What Actual Megabyte Size Should Negative Scans Be

[nickser]

Hi There,
This will be a quick question for you brain boxes out there to answer for a digital duffer like me!!
I have just recieved back from a lab in Guernsey in the UK my first ever digital scans of seven Fuji NPS films.
I requested 18 meg scans for high resolution. (£10.00 per film) so I thought that each film would have a cd of it's own - 18 x 36 = 648 meg. Yet the scans weigh in at 2 to 3 meg a file. And they are grainy as heck when displayed on my 21" monitor. It is like looking at a 1.3 megapixel p&s.
Am I expecting too much, is my maths gone to pot or have they given me the wrong size?
I have tried to contact them today but they appear to be closed for the weekend.

Many thanks,

Paul

 

[craignorris]

Paul,

The files are likely to be JPEG files, which are compressed. Check the file names on the CD for a ".jpg" extension. The compressed data on the CD is decoded back to the full image size when you open the file. To confirm the scanning resolution, just look at the image size. How many pixels wide by how many pixels high? Multiply width pixels by height pixels by 3 and there's your uncompressed file size. The uncompressed file size is shown in the bottom left corner of the Photoshop window. Are you using Photoshop?

The graininess is not unusual for negative film. I gave up using negative film for serious work because it is too grainy. Fuji Reala and Konica Impresa are the exceptions. Use ISO100 slide film if you don't want to see grain in the scans.

Best,

Craig

 

[writing4me]

Hi Paul, Interesting question. That does sounds like they've given you jpegs which I find suitable for doing web sites etc. To get a decent higher res scan of an entire roll at time of processing the film costs a bit more and typically the lab has to be better than the average mini-lab for it to be worth the price. (One of my local Frontier labs has good equipment, but teenagers they employ to cheaply run the stuff do not care about the result, either that or they just havent' been taught they can do a better job). It could be you're lab suffers the same problem of non-interested work force and they didn't follow your instructions. I've found that to get at least Fuji Frontier lab that does a decent job of hi-res scans as each roll of E6 is processed adds about US$30 to the existing charge of processing the slides. (Cheaper for negatives, but they're too grainy for me)

If I want one exceptional image scanned for something better (say I want to use the image in a brochure, or ad) I send it for a drum scan and tell the lab, or scanning service, that I want a final size of say 9"x12-or-whatever" at 300 dpi, output as CMYK TIFF images. A TIFF of that specification in CMYK mode it will be about 38mb in size. A TIFF in RGB mode it will be about 28mb. (The jpegs would be smaller but suffer from compression, the TIFF does not.)

Does that help? -Lynn

 

[ou1954]

>Posted by Paul matthews on Saturday, July 26, 2003 - 1:06 pm: > >Hi There, >This will be a quick question for you brain boxes out there to answer >for a digital duffer like me!! >I have just received back from a lab in Guernsey in the UK my first >ever digital scans of seven Fuji NPS films. >I requested 18 Meg scans for high resolution. (=A310.00 per film) so I >thought that each film would have a cd of it's own - 18 x 36 =3D 648 >Meg. Yet the scans weigh in at 2 to 3 Meg a file. And they are grainy >as heck when displayed on my 21" monitor. It is like looking at a 1.3 >megapixel p&s. >Am I expecting too much, is my maths gone to pot or have they given me >the wrong size? >I have tried to contact them today but they appear to be closed for >the weekend. > >Many thanks, > >Paul

Something is not right here. I get a roll processed at various one-hour=20 places around here and a scan with a CD costs from $3.95 to $5.95 in=20 addition to the regular processing and 4X6 print set.

The $3.95 is a typical price for the one hour on-site service and $5.95 if= =20 they send it to Kodak.

I generally then get a set of prints, an index sheet, and either 1 set of=20 digital files, or from Kodak I get 3 sets of 3 different resolutions and=20 they are always of excellent quality.

Try a different vendor.

DAW

 

[nickser]

Hi folks,

Thanks for all the replies. It appears that they have given me 6 meg files instead of 18 meg files. I will get onto them first thing on monday and get them to sort it out for me pronto!

