Luminous Landscape Forum
Raw & Post Processing, Printing => Digital Image Processing => Topic started by: PeterAit on January 03, 2015, 11:02:07 am
-
I am scanning some 35mm Kodacolor negs using an Epson V750. Is 1200 DPI a good scanner setting to use? I am trying to balance detail extraction with scanning speed. ANy other tips would be welcome - I am totally new to this.
-
The resolution setting will not affect detail extraction unless you make it high enough that pixels are being invented (i.e. exceeds the CCD resolution of the scanner), or too low relative to the size of print you want to make. This question cannot be answered specifically without knowing what size you intend to print and what size is your media.
-
The resolution setting will not affect detail extraction unless you make it high enough that pixels are being invented (i.e. exceeds the CCD resolution of the scanner), or too low relative to the size of print you want to make. This question cannot be answered specifically without knowing what size you intend to print and what size is your media.
I don't agree with what you say. If I scan at too low a resolution, won't there be detail in the negative that is not captured in the scan? I don't want to "invent pixels," of course, but I also don't want to leave real pixels behind!
The point is, if I scan at the scanner's highest optical resolution (not interpolated resolution), I will get all the detail but the scans will be slow and the files huge.
-
If you chose to answers the questions I asked you I can answer your question about the appropriate resolution. I've been doing this stuff since 1999.
-
I don't agree with what you say. If I scan at too low a resolution, won't there be detail in the negative that is not captured in the scan? I don't want to "invent pixels," of course, but I also don't want to leave real pixels behind!
The point is, if I scan at the scanner's highest optical resolution (not interpolated resolution), I will get all the detail but the scans will be slow and the files huge.
Peter - the input resolution should relate to the output resolution and the size of the output. At the output end, the most often cited norm for human visual resolving power is about 300 PPI. So you need the number of input pixels to achieve roughly this output criterion. If you are printing from an Epson professional printer, the usual resolution of the print head is 360PPI (Canon is 300), so if you aim for 360PPI at output, once you know the output dimensions you can derive the input PPI. You will NOT leave detail on the table scanning at an input resolution that is appropriate to the maximum output dimensions you intend to print. I have tested this every which way from Sunday and I can confirm that you do not need to scan at the maximum optical resolution of the scanner, take all that time and end up with SUCH huge files. It's not necessary. (Some people claim that when you then go to downsize it helps clean-up artifacts - I prefer more directly controllable ways of doing this.) You will not leave useful information on the table scanning at the appropriate input resolution. If you are scanning media measuring 1.4 inches by 0.95 inches (35mm), and if you want to print the long side at say 17 inches, the total number of linear output pixels you need for that is 360*17 = 6120 pixels. Your original media long side is 1.4 inches. Therefore the total input pixels per inch you need at the scan stage is 6120/1.4 = 4371. Check: 4371 pixels/inch * 1.4 inches = 6120 total pixels, which divided by 360 PPI output gives you the 17 inches you want to print at that resolution. The formula for right-sizing the input pixel requirements per inch is [print length*output resolution per inch]/media length. If you are printing smaller the input resolution requirement goes down accordingly.You can of course experiment with whatever settings you want to try that deviate from this approach and you'll see what I mean.
-
If you chose to answers the questions I asked you I can answer your question about the appropriate resolution. I've been doing this stuff since 1999.
Who says I want to print? Anyway, that is totally irrelevant to my question--getting the maximum detail from these negs. Look, if you don't know the answer to my question, fine. Then don't answer.
-
Being "totally new at this" SHOULD make you want to work WITH an expert, not act like a jack ass.
How about pulling your panties back out of your crack and be a bit nicer to one of the smart ones on here. All he is trying to do is help and you act like you already know more than he does even tho you are "totally new at this".
Information like what you intend to do with the file is quite helpful and when the vast majority of people who scan negatives or slides makes prints from them, the usual assumption is just that.
Now, how about you start over and provide more info and be nicer. Who knows, you might even learn something ... seeing as how you are totally new at this.
-
Jeez, talk about "no good deed goes unpunished"....
Here's a related V750 question for Mark, and I promise not to attack him for trying to help....
Say I want to use my V750 to scan chromes for a target Epson print size of 17x25 (I don't, but let's just say). My target resolution for a 360 dpi print is 6120 ppi. The Epson's film scanning resolution is spec'd at 6400, but the effective measured resolution is closer to 2300 ppi (http://www.filmscanner.info/en/EpsonPerfectionV750Pro.html). So should I set my scanning resolution to 6400 or to 2400? Is the effective measured resolution I obtain with a 2400 ppi setting going to be actually much less?
-
Who says I want to print? Anyway, that is totally irrelevant to my question--getting the maximum detail from these negs. Look, if you don't know the answer to my question, fine. Then don't answer.
There is no answer to your question based on the information you provided when you asked it. The safest answer I can give you if you don't know or don't want to say what you will do with these scans is to scan at the maximum optical resolution of the scanner and suffer the time and file sizes involves.
-
Jeez, talk about "no good deed goes unpunished"....
Here's a related V750 question for Mark, and I promise not to attack him for trying to help....
Say I want to use my V750 to scan chromes for a target Epson print size of 17x25 (I don't, but let's just say). My target resolution for a 360 dpi print is 6120 ppi. The Epson's film scanning resolution is spec'd at 6400, but the effective measured resolution is closer to 2300 ppi (http://www.filmscanner.info/en/EpsonPerfectionV750Pro.html). So should I set my scanning resolution to 6400 or to 2400? Is the effective measured resolution I obtain with a 2400 ppi setting going to be actually much less?
Hi Alan,
OK - now it gets interesting. I'll be addressing all this in some detail in my forthcoming review of the Epson V850 Pro, and don't worry, I'm not concerned about the attacks. Anyone who doesn't believe me is always free to do all the testing I've done and they'll see for themselves - each to his/her own. So let's go:
The number Patrick Wagner (Scandig) reports is a visually assessed resolving power of the whole scanning system determined with the aid of the line-pairs in the SilverFast USAF 1951 target. (The resolution number the scanner manufacturers give is an ISO compliant value based on the CCD structure - very different things.) I use the same target. It's a good target with one big limitation: it tells you what's happening at the center of the image, but not at the corners. Corner resolution could be worse. Anyhow, I haven't seen any different resolution target for measuring scanners, so that's what a number of us who do this work are using. The other thing about it, which is not the target's fault, because I've examined this target under a 30X microscope and I can confirm it is printed very well even at the finest detail (Group 7 Element 6), is the user's judgment when you try to look for white/black separations between the bars in the elements. You reach a point where they become fuzzy but are still somewhat distinct, and that status can persist over a couple of such blocs, each relating to a different resolution result. So what to do? I've landed on the idea that it's best to use a credible range rather than a point estimate that could be argued either this way or that way. But it's not a large range, so grosso-modo, without being too anal about it, one gets a useful answer.(And by the way, Patrick's eyesight and mine aren't quite the same - or his scanner is different - but I get better from the Epson V750 than he reports - not hugely, but better.
