VERY large enlargements

[vatovec]

In the next year or so i plan to have an exhibition of 120x140cm

(47x55 inch) size prints.

The source is a 6x7 cm BW negative that must be scanned at 6000dpi

(if i`m correct) to have an output file of 47x55 inch @ 300 dpi am i

correct?

That means that a color file that size would be approx. 700MB - for

using Photoshop to color some parts of the prints i need at least

twice as much RAM as the size of the file i`m working with - am i

correct (that means a pc with 1,4 GB RAM!!!).

One file barely goes on one CD - cca. 750 MB!

Are there any other journeys to the desired final prints or i have to

buy a 10.000 USD graphic station and put all my earnings in digital

scanning?

[jwink3101]

Hello<p>I don't know how much help I could give but it appears to me (using Photoshop) the file will be 670mb. You are about correct for the dpi (6000).

<p>

I want to make one point. If you are printing at this size you <b>do not</b> need 300dpi prints. The viewing distance is 184 cm (diagonal). At this much lower DPIs are sufficient. I don't have the exact number but I�m <I>sure</I> that 200dpi would be great. <P>I would love to hear where you buy your CD's because the most I have ever found was 700mb which is still better than the old 650mb. I hope this helps

[jwink3101]

I forgot to add that you might want to patrician a drive and use it solely for a PS scratch disk.

[sakari_m_kel_]

I usually scan 6x7 cm slides with 4000 dpi resulting some 550 MB files (16 bits). Some 5x12 cm panoramic slides give 800-950 MB files. It is true that Photoshop works better with enoug RAM, but it is possible to work with files that large using quite usual PC - maybe 512 MB RAM is convenient.... The price here is that it is very (and I mean VERY) slow - everythig takes time. But, if time is inexpensive, you can do it. Buy a extra hard drive, keep it empty (that means there is absolutely nothing in that drive!) and use it as Photoshops work drive. Do not use the virtual memory of Windows.

 

You have to have something to do while waiting - read a book, have another computer to play with, do some handicraft etc. There will be some 10-15 minutes between every mouseclick, if your computer is slow - but it will do the job.

 

AND - you probably can live with 4000 dpi scans, if those are less expensive. You can upscale that much in Photoshop or use fractal software. Nobody looks at pohotographs of that size at 20 cm distance - and if looks, so what?

 

Sakari

[steven_.]

I know I'm missing something! How did you calculate 6000dpi to create the image size you needed?

[vatovec]

Simple: output should be at 300 dpi so 47inch*300dpi=14100 pixels

So i need to scan 14100 pixels across 6cm(2,36inch) negative so: 14100/2,36=5974dpi cca 6000dpi!

[richard_hasty]

1.5 gigabytes of RAM for a recent PC (DDR 2400 RAM) should be less than $300 USD. This is about the same price as labs I've looked at charge for digital output at 44x55 inches.

[peter nelson]

I would also like to second (third?) the comment that you don't need 300 DPI for prints that big because the viewing distance would be much farther away than a conventional print.<P>

 

There's another reason you don't need such high resolution - the image won't contain any information at that resolution. The best medium format lenses (I have a Mamiya RB67) can resolve about 70 lp/mm on extremely fine grain film, tripod-mounted, optimum f/stop, using a high-contrast test target. That works out to 3556 samples/mm. Real-world photos seldom do even that well. (BTW, the best 35mm lenses can do a bit better - 85 lp/mm). So if you scan at 4000 DPI you're already getting all the detail your negatives are likely to contain. Anything beyond that will needlessly take up disk space.

