Sign in with
Sign up | Sign in
Your question

Nikon D70S JPG file size?

Last response: in Digital Camera
Share
Anonymous
June 21, 2005 10:53:05 PM

Archived from groups: rec.photo.digital.slr-systems (More info?)

I routinely use RAW files, so this is just an academic question.

The Nikon D70S image size is 6 megapixels, which is 18 MB uncompressed.
The manual pg 41 says JPG file compression is:

Fine 1:4 compression, so file should be about 18/4 = 4.5MB
Normal 1:8 compression, so file should be about 18/8 = 2.25MB
Basic 1:16 compression, so file should be about 18/16 = 1.12MB

Given that JPG compression varies some with image content, but mine are
routinely quite a bit smaller, nearly half, like 2.5, 1.2, 0.6 MB.
This is rather different than 1:4, 1:8, and 1:16.

The largest Fine file (frame full of tiny busy detail) that I've been
able to create is about 3.2 MB. These JPG artifacts are not bad, but
worse than I'd expect for 1:4, and of course, it isnt 1:4.

I assume others see the same file sizes? It seems arbitrary what size
they are, except that they state numbers which seem inaccurate.
Anonymous
June 21, 2005 10:53:06 PM

Archived from groups: rec.photo.digital.slr-systems (More info?)

In article <lmZte.1603$G4.331@trnddc09>, Wayne <nospam@invalid.com> wrote:
>I routinely use RAW files, so this is just an academic question.
>
>The Nikon D70S image size is 6 megapixels, which is 18 MB uncompressed.
>The manual pg 41 says JPG file compression is:
>
>Fine 1:4 compression, so file should be about 18/4 = 4.5MB
>Normal 1:8 compression, so file should be about 18/8 = 2.25MB
>Basic 1:16 compression, so file should be about 18/16 = 1.12MB
>
>Given that JPG compression varies some with image content, but mine are
>routinely quite a bit smaller, nearly half, like 2.5, 1.2, 0.6 MB.
>This is rather different than 1:4, 1:8, and 1:16.
>
>The largest Fine file (frame full of tiny busy detail) that I've been
>able to create is about 3.2 MB. These JPG artifacts are not bad, but
>worse than I'd expect for 1:4, and of course, it isnt 1:4.
>
>I assume others see the same file sizes? It seems arbitrary what size
>they are, except that they state numbers which seem inaccurate.

I think that those are the worst-case sizes, which probably
assume all adjacent pixels differ wildly.

I can simply state for my most often used JPEG format, Medium
size, Fine compression, the camera predicts a maximum of 522 images on a
1GB CF card, while in reality, the only time I've actually filled the
card, I got 705 images in reality.

So -- with two assumptions of what is meant by Gigabyte
(marketing (1,000,000,000) vs computer (1073741824)), this says that the
image size under these conditions would be:

Count Marketing Computer
=================================
522 1.92 MB 2.06 MB (predicted)
705 1.42 MB 1.52 MB (real)

Note that one of the changes in the firmware between the D70 and
the D70s was an improvement in the prediction accuracy for the raw (NEF)
shots. It has not improved the prediction accuracy for the JPEG images,
however -- as is shown by the fact that I installed the updated
firmware, and the predicted number of Medium/Fine JPEGs has not changed.

Now -- from a directory which contains 197 of the JPEG images, I
find a total size of 255352022 bytes, or an average of 1.29 MB per image
-- which is closer to the "Marketing" assumption to the size of a GB on
the CF card. (But -- remember that there is overhead in the file
system, including the directory, the FAT, and unused tail ends of blocks
allocated to hold the file.

I've got some Large/Fine images, which are a bit over 3MB.

Enjoy,
DoN.
--
Email: <dnichols@d-and-d.com> | Voice (all times): (703) 938-4564
(too) near Washington D.C. | http://www.d-and-d.com/dnichols/DoN.html
--- Black Holes are where God is dividing by zero ---
Anonymous
June 22, 2005 3:24:51 AM

Archived from groups: rec.photo.digital.slr-systems (More info?)

In article <d99tnn$g4o$1@Fuego.d-and-d.com>, dnichols@d-and-d.com says...

> I've got some Large/Fine images, which are a bit over 3MB.


Yeah, same for my largest ones. I could believe a 3500KB max case, but my
largest are more like 3350KB (and I've tried a bit).

I was careless with my 18 MB computation of 6x3. It should be /1024 =
17625KB. Pointless to use MB since Windows doesnt use MB (but Adobe does).

