Scan Resolution?

[Scott --]

Hey, everyone -

I actually managed (somehow, despite my usual ineptitude) to make a possible sale today. Client's requesting a "high resolution scan", which he defines as 300 dpi at 8-1/2 x 11".

Ok, so if I go into VueScan, select "Scan Resolution: 300 dpi" and go to "Output Size: 8.5 x 11", am I getting what's being asked for? I generally scan at the scanner's max (1200 dpi for my crap-cheapo flatbed) and resize in PhotoShop. And, clearly, I should/will clarify with the client what exactly he needs (and, should my scanner not generate what's needed, they'll scan the slide, anyway), but I'd like to get a handle on this tonight before I make a fool of myself tomorrow. Which is inevitable, anyway, but still... :rolleyes:

Any advice?

Thanks,
Scott

[Eric James]

I would scan at max res and then resize in PS using the crop tool set to the client's desired dimensions and resolution. The "scan once" or "master scan" workflow philosophy is a big time saver and saves wear and tear on your equipment and slides/negs. Some scanners don't provide equivalent output quality at all resolution settings; it's a good bet that your scanner performs well at it's maximum scanning resolution.

Congratulations on your sale.

[Gene McCluney]

I will second the above recommendations. The ONLY time I scan at less than maximum resolution of my scanners is when scanning a 4x5 or 8x10 negative, when full optical resolution of the scanner is overkill, and time consuming.

[Ted Harris]

Scott, is it a 35mm slide that you are being asked to scan, if so does your potential client understand that even the best of the current generation of consumer flatbed scanners doesn't do much of a job on a 35mm slide? You can save both of you a lot of grief if he understands that.

Having said that, the earlier advice is correct. Scan at the maximum real resolution of the scanner and then down rez in PS. The technique to use in PS is to go to Image>Image Size. Next uncheck the "Resample Image" box to maintain the size of the file. Then set the image size.

[JW Dewdney]

I will second the above recommendations. The ONLY time I scan at less than maximum resolution of my scanners is when scanning a 4x5 or 8x10 negative, when full optical resolution of the scanner is overkill, and time consuming.

Okay - sorry Gene - not that it's relevant to the conversation - but that just didn't make ANY sense to me. You use FULL optical res when scanning printed materials (which generally don't need any more than 300ppi) - yet when it comes to scanning FILM - which has alot more than 300ppi to OFFER - that's when you DON'T??

Could you have had them reversed?

[Gene McCluney]

Okay - sorry Gene - not that it's relevant to the conversation - but that just didn't make ANY sense to me. You use FULL optical res when scanning printed materials (which generally don't need any more than 300ppi) - yet when it comes to scanning FILM - which has alot more than 300ppi to OFFER - that's when you DON'T??

Could you have had them reversed?

I didn't put printed materials in my equation. What I meant to say was any film below 4x5 in size, I use full optical resolution of my scanners, which is 4000ppi. For 4x5 I often use 1250ppi, out of 2500ppi available to me. For 8x10, for convenience I use 600ppi frequently, as even that makes a 16x20 300ppi print. There, is that clear? Now, for reflected (print) material, I scan at whatever resolution will result in the final reproduction size requested by my clients.

[Ted Harris]

Gene, what you ar doing may or may not make sense depending on th scanner you are using. Point being that the full advertised optical resolution of all of the consumer flatbed scanners is total marketing hype and th real optical resolution is in the range of 2000 to 2400 spi for all of the current models, much lower if we are talking older scanners. If you are using a high end scanner then it might make sense.

[Scott --]

Thanks for the input, guys.

Specifically, I'm scanning a 4x5 Provia transparency on an old HP 5370C Scanjet flatbed with transparency adapter and VueScan. It's not the best, and I don't know if I can coax something worthwhile out of it; like I said, I'll get the best I can together, forward it on, and if it's not going to cut it, they can drum scan the slide. Which is what'll likely happen, but if there's a way to expidite handling on this (which a digital file would obviously do), then I want to give it a shot.

Looks like I'll scan it at the max optical, maybe a little higher, then downsample it in PS to the requested 300 dpi. Sound right? And I'm guessing I should be working at 48 bit color instead of 24...

