Image size , A technical puzzle.

On 7/10/2015 12:08 PM, Alan Browne wrote:
On 2015-07-09 22:32, PeterN wrote:
I recently submitted two images to a competition. Bothe were 100 ppi and
measured, in pixels, 1020 x 768, both were saved at the same JPEG
compression level. Both files were saved as 8 bit JPEG.

One of the images was a bit over 500 k. The other was over twice the
size.

The size of the second image exceeded the size limit for the
competition, so I saved at as a slightly lower quality, to comply with
the size limit.

Why would there be such a large a difference?

The only thing I can thinnk of is that the smaller image was cropped a
bit more than the larger.

Assuming two different images, the one with the most information
'change' will be larger (file size) than the other for the same pixel
dimensions.

Think of a placid scene, a smooth lake with few reflections...

v. a busy Times Square scene with a lot going on. More information has
to be encoded more often in the image v. the smooth scene where
compression is most efficient.


Gotcha1 Thanks.

--
PeterN

On 7/10/2015 12:20 PM, rickman wrote:
On 7/10/2015 11:37 AM, PeterN wrote:
On 7/10/2015 9:57 AM, Mayayana wrote:
As others have said, details make the difference. JPG
compresses by dropping out data, which is why over
processing shows little rectangles. I just saved a 1020
x 768 JPG at compression level 8 and it's under 5 KB.
But it's just a solid blue field, so it's easy to store that
data as something that boils down to "783360 instances
of pixels of color 0, 0, 256". Interestingly, when I opened
the JPG in a hex editor it turns out that most of the
5 KB is just null byte filler. The "start of stream" marker
that indicates the beginning of the image data is two
bytes, FF DA. The actual image data seems to be only
17 or 18 bytes.



thanks. I just confirmed that concept. Thats a lot for my septgenerian
brain to comprehend. It makes sense.

It is a bit over simplified. JPG is a form of lossy compression which
means on expansion you won't recover the exact bits of the original
image. You will recover an image that is similar to the original in
many ways. The degree of compression trades off against the
"similarity" of the recovered image to the original.

I feel that calling this sort of compression, "dropping out data" to be
inaccurate. That sounds like individual bits or other data is just
eliminated from the image which is not how it works.


OK thanks.

--
PeterN

| I feel that calling this sort of compression, "dropping out data" to be
| inaccurate. That sounds like individual bits or other data is just
| eliminated from the image which is not how it works.
|

I don't entirely understand the method, but
my impression was that it will match up contiguous
pixels in order to arrive at less actual pixel data.
When that goes far enough there are lines and
rectangles visible. I'd call that dropping out
data. If you have two pixels of different hues
and they're matched up to share the same hue,
that's data dropped out that can never be retrieved.
If my understanding is wrong I welcome a correction.

On 7/10/2015 12:47 PM, Tony Cooper wrote:
On Fri, 10 Jul 2015 11:46:05 -0400, PeterN
wrote:

On 7/10/2015 10:33 AM, Tony Cooper wrote:
On Fri, 10 Jul 2015 04:15:29 -0400, nospam
wrote:

In article , Tony Cooper
wrote:

based on his description, it's neccc, which is a rather well known
event with well paid staff.

It's an area-wide umbrella organization that any camera club in the
New England area can join. Most states, or groups of states, have
one. It's not well-known outside of New England any more than the
FCCC is well-known outside of Florida.

There are CCCs all across the country.

I am absolutely amazed that you know the salaries of the people in an
organization clear across the country from you. You wouldn't be
making **** up again, would you?

i do not make up ****.

i know several people who have been neccc staff or speakers at one time
or another and they were paid.

Quite frankly, I don't believe you. I think that you are lying when
you say you know NEFCCC staff.

While I think it's a lie, it does make sense. You've repeatedly
called them all morons in this thread, and it does make sense that you
hang out with morons.

You said "well paid", not "paid". There's a difference. McD counter
help is paid. They are not well paid.


and how is it you know where i live? you wouldn't be making **** up
again, would you?

another thing you got wrong is that neccc is actually well known
outside of new england, as is http://www.swmccc.org outside of
michigan, because they're the two with model shoots. while most
attendees will be relatively nearby, not all of them will.

Model shoots make a program nationally known? Bull****. My local
club (which is a member of FCCC) has model shoots twice a year at the
meetings. The last one had three lighting and background set-ups.

I wouldn't presume to call anyone I don't know a moron like you do,
but I am beginning to think that the people who run the NECCC are
somewhat disorganized.

For the model shoot at the Amherst meeting, the rules state that
models must be sent photographs taken of them in the format of 1024 x
768 with a "dpi of 240".

The rules also state "Photographers should send the Photoshopped
images with the model's name to the individual model."

and "Please feel free to send as many tweaked digital images as you
with to models".