OK, how to interpret this information: what it means is that no matter how many PPI you scan-in, the perceived clarity of the image will not exceed the equivalent of what that target reading indicates. But you still need enough output pixels to optimize what comes out of your printer, otherwise you will be introducing yet another limitation on perceived end-point sharpness. So the arithmetic I provided in a post above remains valid up to the ISO definition of the scanner's input resolution (6400 in the case of the V750), and you would need all that resolution if you intend to make a 17.8 inch print at 360 PPI without resampling. Just know, however that in doing so, the "equivalent sharpness" if I can put it that way, is what the USAF target says it is. Now, the saving grace is that if you look at a 17.8 inch print from the viewing distance it would normally need (i.e. you are not pixel peeping - but enjoying the print), the apparent definition may well look OK, depending on the quality of the media to begin with. From my experience that is usually the most critical constraint on apparent sharpness - the quality of what we start with.
So bottom line: don't worry about the 2300 or whatever - that's the scanner and you have no control over it. For deciding the appropriate input PPI, use the arithmetic relating linear dimensions with PPI at input and output.
-
But you still need enough output pixels to optimize what comes out of your printer, otherwise you will be introducing yet another limitation on perceived end-point sharpness.
Oh, right! Thanks.
-
Peter's asking a perfectly reasonable question. It has nothing to do with what he wants to do with it after it's scanned. You want to scan ONCE and be done with it. Then you remove spots and adjust in post, ONCE. Then you can decide whether to print 8x10 16x20 or just reduce for a small image on the internet. Why would you want to scan and spot and edit separately for each final use? Doesn't make sense; waste of time. So you want to get the max out of it on the first scan. PPI DPI etc is superfluous for the scan.
So Peter, I use the Epson V600 and scan at 2400, 48 bit. The V700 might be slightly different. Try different resolutions and see if you can tell if there's any difference. I don't use ICE so the scan takes about a minute per 35mm picture. When I scan 6x7mm 120 medium format, the files are approx. 200mb files; 37mb files for 35mm color slides. Good luck.
-
Oh another suggestion. Scan flat. Do all your adjustments in post. Again, that will eliminate wasting time on extra scans. Of course you probably have Vuescan or the other non Epson program (I use Epsonscan). Those others do all the adjustments pre-scan. Reading all the problem others have had, I wonder the advantage, if any. Does the machine come with Epsonscan also? I'd try it with both Epson and the other program and see what you like. Also, using Epsonscan, you won't have to learn another processing program; just use PS or whatever afterwards. Note that you'll get 90% of the correction just adjusting Levels (black and white points) after the scan. Then use the other sliders to tweak. It'll take some time getting use too. So don't get frustrated.
-
............ You want to scan ONCE and be done with it. Then you remove spots and adjust in post, ONCE.
You mean YOU want to scan once and be done with it.
But no-one is talking about scanning more than once. Most commonly, one knows in advance what one wants to do with the scan, does it once and doesn't look back. That can save a lot of time and disk space. If one is a professional re-purposing the same images for all kinds of different unpredictable output situations - I agree - it makes most sense to scan at the optical resolution of the scanner and have lots of storage available. Otherwise it is not necessarily optimal. As for dust and scratch removal - doing this in post is slow and inefficient. One is best off using SilverFast's iSRD function for this at the scan stage. Saves scads of time, has refined controls (unlike ICE) and works very well, especially with a scanner such as the Epson V750 which has an infrared channel for detection.iSRD also works on Kodachrome, and can even work on SOME B&W media, depending on the extent of residual silver halide.
-
Mark: The OP indicated he doesn't know yet what he wants to do with it. So for him, scanning it at max optical resolution the scanner can do is best. I'm glad you agree with that. As far as disk space, he'll get 35mb tiff for each slide. He could save if he saves in jpegs but it will only take 37gb to store 1000 35mm slides in tiff. That's nothing for most hard drives today with TB capacity. Anyway, he'll probably blow his brains out by the time he scans 1000 slides, so it won't matter. :) One question you can answer, what's the most real optical resolution can you get from a V750? I'm not familiar with that model. Do you have any figures on the V600? I use 2400.
ICE is good but it increases the time it takes to scan. You still have to go back and check and spot some anyway. That's something Pete can try both ways to see what works for him the best.
Regarding your recommending Silverfast, I think it's incumbent for the purposes of disclosure that you mention that you published and sell a book on using Silverfast and have an interest in it's success. I never used it and it could be the greatest. But I haven't been convinced to make the switch by what others have posted about it. Seems to be very difficult to set up the parameters to get the scan right. And then you have to scan again if it isn't right. That doesn't happen with Epsonscan if you scan flat without any edits during the scan process. Again, that's something that Pete can try both ways since he should have both programs with his scanner.
-
If he finds Silverfast is the way to go, I'm sure your book would be very helpful to him.
-
Mark: The OP indicated he doesn't know yet what he wants to do with it. So for him, scanning it at max optical resolution the scanner can do is best. I'm glad you agree with that. As far as disk space, he'll get 35mb tiff for each slide. He could save if he saves in jpegs but it will only take 37gb to store 1000 35mm slides in tiff. That's nothing for most hard drives today with TB capacity. Anyway, he'll probably blow his brains out by the time he scans 1000 slides, so it won't matter. :) One question you can answer, what's the most real optical resolution can you get from a V750? I'm not familiar with that model. Do you have any figures on the V600? I use 2400.
ICE is good but it increases the time it takes to scan. You still have to go back and check and spot some anyway. That's something Pete can try both ways to see what works for him the best.
Regarding your recommending Silverfast, I think it's incumbent for the purposes of disclosure that you mention that you published and sell a book on using Silverfast and have an interest in it's success. I never used it and it could be the greatest. But I haven't been convinced to make the switch by what others have posted about it. Seems to be very difficult to set up the parameters to get the scan right. And then you have to scan again if it isn't right. That doesn't happen with Epsonscan if you scan flat without any edits during the scan process. Again, that's something that Pete can try both ways since he should have both programs with his scanner.
Alan, just a few factual points:
In his question, the OP did not say anything about purpose and he didn't say he doesn't know what he wants to do with the scans - that's why I asked, because unless you go for maximum CCD resolution, purposes matters to the answer he wanted.
Re the questions you asked me: the CCD resolution of the scanner according to the ISO standard with which it complies is 6400 PPI. This does not tell the story about either the resolving power of the scanning system, or the settings you should use. See my reply to Alan Fairley on all that, just above. I do not know what the ISO standard resolution of a V600 is because I don't have that scanner and never used it. The best way to decide what input resolution you should use for your purposes it to make tests, print the results and compare.
There is no secret I wrote a book about scanning with SilverFast 8. It was publicly announced on this website. I have no commercial interest in LaserSoft Imaging, they can sell as many or as few copies of their software as the market determines and it has zero impact on me. Trust me Alan, you can ask any author with the possible exception of Scott Kelby - you don't get rich writing such books - especially on a topic as arcane as this one with the limited residual market there is for the technology altogether. Selling a few books more or less passes under the radar - way under. I feel blessed and fortunate that I need not even think of depending on this stuff to put food on the table.