[ellis_vener_photography]

Durst Lambda printers only need files that are 200ppi resolution, there is a built in interpolation program that matches the size of the input image to the size of the print being made.<P>

Better than a partition on the same hard drive is to use a second hard drive to use as your scratch disk. This will speed things up considerably.<P>Rather than buy all of this yourself I suggest you find a high quality digital imaging lab to consult with and to make your prints.<P>In the USA I recommend <P><A HREF = http://www.nancyscans.com> Nancyscans</A> <P>or<P> <A HREF = http://www.westcoastimaging.com> West Coast Imaging</A>

[leonard_evens]

Peter,

 

In some circumstances it does make sense to scan at a higher resolution than what the lens can provide. There are a couple of ad hoc rules for calculating the resolution in lp/mm of a system from the resolutions of the components, but they both result in an answer less than any of the components. If all the components of the system have similar resolutions, then the net result can be significantly less than any of the components. But if one component has a much higher resolution, then its contribution to lowering the resoltuion of the system may be minimal. So it makes as much sense to scan at as high a resolution as is practical as it does to insist on as high a quality enlarging lens as possible. If the lens/film resolution ends up being the limiting factor, that is a desirable situation.

 

Having said that, let me add that I doubt if there is a lot to be gained by going to 6000 ppi vs 4000 ppi for large prints from 6 x 7 cm sources, at least if the prints are viewed from normal viewing distances. One can even do reasonably well with somewhat smaller large prints starting with a scanner like the Epson 2450, although I think the results of using a dedicated 4000 ppi film scanner would be clearly sharper.

 

I recently visited the Chicago Art Institute and looked at their Photography exhibits. I was surprised that few of the prints on display were really sharp when viewed very close up. But of course those prints were chosen partly for historical reasons and partly for aesthetic reasons, and technical quality was probably way down on the list of criteria used by those who selected the prints.

[vatovec]

Peter are u sure?

I think that the lpmm ideal numbers should be higher (at least twice as much as u`re stating) since i once read that some lenses outperform classic bw film (tmax being near 120 lpmm i think) - and yes i know 35mm lenses resolve more detail than mf.

By the way i too have the rb67 which lens do you own (i`m stuck with the 90 3.8C and 180 4.5C.

[peza]

Depends on your lens, 6000 DPI is asking for well above average lens, unless your lens is doing around 80-90lp/mm and negative is well developed TechPan or TMAX-100 or APX25 or in worst case PANF-50, it is next to useless unless you look for "nicely" defined grain.<br>

Far biggest influence will have properly done digital sharpening.

[peter nelson]

<I>Peter are u sure? I think that the lpmm ideal numbers should be higher (at least twice as much as u`re stating) since i once read that some lenses outperform classic bw film (tmax being near 120 lpmm i think)</I><P>

 

Oh, the lens <B>definitely</B> outperforms the film! I've seen tests of lenses by themselves (i.e., without film) that went to over 200 lp/mm! That's why I specified fine grain film in my comments. You can also get higher lp/mm tests using special ultrafine grain films used for archiving documents and other exotic (non photographic) uses. Also some people say line-pair per mm and mean that, others say lines/mm, and mean line-pairs, but others say lines per mm and mean LINES - i.e., to them a line-pair is two lines. So you have to read the fine print. Here's a website with lens tests that has a typical range of data for a variety of lenses:<BR>

http://www.hevanet.com/cperez/testing.html<P>

 

 

<I> - and yes i know 35mm lenses resolve more detail than mf. By the way i too have the rb67 which lens do you own (i`m stuck with the 90 3.8C and 180 4.5C.</I><P>

 

I have a 180 f/4.5 and a 65 f/4.5 - neither of which is anything to write home about.

[kelly_flanigan1]

