Date: Sat, 31 Aug 2002 22:02:49 EDT
From: Dan Margulis
Subject: Maximize compatibility/PS7 large file bug?
DanNot sure if the following is appropriate for your list, so will run it by you first.
i posted this on 'PS Digest List' Aug 18, received one reply confirming possible problem without any working remedy.
-------------------------------------------
Am having a problem trying to save layered files in PS 7 with out the composite.
In PS Prefs, have unchecked 'Always Maximize Compatibility. . .'
When then Save or SaveAs is tried, Dialog Box comes up 'Maximize Compatibility, this is also unchecked.
Clk OK, and progress bar starts up 'Generating Full Resolution Composite'.
Presently working on files__res 12, 45 by 39 inches, while working on them they get to 1 to 2GB, so would like to minimize file size and more importantly, time on saves. (7 is useful for these files because of the larger brush sizes available, 2,500Pixels)
Running OS 9.2.1, 1.5GB Ram (995MB to PS), dedicated 15GB Scratch Disk.
The above is experienced with 2 different work stations, 3 separate PS 7s, all with same result, can't seem to save w/o composite, what am i doing wrong?
Have reinstalled 7, along with 7.0.1 updater to no avail.
Also, i believe i read somewhere, sometime, PS 6 has 1GB max Ram allocation, does the same hold true for 7 ?
Regards,
Floyd RominskiDate: Sat, 31 Aug 2002 21:10:13 -0400
From: Floyd Rominski
Subject: Not sure if appropriate
To: Dan Margulis
Date: Sun, 01 Sep 2002 15:16:59 -0000
From: Stephen Marsh
Subject: Re: Maximize compatibility/PS7 large file bug?
Floyd, try disabling all thumbnail and icon previews, and still disable the full resolution composite.Save the file, does the message still come up? If so, note the file size.
Next, turn on the pref or hit the checkbox to accept the filebloat and take the full resolution composite.
Now note the file size, it should be larger than before.
What this should demonstrate is that even though the message comes up when you have the settings disabled in 7.0.1 - there is no bloat from a full res image, only thumbs/full size thumbs/icons etc.
I think the issue may be that the wording is still being used for the full resolution composite - even when this is not being used and only smaller thumbs are being created...perhaps Adobe were short for words?
> (7 is useful for these files because of the
> larger brush sizes available, 2,500Pixels)Have you tried reducing the spacing from 25% to 15% or lower for these very large brushes for less edge artifacting in sweeping strokes?
Stephen Marsh.
Date: Sun, 1 Sep 2002 11:32:00 -0700
From: Jan Steinman
Subject: Maximize compatibility/PS7 large file bug?
> From: Floyd Rominski via Dan>Am having a problem trying to save layered files in PS 7 with out the
>composite.Ah, yes. That's Adobe trying on its Microsoft suit. I hope they look in the mirror before they go out looking like that!
I have complained, as have many others who do BIG work (> 500MB or so). This "feature" is being shoved down our throats for the convenience of future Adobe products. This really sucks!
This "feature" adds a significant amount of time to saving large files. I'm seriously considering checking out the competition (such as it is) because of it. Does anyone know how well Painter, Corel, Canvas, et. al. behave with huge files?
Wouldn't it be nice if some white knight bought the rights to Live Picture and updated it?
--
: Jan Steinman
: Bytesmiths
: 19280 Rydman Court, West Linn, OR 97068, 503.635.3229
Date: Sun, 1 Sep 2002 16:02:43 EDT
From: Dan Margulis
Subject: Re: Maximize compatibility/PS7 large file bug?
Floyd writes,>Am having a problem trying to save layered files in PS 7 with out the
>composite.Having a free G4 on my desk, I made a 1-gig file with an adjustment layer and confirmed what Stephen has already said.
Photoshop 7 is *not* saving the composite. Instead, the problem is that your preferences are set up to generate previews. To make those previews, Photoshop has to generate a temporary composite version for its own internal purposes. And, indeed, for a 1-gig file, you can go out for a sandwich while it's doing that.
Just for sport, I tried opening and resaving the file both ways in PS6. Everything works exactly the same way, *except* that the message "Generating Full Resolution Composite" doesn't appear. Instead, you just get a long, long, long delay before it starts to save the file, which vanishes if you disable previews. But the sizes of the documents saved by PS6 and PS7 were identical.
Dan Margulis
Date: Sun, 01 Sep 2002 23:06:28 -0400
From: Floyd Rominski
Subject: Re:Maximize compatibility/PS7
Stephen / DanThanks for your response, you're saving me time.
File size indeed is same w/icons and no Max.Compatibility, as no to both icons and Max.C.
More importantly there IS a +- 30% savings in time.
e.g.: Working File size-1.69Gb w/icons & no Max.C., saves to 1.52Gb in 7min. 29sec. w/o icons & no Max.C., saves to 1.52Gb in 5min. 17sec. / ballpark 30% less than above.
w/ icons & Max.C. 1.93Gb in 7min. 32sec.
Thanks gents, no icons will be the flow for these lrg. files while working them at home and NYC. The final save for the NYC server will be the only one w/icons for the PrePress House, they want to see the icons.
Yes Stephen, i've been using 5% to 10% and less for brush spacing.
One last Query, does anyone know what the max Ram allocation is for PS 6 and 7, can one go higher than 1Gb? As you know the latest G4 Wind Tunnels hold 2Gb. Do you think they will reduce the noise level in the near future?
Thanks again,
Regards,
Floyd Rominski
Date: Mon, 02 Sep 2002 07:37:30 -0000
From: Stephen Marsh
Subject: Re: Maximize compatibility/PS7 large file bug?
Dan writes:> Having a free G4 on my desk, I made a 1-gig file with an adjustment layer and
> confirmed what Stephen has already said.
>
> Photoshop 7 is *not* saving the composite. Instead, the problem is that your
> preferences are set up to generate previews. To make those previews,
> Photoshop has to generate a temporary composite version for its own internal
> purposes. And, indeed, for a 1-gig file, you can go out for a sandwich while
> it's doing that.A nice (sadly not very fast) batch thumb preview drag-n-drop freeware utility can be found for classic Mac OS here:
http://hyperarchive.lcs.mit.edu/HyperArchive/Archive/gst/grf/drop-preview-103.hqx
Virus check and test that it does not eat your important files (I have had no issues with a few years of use).
Sincerely
Stephen Marsh.
Date: Mon, 2 Sep 2002 11:54:38 -0700
From: Jan Steinman
Subject: Re: Maximize compatibility/PS7 large file bug?
> From: Stephen Marsh
>
>A nice (sadly not very fast) batch thumb preview drag-n-drop freeware
>utility can be found for classic Mac OS here:
Unfortunately, it's a Classic app.
--
: Jan Steinman
: Bytesmiths
: 19280 Rydman Court, West Linn, OR 97068, 503.635.3229
Date: Mon, 2 Sep 2002 11:47:50 -0700
From: Jan Steinman
Subject: Re: Maximize compatibility/PS7 large file bug?
>Photoshop 7 is *not* saving the composite. Instead, the problem is that your
>preferences are set up to generate previews...THANK YOU!, DAN! I'm sure the rest of us big-file-freaks will like that one, too.
It would be nice if there was a threshold one could set -- say, only generate previews if the images is under n megapixels, for example.
--
: Jan Steinman
: Bytesmiths
: 19280 Rydman Court, West Linn, OR 97068, 503.635.3229
Date: Mon, 2 Sep 2002 11:40:34 -0700
From: Jan Steinman
Subject: Re: Maximize compatibility/PS7 large file bug?
