What size scans can I get from roll film (6x4.5)?

[jason_keen1]

<p>Hey all,<br>

I'm looking to do some larger prints 30x40, but I don't want an overwhelming amount of grain or having to interpolate the image by enlarging it. Approximately how much resolution does medium format roll film have? Specifically, in 6x4.5 or 6x7. I have a 6x4.5 Fujica rangefinder but am willing to upgrade to 6x7 or 6x9 if it's going to produce dramatically better prints. Thanks for considering my question. Jason</p>

[craig_supplee]

<p>A lot of variables here, starting with what type of film you use. Naturally the slower the film speed, the less grain, and larger prints are conceivable from it. Next is making sure you are using a very solid tripod and cable release. Then you get to the scanner, which also plays a big role. I have a Microtek M1, which can scan at 4800dpi. On my 6 x 7 slides off my RB67, I can get a starting file size of several hundred (400+ is not uncommon) MB. By the time I'm ready to resize the image, I don't have to worry about adding pixels. A 30 x 40 print looks good to my eyes off this size film.</p>

<p>You might be stretching it a bit to go that big on 6 x 4.5, although I have never tried it.</p>

<p>Another consideration is what file size you can actually send to a lab for an enlargement.</p>

[jason_keen1]

<p>Sorrry, I should have specified, i'm using primarily Provia 100.</p>

[bueh]

<p>A lot depends on the skill and equipment of the lab technician who prints your film. For all practical purposes you can enlarge medium format film as big as you want and the different formats will not produce dramatically "better" prints. While 6x7 is a small but noticeable step up from 645, unless you are dissatisfied with your results there is no need to get a bigger camera.</p>

<p>Large format, on the other hand.... But anything since glass plates went out of style is a compromise in regards to quality, so why even bother about the differences between the various medium format formats. You worry too much, better work on your technique and experience (which can always be dramatically improved) and fret not about the minor issue whether a couple of square millimeters here and there will make your pictures more perfect.</p>

[david_henderson]

<p>You should be able to produce that size print very well, provided that</p>

<ul>

<li>The original transparencies are extremely sharp and well exposed, with contrast well controlled</li>

<li>That your scans are made using appropriate technology operated by skilled resources.For me I'd want drum scans to support prints that size.</li>

<li>That the process of making the file from the scan is carried out well</li>

<li>That the printing technology selected is operated by people who really know what they are doing. </li>

</ul>

<p>I've made 40" sq prints on a Chromira from 6x6 originals that look sharp to the extent that they do not rely on extended viewing distances to look very good.<br>

Looking briefly at the numbers, you're going to need to print at about 200dpi and a 30 x 40 print is going to need a file of about 150 MB (8 bit), so thats comfortably within range of a 4000ppi scan that you can buy relatively easily. Everything from a nikon Coolscan 9000 up will deliver a 4000 ppi scan tthough I'd personally vote for a 16 bit scan and edit, converting to 8 bit prior to print. If you're not used to this sort of process ( and the fact of your question tends to make me suspect you might not be) then I would not be too tempted to buy a scanner and handle the file creation myself- I'd want these done by someone not fighting up a learning curve. I'd also vote for scans on an Imacon at least and preferably a drum scan to drag every bit of detail out of the transparency especially in the shadows. Drum scans have dropped in price a lot and are subject to frequent special offers even from the top labs.</p>

<p>Clearly a 67 requires less enlargement, and at the margin would produce technically better prints. I don't know that you'd see a huge difference vs 645 from the sort of distance you'd need to be at to take a 30" x 40" print in, visually speaking. And I do know that the size you're after is workable from 645.</p>

[lex_harris]

<p>Jason, if my maths is correct then a 40"x30" print at 300dpi requires a scan size of 12000 x 9000 pixels. Image size of 6 x 4.5 is about 56 x 41 mm, so you would need to scan at about 5500 dpi to achieve that scan size. From my experience with scanning 35mm Provia 100F at 6300dpi on an Imacon 646 the grain would definitely be visible at that magnification but not necessarily overwhelming although that could be a matter of taste. You could try scanning a small crop at 5500dpi and print at full size to get an idea of what the grain will be like. Other things to consider are that a scan file of this size will be very big, and it may be beyond some MF scanners. Imacon 646, 848 and 949 scanners for example can't do any more than 4000dpi with 6x4.5 and 3200dpi with 6x6 and 6x7. </p>

