy, and you should not re-encode
such images I expected the visual changes to be more pronounced. I do not
know which quality was used on the output file (just allowed it to use its
default to see what the end result would be), but seeing that the filesize
became less I think I can safely assume it wasn't a 100%.

:-) "incorrectly coded" ? If that where so IE would not be able to
display them either.

Not necessarily; as JJ has said, IE may be repairing too.

Apart from that being a (wild? educated?) guess, seeing that the repairing
doesn't seem to take any noticable time (read: is easy to do), why would a
browser like FF not have incoorporated it too ? Let alone some of Windows
own default programs, like the mentioned picture-and-fax viewer, and paint.

If the "bad" data was in the header, and IE corrects it, then there's be
nothing visible.

I'm sorry, but that is a bit to much "if"-ing to my liking, and *way* to
little to back it up. Possible ? Ofcourse. Likely ? Not really.

(When JJ said "corrupted data", he I'm pretty sure was not distinguishing
between the various parts of the file, such as header and pixel data.)

I don't think so either. But it does not make a difference either way.
If its random corruption than the chance that my batch of 29 images (or the
80 I checked out later) all could be repaired is astronomical small.
*Especially* as the JPEG encoding does not seem to have any error correction
on its data, let alone on the header.

If its *not* random corruption (but maybe just some programs writing
malformed headers) than it should by now be known to a *lot* of people, most
particulary the likes of FireFox and FreeImage.

In short, corruption (either of them) isn't really likely.

Than again, I also have nothing to back that up. :-( :-)

Regards,
Rudy Wieser

P.s.
Of the hunderd images FreeImage (the DLL) could not read I've got only a
single one which also fails to be re-encoded using GDI+

P.p.s.
Although GDI+ can read the "broken" JPG images and write them back to disk
in another format (writing them back in the same format results in a simple
file copy :-\ ), I cannot draw parts of it to a new "graphics" canvas (which
I can do with non-broken images). It seems to indicate that something other
than just an image header is involved ...

In message , R.Wieser
writes:
John,

As the thread proceeds, what you are looking for is becoming clearer (and
possibly changing, though I'm not going to claim that definitely!).

:-) When I posted I was thinking about some image-header related problem.
Only later on I realized that the encoding itself might be it. So although
the problem (and the question) itself hasn't changed, my idea to what might
be causing it has.

That I can do by simply looking at the JPEGs or GIFs header.

You hadn't revealed that you've actually done that (which you are
now implying you have).

Does knowing it bring you closer to knowing what goes wrong ? I don't
think so.

Not me, but it might you: if you've looked at the headers (which to some
extent at least implies you know what you're looking at!), you might
detect a pattern in the rogue images that you don't see in others. (I
presume you've looked at the headers of some problem-free images too.)
[]
Like you, I don't want to load something - especially not something as
complicated as IE (if it's different to the version I have) - just to try
it. But I'm pretty sure IE does not offer much information about images.

Having information is one thing. But that doesn't mean sh*t if you do not
know what you should be looking at. And as I definitily do not know
gathering such info would constitute to a waste of time.

Someone (was it you yourself?) suggested the images might be to more
than 8 bits (256 levels) of greyscale, for example.

And you guys do not seem to know either. Not a single "you could take a
peek at {X}", or a "what does {Y} say" mentioned anywhere. Just a blanket
"go take look" suggestion (of sorts). :-(

I don't think IrfanView touches the registry much, certainly if you don't
let it become the default viewer for filetypes.
....
Basically, IrfanView is a well-written piece of software that I think
you'd actually like

Is it now *my* turn to say "You hadn't revealed that ..." ? :-p

Go on. Give it a try. Even if nothing to do with the current subject
(-:.

Other than pixel-by-pixel comparison, you can't be sure, though; JPG
coding (and thus re-encoding) is very subtle.

Having read about how JPEG encoding is lossy, and you should not re-encode
such images I expected the visual changes to be more pronounced. I do not

IME, the corruption due to re-encoding is a lot less than I'd expected,
on the images I've done it on; almost never visible (to me anyway) at
full-image view, and only visible in certain areas even close up. YMMV.

know which quality was used on the output file (just allowed it to use its
default to see what the end result would be), but seeing that the filesize
became less I think I can safely assume it wasn't a 100%.

I don't _think_ there is such a thing as 100% (i. e. lossless) JPEG
compression, other than just file copying (which isn't JPEG compression
of course). There _is_ lossless _cropping_ of JPEGs, and also lossless
rotation. (The cropping has to be on block boundaries - 16 pixels, I
think.)

:-) "incorrectly coded" ? If that where so IE would not be able to
display them either.

Not necessarily; as JJ has said, IE may be repairing too.

Apart from that being a (wild? educated?) guess, seeing that the repairing
doesn't seem to take any noticable time (read: is easy to do), why would a
browser like FF not have incoorporated it too ? Let alone some of Windows
own default programs, like the mentioned picture-and-fax viewer, and paint.