From: Stephen Marsh
>
>What this should demonstrate is that even though the message comes up
>when you have the settings disabled in 7.0.1 - there is no bloat from
>a full res image, only thumbs/full size thumbs/icons etc...I've got a terabyte of on-line disk storage -- I don't care about file size as much as I do the time it takes while generating the full-res composite.
And if it isn't putting the composite in the file, why the hell does it make me wait minutes while it generates it?
--
: Jan Steinman
: Bytesmiths
: 19280 Rydman Court, West Linn, OR 97068, 503.635.3229
Date: Tue, 03 Sep 2002 17:22:10 -0400
From: "Jerry L. P'Simer"
Subject: Re: Re: Maximize compatibility/PS7 large file bug?
Stephen Marsh wrote:> Now note the file size, it should be larger than before.
>
> What this should demonstrate is that even though the message comes up
> when you have the settings disabled in 7.0.1 - there is no bloat from
> a full res image, only thumbs/full size thumbs/icons etc.This is very true. What is a major PITA is the fact that you still have to wait for a composite to be created even though it does not actually save it. I have to work on very large layer files quite frequently (not by choice) which are often over a gigabyte and have seen this composite take more than 20 minutes to create. I have tried to get rid of this very frustrating event to no evail. It does not seem that the upgrade to 7.0.1 has corrected this error.
Regards,
Jerry P'Simer
Date: Tue, 03 Sep 2002 17:32:00 -0400
From: "Jerry L. P'Simer"
Subject: Re: Maximize compatibility/PS7 large file bug?
Dan Margulis wrote:> Photoshop 7 is *not* saving the composite. Instead, the problem is that your
> preferences are set up to generate previews. To make those previews,
> Photoshop has to generate a temporary composite version for its own internal
> purposes. And, indeed, for a 1-gig file, you can go out for a sandwich while
> it's doing that.I should have read your post before posting. I will try this tomorrow. I'm sure this will save me hours of frustrating grief. I sure don't need the previews bad enough to wait for them.
Thanks much.
Jerry P'Simer
Date: Tue, 3 Sep 2002 19:53:19 -0700
From: Jan Steinman
Subject: Re: Maximize compatibility/PS7 large file bug?
I spoke too soon. I turned off previews, and I still get MINUTES of "Generating full resolution composite" crap on my big images. I DON'T WANT THE DAMN FULL RES COMP!I think it's time to see what the rest of the world has been up to. I've got old licenses of Corel, Painter, and Canvas around somewhere... wonder what they charge for upgrades these days... wonder if they'll give me a break if I tell them I'm FED UP WITH PHOTOSHOP!!!
Will someone please buy the rights to Live Picture and update it?
--
: Jan Steinman
: Bytesmiths
: 19280 Rydman Court, West Linn, OR 97068, 503.635.3229
Date: Thu, 5 Sep 2002 18:03:10 +1000
From: "Stephen Marsh"
Subject: Re: Maximize compatibility/PS7 large file bug?
Hi Jan - I am responding to both your posts in the one reply.>> Unfortunately, it's a Classic app.<<
Whats so unfortunate about that?!<vbg>
I guess you failed to notice my description which noted this app was for classic MacOS.
What works out nice for you (I hope) - is that there is probably more chance of getting the same product made for OSX since there are a lot of hackers out there who love to play. All the classic Mac utility in question did was provided a non-gui source location to drop files onto - it did 'nothing' by itself to make the thumb, it called upon Apple QuickTime to read the file/s and generate the preview image thumb (not sure if QT embeds or if DropPreview does).
P.S. I know us 'classic' uses are living on borrowed time - but I just can't see the changeover to OSX happening quickly in print production - despite Apples moves.
>> I've got a terabyte of on-line disk storage -- I don't care about file size as much as I do the time it takes while generating the full-res composite.<<
I can't help you there.<g> So turn off the bloat, turn off the previews and use your trusty 'vintage' Mac and OS to run DropPreview until you find the solution you need for OSX.
>> And if it isn't putting the composite in the file, why the hell does it make me wait minutes while it generates it?<<
I think as noted by Dan - it seems that the thumb generation process needs a larger flat composite source to make the final thumb from. While we are at it - there are also those annoying 'preparing to save' or 'preparing to apply filter' pre-operation progress bars - they can be just as annoying.
Stephen Marsh.
Date: Wed, 4 Sep 2002 19:05:09 EDT
From: Dan Margulis
Subject: Why such large files
Floyd, Jerry, Jan:A couple of people have asked me, and I don't know the answer: why are you people regularly working on such enormous (1gb+) files? What applications require such a huge amount of data?
Dan Margulis
Date: Thu, 5 Sep 2002 23:38:00 -0700
From: Jan Steinman
Subject: Re: Why such large files
>why are you...
>regularly working on such enormous (1gb+) files? What applications
>require such a huge amount of data?I do huge fine-art prints and murals. I drum-scan 4"x5" Velvia at 5,400 spi, then print it at 300 ppi. I've even combined two large-format images into a panorama, and have combined two over/under exposed large format images in registration for more dynamic range. (I like to shoot into the sun, which doesn't work too well with the 8-9 stops you get out of fine-grained transparency film.)
My biggest work to date has been a bit under 10'x4' (yes, that's in FEET). Some of my "smaller" works are only 24"x36", which is still about a quarter gigabyte for the background, without layers or masks.
--
: Jan Steinman
: Bytesmiths
: 19280 Rydman Court, West Linn, OR 97068, 503.635.3229
Date: Fri, 6 Sep 2002 09:37:15 -0600
From: Les De Moss
Subject: Re: Re: Why such large files
>> My biggest work to date has been a bit under 10'x4' (yes, that's in FEET). Some of my "smaller" works are only 24"x36", which is still about a quarter gigabyte for the background, without layers or masks.>>Out of curiosity, Jan.... It sounds like you are determining file size (needed for a particular output size) by calculating DPI of the output as if the same as PPI in the file. PPI and DPI do not directly relate to one-another. Have you tested output at various (smaller) file sizes? In our lab we've found that file sizes on the order of 1/3 size render comparable results.
i.e: ((L" x W" x L/dpi x W/dpi x bit-depth)/8,000,000)/3
If we were outputting a 4'x6' mural at 300 dpi, 24-bit, the calculation is as follows:
((48 x 60 x 300 x 300 x 24)/8,000,000)/3 = 260MB
Using this calculation, a 24x36" mural would require a file of about 80MB, rather than 233MB.
Regards,
Les De Moss
www.digigraphics.com
Fort Collins, Co
Date: Fri, 06 Sep 2002 18:11:37 -0400
From: "Jerry L. P'Simer"
Subject: Re: Why such large files
Dan,
For me this is not by choice. I regularily do a large amount of retouching for an art director out in California who insists that his beauty shots are a minimum of 400MB. The retouching that I do has to remain fully editable at all times in case he changes his mind about the alterations he is having me make to his images (very large layer files). Often times the amount of retouching that is required is tremondous. To date I have never seen an instance where the final image used needed anywhere near the kind of resolution that is built in at the start. Even the product cards that I have used the images for after final approval have only required about 40 to 50MB of resolution yet there is enough resolution in the images to print a 30 x 40 inch poster at 200 line screen. I have spoken to the art director many times about this but he still insists that he may need this much resolution at some point in time. Since we are paid by the hour for the retouching changes that we make, who are we to complain about file sizes.
Regards, Jerry
Date: Fri, 6 Sep 2002 17:06:57 -0600
From: Les De Moss
Subject: Re: Re: Why such large files
Peter Figen writes> I'm not sure how you arrived at your figures, but there ARE output devices in
> use where you actually can see a degradation; in print quality using less than
> the optimum for that device.Are you defining "Optimum" for a particular device as a 1:1 ratio between ppi and dpi?