[steve_mareno]

<p>Lex is right. I just checked a 4000ppi scan of one of my 6x4.5 negs and it's around 30" on the long end if you're using 300dpi at the printer. If your images are B&W you may be able to get by w/ 250dpi at the printer or even less, but it depends on your printer and your image. I am not sure I would want to rez up the image at these print sizes, but you could give it a try and see. If it were me I would just downscale my prints a wee bit. A 30" print is still a big print.</p>

[charlesp]

<p>Please excuse the slightly off-topic question but is it correct to assume that, given the file size required for such large prints, a digital camera like a Canon 5d would not be up to the task of a 30" long size print? I currently shoot 6x4.5 and 6x7, but I'm wondering...</p>

[lex_harris]

<p> "a digital camera like a Canon 5d would not be up to the task of a 30" long size print?"</p>

<p>I don't have a 5D and I've never tried it so I can't say what a 30" print would look like, but the 12.8MP image from a 5D would be around 4300x2970 pixels in size I guess so if you want to do a 30" wide print at 300dpi or even 200dpi you would have to increase the image size quite a bit through pixel interpolation. I imagine there is going to be some loss of quality as a result. How much loss of quality, especially subjectively, is open to debate - I'm sure there are some far more knowledgeable forum members who could answer this question much better than me :-)</p>

[david_henderson]

<p>You do not need to print at 300dpi for a print of 30" x 40". 300 dpi is not a mantra that you have to apply equally to every print situation. Therefore some of the advice you are getting about file size requirements is making it look more difficult than it is. If you don't accept my word for it, despite the fact I've actually done it , then check out the website of West Coast Imaging, who are about as good at this sort of thing as anyone walking- <a href="http://www.westcoastimaging.com">www.westcoastimaging.com</a>. A 4000ppi scan providing a file size of 170mb is more than enough for the job.<br>

The point of being emphatic here is this. A 4000 ppi 8 bit scan gives you a wide range of commercial sources and prices to get these scans made. For me, as I've said, I'd get a drum scan at 16 bit because for me thats the ultimate in quality. Your view of the costs involved in that might be different, but a 4000ppi scan is relatively easy to buy from a number of sources, scanning routes and prices. Contrary to the advice above, the Imacons and even a Coolscan 9000 are perfectly able to give you all the file size you need for this venture, albeit that I would personally prefer to get drum scans at higher prices. WCI list right now for an 8 bit drum scan up to 200MB (or 16 bit to 400MB)is just under $50. Here in the UK I get cleaned Imacon scans that size for the sterling equivalent of $14. </p>

<p>The 5D question is interesting and here , whilst I use a 5D, I have to join others in speculation because I haven't made large prints from those files. Looking at these files, I would expect that a print that size from a 5D would depend a lot on viewing distance and subject matter for its quality as well as the skill with which the project was executed. I would not expect that photographs with large amounts of detail would work well at all- for example fall foliage pictures with lots of tiny leaves would not respond to interpolation well. OTOH a simple graphic image might. If I really needed to make prints that size from a digital original then I'd be looking at a digital back for a MF camera as a function of sensor size as well as pixel count. The fact that these are always very expensive and often downright unaffordable is what results in me maintaining a scanned film approach for larger prints, and if I were Mr Patrick I'd hang on to my MF film cameras.</p>

[lex_harris]

<p>"Contrary to the advice above, the Imacons and even a Coolscan 9000 are perfectly able to give you all the file size you need for this venture"</p>

<p>As far as I am aware no Imacon scanner will scan above 4000dpi for 6x4.5, or 3200dpi for 6x6 and 6x7. That includes 646, 848, 949, X1 and X5. So with a 645 or 67 image and 200dpi print you just will make it, but for higher print resolutions you will not. I don't have a Nikon 9000 or drum scanner so can't comment on those.</p>