Thanks to all,

Paul

 

[leo_jar]

Hi,
To optimize the scanning, it is better to scan the film (neg/positve) with the resolution 2000x3000 which will produce 18M tiff file. This number is based on the calculation of average 135 lens resolution.
Any higher resolution scanning will do no/little different with the interpolation in Photoshop. I do suggest you can do a test in scanning the 135 film with different resolution and compare the result with the image interpolation in photoshop. you can get better idea.
By the way, the simple rules for scanning
135 film , 18M tiff
120 film , 40M tiff
4x5 film , 50M tiff
8x10 film, 80M tiff.
One more thing, longer scanning time (in flat bed scanner) may result in burr image due to the vibration.

Leo,

 

[mike_nunan]

Leo,

While your suggested file size may be sufficient in terms of "average" lens resolution, it does not take into account the grain feature size of the emulsion itself. I can confirm from direct experience that there are visible benefits to be gained from scanning much larger files from 135, if you are aiming for large reproductions. My Nikon LS-30 scanner has 2700ppi resolution and produces a 28MB file, which by your criterion should already be lens-limited in terms of sharpness. Scans from the new Minolta 5400 are visibly sharper, however (to put things into perspective, that is a 112MB file). I have written a comparison review on

Please, Log in or Register to view URLs content!

which includes a bunch of s&le scans, so you can judge for yourself. I would suggest you take a look at the "Wharf view" s&le, which is shot on ISO100 B&W film using a Planar 50/1.4. The Minolta scan is clearly sharper.

Regards,

-= mike =-

 

[nickser]

Hello Folks,
Further to my last thanks, it has been shown my maths is duff. I do have 18 meg files.
I am just not overly impressed with the result. I will load PS7 (never used it before for digital photographs) and see what transpires from there. The lab said they were looking into changing the scans to improve quality (the 5x7 prints were good, they use the same size files to print out, so they say). Looking at what size the new scanners put a file out at, 18 meg seems inadequate. Maybe the next investment is my own film scanner.

 

[leo_jar]

Hi Mike,
Sharper doesn't mean higher resolution. The limit of 135mm is about 25M. How to testing the film limit is to taking a resolution chart and then scanning the film by different resolution setting.
I was working in Kodak Professional for very long time and concentrated on digital imaging. And there is a principle in designing "bridge" digital system (i.e. film/paper to digital or digital to film/paper). By using drum scanner, you can scan 135mm film to 100M is very sharp image but the information on the film is same to the file in 50M (i.e. the words of the banner in the image can't be seem in 50M image is still cannot be seem in 100M scanned image). One more thing I should need to mention is that the limiting factor for the CCD scanner is the dynamic range not the resolution. Drum scanner is still have the advantage and can nearly restore/record all the information from the film.
Cheers,
Leo

 

[mike_nunan]

Hi Leo,

> Sharper doesn't mean higher resolution

Fair comment, you can't get back information that isn't there in the first place, but I'm reluctant to accept that the 6Mpixel file you're talking about is able to capture all the detail in a good 35mm frame. However, I believe that the 40Mpixel file produced by my Minolta 5400 gets close (it equates to 106 line-pairs per mm). I've never seen MTF values for top grade optics at 100lpmm, but I suspect there won't be much contrast left at those frequencies, even with the best Zeiss or Leica glass. I see that some drum scanner manufacturers see a need for 15000ppi devices, so I imagine there must be some kind of benefit in going that high or they wouldn't sell any. My guess is that the main benefit would be cleaner, more "analogue" reproduction of the grain structure. Nine times out of ten, when I see a print that is obviously digital, it is the gritty nature of the grain that gives it away.

I agree that dynamic range is a big issue for desktop CCD scanners, but it's stunning how far things have come in the last three or four years. Apparently Minolta quote a Dmax figure of 3.8 for the 5400 (ignore the spurious 4.8D figure -- that is just the dynamic range allowed by a 16-bit number expressed as a log) and from what I see on my K-chrome scans, I'm prepared to believe it. I've also been amazed how little difference 16x multis&ling seems to make, which is another sign that the noise floor is pretty low to begin with.