First, the formula for determining file size in MB is straight-forward and long used to determine file size when the convenience of a PhotoShop calculation is not available:
(L" x W" x Lppi x Wppi x bit depth)/8,000,000
An 8x10" uncompressed Image File at 300 ppi, 24 bit equals 21.6 MBPurely as an example, I suggested that Jan experiment with a file size of aprox. 1/3 and inspect the results. From practical experience scanning with our drum scanner and printing with our film recorders, large-format inkjets, and RGB laser devices like the Pictrography, I can suggest with great certainty that using a 1:1 ratio of ppi to dpi is excessive and unnecessary.
My main point is that DPI and PPI do not have a direct correlation to each other. Many people are confused by this. That is, 1 pixel of file data does not equal one dot of ink, or point of exposure. Once ripped and sent to print, a single pixel may result in numerous dots. If you calculate backward from the total number of dots your printer will produce on a given size print, it results in a file size that is wildly larger than what's required for optimal output.
Supplying the output device with data beyond what it can use is a waste of time. Therefore, as I suggested to Jan.... Experiment with file size to determine exactly what's necessary to achieve optimum results. It won't be a 1:1 ratio of ppi to dpi, and depending on the device could be as little as one third.
Les De Moss
DigiGraphics
Date: Fri, 06 Sep 2002 22:14:19 -0400
From: Floyd Rominski
Subject: Re: Why such large files
> A couple of people have asked me, and I don't know the answer: why are you
> people regularly working on such enormous (1gb+) files? What applications
> require such a huge amount of data?
Dan
i posed your question to the sales rep handling this particular campaign. (i might add his habit is to generally have scans made larger then necessary, he hasn't read your books!)
His response was. These series of files, as others in the past, are to be used by the agency internationally for any type of output from magazine, to LRG durotrans,00 etc., etc..
This is not the every day type of work i'm doing, but occasionally these campaigns do come around, thankfully. . . Today it was a 3.3Gb file while working it, reduced to 1.+ after merging some layers.
As mentioned before, they start out at res 12, ballpark 45 by 40 inches. the various layouts show them cropped vert., horiz.. The base file is used for all configurations, with an xtra layer for the particular foreign language inc. English. ;)
----------For others manipulating lrg. files Shangara Singh gave good advise as follows. . .
"The Use Pixel Doubling preference option speeds up the preview of a tool or command's effects by temporarily doubling the size of the pixels (halving the resolution) in the preview. This option has no effect on the pixels in the file; it simply provides faster previews with the tools and commands."
To speed up previews:
1.Do one of the following:
* In Windows or Mac OS 9.x, choose Edit Preferences Display & Cursors.
* In Mac OS X, choose Photoshop > Preferences > Display & Cursors.
2.Select Use Pixel Doubling, and click OK.
------------Hope that helps some fellow sufferers
regards,
Floyd Rominski
Date: Fri, 6 Sep 2002 23:29:27 -0700
From: Jan Steinman
Subject: Re: Why such large files
> From: Les De Moss
>
>>>> My biggest work to date has been a bit under 10'x4' (yes, that's in
>FEET). Some of my "smaller" works are only 24"x36", which is still about a
>quarter gigabyte for the background, without layers or masks.
>
>Out of curiosity, Jan.... It sounds like you are determining file size
>(needed for a particular output size) by calculating DPI of the output as if
>the same as PPI in the file. PPI and DPI do not directly relate to
>one-another.Yes, I think I know the difference. My output DPI is 1440. My input to RIP is 300 PPI.
>Have you tested output at various (smaller) file sizes? In our
>lab we've found that file sizes on the order of 1/3 size render comparable
>results.Yes. I can easily tell the difference between 300 ppi and 100 ppi when RIPped to 1440 or even 720 dpi. In fact, with the help of reading glasses, I can even tell the difference between 260 ppi and 300 ppi.
I calculate my required PPI to RIP based on viewing distance, just as billboard designers do. Much of my work is designed to be viewed at any distance, which for practical purposes, means six inches. This is based on human acuity of vision. I did the research and trig once, but now only recall the results. :-)
The 10'x4' mural was designed to be viewed no closer than about 24", and was 140 ppi, RIPped to 720 dpi. As I recall, it was still almost a gigabyte -- and several gigs before flattening.
--
: Jan Steinman
: Bytesmiths
: 19280 Rydman Court, West Linn, OR 97068, 503.635.3229
Date: Sat, 7 Sep 2002 11:34:52 -0400
From: "john castronovo"
Subject: Re: Why such large files
From: Jerry L. P'Simer
> I have spoken to the art director many times
> about this but he still insists that he may need this much resolution at some
> point in time. Since we are paid by the hour for the retouching changes that
> we make, who are we to complain about file sizes.
Well said !!
As a former photographer, let me add to the discussion that this is an old question for shooters. The decision to shoot a particular format from 35mm, 120, 4x5 or 8x10, is based not on the use in question, but all possible uses.Most importantly, the subject matter makes the choice for us. Images with a lot going on (like a field of trees) can survive smaller formats whereas polished or sharp objects (like a car or machine) require larger formats. Also, anything with type or line art needs to be sharp, which is why we add vector art for the rip to handle where possible instead of adding it to the tiff file. The pictorial portions of the finished ripped file are generally less able to reveal the fact that they came from a lower res scan, but there's no getting around low res line art.
John Castronovo
Date: Sat, 07 Sep 2002 19:10:43 -0400
From: Terry Wyse
Subject: Re: Re: Why such large files
Jan Steinman wrote:> My output DPI is 1440. My input to RIP is 300 PPI.
An Epson inkjet, right? In reality, you should always calculate your dpi at 720. On all Epsons that I can think of, 1440dpi is achieved by simply overlapping the 720dpi dots or spots. 1440dpi simply lays down more but does not effectively increase the resolution of the output.
Based on this, 720/2 or 360dpi is more than enough and most find that 720/3 or 240dpi produces a very high quality print. Personally, I don't see any difference at all till about 720/4 (180dpi). The main thing is adjusting your USM setting (pixel radius mostly) for the specific dpi you're going to use.
Just my $.03 worth,
Terry
__________________________________
WyseConsul
Color Management Consulting
v 704.843.0858
e
__________________________________
Date: Sat, 7 Sep 2002 22:07:35 -0700
From: Jan Steinman
Subject: Re: Why such large files
An Epson inkjet, right?No, a Roland, using Hexachrome inks and Hexachrome RIP.
And I've done double-blind tests. The 1440 output is perceived to be of higher quality by a panel of friends, professional artists, and CUSTOMERS!
Now one can argue until one is blue in the face that one technical point or another has some theoretical merit, but when my customers speak, I listen rather than argue!
--
: Jan Steinman -- nature photography:
: Bytesmiths -- artists' services:
Date: Sun, 8 Sep 2002 22:34:06 -0700
From: "Darren Bernaerdt"
Subject: RE: Why such large files
John & group,John, you certainly bring up a good point, but I'd like to put a further spin on it.
I was speaking with another photographer earlier this week with regard to scanning resolution. The stock photo agency that represents him used to ask for 18MB scans for anything submitted digitally. Now, they want 50 to 60MB.
Has anyone tried some tests to see at what point you are simply sampling the film's grain finer (actually dye clouds since we're talking about color images) versus getting more actual data. I don't think anyone would argue an 18MB scan of a 35mm slide would have less information than a 50MB scan. I seem to remember hearing figures such as:
35mm - about 30MB 2 1/4" square - about 70MB
The concept here, was that scanning at a higher resolution wouldn't yield any additional detail, just a larger file. If we assume a very fine grained film and a subject that will make a difference, what truly are the limits for the these formats and 4x5"? I haven't had the access to a decent scanner to try this out for myself and couldn't justify the cost to send it out to be done.