For example, file size:

Shown by Windows command line DIR, units are bytes ( /1, exact value)

Shown by Windows file Explorer is KB (/1024, about 2.4% smaller than
millions). Its rounding is slightly suspect, it rounds up surprisingly.

MB is ( /1024x1024, nearly 4.9% smaller than millions), but MB is not used
by Windows file Explorer, affecting size comparisons. 1 MB would be shown as
1024KB in Windows file Explorer.

So the point, to be less careless, the maximum probable JPG (3008x2000
pixels, Fine JPG setting) is maybe 3500 KB (but typical might be closer to
2500KB). So for this probable maximum case, this predicts 17625/3500 = 5:1
compression actual, which is 20% instead of 25%. However typical case is
more like 17625/2500 = 7:1 compression. Nikon calls it 4:1.
Related resources
Can't find your answer ? Ask !
Anonymous
June 22, 2005 3:24:52 AM

Archived from groups: rec.photo.digital.slr-systems (More info?)

In article <7l1ue.2728$Uj.2264@trnddc08>, Wayne <nospam@invalid.com> wrote:
>In article <d99tnn$g4o$1@Fuego.d-and-d.com>, dnichols@d-and-d.com says...
>
>> I've got some Large/Fine images, which are a bit over 3MB.
>
>
>Yeah, same for my largest ones. I could believe a 3500KB max case, but my
>largest are more like 3350KB (and I've tried a bit).

Hmm ... perhaps a coarse fabric, perhaps photographed with the
threads forming a diagonal? Or perhaps some millimeter division graph
paper similarly oriented?

>I was careless with my 18 MB computation of 6x3. It should be /1024 =
>17625KB. Pointless to use MB since Windows doesnt use MB (but Adobe does).

Well ... I was working on a unix system, where the sizes are
given to the byte -- unless you ask for the options to select other
units.

>For example, file size:
>
>Shown by Windows command line DIR, units are bytes ( /1, exact value)
>
>Shown by Windows file Explorer is KB (/1024, about 2.4% smaller than
>millions). Its rounding is slightly suspect, it rounds up surprisingly.

It may be telling you the actual disk space used, including the
remainder of disk blocks allocated but not fully filled by the end of
the file.

>MB is ( /1024x1024, nearly 4.9% smaller than millions), but MB is not used
>by Windows file Explorer, affecting size comparisons. 1 MB would be shown as
>1024KB in Windows file Explorer.

Not a problem for my system, at least. :-)

>So the point, to be less careless, the maximum probable JPG (3008x2000
>pixels, Fine JPG setting) is maybe 3500 KB (but typical might be closer to
>2500KB). So for this probable maximum case, this predicts 17625/3500 = 5:1
>compression actual, which is 20% instead of 25%. However typical case is
>more like 17625/2500 = 7:1 compression. Nikon calls it 4:1.

Perhaps just leaving elbow room for extremely atypical images
which compress very poorly.

Personally, I don't really care, as long as I have sufficient
room to store them. :-)

Enjoy,
DoN.
--
Email: <dnichols@d-and-d.com> | Voice (all times): (703) 938-4564
(too) near Washington D.C. | http://www.d-and-d.com/dnichols/DoN.html
--- Black Holes are where God is dividing by zero ---
Anonymous
June 22, 2005 5:05:20 AM

Archived from groups: rec.photo.digital.slr-systems (More info?)

In article <d9a9lh$kqi$1@Fuego.d-and-d.com>, dnichols@d-and-d.com says...

> Hmm ... perhaps a coarse fabric, perhaps photographed with the
>threads forming a diagonal? Or perhaps some millimeter division graph
>paper similarly oriented?

But who would want JPG artifacts for such detailed applications :)  And why
is maximum important? To me, the typical size case is much more important.

> Well ... I was working on a unix system, where the sizes are
>given to the byte -- unless you ask for the options to select other
>units.

Yeah, the file systems do keep record of the exact file size in bytes.
Windows command line DIR shows the exact file size in bytes, and all other
displays are a distortion of that. I dont know why Windows keeps the 1024
concept for files, there is no point of it. This concept is totally
meaningless to disk file size. Even hard disk purchase size is sized in
millions of bytes today too, not in binary MB until we format them.

> It may be telling you the actual disk space used, including the
>remainder of disk blocks allocated but not fully filled by the end of
>the file.