Speaking of multis&ling, scanning at a high resolution then down-s&ling effectively gives it to you without the extended scan times, so in that respect it also makes a contribution to the dynamic range.

Finally, to draw a parallel with the audio industry, when PCM (CD-type) audio began to catch on in the early 80's it was accepted wisdom that 44.1kHz/16-bit was &ly sufficient. Nowadays 96kHz/24-bit systems are fairly standard, 192kHz s&ling rates are becoming commonplace and the shortcomings of the CD audio standard are almost universally acknowledged. I suspect the small files and limited dynamic range of today's low- and mid-range scanners will be looked at in the same way 20 years from now. The trend is already there; for ex&le I know of one movie post shop which is switching from 16-bit to 32-bit IEEE floating point representation for their image files.

Best regards,

-= mike =-

 

[leo_jar]

Hi Mike,
You are right and wrong, only the high contract prepress camera lens has the resolution that you mention. And the scanner with 15000ppi(around 600ppm)is for the use of electronic/chip industry and not for color film scanning. To take 6M pixels in 36x24 film, it is equel to 83 pixel per mm (pixel mean the spot with RGB information). DO you mean 40M file (RGB Tiff) or 40M pixels file (=120M file RGB Tiff)? By the way, if you would like to scan the color of each dye on each layer of film, you may need a higher for 135film (as the Pro Photo CD resolution). But as I state before the "image information" is still around 20M (a group of dye particles carry same set of information due to limit of lens resolution). Kodak has the extensive study in determine which is the optimal resolution for film scanning when Kodak developed PhotoCD system and what I write here was the summary from the result.
CCD is a hot items during these ten year and ton of money had been spent in new CCD technology. So the SN is better and so the dynamic is now better than several year ago. Here is a good ex&le. Imacon scanner is the professional scanner using CCD technolgoy and for desktop publishing. They using Three Ccd for different color scanning. Between the CCD and the film is a ultra high resolution lens for focusing and fit for different size of film. For 35mm film, the high resolution is 222ppm. I do have the experience in checking several scanner with the same scanning resolution, they produce different resolution image. Some scanner need to increase scanning resolution in getting compartable image. There are too many factors affecting the image. But please chk to scan b/W image using color mode to cross chk if the scanner is using color moasic filter over the ccd and do the interpolation color information.
In designing CD audio, the enginner of Philips using a lot of effort in squeezing 70mins audio information into 650M disk so they need to give up many thing such as the information between each s&le data (the information inbetween the s&le is reconstructed afterward. Audio is analog signal and record in digital media. High s&le freq means more accurate reflect the sound. Bit equals to dyanamic of scanner. True is that SACD produces better sound. However, in scanning industry, drum scanner already be here more than 15 years and there is not the issue as cd audio development.

Last but not least, there is a different theory about the film resolution. The true is that the resolution of movie film is not very high. But the random pattern of dye in different frame makes the preceive resolution of the projection image become very high. The IEEE image is for digtal film production and need 32bit (even 48).

 

[mike_nunan]

Hi Leo,

I've definitely seen 15000ppi scanners that are intended for colour repro use (ISTR there are Tango and ICG models that will reach this resolution, although they carry six-figure price tags). However, I tend to agree that the benefit seems questionable with general-purpose film, and that they probably find their most appropriate applications in photolithography and other technical niches.

Yes, I did mean 40Mpixels (=120MB for 8-bit-per-pixel uncompressed RGB tiff). Your point about Kodak's research results for PhotoCD mirrors what I said about 16-bit PCM audio -- in the end both formats were constrained by the practical limitations of their day, and the designers were trying to avoid squandering resources or creating requirements for expensive parts in the implementations. We are lucky that we are not tied to a fixed file format for distribution in the way that audio people have to live with redbook CD-audio, but it makes no sense to stick with a 6M pixel standard for 135 format scans. 83 pixels per mm is 41.5 line pairs per mm, and there are many lenses that will provide useful contrast values at higher frequencies than that (though not up to 100lpmm!)