Darren Bernaerdt
> From: john castronovo
> As a former photographer, let me add to the discussion that this is an old
> question for shooters. The decision to shoot a particular format
> from 35mm, 120, 4x5 or 8x10, is based not on the use in question, but all possible
> uses.
Date: Mon, 9 Sep 2002 09:23:11 -0400
From: "john castronovo"
Subject: Re: Why such large files
> Has anyone tried some tests to see at what point you are simply sampling the
> film's grain finer (actually dye clouds since we're talking about color
> images) versus getting more actual data. I don't think anyone would argue an
> 18MB scan of a 35mm slide would have less information than a 50MB scan.While scanning at higher resolutions is a case of diminishing returns for many reasons, the numbers you show are certainly too low. In 8 bit/color RGB, the average 35mm will usually show improvement in scanning up to about 60 megs and diminishing improvement from there. Of course, the scanner must not have much flare to show the difference, so a clean drum scan is required. An exceptional slide might show improvement all the way up to 90 megs or beyond.
Theoretical limits put the very best scan, of the very best slide, at 16 bit depth, in RGB, somewhere close to three gigs to get everything on the film. (This was worked out by a customer of mine who's into digital photography now and he's got a Ph.D. in physics. I'll ask him for the math if you really need to see it.) We're talking about scanning molecules here, I'm afraid, but if one really wants to get it ALL . . . well, there it is!
John Castronovo
Date: Mon, 09 Sep 2002 16:01:27 -0000
From: "pleistocenehome"
Subject: Re: Why such large files
Hi Darren,We can talk about optimal file size in terms of scanner resolution also. I scan MF at 4000 dpi and am most assuredly recording real detail. I get a ~280 mb 8-bit (or a ~560 mb 16-bit) file from these scans. You can calculate this for MF (6x7) as follows: ~6 sq.in. X 4000 X 4000 X 3 (for RGB) = 288 mb. I have heard that there is usually 5000-6000 dpi worth of real information in a color transparency (lets say Provia) and maybe more. So you can rapidly go to giant file sizes if you want ALL the info on a transparency in your PS file, which you might want if you are planning to make the most detailed very large fine-art prints (say 44" wide on a LF Epson printer) possible.
I have been making 20x25 prints (using 360 ppi input res files on an Epson 10K) from my 4000 dpi scans without any need for resampling. I could make a 35X44 print at 240 input res without any upsampling but haven't tried this yet. Or with modest upsampling I can make as large a print (44" wide) as my printer will make from my MF 4000 dpi scans. Whether I would see the difference in a print this big if I used a 5000-6000 dpi scan is a good question that I would like to answer sometime. I need a really good Tango (or other high-end drum) operator who I can sit down with and make all the prescan corrections. I have not been particularly pleased with the commercial drum work I have gotten because of the 8-bit un-color corrected files I get. I like the results from my Nikon 8000 mo better, because I can do the work in 16 bit the way I want it. ((Also I am probably being really dense, but I have no idea where Les De Moss gets his formula using 8,000,000 for calculating res)
Tom Andrews
http://www.wildlandart.com> I was speaking with another photographer earlier this week with regard to
> scanning resolution. The stock photo agency that represents him used to ask
> for 18MB scans for anything submitted digitally. Now, they want
> 50 to 60MB.
Date: Mon, 09 Sep 2002 14:31:40 -0700
From: Peter Figen
Subject: Re: Why such large files
Darren Bernaerdt wrote:> The concept here, was that scanning at a higher resolution wouldn't yield
> any additional detail, just a larger file. If we assume a very fine grained
> film and a subject that will make a difference, what truly are the limits
> for the these formats and 4x5"?Darren,
Films like Velvia and E100 can easily support a 6.35 micron scanning aperture, which equates to 4000 real ppi. Velvia's grain structure is in the 5-6 micron range, Ektachrome, a couple of microns larger, so it takes a bit more than 4000 ppi to actually resolve the true grain structure of these films. Whether or not the taking lens is as sharp as the film is another matter altogether.
T-Max 100, on the other hand, has a grain structure of around 3 microns, and can actually support a true 8000 ppi, which requires a 3 micron aperture, something which few scanners have, no matter what their claimed optical resolution is. Again, in order to actually utilize the potential of this film, you have to shoot under strict conditions with the sharpest lenses at the sharpest apertures. I've made comparison scans of very sharp T-Max negs comparing the 6.35 micron aperture to the 3.2 micron aperture and have seen a very slight improvement in sharpness at the higher res.
Most 35mm scans are going to hit their optimum in the 60mb - 4000 ppi range, but some can actually benefit from a higher resolution.
In addition, the scanner that is used for making evaluation scans is of prime importance. Most of the scanning tests I've seen comparing films or more recently, comparing film to digital capture, have been done on scanners that maybe aren't the best quality. One of the most popular "professional" models that has been heavily marketed to photographers has been independently tested to actually resolve about half of it's advertised resolution for 35mm. I've done a fair amount of these types of comparisons using a Howtek HR8000 which has had the optical system custom aligned for the best sharpness. It does make a difference.
A related aside: I recently did a film to digital comparison between the Kodak DCS ProBack and shooting the same product with E100SW, and the Ektachrome, scanned on the Howtek, was definitely the winner in terms of overall image quality, sharpness and rendering of fine highlight and shadow detail. The DCS was the winner in terms of production speed, and for smaller uses, it would be fine, but for someone wanting to eke out the last bit of image detail, especially for large images, film still wins. I could have shot Velvia and seen an even greater difference, but I wanted to shoot the type of film that I actually shoot in the studio.
Peter Figen
Peter Figen Photography
Los Angeles, Ca.
Date: Mon, 9 Sep 2002 15:24:05 -0600
From: Les De Moss
Subject: Re: Re: Why such large files
Tom Andrews Writes:>>Also I am probably being really dense, but I have no idea where Les De Moss gets his formula using 8,000,000 for calculating res>>
Dense? That's way to harsh, Tom ;-). I have to redo the math every time someone asks for an explanation...
The product of the formula: (L" x W" x L:ppi x W:ppi x bit-depth) equals the total BIT count of the file.
There are 8 bits to byte, and a million bytes to a Megabyte. Dividing the product by 8,000,000 returns total Megabytes (8 x 1,000,000).
This formula contains zero smoke/mirrors/theory. It is a straight mathematic calculation of file size according to the amount of data it contains.
It was originally published by DIMA/PMA in the early nineties. It's the formula on which cardboard/plastic portable "slide-rule" file-size calculators are based.
Les De Moss
A number of members have written about how higher resolution scanning can in fact resolve additional data. While this is true, the question has to be whether one actually *wants* that data. If there's a lot more data than what's needed for proper output, it can actually be counterproductive.Files that are oversampled are softer, more defocused than ones that are at the proper resolution. Or, if you'd prefer the positive spin, they're smoother, more consistent. Either way, if you scan at double the necessary resolution and print it at the same size as one that was scanned at a normal resolution, there are going to be two different results, enough so that the client may prefer one--or may prefer the other. The lower-resolution files tend to look sharper and crisper, so my experience is that extra resolution is counterproductive in most cases. But, there are certainly some images that will look a bit better with the extra data, so what Jan says isn't surprising.
The whole thing is somewhat similar to the debate over correcting 8-bit vs. 16-bit files, but more significant. Working 16-bit only will cause a smoother-looking file but it has to be a really enormous series of moves and even then you have to look very, very carefully to see it. Files that are at twice the normal resolution, you can see the difference easily with the naked eye.
Dan Margulis
Date: Mon, 9 Sep 2002 17:58:53 -0400
From: Preston Earle
Subject: Re: Re: Why such large files
"Les De Moss" writes:"The product of the formula: (L" x W" x L:ppi x W:ppi x bit-depth)
equals the total BIT count of the file."There are 8 bits to byte, and a million bytes to a Megabyte. Dividing
the product by 8,000,000 returns total Megabytes (8 x 1,000,000)."