<p>No-one is advising that you can't do the job with an Imacon or Coolscan 9000. To print a 645 image at 200dpi a 4000dpi scan will do: 56mm x 41.5mm at 4000dpi = 8818 x 6535 pixels, 40" x 30" at 200dpi = 8000 x 6000 pixels, no problem. A 67 image at 3200dpi will produce a similar scan size, again no problem. But if you want to print at higher resolution (and I'm not suggesting this is necessary) then 4000/3200dpi is not going to give you the required scan size.</p>

<p>Regarding file size, I believe the general wisdom is to scan at 16 bit and then convert to 8 bit after any image corrections have been completed. 16 bit TIFFs of the above pixel size will be about 350MB and on my PC this is manageable in Photoshop albeit some operations are a little slow. As David Henderson has pointed out filesize reduces to 170MB or so after converting to 8bit TIFF.</p>

<p>If you were to aim at 300dpi printing (and I repeat I'm not suggesting this is necessary) then you are working with a 12000 x 9000 pixel file, which in 16bit TIFF will be about 600-700MB in size. Working with files that big on an ordinary PC like mine does require a little patience.<br>

 

<p>"if I were Mr Patrick I'd hang on to my MF film cameras"<br>

Indeed ;-)</p>

</p>

[Ed_Ingold]

<p>Speaking from experience, using a Nikon LS-8000 and an Hasselblad, you make an excellent 20x30 inch enlargement, and 30x40 would require only modest resampling* if you want 300 ppi. (You can get by with less, but it's safer to resample for a print that might hang on a wall rather than a poster or billboard.) I would venture even larger with B&W film, because a little grittiness can be effective in that medium, but not so cool with color.</p>

<p>The usual caveats apply -- use fine grained film, compose carefully to minimize cropping, use a sturdy tripod with the mirror up and a cable release. Mirror shake alone can add an uncertainty of about 4 pixels (at 4000 ppi) at 1/15 second. At 30x40 inches you are entering large format country, so you have to cross all the T's and dot all the I's to make it work.</p>

<p>You can resample a good shot by a factor of 2x. With care, the results look soft without any visible signs or artifacts of resampling. You don't get even close to that figure with what you seek to accomplish.</p>

[Ed_Ingold]

<p>I see no advantage printing from a 16 bit TIFF file. Edit in 16 bit mode to produce a non-resampled master image, then convert to 8 bit when you crop, size, resample and copy to a JPEG print file.</p>

[stuart_richardson]

<p>At least for Epsons, large prints are best made at 240dpi. They have two dpi settings -- 360dpi and 240dpi. It is very difficult to tell the difference with the naked eye, and at the viewing distance for a 40 inch print, there is no way that there would be a visible difference. I have 30x30 inch prints from 3200dpi scans of 6x6 negs on an Imacon 646, printed at 240dpi on an Epson 9800. Nose to the print, and they are still sharp. I have no doubt that using good technique and equipment, you should be able to get good results from 645 at 40". The deciding factor is as mentioned above -- the quality of your slide (and taking lens), a very good scanner, either dedicated film scanner or drum scanner, and someone who knows how to properly prep and sharpen a scan for large digital printing. File preparation is not a trivial matter when you are making enlargements this big. </p>

[patrick j dempsey]