For very large prints; 150 pixels/inch to 200 is usually plenty; <b>for photos</b> . <BR><BR>If the enlargement is a map; it WILL be looked at up close; it has dinky type; or alot of detail; or is a wall map or aerial photo used for planning. Many inkjet plotters today are much larger than decade old 36" wide models. I have seen maping outputs that are a combination of aerial photo; plus tax map parcel data that are 42 and 54 inches wide; at 600ppi. The lines & text used many times are real small; and dont lool sharp when printed at 300 ppi; versus 600 ppi. <BR><BR>The detail obtained from a 6x7 B&W scan is combination of the quality of the lens; whether it was focused correctly; the film; and the scanner. Each item can be the weakest link.If one of the 4 is perfect; and another is poor; the weakest link ruins the image. In B&W; with slow films; the film has a higher MTF than the lens. <BR><BR> This example is a 6x7 to 120x140 cm; or a<b> 20X enlargement</b> . Our business enlarged microfilm aperture cards by 40X three decades ago; with a special Schneider Componon-M lens optimized for large ratios. I enlarged 35mm negatives to 42x63 inches; using the full wide of the 42 inch paper. These are a bear to develop in trays!<bR><BR><b>Not all Pixels/inch numbers are the same; a drum scan by a pro lab will beat a home Epson scanner; even if the ppi are the same settings. </b><BR><BR>The limiting factor is your technique in making a sharp negative. Many scans above 2000 to 3000 ppi pull little extra detail;because of camera vibration; focus off very slightly; poor lighting; which yields poor contrast. <BR><BR>Not to start too much of a war here; but I feel very few members understand that lens resolution numbers are done with charts with a contrast ratio which is very very high; ie 1:1000 . The tests I have done with 1954 USAF charts with 1:10 contrast are RADICALLY different; and always the lens test data is WAY lower. Real world photos with 1:1000 contrast happens in mapping; but not the photos most all photographers here do. The normal photo scene contrast means that a 20X enlargement is not going to be as sharp as lens resolution numbers would indicate.<BR><BR>Using a large format of 4x5, 5x7, or 8x10 will make a better print. <BR>

[peter nelson]

Kelly Flanigan says: <I>but I feel very few members understand that lens resolution numbers are done with charts with a contrast ratio </I><P>

 

Anyone who read my response to this question did!<BR>

<B> 8-)</B><BR>

I wrote: <I> <B>. . . </B>using a high-contrast test target</I>

[kelly_flanigan1]

Lens test data has been in line pairs/mm since the targets were printed in 1954. In Optical groups; we Engineers use the term lines/mm many times; as a quicker way to say the term line pairs/mm. Since we have looked at vast numbers of negatives under microscopes; at the pairs of lines for max resolution; we know there are white and black lines to each part of the target. Not until the Internet came out did this line pairs/mm verus lines/mm get mentioned. It is because people today dont understand the target; and have never used one either. In reading the targets; one gets the minimum group and pair numbers; and records the horizontal and vertical data. Only when done is the data converted to line pairs/mm.

[kelly_flanigan1]

A typical image on a negative may have 50 line pairs/mm to be considered very very good. Most people get far less than this; because tripods; exposure; etc are not used. <BR><BR>In actual resolution testing; the resultant resolution varies alot with exposure. Overexposure ruins the resolution alot. A lens test maybe run with many different exposures; and only the best data used by the marketing dept. In line work; with high contrast lith films; the resolution can be only 30% of max; if the exposure is off 1/2 stop. We ran resolution tests on several enlarging lenses; and varied the exposure(time) 1/3 stop for each fstop tested. Here the high contrast high resolution film will give high resolution numbers; but the exposure in several cases was varied +/- 10% to give even better data. When using normal contrast films on the same lens; the resolution is less; and the exposure latitude is more. A 1946 127mm F4.5 Kodak Ektar on my 4x5 tested in at 80 line pairs/mm; at the center; at is best aperture of F22 ; at a 1:10 copy ratio.

[peza]

I played this lp/mm game with all my lenses and under microscope I could see WAY more than in 4000DPI scans optimaly sharpened and adjusted contrast. Winner's test is attached and it went well over 80 lp/mm which is top resulution visible in test I did -and scanner can't resolve this (Canon FS4000US). But this is 35mm negative (TMAX-100, ID-11) and something I call top notch lens. Only 135/2.0L and "big whites" are better.<br>

What concerns your lens - it would be big exception when it would go over 80lp/mm if you can believe tests on internet, one here: <a href="http://www.hevanet.com/cperez/MF_testing.html">http://www.hevanet.com/cperez/MF_testing.html</a>.