So much easier just scanning for web use, I tell ya. :rolleyes:

[JavaDuke]

Scott,

Don't scan higher, just use the native optical resolution of the scanner. Resampling algorithms of Photoshop are way better than any scanner firmware or scanning software. For example, if your native scanner res is 2500 dpi, scan at this resolution, then open the file in Photoshop, open the Image Size dialog and resample your image to the required size and resolution. Some people recommend downsampling the image with lots of fine details gradually, while some are ok with doing it in one step. I think in your case it's ok to do it in one step, since your print will be only 8.5 x 11".
Hope this helps.

P.S. Oh, and yes, 48 bit color (i.e. 16-bit) Adobe RGB. You can convert it to 8 bit sRGB last step before you send it to the printer.

[Scott --]

Ok, got a decent scan at the max native resolution (1200 ppi), sized to a bit more than 8-1/2 x 11 to cover the aspect ratio. Downsampled it in PhotoShop, and the image looks nice. Have run off a slew of format and USM questions to the publisher (essentially confirming that they'll want the raw scan film at their requested resolution, vs. me doing any USM or anything). We'll see what happens.

Thanks for all the input.
Scott

[Leonard Evens]

As others have indicated, you should scan at your scanners maximal optical resolution. That will determine the number of pixels you get. A typical 4 x 5 frame is about 96 x 120 mm, which at 1200 spi (samples per inch) will yield an image of size roughly 4500 x 5600 mm. (You will probably get a bit less because of cropping.) If you print that at 8.25 x 10.3 inches, you will be sending the printer roughly 545 ppi, which is higher than the requested 300 ppi. (Divide the number of pixels along a dimension by the linear dimension. If the aspect ratio is maintained, either dimension can be used.) That will avoid any hint of pixelization in the print, which may be what the client has in mind.

Unfortunately, although your scanner is collecting a large number of samples, it probably is not doing very well at preserving fine detail. Scanning at 1200 spi, which is equivalent to roughly 47 samples per mm can in principle deliver a resolution of half that or about 23 lp/mm. To get an 8 1/2 x 11 print out of that you need to enlarge a bit over two times which would reduce the resolution to about 10 lp/mm in the print. That would be more than enough provided your scanner actually provided that resolution in lp/mm, but you would be lucky if it actually delivered half of that. So the best you should expect in the print might be about 5 lp/mm, which may be just barely acceptable.

Your client may or may not be happy with what he gets, depending on just what he looks at. I suspect issues like composition and how you use DOF will make more difference with respect to his impression of sharpness than the above analysis would suggest. Also applying a moderate degree of sharpening in your photoeditor may help, but don't overdo it.

[Scott --]

Just heard from the client, and the scan was good! Scanned at the native optical max resolution (1200 ppi), adjusted curves ever so slightly, and left the USM to the publisher. They were happy.

Not bad for a $30 flatbed HP! :D

Thanks for the help.
Scott

[Neal Shields]

However, even with a slide, I expect you are throwing away a lot of dynamic range with a flat bed scanner. I did some personal compairisons between comercial drum scans and my own flat bed to compair resolution but what astounded me more than the additional resolution was the additional dynamic range that the drum scanner captured.

If I were trying to establish a reputation and a business, I would not stop with what made the customer happy, if personal economics permited.

After all the purpose of a photograph is usually to show it to other people so many more people are likely to see it besides your customer.

[Kirk Gittings]

Neal et al.

If you have a chance, check out Silverfast's new Multi-exposure, multipass feature. It dramatically increases dynamic range on flatbeds.

[Gary L. Quay]

I have a related question. I scanned some old 35mm slides today. I scanned them at 300 dpi at 8x10 output. I had heard that one could, instead of making one huge file at 4800 dpi, one could make smaller files, one for each size of the eventual output. My printer will only do up to 8.5x11, but I want to get an Epson Pro 4000, which will do 16x10. Is there any reason I can't scan a 4x6, a 5x7, an 8x10, and an 11x14 seperately at 300 dpi each from a slide (apart from normal resolution issues with 35mm) and get sharp images to print? At 300 dpi, these would still take up less space on my hard drive, and load much faster. At 4800 dpi (max optical on my Epson 4870), my computer (an eMac) takes geologic time to make the simplest color corrections.

I shoot mostly medium and large format nowadays. So it may not be as much of an issue. I don't know.

Thanks
--Gary

[JW Dewdney]

Boy - if someone's paying you for it, Scott - and it's a transparency - I sure wouldn't be using a scanjet!! Do yourself a favour -and shell out $30 for a good drum scan. It'll look WAY better - and the better looking file will do a little advance advertising for you, too.