<p>Using slide film, it is very likely you will run into resolution issues with the scanner before you hit the limits of the film. I have heard that many scanners interpolate data to reach their "highest" resolution which may cause somewhat soft scans. There are many factors that go into scanning and you need to really be scanning with your final print size in mind. High quality prints are typically made at 300dpi, although if you plan on being far away from the prints, they can be smaller resolution like 150dpi and still be fine... billboards and restaurant signs can be made even from 35mm slides because of extremely low dpi interpolated into large dots that optically blend together. If your scanner is capable of 2400dpi scans... then starting with a slide 2.25" x 1.8" you will end up with a roughly 5400 x 4300 pixel (23MP) file. At 300 print dpi that translates directly down to 18" x 14.3". A 6x9 slide would give you double that at 18" x 28.6" for the highest resolution print, with larger prints totally reasonable at lower print resolutions. This might be confusing because you see professional print shops making prints those sizes from 8 and 10MP digital cameras, using roughly half or even one third of the data available in that scan. They can do this because they rely heavily on interpolation software to "uprez" a file to the print size they want. If you start from a drum-scanned file at 8000dpi then you would have 3 times the data to start with, although at that resolution you will clearly be hitting the limitations of both the film structure as well as the line resolution of your lens.</p>

[patrick j dempsey]

<p>Oh and BTW, using traditional wet printing techniques, much larger prints should be capable, but good luck finding a lab that will do it. Remember... slide film was originally intended to be projected, and in the days of optical projection, 35mm slides (1/3 the area of your 645) were regularly used to cover WALLS. Go figure. </p>

[robert lee]

<blockquote>

<p>Using slide film, it is very likely you will run into resolution issues with the scanner before you hit the limits of the film.</p>

</blockquote>

<p>Actually, reversal film resolution is fairly low. Velvia is amongst the sharpest available, and it's 50% MTF is just about 50 cycles/mm. Practically, this means a 4000dpi scan from a Nikon or better picks up just about all the spatial information there is on the media.</p>

[dave_redmann]

<p>Wow, a lot of people are missing the very basic fact that, in most cases, the film cannot record enough detail to give you a 30x40-inch print at 300 ppi. There's just no point in scanning at the super-high resolutions required, even if you have an Imacon or drum scanner that can do it. Statements like, "Using slide film, it is very likely you will run into resolution issues with the scanner before you hit the limits of the film," are just flat-out wrong.<br>

Somebody mentioned Provia 100F, which is a good place to start (this analysis does not apply to all films, with T-Max 100 being an example of a notable exception). The Fuji data sheet's MTF curve (I'm looking at AF3-036E<strong></strong>) shows the limit of resolution, judged by the common criterion of 50% MTF response, is about 40 lp/mm. That's just over 2000 ppi. The curve stops entirely at about 60 lp/mm (roughly 3000 ppi), at which point the MTF response is down to about 25%. And of course to achieve this response, you need perfect lenses, film flatness, technique, etc. So assuming a good drum scan or the equivalent, scanning at 4000 ppi, to say nothing of 5500 ppi, is likely to yield little to no additional image detail, compared to 3000 ppi, but will resolve more grain.<br>

I realize that you probably want to scan at a somewhat higher nominal resolution than the film can deliver, to account for the fact that the scanner's effective resolution may be somewhat less that its nominal resolution, but I'd submit that any scans over 2500 to 3000 ppi are going to start picking up more film grain than real image detail.<br>

But don't fret too much. If you scan 645 film at 3000 ppi, you'll get about 4900 x 6600 pixels, which is enough for about 163 ppi, which I think will look fine at 30x40 inches, even at relatively close viewing distances. But you will probably see some grain, although not bad. Although 300 ppi may be ideal, for larger prints, in my experience 175 ppi can look quite good, and I've got some big prints at 120 ppi that look decent from a distance.</p>

 

[mauro_franic]

<p>Dave,<br>

No mater what white papers say, I can substain from direct experience the films like Velvia and TMX, to name a few, capture more detail than a Coolscan 9000 can retrieve (which is close to its nominal 4000dpi). This is 90+ TRUE megapixels from 6x7. I mentioned TRUE megapixels because DSLRs for example only capture around 2/3rds of the rated megapixel count.</p>

<p>You can make outstanding 30" prints from 6x7 film + Coolscan combination.</p>

<p>No DSLR will compare in detail to a good scan of 6x7 Velvia and TMX on a 30" print.</p>

<p> </p>

[evphotography]

<p>I have made prints 20x30 from 6x4.5 film with 300MB scans that look beautiful, can not see any grain the image at print size. Probably anything larger would be pushing it, 6x7 you could probably go bigger.</p>