If the "bad" data was in the header, and IE corrects it, then there's be
nothing visible.

I'm sorry, but that is a bit to much "if"-ing to my liking, and *way* to
little to back it up. Possible ? Ofcourse. Likely ? Not really.

If the source of your strange images is unusual, then it is possible. I
think you did hint that it might have been the output of a particular
scanner (driver), which, if not a popular make/model, might produce
output that complies with the standard (or possibly with a draft version
thereof), but in a way that isn't often seen.

(When JJ said "corrupted data", he I'm pretty sure was not distinguishing
between the various parts of the file, such as header and pixel data.)

I don't think so either. But it does not make a difference either way.
If its random corruption than the chance that my batch of 29 images (or the
80 I checked out later) all could be repaired is astronomical small.

Agreed, actual corruption likely to be in the image part of the data.

*Especially* as the JPEG encoding does not seem to have any error correction
on its data, let alone on the header.

If its *not* random corruption (but maybe just some programs writing
malformed headers) than it should by now be known to a *lot* of people, most
particulary the likes of FireFox and FreeImage.

See above - not if it's from a rare source. (Can you tell us what the
origin _was_ for these 29 rogue images? That might jog someone's memory
["oh yes, I remember now, images from xyz scanner model n never "played
nice" with any but their own viewer and IE"]!)

In short, corruption (either of them) isn't really likely.

Than again, I also have nothing to back that up. :-( :-)

(-:

Regards,
Rudy Wieser

P.s.
Of the hunderd images FreeImage (the DLL) could not read I've got only a
single one which also fails to be re-encoded using GDI+

Is it 29 images or 100?
[]
--
J. P. Gilliver. UMRA: 1960/1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

The early worm gets the bird.

John,

Not me, but it might you: if you've looked at the headers (which to some
extent at least implies you know what you're looking at!),

I know how to display the header data, and that (at least in GIF images)
*some* of the fields are effectivily duplicates (not sure abot JPEG
anymore).

you might detect a pattern in the rogue images that you don't see in
others.

Yeah, I could try to go that way. Though it assumes that there is an
actually detectable pattern, and not something subtile or certain
combinations of (not always the exact same) of values. In short, it would
take quite some time, and lots of guessing. And its not even certain that
the problem is located in those headers ...

Which is pretty-much why I posted here.

Someone (was it you yourself?) suggested the images might be to more than
8 bits (256 levels) of greyscale, for example.

That wasn't me, but I gave a reason why such a thing could happen (in short:
old, limited capable viewers). Stemming from never having heard about GIF
images with more than 256 colors (which is exactly why PNG was created :-) )

Go on. Give it a try. Even if nothing to do with the current subject (-:.

I might, even if I currently (do not seem to) have no use for it. As you
said, I like well-behaving software. :-)

I don't _think_ there is such a thing as 100% (i. e. lossless) JPEG
compression

There is though. As far as I know the "quality" has to do with how similar
the colors of a certain square of pixels need to be to combine them into a
block. With %100 indicating they need to be *exactly* the same.

So yes, AFAIK JPEG is able to store an image in a loss-less way. It comes
at a (high) cost though: the filesize will normally exeede that of a GIF or
PNG - because it doesn't use compression (but just "similar color"
collapsing - which at 100% quality is effectivily disabled).

If the source of your strange images is unusual, then it is possible. I
think you did hint that it might have been the output of a particular
scanner (driver), which, if not a popular make/model, might produce output
that complies with the standard (or possibly with a draft version
thereof), but in a way that isn't often seen.

I have no clue if the involved "bad" images came from the same or similar
encoder programs, it was just something that crossed my mind as a plausable
explanation - because of something you (better than I did) described as
"possibly a draft version".

Agreed, actual corruption likely to be in the image part of the data.

That, and the chance that *all* the header data (regardless of where the
corruption struck in it) can be recreated from the surrounding/following
(image) data is not too high either.

If its *not* random corruption (but maybe just some programs writing
malformed headers) than it should by now be known to a *lot* of people,
most particulary the likes of FireFox and FreeImage

See above - not if it's from a rare source.

More likely that it its known, but either considered too insignificant or
just not being part of the specs anymore.

But than its funny that IE still supported it though (XP, and IE that comes
with it, is already rather old). Maybe one of MicroSofts own extensions to
the specs ? Would not surprise me in the least.

Can you tell us what the origin _was_ for these 29 rogue images?

Nope. Sorry. Just that there where multiple sources. A quick peek at
possible metadata did not turn up anything either.

Of the hunderd images FreeImage (the DLL) could not read I've got only a
single one which also fails to be re-encoded using GDI+

Is it 29 images or 100?

Both. The first scan I did was a limited one (because having too much to
look at is as deadly as having too little) and turned up 29 images. A
later, full scan revealed the others.

Regards,
Rudy Wieser