Lost App

  • Thread starter Thread starter BeeJ
  • Start date Start date
Not necessarily - they extended the jpg standard in the early 90's to
include a lossless mode. That said, I don't know if it is capable of
compressing a 4MB bit mapped image down to 125KB...
Click to expand...

Be realistic for heaven's sake. How can 4GB be converted losslessly to
125KB? If that were possible, I should be able to convert 125 cents to
125K bucks!

They are referring to lossless rotation of images as I mentioned before. --
choro
*****
 
yeah, just got a lossless from 4.1MB down to 23kB using IrfanView 4.25
operating on a simple .bmp
Click to expand...

You must be out of your mind or else you don't know what you are talking
about!

But we know that .bmp files, just like .tif files, are fairly massive.--
choro
*****
 
Robert said:
yeah, just got a lossless from 4.1MB down to 23kB using IrfanView 4.25
operating on a simple .bmp
Click to expand...

There are various ways to reduce the amount of information in a picture.

1) Reduce the pixel dimensions. Take a 1600 by 1200 image (2 million pixels,
6 million bytes in 24 bit color) and reduce it to 320 by 200 pixels.

That ruins the visual appearance of the picture, so I guess I'd still
class such a method as lossy. There is less info in the picture than
originally. But no compression code is involved (no LZW for example).

So if that's what you're doing, changing the resolution, that can help
achieve more than 3:1 improvement. As long as the image is viewed on
a small screen, or printed on wallet sized photo paper, it might still
be acceptable.

2) Actual compression. If you start with the 1600 by 1200 image and
compress it losslessly, the file might be reduced from 6MB to 2MB.
You can get roughly 3:1 compression, without changing the pixel dimensions
of the picture. If you seek a higher compression ratio, more information is
discarded. JPEG does this, in the frequency domain, tossing higher
frequency (sharper edges in image) content. Once the lossy compression
reaches 144:1, it's getting pretty useless. Video compression works the
same way, using "macroblocks" and DCT to work in the frequency domain.
The compression method throws away information that doesn't bother
the human eye too much. Especially in video, you can throw away a lot
of info, and still have usable content. (In video, 100:1 would not be
unusual.) Still pictures, your eye and brain are less willing to
compromise, because there's plenty of time to detect the problems with it.

See the examples (JPEG "tombstone" images) about 70% of the way down
this web page. The samples show what happens when the resolution is
held constant, and the "Q factor" is varied. Q=100 is a compression
ratio of 2.6:1, and is still slightly lossy.

http://en.wikipedia.org/wiki/JPEG

One kind of error in JPEG, is roundoff error in the color space.
You can start with a 256 color image, run it through JPEG. Later,
when viewing the JPEG, your viewer application will request the
screen operate in 24 bit or 32 bit mode, for best results. If you
use a program which counts "color palette entries", the output
from JPEG might take 100,000 colors to represent. What you'll
find is there are a number of colors which are very close to
one another. They all should have been the same color, but due
to math roundoff errors, the color palette refers to the pixels
as being different colors. So that's another kind of degradation
which is not apparent to the human eye, and is only of theoretical
interest. If you need to convert the 100,000 color image back
to 256 colors, it's not hard to do with the appropriate quantizer
code, so it's really not a big deal. It's just an example of
another aspect, not widely considered, of lossy operation. The
colors were not preserved with good accuracy. The human eye
isn't good enough to detect the difference.

Paul
 
yeah, just got a lossless from 4.1MB down to 23kB using IrfanView 4.25
operating on a simple .bmp
Click to expand...

You're apparently starting with an image that lacks any significant
detail or something. Otherwise, I don't see how you're getting that
much *lossless* compression out of it.
 
You're apparently starting with an image that lacks any significant
detail or something. Otherwise, I don't see how you're getting that
much *lossless* compression out of it.
Click to expand...

The image of a blank sheet of paper under diffused lighting with no
shiny spots, may be?!--
choro
*****
 
The image of a blank sheet of paper under diffused lighting with no
shiny spots, may be?!--
choro
*****
Click to expand...

I had the same idea, only in reverse, such as a photo taken inside a
dark shoe box in a dark closet in the middle of the night, without
flash, of course. *shrug* Who knows, but he's getting some amazing
compression ratios even for lossy, let alone lossless.
 
I had the same idea, only in reverse, such as a photo taken inside a
dark shoe box in a dark closet in the middle of the night, without
flash, of course. *shrug* Who knows, but he's getting some amazing
compression ratios even for lossy, let alone lossless.
Click to expand...

He's obviously discovered the "Black Hole" method of compacting image
files, don't you think?--
choro
*****
 
On 05/10/2012 20:42, Robert Macy wrote:
...snip....

You must be out of your mind or else you don't know what you are talking
about!