Not to put too fine a point on it - but I'm actually quite agnostic about software - for me, it's whatever works best in a systematic workflow for getting quality and efficiency. iSRD, as an example, happens to be a really good tool. It does not slow down scanning and from my experience with it, I'd say it eliminates just about all the crap you need to eliminate without impairing the image one iota. There can be "outliers" it doesn't pick up and those one deals with in LR or PS. Not to toot my horn inordinately, but a major emphasis of my book is about integrated workflows that combine SilverFast with Lightroom and/or Photoshop. Each of these applications have their strengths, weaknesses and optima, so after a great deal of research I've shown how many ways there are to skin this cat (and there could well be more, who knows). There are no pat answers - just different ways of doing things that have their pros and cons. This happens to be one particular area where dogma should take a back seat.
Finally, the nice thing about all this software is that the companies allow us to download and demo them. I encourage anyone who asks me to do just that - its free except for the time it takes to install, learn and play, see what floats your boat. Above all, I discourage being overly influenced by other peoples' problems on the internet. Remember Dragnet in the old days - 8 million souls, 8 million stories.........
Cheers
-
Speaking of getting rid of spots, LR5 has a neat feature. When you click on Spot Removal, you can check a box on the bottom bar that allows spots to "jump out" so you can easily find them to get rid of. The new Spot removal tool also allows you to drag the tool if the the spot is elongated so the replacement will match exactly. It doesn;t only have to be a spot.
-
Back to the OP, seems to me he should alternatively scan for the largest conceivable output size he can foresee (8x10, 20x24, ?x?) or at the maximum rather scanner is capable of. If the OP searches some of the drum scanning threads here, he will find some intense discussion what resolution is required for a scan that captures all the information film contains. If I recall correctly, that figure was comfortably above what the V750 is capable of.
-
Drum scanning is a different proposition altogether. Given that the OP has an Epson V750 I think he should experiment with several resolution settings using his own media and see for himself what difference it will make to the output he wants. Doing one's own research and experimentation, especially when it is so easy, is the best way of being satisfied about what works best relative to the need. I'll end-up on this thread by also hinting that one needs to consider how much can actually be pulled off a piece of Kodacolor - have a look at papers by Tim Vitale and others about the effective resolution of these materials. It's a complex topic on its own as the chemistry has several aspects (silver particles, dye particles, clumps etc.) that influence resolvable detail in different ways, and then there is the quality of the original capture - camera, lens, exposure technique, steadiness etc. etc. The media itself is often the critical variable and when that's the case scanning in more pixels than of benefit to the resolvable detail in the media won't do any harm, but it's just a waste of time and storage - yes storage is cheap, but time has real value - at least to me; bottom line: do the research with your own stuff and decide what you like best.
-
I am scanning some 35mm Kodacolor negs using an Epson V750. Is 1200 DPI a good scanner setting to use? I am trying to balance detail extraction with scanning speed. ANy other tips would be welcome - I am totally new to this.
Hi Peter,
I've been scanning film for a long time, probably close to some 20 years, but less since my digital cameras files exceeded 16MP (roughly(!) matches 4000PPI). There is still a lot of misinformation going round about how scanners and film interact, but it is essentially quite simple once you understand the factors that play a role. Let's concentrate on resolution for now, since that's you main question.
First we need to establish what resolution the film actually caught. If detail isn't resolved on film, you won't resolve it in a scan. This immediately becomes a question that's hard to answer, because it depends on the interaction of the film (slow thin emulsion, low or high contrast), the camera optics (focus/diffraction/aberrations), and shooting conditions (tripod/handheld).
However, we can empirically demonstrate (http://www.clarkvision.com/articles/scandetail/) that it is possible to extract more detail from film with scanning densities of upto 6000-8000 PPI, presuming that such level of detail was captured on film to begin with. That does also help to reduce grain-aliasing (http://www.photoscientia.co.uk/Grain.htm), but that's a somewhat different subject.
Now, the difference between, say, 4000 PPI and higher sampling densities is in the range of diminishing returns but it still helps. However, this also presumes a scanner is being used that is capable of delivering high MTF at those resolutions, and this is where things change quite a bit with the Epson V7xx and V8xx scanners when compared to dedicated filmscanners or especially drumscanners.
One major obstacle is getting the film in the plane of best focus, which includes height of film in the film holder, and film warp or even movement during the scan due to heat build-up. Then there is the quality of the scanner optics (lens/prism/mirrors whatever is included), and the relatively poor protection against veiling glare (also because a large area of the film is exposed while only a small strip is being scanned).
Another obstacle is that it's hard to predict how the MTF of the film image and the MTF of the scanner (assuming perfect focus and a clean lens) will combine into a system MTF.
One possible way of knowing the practical upper limit on scanning PPI with a particular scanner is by determining the limiting resolution of the scanner itself, and then assume a perfect film image (which it isn't). There is an ISO 16067-2 procedure (http://www.iso.org/iso/home/store/catalogue_tc/catalogue_detail.htm?csnumber=35201) that describes the best way of doing that, and it involves scanning a slanted edge which then allows to produce an MTF curve which should tell us where we can draw the line on practical resolution limits.
The 1951 USAF resolution test chart which was mentioned, is designed to be used with analog recording systems (i.e. film), and is not that well suited for discrete sampling systems such as line scanners and digital camera sensors. It's main issues are that the contrast is unreasonably high (lower contrast will already be unresolved when higher contrast can still be resolved) and, more importantly, the bar patterns may either align with the sensel pitch of not. That makes it sensitive to positioning which can result in up to a factor of 2 resolution difference between tests.
Another and much more practical method, if one can still produce a film image with the equipment and film/processing, is by scanning a (sinusoidal) grating at various rotations as captured on the actual film. My free radial grating (Star) test target (http://www.openphotographyforums.com/forums/showthread.php?t=13217) is very well suited for such tests. It allows to visually compare performance, and it allows to quite accurately quantify the limiting system resolution of capture+scanning operations, all the way up to the Nyquist frequency of the scanner/digitizer.
Here (http://bvdwolf.home.xs4all.nl/main/foto/scan/se5400/se5400-1.htm) is an example of a test I did on a much older target design with a dedicated film scanner. Those old webpages were not maintained/expanded when a newer model scanner was introduced, since I received fewer questions that needed to be specifically answered for that first generation 5400 scanner. However, it still shows that the scan resolution effectively increased with scanning/sampling density on those dedicated film scanners for home use. BTW, an effective system resolution of 76.1 lp/mm on that SE5400 resembles a direct digital camera capture with an approx. 6.4 micron sensel pitch.
Concluding (I'm jumping ahead to it a bit), we can assume that there is a benefit to scan at at least 2400 PPI on the Epson scanners, with diminishing returns for resolution all the way to 6400 PPI. It will help to reduce grain-aliasing a bit at the higher scan densities, but the limited scan resolution (and relatively diffuse lighting) already have a 'positive' effect on the visibility of noise.
Personally, I would scan at full resolution capability, and down-sample for storage. The amount of down-sampling will determine the level of loss we can tolerate for the intended use.
Cheers,
Bart
-
Peter, I have Epson V750. About a year ago I scanned about 3000 color negs. Here is what I found out.
- Most important: don't start a big scanning task until you feel comfortable about what you are doing. Do tests with several settings including resolution and focus (and software).