------------------
Isn't every pixel a byte, such that the divisor is 1,000,000 rather than 8,000,000, or, more accurately, isn't the divisor 1,024 x 1,000 =1,024,000?Preston Earle
Date: Mon, 9 Sep 2002 16:55:38 -0600
From: "Les De Moss
Subject: Re: Re: Why such large files
Preston Earle Writes:> Isn't every pixel a byte, such that the divisor is 1,000,000 rather than
> 8,000,000, or, more accurately, isn't the divisor 1,024 x 1,000 =
> 1,024,000?No. A pixel may contain anywhere from 2 bits to 16 bits of data, depending on the "Bit-Depth" of the image. Hence the debates over 8 or 16 bit image quality.
A line image (Pure black, pure white) is a 2-bit image. The bit or "switch" is either on or off. A BW Grayscale image is 8 bit. Move to RGB, which is in reality three BW grayscale images, and you have three, 8-bit channels, for a total bit-depth of 24. Work in 16-bit RGB color, and your bit count goes to 48 (3x16)
Les De Moss
Date: Mon, 09 Sep 2002 18:20:49 -0400
From: Terry Wyse
Subject: Re: Why such large files
on 9/9/02 5:31 PM, Peter Figen wrote:> Films like Velvia and E100 can easily support a 6.35 micron scanning aperture,
> which equates to 4000 real ppi. Velvia's grain structure is in the 5-6 micron
> range, Ektachrome, a couple of microns larger, so it takes a bit more than 4000
> ppi to actually resolve the true grain structure of these films. Whether or not
> the taking lens is as sharp as the film is another matter altogether.I did some searching and find numbers in the 100-160 l/mm resolving power of these films. T-Max was around 200 l/mm.
160 l/mm (Velvia) is indeed then 4064ppi. It was also interesting that most of the color neg films were way down in the 50 l/mm range.
There was also some interesting notes on true "system" resolution (lens + film combo) which is typically much lower, closer to 1/2 to 2/3 of the film's resolving power. As a practical matter, I would think 2,000-3,000ppi would be a reasonable upper limit for most scanning.
Here's the link: http://people.smu.edu/rmonagha/mf/lenslpm.html
Interesting stuff.
Regards,
Terry__________________________________
WyseConsul
Color Management Consulting
v 704.843.0858
__________________________________
Date: Tue, 10 Sep 2002 02:36:44 -0000
From: "pleistocenehome"
Subject: Re: Why Such Large Files
Hi Dan,
Ah, this will provoke more tests. I regularly scan at my scanner's highest possible resolution (MF at 4000dpi on a Nikon8000) and then downsample to make prints in smaller sizes, keeping the original file for those gigundo prints in my future. Maybe I need to do several scans at once at various resolutions. Now this sounds really time consuming (not to mention thef extra HD and CD space for storage). So I will need to test this and hope I don't see any of the differences you mention. Maybe I am compensating for some lack of crispness in smaller (downsampled)l prints with extra USM. So many variables! Thanks,
Tom Andrews
http://www.wildlandart.com> A number of members have written about how higher resolution scanning can in fact resolve additional data. While this is true, the question has to be whether one actually *wants* that data. If there's a lot more data than what's needed for proper output, it can actually be counterproductive.
Date: Mon, 9 Sep 2002 20:28:48 -0700
From: Jan Steinman
Subject: RE: Why such large files
>Has anyone tried some tests to see at what point you are simply sampling the
>film's grain finer (actually dye clouds since we're talking about color
>images) versus getting more actual data.Hoo boy, are you opening a can of worms!
>I seem to remember hearing figures such as:
>35mm - about 30MB
>2 1/4" square - about 70MBThere are so many details to consider! The type of film, whether a tripod was used, the quality of lenses, the shutter speed -- even subject matter! It is simply impossible to state with any certainty that "a 35mm frame contains x amount of data."
Using Genuine Fractals, I have produced 1/4 GB files from 35mm film that, when printed 24"x36", experienced photographers walk up to, examine closely, and say, "What large format film were you using?" It's certainly not fair to say that frame of film contained 250 MB of useable data, and yet it passed the "Turing Test" in that the end result was imperceptably different from film that actually contained 250 MB of data.
I've also got some 35mm frames that should not be sampled at more than about 900 spi! (That's only about 3MB.)
For each image I drum scan, I try different apertures until the grain just goes away. Some combinations of film, lens, shutter speed, etc. can max out my scanner at 5,400 spi, but that resolution would only provide very detailed images of the actual grain structure in other cases.
>If we assume a very fine grained
>film and a subject that will make a difference, what truly are the limits
>for the these formats and 4x5?First, I think the assumptions are faulty. Each image is unique. The best you can hope to do is to make such a determination for *identical* images, shot under *identical* conditions, and then all you've done is measured how much data is in your test image, NOT how much you could expect out of some other image!
That said, there will be people who insist on trusting numbers more than results. Here's a couple references for you:
http:/s.qwest.net/il.htm
http:/ren.com/>--
: Jan Steinman -- nature photography:
: Bytesmiths -- artists' services:
Date: Mon, 9 Sep 2002 20:41:04 -0700
From: Jan Steinman
Subject: Re: Why such large files
From: "Les De Moss"
>
>The product of the formula: (L" x W" x L:ppi x W:ppi x bit-depth) equals the
>total BIT count of the file.
>
>There are 8 bits to byte, and a million bytes to a Megabyte. Dividing the
>product by 8,000,000 returns total Megabytes (8 x 1,000,000).
>
>This formula contains zero smoke/mirrors/theory. It is a straight mathematic
>calculation of file size according to the amount of data it contains.Actually, a megabyte is generally measured as 2 to the 20th power, which is actually 1,048,576 bytes. (Although storage device makers generally "cheat" by using 1,000,000, and international standards bodies have reacted by calling 2^20 a "Mebabyte." Argh.)
And unless it is strictly a "raw" data file, standard file types always include some overhead. For example, the Tagged Image File Format (TIFF) can have any amount of "extra" data in it, such as entire ICC profiles, which can bloat the file significantly.
Use the formula as a lower bound, not an exact amount!
--
: Jan Steinman -- nature photography:
: Bytesmiths -- artists' services:
Date: Mon, 9 Sep 2002 20:59:42 -0700
From: Jan Steinman
Subject: Re: Why such large files
From: Terry Wyse
>There was also some interesting notes on true "system" resolution (lens +
>film combo) which is typically much lower, closer to 1/2 to 2/3 of the
>film's resolving power.The research I've done indicates that the system resolution is equal to the inverse of the sum of the inverses of the individual components.
So if your film resolves 120 l/mm, and your lens resolves 120 l/mm, and you scan at 120 l/mm (or 240 samples per mm, 6,000 spi), then your overall resolution is a mere 40 l/mm! 1/(1/120+1/120+1/120)
The subtle point that is going to draw lots of fire here, is that your scanner is part of the overall system response, and for it to have insignificant impact on overall system response, it must have MANY TIMES the resolving power of the other components.
The above example approaches an 80 l/mm resolution as the sampling interval approaches infinitesimal, which for practical purposes, is 10 times the response of the best system component. This theory indicates there may be benefits to sampling as high as 1200 l/mm, or 60,000 spi!!!