[kelly_flanigan1]

<p>Customers want a high dpi/ppi scans because its really an emotional issue; like high GPA; one was a good boy on Santas List.<br>

<br /> How much info is in a piece of film will be asked until the end of time.<br>

<br /> Its like worrying how many ears of corn there are in Iowa; or shoes per woman.<br>

<br /> This question is 2 decades old; actually older..<br>

<br /> One thing I have noted in printing the publics images is that the folks who dwell on this issue of megapixels per film area input the absolute WORST inputs for giant images; while carefree dont care folks often have the best inputs. Its like folks brains are so focused on granite block test data they have no brain power left about make an image with IMPACT.<br>

<br /> Its hard for many folks to run a small test to see what works; many times the upsizing of files just makes a bloaded file thats little different than the orginal; but its just more bloaded; ie useless info. As Beavis says one cannot polish a turd.<br>

<br /> Roll films that are 6x4.5cm can be just a Brownie Bullets Verichrome; where our Epson 1200U flatbed is totaly overkill; or the input might be a very sharp; or just so-so 6x4.5 Fujica rangefinder's negative; where a drum scan or Nikon Scan might be the hot trick for the sharp one; or the Epson 2450/3200 overkill for the so so one.<br>

<br /> A few folks might get 50 line pairs per mm on film at times; the peanut gallerys images of inputs are often close on some; and Nikon 9000 is overkill.<br>

<br /> The world mostly does not shoot at F8; or with tripod/granite block; or even get the focus exact.<br /> The world images are not best case examples.<br>

<br /> The concept of design margin seems to have been sucked out of folks brains; ie having a safety factor; ie NOT cooking the books.<br>

<br /> If one uses a larger negative and a lower enlargement; one gets better tonality.</p>

[david_henderson]

<p>Eric. Assuming your scan numbers refer to 8 bit, do you think its likely that your printer will have reduced the scan to its native resolution (360dpi/300dpi/240 dpi?) rather than use all the information you fed it? I ask because on the face of it you could have printed at not much short of 600dpi with such a large file. If so you could arguably have made significantly bigger prints from the same scans- or indeed got a much smaller scan to start with, rather than concluding that you are "on the edge". </p>

[Ed_Ingold]

<p>Franic said <em>"I mentioned TRUE megapixels because DSLRs for example only capture around 2/3rds of the rated megapixel count."</em></p>

<p>What is the basis for this ridiculous statement?</p>

[Ed_Ingold]

<p>I have made many enlargements from scanned, 6x6 Fuji Reala. The grain is barely visible at 16x20 inches and not objectionable at 30x40 inches. While Velvia has less visible grain (it is fuzzed up as dye clouds, except for the ubiquitous "pepper grain" effect), neither does it have exceptional resolution. Its touted resolution using using special targets is due solely to its inherent high contrast, and is comparable to Reala (80 lp/mm) with subjects having normal contrast.</p>

<p>At 4000 ppi, the resolution of a Nikon LS-8000/9000 is comparable to that of most film (excluding Technical Pan, for example). The composite resolution is less than either (q.v. root-sum-square estimation of uncertainties), closer to about 60 lp/mm. Obviously the finest detail will be degraded in a huge enlargement, however resolution is seldom the most important aspect of a well-executed photograph or print. Subject, composition and tonal range rank higher in my estimation (unless you really goof up).</p>

<p>If you don't see grain at these sizes (8x to 15x), you aren't scanning (or enlarging) very carefully. If you object to moderate grain, enlarge less or use large format film (or digital capture).</p>

[nealcurrie]

<p>What Mauro is referring to is how much data can be captured with a dSLR. Obviously the sensors are capable of capturing data, but due to lens quality (and possibly diffraction), a 10MP camera cannot capture 10MP of discernible data. Just like techpan is rated at 320lp/mm, but you'll never get this if you're using a lens to make the picture. Our lenses aren't up to the task of getting all the use out of the digital sensors.</p>