<br><br>

Quality scan is important - as you see here, altough dedicate film scanner - still can't show everything it theoretically could, but about 10% less. Wet drum scan is must in your case because of grain issues.<br><br>

I wish you had taken them with Hasselblad 250/5.6 $$$$SA, than there would be plenty of lp/mm's:)

 

Can you tell us which negative you'd like to enlarge that much ?

[vatovec]

Not a specific negatve, some are taken some are yet to be taken - if u look at my Primavera folder - that should be it - i have an exhibition starting on the 21st march - those are 18x24cm fotographs - next year i plan to make some more monumental exhibition of the same subject not sure yet about the title,...

6x7 bw negs probably taken on tmx100 or apx100 (if not TP)!

[david_bradley1]

David,

<P>

The discussion seems to have drifted away form your original questions. I also work on 600MB - 1.1GB files on a daily basis. Here are some observations:

<UL>

<LI>I agree with the other posters that 300 DPI is not necessary for the images and image sizes you are describing. I looked at the portfolio you mentioned, and the images are striking because of their concept but not really because of extremely fine detail. It's the compelligness of the image that will win the day, not the extreme fine technical detail. I have some technically weak but very compelling mountain climbing images that I took with a DC4800 digital camera, intending them to be only personal snapshots. But a few turned out so well that I've blown them up fairly large and printed them, at times, as low as 180DPI. I personally can tell that the technical quality is low, but these images generate many more involuntary cries of "Wow!" when people walk into the room than some of my ultra-high quality large format images.

<LI>You mentioned you would be scanning B&W. In that case, I would suggest getting 16-bit monochrome scans instead of 8-bit RGB scans. This will a) decrease the size of your scans by 1/3 (16-bits/pixel vs. 26-bits/pixel) and b) avoid any solarization effects you might get by starting with only 256 levels of grey.

<LI>The industry standard right now seems to be to use 300MB scans for medium and large format originals. As far as I can tell, this is due to hardawre and software limitations, not scanning or film limitations. Photoshop can't handle more than 2GB of RAM no matter how much your computer has available. (See several discussion about this at <A TARGET="_new" HREF="http://www.adobeforums.com">www.adobeforums.com</A>.) Therefore, speed really slows down when working with files larger than 300MB or so. (Adobe recommends that RAM size be 5x larger than image size.) I recommend that, to start with, you not exceed this size; see if it's you are satisfied with the results before you start to make life more difficult for yourself. This means getting a 600MB 16-bit grayscale scan, doing some initial levels, then comverting it to a 300MB 8-bit file and doing the rest of your work on it.

<LI>In my experience dealing with very large Photoshop files, you're going to fill up your RAM pretty quickly, and after that your speed will be limited by the speed of your scratch disk. So it's worthwhile to speed this up as much as you can. I know that recommendations for specific hardware have a short lifespan, but I will make some specific remendations in this case. For a PS scratch disk you want hardware that will support a very high bandwidth, but you don't care very much about seek time. There is a disk out right now, the Maxtor 40GB DiamondMax Plus 8 that has the highest bandwidth of any IDE drive currently available (over 60GB/second) and is also one of the cheapest (about $70, but some retailers provide coupons that lower the price to $40). It is also small, requires relatively little power, and generates relatively little heat. It is cheap because it has a small capacity and has a slow seek time. But it is, in my opinion, the best Photoshop scratch disk current available (unless you want to go SCSI). If your computer has room for more than two disks, I would suggest you consider adding TWO of these disk in a small RAID configuration. I use two of them with a $100 Highpoint 404 RAID controller and I get over 115MB/sec sustained throughput to my PS scratch disk, and the total cost was only abut $250. This is about as fast as one can get from a 32-bit/32-MHz PCI slot. However, to reiterate what others have said, it's best to max out your machine's RAM capability first. For the file sizes you're talking about, I wouldn't recommend anything less than 1.5GB. Fortunately RAM is cheap now.

</UL>

My two cents.