- At first I also thought that scanning ONCE is the right way to go (as Alan suggested). After doing my tests I decided that scanning TWICE is my way.
- FIRST, I scanned all the 3000 negs. Epsonscan. Batch mode. Resolution of 1200 PPI. JPEG. For me, I found this to be good balance of resolution ('quality'), scanning time and disk space.
- SECOND, I picked 'the best' (not many, I'm afraid…). EpsonScan or Vuescan. Resolution of 2400 or 4800. TIFF 16bit. This took more time and a LOT more disk space, but there wasn't very many 'the best' images to scan. I plan NOT to do a THIRD scan. Hopefully my SECOND scan is good enough.
- I also tried SilverFast and found it, to put it politely 'not for me'. Btw, if anyone can produce a successful batch scan with SilverFast in 30 minutes, I'll buy him/her a lunch. Vuescan is very good in individual scans, but not in batch scanning. I found EpsonScan the most intuitive of the three.
My two cents
/fox
-
- I also tried SilverFast and found it, to put it politely 'not for me'. Btw, if anyone can produce a successful batch scan with SilverFast in 30 minutes, I'll buy him/her a lunch. Vuescan is very good in individual scans, but not in batch scanning. I found EpsonScan the most intuitive of the three.
/fox
Which version of SilverFast did you try? Was it 6.6 or 8?
Do you live in Toronto and have a Visa or Mastercard for the lunch? You'll need it. :-)
-
It was 6.6. The comments about v8 that I have read are more or less the same as 6.6.
I live in Europe ;). And what I really meant in my offer: "in the first 30 minutes with SF". But for you Mark, I'll buy you a lunch anytime, if we meet (I'll choose the place).
-
You should do more than read comments. nothing replaces trying it - the demo of version 8 is free. You'll be reading more about batch scanning anon, so stay tuned. I think you are wise to hedge your bets by controlling the lunch venue :-)
-
Ah yes, you said "anytime" for lunch - that's gracious of you. Much appreciated. :-)
-
Another and much more practical method, if one can still produce a film image with the equipment and film/processing, is by scanning a (sinusoidal) grating at various rotations as captured on the actual film. My free radial grating (Star) test target (http://www.openphotographyforums.com/forums/showthread.php?t=13217) is very well suited for such tests. It allows to visually compare performance, and it allows to quite accurately quantify the limiting system resolution of capture+scanning operations, all the way up to the Nyquist frequency of the scanner/digitizer.
Here (http://bvdwolf.home.xs4all.nl/main/foto/scan/se5400/se5400-1.htm) is an example of a test I did on a much older target design with a dedicated film scanner. Those old webpages were not maintained/expanded when a newer model scanner was introduced, since I received fewer questions that needed to be specifically answered for that first generation 5400 scanner. However, it still shows that the scan resolution effectively increased with scanning/sampling density on those dedicated film scanners for home use. BTW, an effective system resolution of 76.1 lp/mm on that SE5400 resembles a direct digital camera capture with an approx. 6.4 micron sensel pitch.
Further on this note, I think I've found the original (Provia III RDP) film frame from July, 2002. I've scanned it with my Epson V700 using VueScan Pro as scanning software. I've been using VueScan since 2001 (paid once, free upgrades since then), and it still allows me to scan with all scanners I've owned since then. VueScan offers scanning resolutions for the Epson V700 of 6400 / 3200 / 1600 / 800 / 400 / 200 / 100 PPI, and its IR cleaning works fine with the V700.
I've scanned the film strip in the standard Epson 35mm filmstrip holder. If I were seriously contemplating scanning 35mm film with that scanner, I would get a better film holder (http://www.betterscanning.com/scanning/models/vseries.html), but I have access to a dedicated film scanner that delivers higher resolution scans, so I only use the Epson for scanning larger film sizes.
As it turns out, my scanner produces a difference between horizontal resolution (the 'Fast scan' direction), and vertical resolution (the 'Slow scan' direction), the latter being overall some 25% higher. The chart's stars are scanned perfectly symmetrical, so the resolution difference is not due to differences in scanning density, but rather optical, or due to the lack of film flatness.
The first challenge would be to find the best height for the filmstrip holder, in order to achieve the best focus. However, that would take too much time, so I'll just settle for the results with one setting. That will already answer part of the original question, namely; does it matters much at which resolution one does the scanning?
Well, as I said earlier, a higher resolution can be achieved with higher PPI settings, although the gain gets more gradual as one reaches the maximum settings. Since we already know that with a dedicated scanner we can achieve something approaching an 80 cycles/mm resolution, how does the V700 do with the default filmholder settings?
@ 200 PPI 4.0 x 3.9 cycles/mm (H x V), this would be almost adequate for contact print resolution.
@ 400 PPI 8.4 x 8.4 cycles/mm (H x V), a 35mm frame becomes 567 x 378 pixels, with excellent contact print quality.
@ 800 PPI 15.6 x 18.0 cycles/mm (H x V), a 35mm frame becomes 1134 x 756 px, with some 3.6x magnification potential.
@1600 PPI 37.6 x 41.0 cycles/mm (H x V), a 35mm frame becomes 2268 x 1511 px, with some 8.2x magnification potential.
@3200 PPI 44.0 x 50.1 cycles/mm (H x V), a 35mm frame becomes 4535 x 3024 px, with some 10x magnification potential.
@6400 PPI 47.5 x 60.2 cycles/mm (H x V), a 35mm frame becomes 9071 x 6047 px, with some 12x magnification potential.
The mentioned magnification potential assumes something like up to 5 cy/mm in final output, which can be viewed from short distances.
As can be seen, somewhere between 1600 and 3200 PP, the proportional increase of resolution with scanning density starts to drop off, but resolution keeps increasing with scanning density, because there is more resolution in the film than the scanner resolves. Therefore increasing the scanning density will produce more resolution, although the available resolution potential is not reached. Maybe this can be improved a bit by optimizing the filmholder and its positioning, and there will of course also be some variance between actual scanner copies, some are likely better than others.
How this turns out for the newer V8xx line remains to be seen, but a test with a star chart is usually quite revealing (and quantifiable in a useful way).
Cheers,
Bart
P.S. This also suggests that the 6400 PPI scan can be downsampled to some 75% of the scanned dimensions in pixels, where it will have a similar resolution as a dedicated film scanner can achieve directly. It should be possible to go even a bit further down without significant losses, but more acceptable storage requirements.
-
Epson scan on my V600 has settings from 50 to 12800. I won;t list them all but there there are these settings in the middle 1200, 2400, 3200, 4800, 6400. I use 2400 with my V600 Does the Epson scan on the V700 allow the same settings?
-
Epson scan on my V600 has settings from 50 to 12800. I won;t list them all but there there are these settings in the middle 1200, 2400, 3200, 4800, 6400. I use 2400 with my V600 Does the Epson scan on the V700 allow the same settings?
Hi Alan,
I don't know, I only have VueScan installed on my computer, maybe someone else knows. Maybe (probably) some of these intermediate resolutions are resampled from actual samples. One would need to examine the real resolution as measured with e.g. a star chart image on film, to see if these produce real resolution, or resampled resolution differences.
