Jump to content

Calculating file size from scan


emrys

Recommended Posts

I am contemplating picking up a used drum scanner to scan 4x5

trannies. Obviously my primary reason for wanting a drum scanner is

to be able to produce large fine art prints, and I appreciate that

the file sizes in involved will be considerable.

 

What I am trying to come up with is a formula for working out file

sizes in advance, based up color depth and resolution. I would

typically be looking to output on something like a Lightjet 5000 and

I understand that the print resolution here is ~300dpi.

 

I understand that a drum scan will be typically a 48-bit color image,

therefore each pixel will be 48 bits = 6 bytes. By way of example, if

I scanned a 4x5 image at 1000dpi, 48-bit color, I would expect a file

size of:

 

(1000*1000)*(4*5)*6/(1024*1024) = 114Mb

 

Is this right, or way off track?

 

Answers much appreciated,

 

Mike

Link to comment
Share on other sites

Your math is approximately right, but let me make some comments.

 

Even in lossless formats like tiff or png, the files are usually compressed somewhat when written out to disk. This can reduce the file size signficantly.

 

1000 ppi is pretty low for scanning resolution. The largest print you could make that way would be just a bit larger than 11 x 14 inches. If you are not going to use a higher scan resolution than that, it is not clear you are going to gain much by using a drum scanner. You ought to get quite good results even from the new Epson 3200 flatbed.

 

I don't know anything about the Ligthjet 5000, but are you sure it uses 48 bits? If not, the file size could be reduced by a factor of two.

Link to comment
Share on other sites

Very good points.

 

Yes, the Lightjet will print a 24-bit image and not a 48-bit image. I had forgotten about the whole compression issue, and I realise that this will alter the file size drastically. Bit of an oops on the 1000dpi but I was only using it for easy maths. I understand that top of the range drum scanners can scan at up to 12000 dpi.

Link to comment
Share on other sites

For us non-mathematicians, just open a New file in Photoshop or

comparable image editor, enter the parameters (bit depth, resolution,

size-pixels, English, metric, whatever) and, voila!, you have your

answer. If I scan a so and so neg at X res how large will the output

file; if I want to output a y size file at z dpi-all easily arrived

at. GOODLUCK>>

Link to comment
Share on other sites

Finally, someone with basic math skills. Also note that you can define enlaregments in terms of dpi as well. That is, if you scan at 2400dpi and want to print at 300dpi, the enlargement is 8X. For the 4x5 film that would give you 30x38 inch prints. At 12000dpi, you get 40X enlargements or 150x190 inch prints.

 

Having said all that, I would recommend a scanner that does about twice the resolution you need. If the math says you need 2000dpi, choose a scanner that will do about 4000dpi. Also know that you will start seeing the grain of the film at 3000 to 4000dpi.

Link to comment
Share on other sites

my experience is a little differnt from what you guys seem to be saying. i work with a very good digital imaging person and he indicates that you should use different scanning techniques depending on the type of application or printer you will be using. for scans to be sent to a publisher, they should be scanned at twice the print resolution (ie, 600dpi for a 300 line halftone print) and at the correct physical size of the published image (5x7, 8x, whatever). his company uses an optical printer in conjunction with a durst sigma scanner, and he uses the same procedure when making scans to be printed on his optical printer - he scans to the maximum resolution fo the printer (274dpi) at the physical size of the print to be made (16x20, whatever). however, he said that the procedure for inkjet printers is very different, and that it takes extremely high resolution scans lilke you guys are talking about for inkjets to reproduce the same quality prints - he does not use this process because the resulting files are so huge they become extremely difficult to manipulate.
Link to comment
Share on other sites

Just a detail regarding the use of 16 bits/color instead of 8bits/color. It is true

that the printers accept almost only 8 bits/color pixels, but if you intend to

process the images before printing them, then 16 bits makes a lot of sense !

Even if the file size is larger...

 

As an example: you have a nice, very smooth blue sky on your transparency

going continuously from 'very dark/ blue to 'exactly middle' blue. Let's assume

it is the same blue as the 'B' of the 'RGB' filters of your scanner. Then, if you

scan it with 8bits/color, you will end up with 128 levels for your whole sky

tones. This is not enough... Moreover, if you want to lighten a little the tones,

you will end up with a very bad looking staircase effect.

 

The solution is to scan in 16 bits, getting the largest number of colors, do the

color and tone adjustments with the curves tool of your image processing

software, then convert it to 8 bits before saving and/or printing. My personal

choice is to save a 16 bits mostly raw version _and_ an 8 bits processed

version. This is not so much additional space on a DVD-R, and then I avoid

rescanning the original if I need to reprocess it.

 

BTW, I confirm that a 4x5" scan is ~290 Mb @ 2400 dpi with 8 bits/color, and

~580 Mb with 16 bits/color.

Link to comment
Share on other sites

Keep this simple! Start your calculations with the output device you will use. If it needs 300 ppi at actual print size then use that to get the size you need to make the reproduction. Heres a simple version - output device pixels X width of final reproduction (the print). i.e. - 300 pixels X 20 inches = 6000 pixels wide. You need a finished file thats 6000 pixels wide. Figure in 24 bits, there are no printers that I'm aware of that use 48 bits, so you will reduce your 48 bit files to 24 bits before you print. You're probably wondering what happened to the length? Just let it fall when you set the crop in the scanning software.
Link to comment
Share on other sites

For me, 16 bit colors scans is a way of preserving as much image data as possible in the prospect that future versions of Photoshop will be supporting layers in 16 bit. But editing in 16 bit doesn't make sense until then for layers are mandatory in my way of working and are more important towards the end results than the extra bits. Just my opinion, others may be able to do it right first time with a minimum of editing.
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...