(sorry - I hadn't seen the later posts - was responding from the first page.)

[Gene McCluney]

Gene, what you ar doing may or may not make sense depending on th scanner you are using. Point being that the full advertised optical resolution of all of the consumer flatbed scanners is total marketing hype and th real optical resolution is in the range of 2000 to 2400 spi for all of the current models, much lower if we are talking older scanners. If you are using a high end scanner then it might make sense.

For all 35mm and 120 roll film formats I am using a Nikon SuperCoolscan 9000ed which is a "true" 4000ppi. For 4x5 and 8x10 I am using an Microtek Artixscan 2500f, which offers a "true" 2500ppi optical resolution. So for my statements, the methods are "true". I have been scanning for years now, and I know the resolution of my scanners. While these are not "drum" scanners, they are top quality scanners in their respective fields. The Artixscan 2500f particularly has been very highly regarded over the years, and we all know the Nikon 9000ed is the top medium format scanner under $6000.

[Kirk Gittings]

I believe Ted paid less than this for his professional flatbed. A demo model of course.

[Ted Harris]

Gene, no argument with wht you are doing with the scanners you are doing it with. Problem is that lots of folks get suckered by the advertised resolution of consumer scanners in teh under 1000 USD range. The Artixscan 2500 is a great machine for its price and it is unfortunate that Microtek decided to stop selling it in th North American market. It is still available in some European and Asian markets and, as of the last time I asked during the winter, was still very much in production.

Gene is right that there is absolutely nothing available new today that will outperform the 9000 for 35mm and MF film at under 6000 USD, actually for under 13,000 USD new. You will find plenty on the used market in the 4000 - 9000 USD range used but care needs to be exercised in the purchase to be sure that the scanner is fully working and has the necessary software. Example, there is a Creo IQSmart2 for sale on eBay right now with a Buy It Now price of 1200. This is a great deal IF IT WORKS but the problem is that it is without software and hasn't been tested. Lot's of high end scanners get sold this way and the buyers end up paying a lot more than they thought to get them working, often still a good deal but the initial 1000 purchase often ends up as a 4000 - 6000 expenditure before the scanner i working.

I decided not to go the "pig-in a poke" route and wateched and questioned and tested and waited until I found a fully operational excellent scanner at a reasonable price. It took me a year and a half of looking and waiting and even then I got lucky. As Kirk observed I got my Cezanne for a bit under 6000 and it was a fully operational company demo that had been well cared for and serviced and came with all the necessaries to get going. You can spend less but you take a chance. Maybe it will be worth the chance if the price is low enough and you have the time to make the machine right.

BTW the Artixscan 2500 that Gene uses is one of the best machines you can find used for good prices. They are often available for under 500 and, while still not up to the quality of a high end scanner by a long shot are far better than the current new consumer offerings.

[Gene McCluney]

BTW the Artixscan 2500 that Gene uses is one of the best machines you can find used for good prices. They are often available for under 500 and, while still not up to the quality of a high end scanner by a long shot are far better than the current new consumer offerings.

I purchased this scanner brand new for over $2500. At the time I bought it, the dealer kept trying to push me to get the Artixscan 1800. I was absolutely firm in my belief that the 2500 was a better machine, and it has proven itself to me over and over for 4x5 scanning, as well as 8x10 scanning. The only issue I have ever had with it (over the years) was some dirt in the optical chain that was showing up as some streaks in the scans, but I solved this issue. This scanner actually has 2 optical "chains" One is a high-resolution chain capable of 2500ppi, but only for images 4x5 or smaller, the other chain is 1250ppi for images up to 8x10 or so. 1250ppi is plenty good enough for an 8x10 negative, for any practical print size.

[Ted Harris]

Gene, correct. The only thing the 1800f does better than the 2500f is scans larger than 4x5 ands that is only if you have all the paameters set absolutely correctly.

BTW, for those that are searching for an excellent scanner at a bargain price everything Gene says about the 2500f is basically true of the earlier 2500 which differs from the "f" model largely in terms of connectivity ... it uses a SCSI connection only. To a lesser extent also true of the Agfa Duoscan 2500 which is simply a 2500 rebadged by Agfa with firmware that couples it to Agfa's softwware. In the instance of the Agfa machine you are well advised to buy Silverfast Ai.