Cheers,
Bart
-
What do you mean re-sampled resolutions. Does the V700 do that?
-
Scanners will scan at their own set of discrete intermediate resolution steps up to the maximum resolution of the CCD as reported pursuant to ISO 14473. The steps Bart cites are determined by the scanner and not the scanning application. The scanning application works with them. If you specify a resolution setting in the software that lies between two such settings, the software will resample to cohere with either the step above or the step below in providing you with the resolution you have selected.
-
So to Alan's question - no. The scanner doesn't do the resampling, the software does. So, for example, let us say the two discrete steps the scanner provides are 3200 and 6400, but you want 4800. In the software you select 4800. The software knows for that scanner there is no such discrete setting, so if it's smart software, it will scan at 6400 and then downsample its 6400 result result to the 4800 you requested.
But you know, while it's good to understand what's going on under the hood (and sometimes necessary for diagnostic purposes), the most important thing is to actually see what difference this or that setting makes to the printed output. That is really the bottom line. Often times imagined differences inferred from theory just don't show up in comparative prints - depending. That is why, going back to the post that started this thread, I would always recommend empirical testing starting with real photos typical of the media one starts with and ending with the output for which the scan will typically be used. This is real painless (except for some time, paper and ink) and you know for sure what you're about.
-
What do you mean re-sampled resolutions. Does the V700 do that?
According to Ed Hamrick, the author of Vuescan and who reverse engineered the command set of all scanners to allow driving them with optimal settings, the V600 (http://www.hamrick.com/vuescan/epson_perfection_v600.html), the V700 (http://www.hamrick.com/vuescan/epson_perfection_v700.html), and the V800 (http://www.hamrick.com/vuescan/epson_perfection_v800.html) series have an optical resolution of 6400 PPI maximum. That means that only integer fractions of that can be exact samples, and other intermediate resolutions are an interpolated, version of the native sensor resolution.
So, 6400, 3200 (skip 1 or bin 2pixels), 2133 (skip 2 or bin 3 pixels), etc. would be integer fractions, with 6400, 3200, 1600, 800, etc. being exact halves (skipping to or binning of double numbers of pixels). In the fast scan direction (along the linear scanlines) there is not much one can do to change resolution other than skip or bin pixels and remain exact, but that doesn't influence scan speed. In the slow scan direction the stepping motor can skip line positions (within the accuracy of the stepping motor interval), and that would affect overall scan speed.
Since the data interface is the slowest factor, one can usually get an idea about skipped (slow scan direction) lines from the scan speed. If the scan speed changes, the resolutions in between will be mathematically interpolated. If the scan speed remains the same, then different resolutions will be interpolated. The most accurate way of determining what is going on, is to measure the actual captured resolution.
Cheers,
Bart
-
So to Alan's question - no. The scanner doesn't do the resampling, the software does.
In general that's correct, although there are sometimes instructions in the scanner's firmware that allow to bin pixels and average them, before sending them to the interface with the computer. That could be interpreted as interpolation, albeit with a very coarse approach (although it also reduces noise). Epson is also known for using staggered line sensors in some of their models, where sensel lines are offset by half a sensel position, and 6 line sensors (2x Red, 2x Green, 2x Blue) are used to capture the 3 color bands. That also can be used for achieving some of the intermediate resolutions.
The question however, I agree, is mostly academic. It really only matters what comes out of the scanner, and that is measurable. And if it's significant enough for the output size and viewing distance, it may be visible.
Cheers,
Bart
-
Bart How do you measure the actual interpolated resolution?
How would I use my 8 1/2" x 11" printer to see what is the actual resolution that's the best to scan with assuming that I would want to enlarge later with an outside print service? Should I crop a portion of the picture?
-
Bart How do you measure the actual interpolated resolution?
Alan, the IMHO easiest way is to use a Star target. One very well focused shot on low ISO film of a printed target can serve as a master file for any type of scanning situation. If you want a very accurate result, you should use the same film and lens as you want to answer the question for (because some lenses are more blurry than others), but otherwise use the best. Shoot the target (printed at 600 or 720PPI) from a distance of between 25 and 50x focal length (50x is better for low ISO film). Exact distance is not important, the blur diameter will always be the same if the lens is reasonably well corrected for the shooting distance.
The diameter of the target's central blurred region (usually circular or elliptical) of the scan in pixels, will allow to directly quantify the actual resolution in cycles/mm. The formula to use for scanning scenarios is (depending on the number of cycles of the target itself):
Cycles/mm = 144 / pi / (pixels / scanPPI * 25.4) ,
or pre-calculated and simplified for a 144 cycle target that becomes = (1.80459 * scanPPI) / pixels .
How would I use my 8 1/2" x 11" printer to see what is the actual resolution that's the best to scan with assuming that I would want to enlarge later with an outside print service? Should I crop a portion of the picture?
You would typically aim at achieving a resolution on output of 5 - 8 cycles/mm at reading distance of, say, 12 inches. At twice that distance you only require half the resolution, etc. So you don't really need to print if you know the resolution of the scan, and the magnification factor for a certain output size.
If you want to print an uncropped 35mm frame scan, 24mm x 36mm = 0.945 x 1.417 inch, it would require an 8-9x magnification to reach 8 1/2" x 11" depending on the crop due to aspect ratio. So you'd want to start with a scanned resolution of 5 cy/mm x 9 = 45 cy/mm for uncompromised print quality (72 cy/mm if you want 8 cy/mm in output after 9x magnification). Letting the print software interpolate to the native resolution of the printer will allow to output sharpen at the final size before sending the file to a print service, or you can let them handle it if they know what they are doing ...
As long as the scan has the required resolution for the output size and viewing distance, the printing become easy. Larger output sizes require higher scan resolution, unless the viewing distance increases proportionally.
Cheers,
Bart
-
I know this maybe different than the OP tools at hand, and it looks like his question was answered with a cement truck of possible answers. (If it helps the OP, Knowing intention of the image does set you up for efficient workflow).
But what about photographing the slides? Using nonAA 30- 40mpixel db?
Well, I don't want to interpret the resolution, but just the 2 methods. I was thinking of getting a slide attachment for my setup, or making one. I know keeping things flat and a few things that will be needed to take in consideration. I guess my Q is more on the ability of the 2 techs on the quality they can produce.
This might help in some efficiency depending on workflow as well? Any thoughts?
-
I know this maybe different than the OP tools at hand, and it looks like his question was answered with a cement truck of possible answers. (If it helps the OP, Knowing intention of the image does set you up for efficient workflow).
But what about photographing the slides? Using nonAA 30- 40mpixel db?
Well, I don't want to interpret the resolution, but just the 2 methods. I was thinking of getting a slide attachment for my setup, or making one. I know keeping things flat and a few things that will be needed to take in consideration. I guess my Q is more on the ability of the 2 techs on the quality they can produce.
This might help in some efficiency depending on workflow as well? Any thoughts?