What all this says to me is that results are much more important than theory. Or the corollary: anyone with a college text on information theory can prove anything he desires. :-)
--
: Jan Steinman -- nature photography:
: Bytesmiths -- artists' services:
Date: Mon, 09 Sep 2002 22:55:32 -0500
From: Bob Smith
Subject: Re: Re: Why Such Large Files
When scanning with a CCD scanner your method is fine and I doubt that you'd see any image quality benefit to scanning at a lower res for small images. A CCD scanner by design has a fixed sampling rate (number of pixels on the chip) in one direction. Anything less than that involves some sort of scheme to down sample the captured data that's probably not much different from what is achieved by down sizing a max res scan in an image editor. I think the only advantage to scanning at low res with a CCD scanner is processing speed. A drum scanner on the other hand can actually sample at variable rates and thus yield slightly different results when scanned at lower res vs sampling down a high res scan.Bob Smith
pleistocenehome wrote:
> Ah, this will provoke more tests. I regularly scan at my scanner's
> highest possible resolution (MF at 4000dpi on a Nikon8000)
> and then downsample to make prints in smaller sizes, keeping
> the original file for those gigundo prints in my future.
Date: Tue, 10 Sep 2002 09:59:42 -0000
From: Stephen Marsh
Subject: Re: Why such large files
"Darren Bernaerdt" wrote:> I was speaking with another photographer earlier this week with regard to
> scanning resolution. The stock photo agency that represents him used to ask
> for 18MB scans for anything submitted digitally. Now, they want 50 to 60MB.Although this thread has moved into the second paragraph of Darren's original post (in answer to the question), I would like to return to the first paragraph of the post which spawned this thread - if I may.
Resolution and resolving image or even substrate detail is obviously part of the issue, but it is probably due to creative flexibility that larger digital files are required to be submitted.
Creative decisions often call for specific areas to be enlarged for cropping or compositing etc. With a hardcopy the traditional method has been to use a positional and then scan a final high res to the final size (or very slightly larger or slightly higher res - say res14 instead of res12). Digital supply of images often does not have this flexibility - the presented data may not have the desired final quality when digitally enlarged.
In one setting I worked in, I would scan at double the resolution and then resample down to the final intended output size (flatsheet conditions up to #175 lpi coated). This was due to the need for creative flexibility for the press work design - and many images were often used for large format inkjet banner output. There was some minimal sharpness loss in this workflow, as the resampling/sharpening was not as good as the original optical scan to the correct size/resolution - but the halftone screen and the printing process can be very forgiving and the differences were only subtle and obvious when both methods were compared side by side.
Stephen Marsh.
Date: Tue, 10 Sep 2002 08:38:33 -0400
From: Terry Wyse
Subject: Re: Why such large files
Peter Figen wrote:> Films like Velvia and E100 can easily support a 6.35 micron scanning aperture,
> which equates to 4000 real ppi. Velvia's grain structure is in the 5-6 micron
> range, Ektachrome, a couple of microns larger, so it takes a bit more than 4000
> ppi to actually resolve the true grain structure of these films. Whether or not
> the taking lens is as sharp as the film is another matter altogether.I did some searching and find numbers in the 100-160 l/mm resolving power of these films. T-Max was around 200 l/mm.
160 l/mm (Velvia) is indeed then 4064ppi. It was also interesting that most of the color neg films were way down in the 50 l/mm range.
There was also some interesting notes on true "system" resolution (lens + film combo) which is typically much lower, closer to 1/2 to 2/3 of the film's resolving power. As a practical matter, I would think 2,000-3,000ppi would be a reasonable upper limit for most scanning.
Here's the link:
http://people.smu.edu/rmonagha/mf/lenslpm.htmlInteresting stuff.
Regards,
Terry__________________________________
WyseConsul
Color Management Consulting
v 704.843.0858
__________________________________
Date: Tue, 10 Sep 2002 08:36:15 -0600
From: Andrew Rodney
Subject: Re: Re: Why Such Large Files
Dan writes:
> A number of members have written about how higher
> resolution scanning can in fact resolve additional data. While
> this is true, the question has to be whether one actually *wants*
> that data. If there's a lot more data than what's needed for proper
> output, it can actually be counterproductive.They do indeed! One huge benefit of over scanning film (other than the fact you have greater flexibility to use the file latter) is you can greatly reduce noise in the resulting scan. This is because noise is almost always single, random pixels in the dark regions of the scan. When you resample down four pixels to one, you lose that noise from resample averaging.
In the old days, the prevailing beliefs was to scan an image as many times as one would need to output. That usually was in effect to help the bottom line and help pay off a dinosaur of a drum scanner. Counter productive and a great way to continue to damage film (the only real original at this time).
Scan once, use many. Scan at the highest resolution you think you'll need in a well behaved RGB space (perhaps archive the raw scan with a profile of the scan itself, convert latter). Sample down for your needs.
You have to scan a catalog of 100 widgets on a white bkgnd which will be reproduced at 2x2" and likely NEVER be used again? Well scanning at max resolution is kind of dumb. But if you are working on a file that needs a good deal of Photoshop work, or you know will likely need to be repurposed, scan at max rez, high bit and you can safely archive the film.
Andrew Rodney
Date: Tue, 10 Sep 2002 11:29:52 -0300
From: André Borges Lopes
Subject: Re(3): Why such large files
Les De Moss wrote:> A line image (Pure black, pure white) is a 2-bit image. The bit or "switch"
> is either on or off.I believe that a line image is actually a 1-bit image. This single bit can be
set ON (black) or OFF (white or transparent). In a bitmapped binary image, it's
represented as:
0 - White/Transparent
1 - BlackA 2-bit image can hold 4 tones (black, white and two shades of gray, or four indexed colors). For example:
00 - White
01 - 33,3% Gray
10 - 66,7% Gray
11 - BlackRegards;
André Borges Lopes
Bytes & Types
São Paulo, Brazil
Date: Tue, 10 Sep 2002 09:08:55 -0400
From: John Castronovo
Subject: Re: Re: Why such large files
I'm in complete agreement, Jan. In a similar fashion, we've known in the lab for decades that transparency dupes need to be enlarged to at least twice their original size to keep details, so a 35mm needs to go up to 70mm or 4x5. A same size dupe, even contacted, can never copy the original's resolution. The entire system response comes into play, and whether it's film or digital, physics won't be cheated.j. castronovo
> The research I've done indicates that the system resolution is equal to
the inverse of the sum of the inverses of the individual components.
Date: Tue, 10 Sep 2002 08:23:57 -0600
From: "Les De Moss"
Subject: Re: Re: Why such large files
Preston Earle Writes:> "There are 8 bits to byte, and a million bytes to a Megabyte. Dividing
> the product by 8,000,000 returns total Megabytes (8 x 1,000,000)."
> ------------------
> Isn't every pixel a byte, such that the divisor is 1,000,000 rather than
> 8,000,000, or, more accurately, isn't the divisor 1,024 x 1,000 =
> 1,024,000?Preston:
You got me thinking. Yes, the exact divisor is 8,388,608 (1024x1024x8)This can be checked by creating an RGB file in Photoshop, 8x10" @ 300ppi and reading the document size, 20.6MB (20.5994).
The formula was rounded to 8,000,000 for ease of use/memory.
Regards,
Les De Moss
www.digi-graphics.com
Date: Tue, 10 Sep 2002 08:51:59 -0600
From: Les De Moss
Subject: Re: Re: Why such large files
Jan Steinman Writes:> Use the formula as a lower bound, not an exact amount!
Not to beat a dead, and decaying horse ;-)... the formula provides the "image" size and is useful in determining the amount of image data needed for a particular purpose. Obviously, file size will be affected by the overhead you mentioned, as well as alpha channels, and whether or not it is compressed.
I originally posted this formula in response to a comment on this thread incorrectly suggesting that the optimum file size (more accurately, image data) ought to be calculated backward from the total DPI of the output device. The formula provides an easy way to make such calculations.