But we know that .bmp files, just like .tif files, are fairly massive.--
choro
*****
Click to expand...

With initial respect, experiment, try. Quit arguing about the origin
of the Nile. Go LOOK for it!

I said I just DID it. And, you tell me I'm "...out of my mind..." I
have experimentation results. Data. You have words. If I were lurking,
I would trust experimentation, data, NOT words.

Now, did you try it?
 
snip....

You're apparently starting with an image that lacks any significant
detail or something. Otherwise, I don't see how you're getting that
much *lossless* compression out of it.
Click to expand...

YES!! of course. A PICTURE in .bmp does not reduce much. Line art,
schematics, etc reduce substantially.

IrfanView ver 3.2 used to blurr the line art - as expected. But, In
ver 4.25 the author changed IrfanView for the better. So, I use it to
compress .bmp line art files for attachments to emails to people whose
systems won't let them receive .zip files, but will allow them to
receive .jpg
 
YES!! of course. A PICTURE in .bmp does not reduce much. Line art,
schematics, etc reduce substantially.

IrfanView ver 3.2 used to blurr the line art - as expected. But, In
ver 4.25 the author changed IrfanView for the better. So, I use it to
compress .bmp line art files for attachments to emails to people whose
systems won't let them receive .zip files, but will allow them to
receive .jpg
Click to expand...

Ok, good, line art goes a long way toward explaining things, but
still, I have serious doubts that the compression is lossless. Try
reversing the procedure, (jpg back to bmp), then do a binary
comparison between the original and the newly created bmp. I doubt
they'll be identical.
 
With initial respect, experiment, try. Quit arguing about the origin
of the Nile. Go LOOK for it!

I said I just DID it. And, you tell me I'm "...out of my mind..." I
have experimentation results. Data. You have words. If I were lurking,
I would trust experimentation, data, NOT words.

Now, did you try it?
Click to expand...

No doubt you have invented the impossible where all others have failed.
How you also invented the perpetual motion machine? Yes, of course, I
know; you are now working on how to stop it!--
choro
*****
 
Ok, good, line art goes a long way toward explaining things, but
still, I have serious doubts that the compression is lossless. Try
reversing the procedure, (jpg back to bmp), then do a binary
comparison between the original and the newly created bmp. I doubt
they'll be identical.
Click to expand...

May be the chap needs new reading glasses! ;-)
That would explain things.--
choro
*****
 
Robert said:
With initial respect, experiment, try. Quit arguing about the origin
of the Nile. Go LOOK for it!

I said I just DID it. And, you tell me I'm "...out of my mind..." I
have experimentation results. Data. You have words. If I were lurking,
I would trust experimentation, data, NOT words.

Now, did you try it?
Click to expand...

There's a way to check for loss.

Using Photoshop (or another image editor with arithmetic capabilities),
load the original picture. Then load the "compressed" picture. Resize
the compressed picture, so the pixel dimensions are the same. (They have
to be the same, so you can subtract them.) Now ask the tool to do "A-B"
or subtraction of one image from the other. If the images were identical
(i.e. lossless), the difference would be a uniform black result
(difference of zero). If the computed difference between the images
results in non-zero pixels, then the images are not the same, and
there is a loss detected.

Paul
 
BeeJ said:
Sigh. Stolen thread. Overloaded with OT.
Click to expand...

I've been watching for your return, I stopped reading all the OT posts
that weren't threaded to your replies. As has been mentioned Irfan
Viewer and XnView are good viewers with some editing ability (I use
Irfan), but that doesn't answer your original question. I don't recall
seeing a specific "Darken background" menu option in apps I have used,
and a quick google doesn't turn up anything specific. Can you remember
anything else about the program... other features, any part of the name,
the logo?
 
I've been watching for your return, I stopped reading all the OT posts
that weren't threaded to your replies.  As has been mentioned Irfan
Viewer and XnView are good viewers with some editing ability (I use
Irfan), but that doesn't answer your original question.  I don't recall
seeing a specific "Darken background" menu option in apps I have used,
and a quick google doesn't turn up anything specific.  Can you remember
anything else about the program... other features, any part of the name,
the logo?
Click to expand...

I responded suggesting IrfanView, because of experience using it to
'lighten' the images I get from Real Estate postings. Either their
machine, or mine, ALWAYS made the images so dark that it was
impossible to even see what you're looking at. Kodak Editor kind of
works, most images are now incompatible with it. Using Irfanview, load
photo, shift gamma to 3.00, adjust brightness, and VOILA! the dark
background suddenly reveals all kinds of details, impossible to see
before.

Often simplly invoking 'Auto adjust colors' works enough.
 