Yes, there's lots of thoughts : http://www.luminous-landscape.com/essays/scannerless_digital_capture_and_processing_of_negative_film_photographs.shtml (http://www.luminous-landscape.com/essays/scannerless_digital_capture_and_processing_of_negative_film_photographs.shtml)
I'll add to that: there is no digital back included in this. I have one, but the cost of a Phase One macro lens for doing this work would be about 4000 dollars and I'm not sure it can fill the frame without tele-extenders. This is something I was intending to look into but the cost is "a bit" of a disincentive. If you're going to do this, you want the very best lens quality your camera can accommodate, because the ultimate constraint on the quality of results will be the lens. Other issues such as flatness, alignment, lighting etc. are much more easily dealt-with. Another really interesting inference from the work we did is that a very high quality film scanner of yester-year will give moderately priced digital camera set-ups a good run for the money.
-
As can be seen, somewhere between 1600 and 3200 PP, the proportional increase of resolution with scanning density starts to drop off, but resolution keeps increasing with scanning density, because there is more resolution in the film than the scanner resolves. Therefore increasing the scanning density will produce more resolution, although the available resolution potential is not reached. Maybe this can be improved a bit by optimizing the filmholder and its positioning, and there will of course also be some variance between actual scanner copies, some are likely better than others.
In order to verify my findings, I also scanned another test film shot from April 2002, on Ektachrome Elite (EB), also with a star target. The shot was made with another lens (and possibly aperture), and the combination of lens/focus/film produces a lower overall resolution scan. However, since I've not optimized scanner focus and film flatness for 35mm scans, I'd rather not focus on the absolute resolution limit for now.
Instead, it is IMHO more interesting to see (attachments) that a similar relative resolution pattern emerges with increasing sampling density. Up to 1600 PPI the resolution increases proportionally with sampling density. Double the PPI, and resolution also approx. doubles. Then between 1600 and 3200 the curve flattens and only smaller, sub-proportional, increases of resolution can be achieved by denser sampling, although resolution does keep increasing.
Cheers,
Bart
P.S. It also looks like it that intermediate Custom PPI settings in VueScan, produce even more moderate increases (probably due to smooth resampling), so it would seem to be better to scan higher, and down-sample.
P.P.S. I've also added a scan resolution test overview for Sensia (RA), which essentially shows the same pattern.
-
the cost of a Phase One macro lens for doing this work would be about 4000 dollars and I'm not sure it can fill the frame without tele-extenders.
Just to add a bit to the topic drift, you could use a bellows plus something like the Rodenstock APO Rodagon D 75mm F4 which is optimized for 1:1 and covers 6cm x 6cm. You could find a clean copy of the lens for $300-350. Admittedly with an MDB you would not be working at its optimal reproduction ratio... You'd also need to drop a couple hundred on the bellows and rig up an adapter (or else press a good view camera into service).
-
Yes, Alan, good suggestion - that could be workable, but would need to research how it would be attached to a Phase One camera and what accessories would be needed to fill the frame with the media at least on one dimension - can't do it for both because the aspect ratios won't match, but that's OK. Means going completely manual, perhaps best tethering the camera to a laptop, experimenting with focus and exposure till one gets it exactly right for each media size and type. No doubt doable and probably wouldn't cost the sky, except for the front-end time. The issue is how much time and money one needs to spend before knowing how worthwhile. I can surmise it may be excellent, but one doesn't know till after a fair bit of experimentation. To keep in mind!
-
Mark, the biggest factor in this type of thing is that what one is chasing are pretty much marginal improvements I think....
-
After a point, for sure. When going manual fitting manual third-party stuff on a camera never conceptualized for it, I think there's a fair bit of essential work before one gets to the point where after the improvements are marginal. To some extent this kind of thing, from what I've seen, is binary - either you get it right or you don't for some identifiable reason - and it's pretty obvious; but after it's fixed further change does become less significant. This is where (good) scanners have some advantage, in the sense that the package is pretty much fixed and you kind of know what you're getting from the get-go.
-
Well, thanks to all for the many responses. I feel like the guy who asked the time and was told how to make a watch!
-
Well, thanks to all for the many responses. I feel like the guy who asked the time and was told how to make a watch!
Hi Peter,
I also hope you got an answer to your question that you can use ...
In fact, I've also installed the Epson Scan utility, just to verify whether the somewhat strange results I got with Vuescan are not the product of a bug or something like that. Well, Epson Scan confirmed the linear increase with scan density for the lower PPIs. up to 800 PPI, and the same sub-proportional increase for the higher PPIs from 3200 to 6400. However, where VueScan also allows 1600 PPI at double the resolution of 800 PPI, Epson scan only offers a 1200 PPI or a 2400 PPI, with the later being similar to an interpolation between VueScan's 1600 and 3200 PPI, but 1200 PPI perhaps being a bit worse than possible (see attachment). VueScan clearly extracts more resolution from the film at 1600 PPI, where Epson Scan would require 2400 PPI to slightly improve over that because it doesn't offer the 1600 PPI option.
Concluding one could say that VueScan seems to have a pivot point at 1600 PPI, and Epson scan at 800 PPI, beyond which only more sub-proportional resolution gains can be expected. Of course VueScan's 1600 PPI will produce smaller output sizes than Epson Scan's 2400 PPI that would be needed to approximately match the relative resolutions, which is useful if you want to maximize resolution at a relatively small size.
In addition, Epson Scan offers 9600 and 12800 PPI modes, but that produces virtually identical resolution to 6400 PPI, although with hugely larger files. The only use I currently see for such settings is when one does need to deliver such huge files. In that case it may (marginally) help to actually scan (by smaller stepper motor movements in the slow scan direction) the pixels instead of just interpolating them.
Do note that these measurements I made were based on 35mm filmstrips (with all the non-flatness and positioning issues that come with it), and that presumably that triggers the use of the higher resolution (6400 PPI) lens of the dual lens system. I presume that the lower resolution (4800 PPI) lens is used for a wider field of view, since the linear sensor arrays have a fixed length.
All this was a bit of a surprise to me (I never tested 35mm film strip scanning before on the V700 because I have better equipment for that), because I expected that resolution gains would gradually reduce as we approached the resolution limits dictated by the optical scanner MTF. Instead, there seems to be a pivot point where proportional resolution increases abruptly turn to sub-proportinal increases, and that at a relatively low level. This seems to indicate something mechanical (such as switching between lenses), and therefore might work out differently between actual copies of the scanners. It would be helpful if others could attempt similar measurements of e.g. a star target caught on film which allows real resolution limit quantifications.
Cheers,
Bart
-
Bart: So what's the best ppi to scan at on 35mm and on 120 using Epson Scan?
-
Bart: So what's the best ppi to scan at on 35mm and on 120 using Epson Scan?
Alan,
Best (resolution wise) is 6400 PPI (also) on Epson Scan, if you need it for the intended output size.
A compromise on Epson Scan would be 2400 PPI (extracts some 80% of available resolution), beyond which resolution can only increase a little bit while file size quadruples per full step. The difference between 2400 PPI and 6400 PPI is still significant, but only if you need it for a given output size. Higher than 6400 PPI has no significant effect on resolution.
For storage it's best to scan high and down-sample to the lowest resolution you are likely to need.
Cheers,
Bart
-
Alan,
Best (resolution wise) is 6400 PPI (also) on Epson Scan, if you need it for the intended output size.