I didn't expect the formula to became the subject of scrutiny, but since it has, the exact calculation is (L" x W" x L:ppi x W:ppi x bit depth)/8,388,608. *This will indeed provide an exact measurement of image data*. This can be confirmed in Photoshop by creating a file, then reading its size.
Regards,
Les De Moss
www.digi-graphics.com
Date: Tue, 10 Sep 2002 17:04:45 -0400
From: "Preston Earle"
Subject: Re: File sizes, was: Why such large files
"Les De Moss" wrote:"You got me thinking. Yes, the exact divisor is 8,388,608 (1024x1024x8)
" This can be checked by creating an RGB file in Photoshop, 8x10" @ 300ppi and reading the document size, 20.6MB (20.5994).
"The formula was rounded to 8,000,000 for ease of use/memory."
---------------------
OK: 8" x 10" x 300ppi x 300ppi x 3 colors = 21,600,000 (This would be
bytes)21,600,000 / 1,024 = 21,094 (This would be Kilobytes, Kb)
PS 7.01 reports an 8x10 300ppi .tif image without a profile attached to be 21,101Kb. "File Properties" reports the size to be 21,607,256 bytes. I suppose the 7256 bytes is file overhead. 21,607,256 / 1,024 = 21,100.8 (or 21,101 among friends). This all makes sense to me.
Where does the 8 come in?
Preston Earle
PEarle@triad.rr.com
Date: Tue, 10 Sep 2002 23:45:09 -0700
From: Jan Steinman
Subject: Re: Why such large files
From: "Les De Moss"
>Not to beat a dead, and decaying horse ;-)...Thwack! Bam! Thunk! :-)
>I didn't expect the formula to became the subject of scrutiny...
on THIS mailing list? You should know better! :-)
I think we're in heated agreement. Les did use "file size" in his message, which I'm also guilty of confusing with "image size" from time to time.
--
: Jan Steinman -- nature photography:
: Bytesmiths -- artists' services:
Date: Tue, 10 Sep 2002 22:22:46 -0400 (EDT)
From: MIke Demyan
Subject: Why such large files? more on resolution
Terry's comments reminded me of a page I once saw on resolution. I can't vouch for any of what he says, because when you get down to microns, you begin to lose me. But this site makes a tricky subject a bit clearer.http://www.theimage.com/photography/photopg1.htm
It goes into a lot on digital as it compares to film.
Someone once told me the resolving power of a camera lens. I seem to recall it was very close in size to the size of a pixel in a ccd.
-Mike Bevans
Date: Wed, 11 Sep 2002 07:50:18 -0400
From: John Rawlins
Subject: Re: Why Such Large Files
Having read the numerous posts over the past week on this issue... one of the biggest reasons for large files has been totally overlooked. Let me state in advance before I get into this, that the following "generalized" statement regarding my opinion of file sizes is not directed at most of this groups members, since most of you are far more advanced in color theory than I. My statements are based from having to deal with this on a day to day basis.My point being, that most of the people doing large files don't have a clue what they are doing, or why they are doing them. Unfortunately as well, this may also apply to a number of the art directors and clients asking for these large files. As a commercial printer that gets in hundreds of supplied scans on a weekly basis, we get to see it all. From the constant barrage of 72 dpi RGB's to the 8000 dpi monster files. My most favorite was a 1/2" x 1" scan we received at 4000 dpi that was supposed to reproduce at 9x12". They had all the logic and reasoning why they did this... but I won't get into that one.
Unfortunately, many of those inexperienced in color think that more resolution will solve the ills of the $99 desktop scanners, not understanding gradation, not understanding unsharp masking, and not understanding why their printed jobs look like junk. We have a standing joke in the prepress department that the more crappy the color, the higher resolution they will use and the more expensive proofing they will demand. They will pay for an $85 two page Kodak Approval because they don't like THEIR color on an $8.00 Epson10M. It makes no sense, but we smile and go on.
Having been electronic prepress from its inception in the 70's, and growing up on the first high end drum scanners, and on into high end flatbeds, our experience has taught us what we can, and cannot get away with, scanning resolution wise. Drawing the line for "overkill" resolution is far lower than most would imagine. As one of the posts stated, scanning or digitally capturing 200 catalog images at maximum resolution is a bit absurd if you pretty much know that they will never be used again. It would be far less costly to reshoot the ones needed later, at the size needed.
Very high resolution scans have their place when needed, but its the exception rather than the rule. The type of original, subject matter, final reproduction size and any future intent, determine what scanner we put it on, and how big we shoot it. You will not be able to match my 200 dpi drum scan of a 35mm Velvia transparency at 2000% with your (or my) flatbed scanner at any dpi. But you can do the scan in 1/4 of time it takes me to mount it and scan it and it will look pretty good. That's why most our scans are now done on a flatbed.
We have done numerous printed tests to experiment with different resolutions, reproduced at different screen rulings with varied subject matter. In my opinion, when reproducing at 100% reproduction size of the file size, the human eye just cannot discern the difference between a 300 dpi and a 400 dpi file. And depending on the subject matter, I could throw in some at 200 dpi and probably get away with it if I wanted to. And why would I want to? Remember guys the poor printer has to RIP these files! Take a 40" form of two dozen images at 600 or 800 dpi that someone has scaled and rotated them all in Quark and it will gag the heartiest of RIP front ends... or at least slow it to a crawl. My terabyte of online goes away quick with a hundred or so jobs in progress. Then when they are done, I have to have the tapes to store them all on!
For the prepress guys and the printers in the group... I'm sure you all can relate to where I'm coming from on this issue. I'm not advocating cutting corners and sacrificing job quality to make the printers lives easier. If repurposing is the reasoning for the large files to start with... just make sure I get the small version. If in an attempt to achieve better quality, and end results is the reasoning for the large files... please consult with the people that have to deal with them to really determine if it is really helping or just adding unnecessary time and expense and your part and ours.
Just an opinion from the last link in the chain.
John Rawlins
Date: Wed, 11 Sep 2002 17:13:29 -0000
From: "andrelopes2"
Subject: Re: File sizes, was: Why such large files
Preston Earle wrote
> Where does the 8 come in?
It comes from 8 bits forming a byte.
Using the suggested formula with a regular RGB file (3 colors / 8 bits per color) you can set the "bit depht" as 24 bits (3 X 8) or either as 3 bytes.
If you use the depth number in bits (24), you must divide the result by eigth to get the correct file size in bytes.
If you use the number in bytes (3) - as you've done in your example - the result doesn't need this division.
Regards.
André Borges Lopes
Bytes & Types
São Paulo, SP - Brazil
Well said! I enjoy hearing from the last guys in line because that's what
counts. Everything up to that point is merely some good ideas on a monitor. Though I enjoy reading the news here and some of the fine points have been very helpful in very specific areas, as a whole I need to understand principles upon which I can discern what details will help or hinder me. Well written, John!Mike DeSantis
DeSantis Photography
Date: Wed, 11 Sep 2002 17:03:10 -0400
From: Terry Wyse
Subject: Re: Why such large files? more on resolution
The quote you attributed to me was actually somebody else but I did confirm that based on the resolving power of the film alone, it did seem to support the notion of scanning the film at 4,000ppi.About lens resolution: the link I had in my post noted that lenses are somewhere in the 2-4x finer resolving than film. I think 300 l/mm was listed as fairly common.
The point I was trying to make was about SYSTEM (lens + film) resolution. According to the info I read, system resolution will ALWAYS be lower than either the film or lens resolution. One would assume that the effective resolution would be the lowest common denominater, either the film or the lens is the weak link. Turns out the effective resolution is lower than either. So this still raises the question of how much scanning resolution is enough. Also consider that photographic subjects (not high contrast resolution targets) are relatively low contrast, further lowering the real resolution of the film. I'd still say that scanning at much over 2,000ppi is not capturing any more photographic detail. This is sort of where the argument for digital comes in that it is nearly as "sharp" as film given all these factors. And you could even consider the scanner itself as part of the system resolution equation. Food for thought.