Possibly it may be erring that that direction, but it doesnt compute slack.
I think the Windows Explorer simply divides file size by 1024, and then
(usually) rounds up to the next unit of 1K, regardless of the disk cluster
size.

> Personally, I don't really care, as long as I have sufficient
>room to store them. :-)

Me too, and RAW has other advantages too - the white balance control is
wonderful. Plus no artifacts nor editing concerns. I was just wondering why
the Nikon FINE file really isnt very close to the 4:1 claimed for it. I'd
like to have a better JPG choice, for just in case someday maybe.
June 22, 2005 2:02:28 PM

Archived from groups: rec.photo.digital.slr-systems (More info?)

Wayne wrote:

<snip>
> The largest Fine file (frame full of tiny busy detail) that I've been
> able to create is about 3.2 MB. These JPG artifacts are not bad, but
> worse than I'd expect for 1:4, and of course, it isnt 1:4.
<snip>

What jpeg artifacts? I've been using a D70 for a while now, and at fine
setting, I don't recall ever seeing any evidence of jpeg artifacts on
screen, let alone on a print.
Maybe you can upload an example for us to see.
There might be good reasons to shoot in raw mode, but jpeg compression
artifacts in the D70 "fine" mode isn't one of them IMO.
Anonymous
June 22, 2005 2:02:29 PM

Archived from groups: rec.photo.digital.slr-systems (More info?)

In article <1119391282.787380@ftpsrv1>, nomail@nomail.com says...

>What jpeg artifacts? I've been using a D70 for a while now, and at fine
>setting, I don't recall ever seeing any evidence of jpeg artifacts on
>screen, let alone on a print.
>Maybe you can upload an example for us to see.


There are always artifacts in JPG, and certainly Fine size has them too.
Perhaps splitting hairs, but zoom the screen to 400% actual size, and then
they show clearly around the sharp edges. Not nearly as bad as Normal or
Basic, but they are there, and easily seen.

Note that looking at 400% is not a trick condition. It is just a learning
exercise, which helps you learn to see them at 100%. It helps you realize
what is in fact actually present.

Does this small degree of artifacts matter at 100% or printed?
No, not very much. Certainly the Fine JPG is a high quality JPG.
But certainly it is also still a JPG. The artifacts are definitely
present, regardless if you realize it or not.

This was the source of my original 4:1 question. I wish Fine were actually
as good as 4:1 would predict, but it isnt 4:1. Typically more like 7:1.

>There might be good reasons to shoot in raw mode, but jpeg compression
>artifacts in the D70 "fine" mode isn't one of them IMO.

Right, JPG is a wonderful reason to use RAW instead. RAW size is only
about 2X JPG file size (typical), not a high price to pay for what you
get. Also the RAW white balance control is fantastic too.
Anonymous
June 22, 2005 2:02:29 PM

Archived from groups: rec.photo.digital.slr-systems (More info?)

frederick <nomail@nomail.com> wrote:

| Wayne wrote:
|
| <snip>
|| The largest Fine file (frame full of tiny busy detail) that I've been
|| able to create is about 3.2 MB. These JPG artifacts are not bad, but
|| worse than I'd expect for 1:4, and of course, it isnt 1:4.
| <snip>
|
| What jpeg artifacts? I've been using a D70 for a while now, and at
| fine setting, I don't recall ever seeing any evidence of jpeg
| artifacts on screen, let alone on a print.
| Maybe you can upload an example for us to see.
| There might be good reasons to shoot in raw mode, but jpeg compression
| artifacts in the D70 "fine" mode isn't one of them IMO.

I agree with you.....

PC
June 22, 2005 6:21:18 PM

Archived from groups: rec.photo.digital.slr-systems (More info?)

Wayne wrote:
> In article <1119391282.787380@ftpsrv1>, nomail@nomail.com says...
>
>
>>What jpeg artifacts? I've been using a D70 for a while now, and at fine
>>setting, I don't recall ever seeing any evidence of jpeg artifacts on
>>screen, let alone on a print.
>>Maybe you can upload an example for us to see.
>
>
>
> There are always artifacts in JPG, and certainly Fine size has them too.
> Perhaps splitting hairs, but zoom the screen to 400% actual size, and then
> they show clearly around the sharp edges. Not nearly as bad as Normal or
> Basic, but they are there, and easily seen.
>
<snip>
Are you sure that this isn't the result of sharpening?
Anonymous
June 22, 2005 6:21:19 PM

Archived from groups: rec.photo.digital.slr-systems (More info?)