A compromise on Epson Scan would be 2400 PPI (extracts some 80% of available resolution), beyond which resolution can only increase a little bit while file size quadruples per full step. The difference between 2400 PPI and 6400 PPI is still significant, but only if you need it for a given output size. Higher than 6400 PPI has no significant effect on resolution.
For storage it's best to scan high and down-sample to the lowest resolution you are likely to need.
Cheers,
Bart
Higher than 6400, more than not having any effect, would probably degrade output quality because it exceeds the resolution of the sensor and triggers software making-up pixels to upres, which would show as degraded IQ if taken too far. I've been reading attentively your observations about the declining incremental improvement of resolution as the setting increases beyond 2400. If you were to actually print some of these intermediate outcomes up to 6400 your observations would be reinforced, confirming an approach to "right-size" rather than necessarily maximize - unless of course one thinks the much larger files may one day come in handy.
-
Higher than 6400, more than not having any effect, would probably degrade output quality because it exceeds the resolution of the sensor and triggers software making-up pixels to upres, which would show as degraded IQ if taken too far.
Hi Mark,
I'm not sure about degradation, but there will at least be no resolution increase if only interpolation is applied. There is hypothetical gain to be had if one uses the 9600 PPI stepping precision that the mechanical drive is capable of in the slow scan direction, but 6400 PPI in the fast scan direction is limited to that by hardware, the length of the sensor scan-lines. So at best there would be some asymmetrical resolution gain possible. However, it looks like it that the 9600 PPI stepping precision is used for accurate placement rather than oversampling, because I have not really seen an increase of resolution in that slow scan direction, it's just unchanged like in the fast scan resolution. More pixels yes, more resolution no.
I've been reading attentively your observations about the declining incremental improvement of resolution as the setting increases beyond 2400. If you were to actually print some of these intermediate outcomes up to 6400 your observations would be reinforced, confirming an approach to "right-size" rather than necessarily maximize - unless of course one thinks the much larger files may one day come in handy.
Well, there is an actual (although slower growing) incremental resolution available above 1600/2400 PPI. I believe that it is available as long as one doesn't exceed a total scan-area width of approx. 5.9 inches, and length of 10 inches. Wider and longer would switch to the '4800 PPI lens', to increase the Field of View that's projected on the sensor. That 6400 PPI incremental resolution means that the effects of things like grain-aliasing will also be reduced. Also the electronic and shot noise of the scanner sensor, will reduce with down-sampling, so there is a benefit to sampling at 6400 and down-sampling to the requirement (output resolution or storage size).
But I agree that one should still keep an eye on the ultimately required resolution, which is why I earlier mentioned the measured magnification potential (assuming a targeted output requirement of 5 cycles/mm), but that does require to measure the actual resolution that one's combined film+scanner MTF is capable off. My measurements of 3 different films show 3 different levels of resolution. although they do seem to be within a certain bandwidth. Maybe they were caused by different lens quality and levels of focus accuracy at shooting time, maybe it is the resolution or MTF of the film.
The measured resolution limits of 40 - 60 cycles/mm would loosely translate to 2032 - 3048 PPI real resolution, or an 8 - 12x magnification potential of the film size to output size that allows relatively close inspection.
Cheers,
Bart
-
And all this time I thought to scan at the highest native resolution(optical max) in the max bits your scanner can do (no software interpolation/oversampling).
3 pages later, I 'm not so sure :-)
So about that camera rig....
-
And all this time I thought to scan at the highest native resolution(optical max) in the max bits your scanner can do (no software interpolation/oversampling).
Phil, nothing wrong with that. A 6400 PPI scanning density setting will actually pull out some 2032 - 3048 PPI real resolution, setting a lower scanning density will pull out less resolution. You'll pull the most resolution of what's there out of the film at 6400 PPI, and probably lose little if you then down-sample that to 75% or even 50% of the scanned dimensions for a more compact file size to archive.
Cheers,
Bart
-
Well, I've been scanning at 2400 PPI, 48 bit, with the V600, so I guess I'm OK for he most part. Thanks for the input.
-
Sorry if I'm a bit Johnny-come-lately.....but this is a discussion that's very near-and-dear. My friend and colleague Mark Segal directed me here as he thought I might find the discussion "interesting" based on some questions I was asking about a certain 1951 USAF target and it's real resolution.
Anyway, I've been rather immersed the past several days in finding good solid information on the resolution of various films, how this translates into scanning resolution, real detail ("system" resolution) and so on. I'm coming at this from a slightly different direction as I'm interested in how it relates to drum scanning....you see, I have a Screen 1045ai drum scanner in my office that I occasionally entertain myself with along with a gaggle of medium format gear.
What REALLY brought this to a head recently was a local photography colleague and one of the organizer's of our local Charlotte Photography Meetup Group that wanted me to try scanning a 4x5 sheet of something called ADOX CMS 20.....some sort of microfilm that's sold in sheets and 120 rolls. This film is claimed to have something like 800 lp/mm or 40,000 [cough] [gag] ppi resolution. Well, my scanner can only crank up to 8,000ppi, well under what the film is supposedly capable of.....but I assumed that the system rez (lens basically) would poop out long before we hit my scanner's top speed of 8K ppi. What I found was interesting.
One of my tests was scanning an area of small detail (fabric texture....attached) at 1K, 2K, 3K, 4K, 6K and 8K ppi. Similar to Mr. Van der Wolf's findings, I found detail improved quickly up to about 4K ppi with smaller improvements after that....but improvements nonetheless. What really surprised me was that there was real image detail improvement right up to the 8K limit of my scanner. Now, I've seen "improvements" before on my own films (Ilford Delta 100/400) but once past about 4K, it was really only improvements in grain structure, not actual image detail (side note....when I did tests of my own film, it was shot with my Pentax 645NII...I plan to do another test with my Pentax 67II to see if I can get some improved image detail at resolutions about 4K). In this case, it wasn't simply improvements in grain rendering (there WAS NONE with this film!). So either my colleague was using an excellent lens (likely) or, given the right film, lenses can perform better than I'd been led to believe (I hear numbers like 80-100 lp/mm bandied about for lenses).
Some fun facts about this drum scanner....
It has multiple discrete drum speeds that corrolate with resolution.....the higher the resolution, the slower than drum has to spin in order for the system to handle the data output. The "system" in this case is an aging Graphite PowerMac running the 90s version of OS 9.2.2! The drum speed is not stepless so there's always a transition point from one rez to another. For example, at 4000ppi drum speed is 900rpm....@ 4001ppi, it drops to 600rpm. What was interesting is that the quality slightly improves at the transition to the slower speed......at the limit of the faster speed (4000ppi) there's some grain "smearing"...but when you set the rex to 4001 and trigger the drum speed reduction, grain sharpens up. As a result, I typically only scan at 3001, 4001 and 6001 ppi resolutions.
This scanner has a 16,384 lineal pixel limit. So, for a 4x5" sheet film, your maximum scanning rez is 16,384/5" or 3,276ppi. So, if you really need a 4x5 scanned at, say, 6,000ppi, you have to cheat by scanning the sheet in sections and stitching it later in Photoshop...works like a charm.