Photoshopped? Tweaked?

The general understanding of the use of "Photoshopped" (a usage
deprecated by Adobe) is "highly manipulated in post". So, I guess you
can "tweak" a model's photograph and give her a boob job.


Many, but not all of The NECCC speakers are paid, but not by NECCC. Many
are sponsored, i.e. Nikon or Canon pay them. In some cases the speakers
appearances are for the opportunity to promote their workshops. In any
case, speakers are irrelevant to the point. The speakers are not the
ones who determine the competition rules. Those who do are unpaid
volunteers.

That's pretty standard for this type of organization. Our group
schedules speakers who are paid by Canon, Nikon, and some of the
software companies. We also schedule non-paid speakers.

Some of the book-writing national names will appear for no fee, but
they hold a for-fee workshop in conjunction with their visit. The
for-fee workshops are, of course, optional but they are usually well
attended.

We have Mike Wacht and Adam Jones (Canon "Explorer of Light")
scheduled. That lady you know that wrote the book on infared
photography was a recent speaker. Can't think of her name. Bryan
Peterson spoke last year, and Joe McNally spoke some time ago.

I think you are talking about Deb Sandige. She does long exposure and
infrared.

At just about all the lectures I've been to I have found the presenters
to be quite approachable, and helpful.


McNally was a judge at one of the competitions, and one of the most
articulate critiquers we've had.

He is one of the most sought after speakers. We would really like to get
him as a judge for our club. I have no clue who to call, it would depend
on whether he is sponsered.

--
PeterN

| What you are describing is just a matter of lowering resolution. That
| may happen when the compression is set high, but there is more to JPG
| compression than just reducing the resolution.
|

I was just reading about quantization. I got
curious after this discussion and realizing I
don't really know how it all works. It seems
quantization requires a math background to
*really* understand. Or at least knowledge
of many technical terms. I can see how you
mean that my definition was simplistic. It doesn't
seem to be terribly far off in principle, though.
Blocks of the image get simplified, making them
ZIP more efficiently.

But that brings in a new can of worms: How
do ZIP and similar methods work? Apparently
compression benefits from patterns, which are
increased with quantization. That might also
help explain why a ZIP (or a JPG) can't be notably
shrunk when put into another ZIP. Any reducible
patterns have presumably already been reduced.
Again, I only understand the actual process in
a general sense.

On Fri, 10 Jul 2015 08:24:14 -0400, PeterN
wrote:

I really wish someone would explain why two images with the same pixel
dimensions, and saved at the same compression level, would not be the
same size.

JPG incorporates an image compression algorithm. The size of the end
result depends on the complexity of the image to be compressed. A
comples image (say a landscape filled with trees) would not compress
as easily as, say, a uniform white image (or a uniform pink image, for
that matter).

Real-world images each have different levels of complexity and hence
are compressible to different extents. That's why when you compress
two different images you will end up with two different end results,
even though you have subjected them to exactly the same process.
--

Regards,

Eric Stevens

On Fri, 10 Jul 2015 08:05:40 -0400, PeterN
wrote:

On 7/9/2015 11:09 PM, nospam wrote:
In article , PeterN
wrote:

I recently submitted two images to a competition. Bothe were 100 ppi and
measured, in pixels, 1020 x 768, both were saved at the same JPEG
compression level. Both files were saved as 8 bit JPEG.

there is no ppi in a jpeg file.

there is a tag that *suggests* an initial size, such as for a page
layout app (and that tag may not necessarily be used, depending on the
app), but other than that, the tag is meaningless.

ppi only matters when printing.

Yes. They intend to print certain images.

But the 72ppi they require fit no printer but that was the common ppi
of CRT screens.



anyone who requests a jpeg file at a specific ppi has no clue.

See above.


the pixel dimensions are what matters, and clearly they're stuck in the
1990s if they want it at 1024x768.

See rule 22.
--

Regards,

Eric Stevens

In article , Mayayana
wrote:

But that brings in a new can of worms: How
do ZIP and similar methods work?

zip is lossless. jpeg is lossy. two different things.

Apparently
compression benefits from patterns, which are
increased with quantization. That might also
help explain why a ZIP (or a JPG) can't be notably
shrunk when put into another ZIP. Any reducible
patterns have presumably already been reduced.
Again, I only understand the actual process in
a general sense.

that is correct, and most of the time, the result will be slightly
larger if the original is non-compressible.

In article , Tony Cooper
wrote:

Evidently, the difference is that the group has published a different
set of rules for the competitions at the conference to be held in July
in Amherst and the other competitions during the year.

So, the conclusion is that the file size is limted for some
competitions but not limited at others. You have to define NEFCCC
entries for which competition.

Tony, I gotta agree with nospam. That is for a different competition,
the interclub competition. Not the NECCC atendee's competition.