After serious thinking Robert Macy wrote :
I responded suggesting IrfanView, because of experience using it to
'lighten' the images I get from Real Estate postings. Either their
machine, or mine, ALWAYS made the images so dark that it was
impossible to even see what you're looking at. Kodak Editor kind of
works, most images are now incompatible with it. Using Irfanview, load
photo, shift gamma to 3.00, adjust brightness, and VOILA! the dark
background suddenly reveals all kinds of details, impossible to see
before.

Often simplly invoking 'Auto adjust colors' works enough.
Click to expand...

Yes, there are many apps that will do similar but with multi-step
operations. The one I had did it with one click. Then click again to
get more etc.

I have googled many times an found nothing.
I have even started up old PCs to see what I had there.
I have one more old PC to look at.

Unfortunately I can not remember any thing more about the app.
This was from about six or seven or more years ago that I used it.
Probably on Windows 2000.
I have had may PC crashes and rebuilds over the decades (yes, decades.
Altair or IMSAI ring any bells?) so maybe it is gone forever. Tragic.

If I knew the logic for darken background I could write software to do
it but I am not sure how that effect differentiates foreground from
background.

Maybe:
stuff around the edges is background 'cause you usually point at and
center the foreground.
sample stuff around the background and see what range is there.
anything below that range of color/saturation/brightness level you
further darken.

Does that seem right?

I may just try your suggestion with GIMP (then Irfanview) and see what
happens.

In one app I have there is a slider for shadow areas that will darken.
Maybe that is what the one-click app sort of did.

I have Corel, Nero, GIMP, Elements and MS to play with.
Problem is,I have a lot of photos to "correct" so that one-click app
was wonderful.
 
BeeJ said:
After serious thinking Robert Macy wrote :

Yes, there are many apps that will do similar but with multi-step
operations. The one I had did it with one click. Then click again to
get more etc.

I have googled many times an found nothing.
I have even started up old PCs to see what I had there.
I have one more old PC to look at.

Unfortunately I can not remember any thing more about the app.
This was from about six or seven or more years ago that I used it.
Probably on Windows 2000.
I have had may PC crashes and rebuilds over the decades (yes, decades.
Altair or IMSAI ring any bells?) so maybe it is gone forever. Tragic.

If I knew the logic for darken background I could write software to do
it but I am not sure how that effect differentiates foreground from
background.

Maybe:
stuff around the edges is background 'cause you usually point at and
center the foreground.
sample stuff around the background and see what range is there.
anything below that range of color/saturation/brightness level you
further darken.

Does that seem right?

I may just try your suggestion with GIMP (then Irfanview) and see what
happens.

In one app I have there is a slider for shadow areas that will darken.
Maybe that is what the one-click app sort of did.

I have Corel, Nero, GIMP, Elements and MS to play with.
Problem is,I have a lot of photos to "correct" so that one-click app
was wonderful.
Click to expand...

Yeah, there are little features in some older apps that I just don't see
in apps I use now. In 1998 I bought a Visioneer scanner with PaperPort
software (before ScanSoft took over the software), and it had some
editing features for the scanned images. One feature was a one-click
button to slightly straighten a tilted image, in tiny increments with
each click.... it was very handy, and I don't see it in any software I
use now.
 
BeeJ said:
After serious thinking Robert Macy wrote :

Yes, there are many apps that will do similar but with multi-step
operations. The one I had did it with one click. Then click again to
get more etc.

I have googled many times an found nothing.
I have even started up old PCs to see what I had there.
I have one more old PC to look at.

Unfortunately I can not remember any thing more about the app.
This was from about six or seven or more years ago that I used it.
Probably on Windows 2000.
I have had may PC crashes and rebuilds over the decades (yes, decades.
Altair or IMSAI ring any bells?) so maybe it is gone forever. Tragic.

If I knew the logic for darken background I could write software to do
it but I am not sure how that effect differentiates foreground from
background.

Maybe:
stuff around the edges is background 'cause you usually point at and
center the foreground.
sample stuff around the background and see what range is there.
anything below that range of color/saturation/brightness level you
further darken.

Does that seem right?

I may just try your suggestion with GIMP (then Irfanview) and see what
happens.

In one app I have there is a slider for shadow areas that will darken.
Maybe that is what the one-click app sort of did.

I have Corel, Nero, GIMP, Elements and MS to play with.
Problem is,I have a lot of photos to "correct" so that one-click app was
wonderful.
Click to expand...

Is the technique "dodge and burn" ?

http://www.photoshopessentials.com/photo-editing/dodge-burn/

http://www.photoshopessentials.com/images/photo-editing/dodge-burn/photo-exposure-corrected.jpg

If so, perhaps you should be looking for "auto dodge and burn".

Although, such a tool is going to have to recognize what is
in the foreground, what is part of the background, and paint
them accordingly. An automated tool might end up with a
counter-intuitive result (i.e. you can't trust it).

If your old tool just "adjusted curves", then that wouldn't be
nearly as effective.

Paul
 
Back
Top