Terry
on 9/10/02 10:22 PM, Mike Demyan wrote:
> Terry's comments reminded me of a page I once saw on resolution. I can't vouch
> for any of what he says, because when you get down to microns, you begin to
> lose me. But this site makes a tricky subject a bit clearer.
>
> http://www.theimage.com/photography/photopg1.htm
>
> It goes into a lot on digital as it compares to film.
>
> Someone once told me the resolving power of a camera lens. I seem to recall
> it
> was very close in size to the size of a pixel in a ccd.__________________________________
WyseConsul
Color Management Consulting
v 704.843.0858
__________________________________
Date: Wed, 11 Sep 2002 17:02:55 -0700
From: Peter Figen
Subject: Re: Why such large files? more on resolution
Terry Wyse wrote:> The point I was trying to make was about SYSTEM (lens + film) resolution.
> According to the info I read, system resolution will ALWAYS be lower than
> either the film or lens resolution. One would assume that the effective
> resolution would be the lowest common denominater, either the film or the
> lens is the weak link. Turns out the effective resolution is lower than
> either. So this still raises the question of how much scanning resolution is
> enough. Also consider that photographic subjects (not high contrast
> resolution targets) are relatively low contrast, further lowering the real
> resolution of the film. I'd still say that scanning at much over 2,000ppi is
> not capturing any more photographic detail. This is sort of where the
> argument for digital comes in that it is nearly as "sharp" as film given all
> these factors. And you could even consider the scanner itself as part of the
> system resolution equation. Food for thought.I agree with your basic premise, but I don't necessarily agree with you conclusions. Having scanned a ton of film at hardware resolutions ranging from less than 1000 ppi all the way to 8000 ppi, I can say without a doubt that, if you are making large prints, there is a definite advantage to scanning at resolutions higher than 2000 ppi. 4000 ppi does seem to be about what most transparency films actually give you. Negs ARE closer to a real 2000 ppi. You can easily see the difference, IF (big if) you have a scanner that can actually resolve the extra detail. The scanner is most definitely part of the optical system, maybe even the most important part, which may also be part of the reason my film to digital comparisons have slightly different results that most that I have read.
As far as digital, while it is grainless, the single shot 35 mm cameras (I've been testing a DCS760 today) aren't even close to film. Just judging from viewing the preliminary captures, and knowing what I get from scanning film, there is still no contest, but I'll wait till I get the film back to scan for a more detailed comparison. The film wins for detail and overall image quality. The digital capture wins for production speed. They are both just tools, which have to be chosen for their respective values.
Peter
Date: Thu, 12 Sep 2002 09:38:23 -0400
From: Jonathan Clymer
Subject: Re: Why such large files? more on resolution
From: Peter Figen
>
> As far as digital, while it is grainless, the single shot 35 mm cameras (I've
> been testing a DCS760 today) aren't even close to film. Just judging from viewing
> the preliminary captures, and knowing what I get from scanning film, there is
> still no contest, but I'll wait till I get the film back to scan for a more
> detailed comparison. The film wins for detail and overall image quality. The
> digital capture wins for production speed. They are both just tools, which have
> to be chosen for their respective values.The conversations about film vs. digital always compare single or three shot chips (usually 12 or 18MB) and film. The files I have seen from large format scanning digital backs (original capture of 300+MB) have astounding quality. It's hard for me to imagine how scanned film could equal this level of quality, but I have not done any serious comparisons. I'm wondering if any one else can intelligently comment on this.
Jonathan Clymer
Date: Thu, 12 Sep 2002 10:02:33 -0600
From: "Les De Moss"
Subject: Re: Re: Why such large files
From Jan Steinman:
> on THIS mailing list? You should know better! :-)I honestly don't know what I was thinking...:-)
Said scrutiny did uncover a question about the accuracy of the Image Size Formula, when dealing with images other than 8 or 16 bit.
Is it correct that a single pixel, whether 1,2,4 or 8 bit, is stored as one byte? And that pixels with a bit-depth between 10 and 16 are stored as 2 bytes? If so, the formula for determining image size is inaccurate for anything other than 8 or 16 bit image files, or 1-2 bytes per pixel.
Les De Moss
www.digi-graphics.com
Date: Thu, 12 Sep 2002 18:25:31 -0000
From: "satusr"
Subject: Re: Why such large files? more on resolution
Jonathan Clymer wrote:
> The conversations about film vs. digital always compare single or three shot
> chips (usually 12 or 18MB) and film. The files I have seen from large format
> scanning digital backs (original capture of 300+MB) have astounding quality.
> It's hard for me to imagine how scanned film could equal this level of
> quality, but I have not done any serious comparisons. I'm wondering if any
> one else can intelligently comment on this.
>We shoot original artwork, and catalog products using a Betterlight Super 8K 4x5 scan-back. We can shoot files sizes exceeding 600 meg.(if we wanted to). Image quality and sharpness are FAR SUPERIOR to any drum scan from any size transparency, hands down. Its not even a fair comparison anymore, film loses everytime. The scanback hardware and lighting to do the job right is a bit pricy, but its really the only way to go anymore.
John Rawlins
Date: Thu, 12 Sep 2002 16:02:10 +0100
From: Paul Fawley
Subject: Re: Why such large files? more on resolution
on 12/9/02 2:38 pm, Jonathan Clymer wrote:It's hard for me to imagine how scanned film could equal this level of quality, but I have not done any serious comparisons.
FYI
a client recently complained about the grain from a 5x4 ektachrome tranny scanned and printed to A3, the image was of office furniture in a large set.
Given a 4shot capture off an Imacon 3020 he was very pleased with the 'Clean' look to the image, and the similar degree of detail. Given a 16 shot capture, there was no contest..add to that the post production available from the original camera file... there aren't many still life subjects that are better shot on film.
Date: Thu, 12 Sep 2002 12:14:17 -0700
From: Peter Figen
Subject: Re: Re: Why such large files? more on resolution
satusr wrote:> We shoot original artwork, and catalog products using a Betterlight
> Super 8K 4x5 scan-back. We can shoot files sizes exceeding 600 meg...
> (if we wanted to). Image quality and sharpness are FAR SUPERIOR to
> any drum scan from any size transparency, hands down. Its not even a
> fair comparison anymore, film loses everytime. The scanback hardware
> and lighting to do the job right is a bit pricy, but its really the
> only way to go anymore.There's no doubt that the very high end scanning backs are better than film. The comparisons I'm talking about are single shot cameras and backs. Scanning backs are great when nothing moves, but for those of us who shoot people or do things like rack focus or shake the camera during the exposure, they don't work. My concern is finding where the state of the art in single shot cameras lies, and so far, film still is better in my opinion, for overall quality, but not for production speed.
Date: Fri, 13 Sep 2002 17:31:46 -0500
From: Al MacDonald
Subject: Re: Re: Why such large files? more on resolution
Even though I have been shooting digital for 7 years with a MegaVision T2, up until last April I was still shooting 50 to 100 sheets of 8x10 film per month for packaging and large reproduction (11x 17 to 11 x 24 final size) jobs....we purchased a Sinar 44HR...96 meg 16 bit files in 4 pop mode....have not shot one sheet of film since....color fidelity and detail is awesome. There is no good reason to shoot film in a commercial environment anymore. For "art" sake I still like film.....always will!
Al MacDonald
Shaughnessy MacDonald, Inc.
Adobe Photoshop training classes are taught in the US by Sterling Ledet & Associates, Inc.