I agree fully on the other points about the impact of the scanner optics and the type of CCD implementation. Light source can also be added to that list (I've never had a close look at Imacon scans of B&W film, but I imagine they must be super-hard with the type of light source they use). Also, it's very interesting what you say about the time-domain effect of movie film projection creating an increase in perceived resolution, that makes a lot of sense.

TTFN

-= mike =-

 

[leo_jar]

Hi Mike,
You may not believe that the light source of Imacon is a on-the-shelf fluoresce tube!! The spectrum is not continuous. However, the matching table is good and can normalize the data.
Kodak's highend scanners are using
1. halogon/tunsgten (older model
2. LED
The 1. is the expensive solution since a big heavy duty powersupply is needed to supply stable voltage to the l& so the color temp can be maintained.
2. LED is a newer solution of the light source.

However, cool light is the best option when considering cost, lifespan and stability. So, why all consumer/prosumer scanners are using cool light source.
But Scitex and Imacon are using coollight source very well but still not the best.
Recently you may know that the digital camera is going to 1x million pixels. Using same philosophy, the optimal resolution of digital SLR camera is 18million pixels (3 x 6millions).

Let share

Leo

 

[ou1954]

>Posted by Leo Jar on Wednesday, July 30, 2003 - 7:01 pm: > >Hi Mike, >You may not believe that the light source of Imacon is a on-the-shelf >fluoresce tube!! The spectrum is not continuous. However, the matching >table is good and can normalize the data. >Kodak's highend scanners are using >1. halogon/tunsgten (older model >2. LED >The 1. is the expensive solution since a big heavy duty powersupply is >needed to supply stable voltage to the l& so the color temp can be >maintained. >2. LED is a newer solution of the light source. > >However, cool light is the best option when considering cost, lifespan >and stability. So, why all consumer/prosumer scanners are using cool >light source. >But Scitex and Imacon are using coollight source very well but still >not the best. >Recently you may know that the digital camera is going to 1x million >pixels. Using same philosophy, the optimal resolution of digital SLR >camera is 18million pixels (3 x 6millions). > >Let share > >Leo

I can't quite understand the last paragraph. Can you re-state it?

Thanks,

DAW

 

[leo_jar]

Hi DAW
Due to the limitation of the lens resolution (since the digital SLR camera is still using conventional lens), the optimal pixels number is 18million. In real life, the pixel count is based on the total of pixel number and they are monochromic sensitive. So RGB filters should be placed on each pixel. By mathamatics calculation (interpolation), the camera will reproduce the full color information on each pixel. (so why the cameras will have different in image quality even they are using the same CCD. They may using different method in calculation)

thanks,

leo

 

[nickser]

Hi Folks,
I got in touch with the firm and they are 18 meg scans. I also bought and Epson 1290s and printed a couple of them out on the supplied epson paper. They are quite marvelous, maybe not as good as a traditional darkroom, but very acceptable. I will not bother sending them to the lab to have them processed. I will leave the lab for the really good/big stuff if I ever need to go that route again

Thanks for the information.

Paul.

 

[noel_greene]

Help please
I am completely new to digital printing and I have trouble sizing images to fit paper size
I had some images scanned and the files are 50mb
I open the image in Photo Elements and print preview shows me a 1.5 x 1.5 inch image in the middle of the screen

I cannot get the image to adjust to A4 size without the scale going to 500%+ and I get a piece of the image printed

I may be doing something very elementary incorrectly

how can I correct this problem. I have tried all boxes in Page Set UP , res&le image, Resizing Image etc

thanks

when I go to print them I cannot get the files

 

[John_bird]

Hi if you are useing photo shop, and I assume elements works the same way. First you must resize the canvers or the image will be croped to fit on the original canvers. If you are useing an epson printer and prin your original file useing epsons soft ware the image will be re sized automaticaly.
Hope this helps John.
ps I use a CanoScan 2700f and the twain inerface on that can be used to out put a printer file as opposed to a direct(100%)2700 dpi file