In article <1119406811.744884@ftpsrv1>, frederick <nomail@nomail.com> wrote:
>Wayne wrote:
>> In article <1119391282.787380@ftpsrv1>, nomail@nomail.com says...
>>
>>
>>>What jpeg artifacts? I've been using a D70 for a while now, and at fine
>>>setting, I don't recall ever seeing any evidence of jpeg artifacts on
>>>screen, let alone on a print.
>>>Maybe you can upload an example for us to see.
>>
>>
>>
>> There are always artifacts in JPG, and certainly Fine size has them too.
>> Perhaps splitting hairs, but zoom the screen to 400% actual size, and then
>> they show clearly around the sharp edges. Not nearly as bad as Normal or
>> Basic, but they are there, and easily seen.
>>
><snip>
>Are you sure that this isn't the result of sharpening?

They are most obvious when photographing file detail drawings
black on white, and then zooming in to see one of those details. In
particular, if you have two lines intersecting at an angle less than 90
degrees, you will see splotches of pixels in the supposedly white area.
This can make small print in such a drawing extremely difficult to read.

So -- for such material, I tend to choose TIFF instead of JPEG,
as it is a lossless format. JPEG compression is *always* a lossy
format. so what is displayed is not what was photographed.

Actually -- for that purpose, two-color GIF is the best format,
but there was that problem with a patented algorithm, so it is not
always an option.

With color photographs (for which JPEG was designed), you won't
see much in the way of problems until you try to zoom (or crop) into
fine detail.

Enjoy,
DoN.
--
Email: <dnichols@d-and-d.com> | Voice (all times): (703) 938-4564
(too) near Washington D.C. | http://www.d-and-d.com/dnichols/DoN.html
--- Black Holes are where God is dividing by zero ---
Anonymous
June 22, 2005 6:21:19 PM

Archived from groups: rec.photo.digital.slr-systems (More info?)

In article <1119406811.744884@ftpsrv1>, nomail@nomail.com says...

>Are you sure that this isn't the result of sharpening?

No, I dont think it was sharpening. I'm not aware of ever enabling
sharpening, and it wasnt right at the edges anyway, instead very slightly away
from the edges, the surrounding smudged artifacts typical of JPG.

But, I have to say that I am unable to find those images now. I routinely use
NEF, and was just playing with the temp JPGs. I was pretty disappointed
initially, but of the JPG I can find now, I must admit they are pretty good,
better than I remember. I'm not sure what the confusion is, so I'm going to
have to repeat that.

But it is still clear that the claimed 4:1 compression ratio for Fine appears
to be noticeably smaller than 4:1, typically perhaps 7:1.
Anonymous
June 23, 2005 4:01:22 PM

Archived from groups: rec.photo.digital.slr-systems (More info?)

Usually I get even smaller files than you. I often shoot objects on a
featureless black background, and in this case JPG Fine can be as
small as 1-2MB. Large featureless areas compress down to almost
nothing.

Different software/firmware often use their own ways to specify JPG
compression, which adds to the confusion. JPG Fine in the D70s seems
approx. equivalent to Photoshop level 10 (out of a maximum of 12) and
99% (out of max 100%) in ThumbsPlus. The Coolpix 5700 JPG Fine is
different from the D70/D70s JPG Fine (the former seems to be about
Photoshop level 8 and ThumbsPlus 78%).

On Tue, 21 Jun 2005 18:53:05 GMT, Wayne <nospam@invalid.com> wrote:

>I routinely use RAW files, so this is just an academic question.
>
>The Nikon D70S image size is 6 megapixels, which is 18 MB uncompressed.
>The manual pg 41 says JPG file compression is:
>
>Fine 1:4 compression, so file should be about 18/4 = 4.5MB
>Normal 1:8 compression, so file should be about 18/8 = 2.25MB
>Basic 1:16 compression, so file should be about 18/16 = 1.12MB
>
>Given that JPG compression varies some with image content, but mine are
>routinely quite a bit smaller, nearly half, like 2.5, 1.2, 0.6 MB.
>This is rather different than 1:4, 1:8, and 1:16.
>
>The largest Fine file (frame full of tiny busy detail) that I've been
>able to create is about 3.2 MB. These JPG artifacts are not bad, but
>worse than I'd expect for 1:4, and of course, it isnt 1:4.
>
>I assume others see the same file sizes? It seems arbitrary what size
>they are, except that they state numbers which seem inaccurate.
!