I figured that out, but you did not tell us what the competition was,
and that it was different from the interclub competition. nospam
didn't seem to know either, since he just said it was probably an
NECCC entry.

while i wasn't 100% sure which contest it was, the evidence *clearly*
pointed to the neccc contest i originally linked, which turned out to
be exactly correct.

I don't think he was aware that there are two sets of
rules, but he won't admit this.

suffice it to say that i know *much* more about neccc than you do.

In article , Tony Cooper
wrote:

based on his description, it's neccc, which is a rather well known
event with well paid staff.

It's an area-wide umbrella organization that any camera club in the
New England area can join. Most states, or groups of states, have
one. It's not well-known outside of New England any more than the
FCCC is well-known outside of Florida.

There are CCCs all across the country.

I am absolutely amazed that you know the salaries of the people in an
organization clear across the country from you. You wouldn't be
making **** up again, would you?

i do not make up ****.

i know several people who have been neccc staff or speakers at one time
or another and they were paid.

Quite frankly, I don't believe you. I think that you are lying when
you say you know NEFCCC staff.

i never said i know any nefccc staff.

prior to this thread, i never heard of nefccc, which is the northeast
florida cancer control collaborative, something else entirely.

however, i do know people who have been or are affiliated with neccc.

While I think it's a lie, it does make sense. You've repeatedly
called them all morons in this thread, and it does make sense that you
hang out with morons.

more of your insults.

You said "well paid", not "paid". There's a difference. McD counter
help is paid. They are not well paid.

you're desperately trying to argue for the sake of arguing.

and how is it you know where i live? you wouldn't be making **** up
again, would you?

another thing you got wrong is that neccc is actually well known
outside of new england, as is http://www.swmccc.org outside of
michigan, because they're the two with model shoots. while most
attendees will be relatively nearby, not all of them will.

Model shoots make a program nationally known? Bull****.

yes they do, particularly for those who do model photography.

on the other hand, someone who shoots landscapes, sports or wildlife
might not care as much.

My local
club (which is a member of FCCC) has model shoots twice a year at the
meetings. The last one had three lighting and background set-ups.

they do that because it's popular.

neccc has roughly 20 models, with 3 indoor settings and multiple
outdoor settings. there are also additional indoor settings with
character models such as clowns, sailors, ballerinas, animals, still
life and more.

there are sometimes special events, such as what they have planned this
year, which is a revolutionary war reenactment:
http://neccc14.neccc.org/2015_conf/2015_Special_70th_Events.pdf

swmccc has model shoots that are done off-site, which is actually
better since the neccc setting is a feeding frenzy with non-ideal
settings (at least outside).

I wouldn't presume to call anyone I don't know a moron like you do,
but I am beginning to think that the people who run the NECCC are
somewhat disorganized.

let's see you put on a conference of that size.

i call people morons when they do stupid things like insist submissions
be 1024x768 with a 'resolution of 100'.

'resolution of 100' is meaningless. whoever wrote that is an idiot. at
least they could say pixels per inch.

For the model shoot at the Amherst meeting, the rules state that
models must be sent photographs taken of them in the format of 1024 x
768 with a "dpi of 240".

how is that disorganized?

they say that because the images will typically be emailed and their
goal is to limit the size of email attachments. the 'dpi of 240' is
meaningless. it can be anything.

there are a lot of idiots who click send on a full 24-36 mp image,
possibly filling the recipients inbox and blowing any data cap if
they're on mobile device.

the photos are to help the model build a portfolio, not for exhibition
in a gallery. if a model wants a higher resolution version or maybe
additional photo shoots, they can contact the photographer directly.

before digital, prints were mailed to someone at neccc, who then
distributed them to the models. that way, the models didn't need to
publicize their address. there were also size restrictions on that too.


The rules also state "Photographers should send the Photoshopped
images with the model's name to the individual model."

and "Please feel free to send as many tweaked digital images as you
with to models".

Photoshopped? Tweaked?

yes.

people often make minor adjustments (i.e., tweak) their photos, such as
white balance, exposure adjustment, leveling horizons, cropping, minor
retouching, etc.

some people go further and explore creative options.

The general understanding of the use of "Photoshopped" (a usage
deprecated by Adobe)

so now you're finally admitting that? hilarious.

a year or two ago i said the same thing but you argued it was common
usage and not up to adobe. yet another one of your contradictions. it's
hard to keep up.

is "highly manipulated in post".

no it isn't. you made that part up.

'photoshopped' simply means edited in photoshop, however, it has become
a generic term for when photoshop itself was not used.

all they are saying that you don't need to send an unadulterated raw.
you can (and actually should) send a retouched and modified version.

So, I guess you
can "tweak" a model's photograph and give her a boob job.

yes, someone could. so what?

if that's the type of work you'd send to a model, then i would suggest
landscapes or flowers.