Scanning time.....for a 4x5" sheet @ 3000ppi, it takes nearly 90 min. to complete the scan........and for a 6000ppi scan in sections, it took me nearly 9 hours of scan time! The one saving grace is the excellent batch scanning capability of the software. You can basically batch up a drum full of scans (or sections of a 4x5) and walk away for hours....or overnight in my case.
In an interest to staying on topic and perhaps adding to the discussion, I'm more of the mind of scanning once at the highest practical resolution (typically 4001ppi) and then downsample for printing. Since it takes something like 35-45' to scan a 6x7 format, I'd rather do it just once. :)
My own personal workflow is, after development, I sleeve all my negs and scan them on my Epson V750 (yes, I have one of those too)....that becomes my contact sheet. After looking through the images, I then pick out the best (typically only 20-30% of each roll) and go through the process of wet-mounting and drum scanning.
Regards,
Terry
-
Terry: Interesting process. Can we see some of the results of both scanners?
-
I don't really have anything from my Epson V750 but I do have something from the drum scanner. Attached is a screen shot of several 100% views in PS ranging from 1,000 ppi (right) to 8,000 ppi (left). You can look at the filenames on the tabs for the resolution.....and see that the 8,000 ppi scan shows just a bit more detail than the 4,000 and 6,000 ppi scans. These were scans from the Adox CMS 20 film (4x5). I should note that these are with NO sharpening applied, either at the scanner or in Photoshop.
I'll see if I can come up with something from the V750 tomorrow.
Terry
-
It seems that the focus seems to degrade moving to the right where there is less resolution. But why should that be? I can understand less resolution. But the results on the right are also out of focus.
Do you have any examples of people and places, full photos that we can compare rather than USAF type charts. That would be more helpful in understanding what we're looking at.
-
The part of the scan you're seeing is the texture of a blouse.....so it's real, not a USAF chart. I can't show the entire image since it's of a person and I haven't asked permission.
They're not out of focus. What I did was take each scan and sample them all up to 10000 ppi so they were all the same relative size. So the lower resolution scans only appear out of focus. It's difficult to show them all at the same size at their native scan resolutions.
If you'd like me to show a screen shot at their original un-sampled resolutions, I can do that.
Terry
-
The part of the scan you're seeing is the texture of a blouse.....so it's real, not a USAF chart. I can't show the entire image since it's of a person and I haven't asked permission.
They're not out of focus. What I did was take each scan and sample them all up to 10000 ppi so they were all the same relative size. So the lower resolution scans only appear out of focus. It's difficult to show them all at the same size at their native scan resolutions.
Hi Terry,
Thanks for joining the discussion,and for adding your examples. The low and high contrast features of the fabric provide a useful spread of feature contrast, which would not show in the 1951 USAF chart comparisons (although there exist also low contrast versions). It's obvious that for a fair visual comparison, one needs to scale (up) the lower sampling densities. Human vision is not that good at comparing both resolution and scale differences at the same time.
It's that difficulty of scale differences, and scanner methods, that led to the modified ISO procedures to objectively determine the Spatial Frequency Response or Modulation Transfer Function (SFR/MTF) of scanners. They more recently also included the star chart measurements in their resolution measurement methods for digital cameras, but it is hard to directly produce such a sinusoidal version with enough detail for scanners.
For us it is a bit more practical, because we need to establish a system performance of camera optics / film / scanner, which is different from looking at a scanner in isolation. For such a system performance it is very simple to take a photograph (a reduced size, and hence very high resolution copy) of such a star target which challenges even the best resolving component of the imaging chain, the camera lens. Then the combined/cascaded MTFs of all components in the system will allow to determine a rather objective metric for the limiting resolution (at a contrast level one can choose). That metric should then ideally match the required resolution goal one has for a specific use of the image.
Drum scanners typically have a spot size that can be varied, but CCD scanners usually have a fixed sensel pitch (with low fill factor) and sometimes variable lighting (ranging from collimated to diffuse) that influence the resolution (and graininess) characteristics when scanning film. But they also have that diminishing returns of resolution increase with higher sampling densities (and smaller spot sizes for more contrast), but up to some 8000 PPI it can be demonstrated that more resolution will be extracted from film, assuming the detail was captured to begin with (so low ISO, fine grained and thin emulsion, using high quality camera optics and good technique).
Cheers,
Bart
-
Perhaps I'm missing something here. I used the Epson V700 a while back to scan some old slides and negatives, and I was rather concerned about test reports on some sites that claimed the real optical resolution of the V700 is not 6400 ppi but 2300 ppi, and that there was no point in scanning at resolutions higher than 2400 ppi. One would merely produce larger files with no increase in detail.
That didn't seem right to me. If it was true, then one could severely criticise Epson for conning the public and wasting everyone's time.
So I carried out my own tests on the sharpest of my slides and negatives, scanning the same negative twice, once at 2400 ppi and again at 6400 ppi, then compared detail, grain and resolution after appropriate processing and sharpening in Photoshop.
Now, an important principle when comparing image detail is to ensure one is comparing same size images. It's a nonsense to compare different size images. So one either interpolates the 2400 ppi scan to 6400 ppi for comparison purposes, or one downsamples the 6400 ppi image to 2400 ppi.
The results of my tests were quite clear. The 6400 ppi scans were always either noticeably sharper and more detailed, but with equal noise, or equally sharp but with lower noise.
There was no way I could get the 2400 ppi scan to match the downsampled 6400 ppi scan in its balance between noise and detail.
So my conclusion was, even if you calculate that a 2400 ppi scan is sufficient for your print size at, say 300 dpi, a 6400 ppi scan which has had more sharpening applied before downsampling, will always produce better results in terms of noise and/or detail, using the Epson V700 with 35mm film.
-
Perhaps I'm missing something here.
Hi Ray,
No you're not missing anything. Scaanning at a higher sampling density will allow to get more detail out of the same film, and it has a positive effect on the way that grain (or dye clouds) is rendered. However, it does take longer to scan, and the resulting file size will be larger.
There was no way I could get the 2400 ppi scan to match the downsampled 6400 ppi scan in its balance between noise and detail.
So my conclusion was, even if you calculate that a 2400 ppi scan is sufficient for your print size at, say 300 dpi, a 6400 ppi scan which has had more sharpening applied before downsampling, will always produce better results in terms of noise and/or detail, using the Epson V700 with 35mm film.
Correct, that's also what the conclusion of most other people is. One may quibble over the amount of difference that can be achieved and if it's worth the extra time and storage space, but the extra quality is there for the taking.
Cheers,
Bart
-
I scan MF 120 6x7 on my V600 using 2400 ppi and 48 bit. An image file is about 200mb. Scanning at 6400 would give me a file 7x as large about 1.4gb. At 4800 the file is 4 times as large at about 800gb which I think is the limit for Lightroom. Does anyone recommend something over 2400 which I had heard back when was the most you could get out of the V600? Advice?
Also, let's say I try different ppi's. What do you look for when comparing to see when there's no more resolution from the scanner? Advice?
As an aside, the Dmax might be the bigger issue. On the V600 is 3.4 and on the V700 it's 4.0. So shadow areas will scan better the higher the Dmax. When I scan, there always are huge amount of sharpening that are required either by the scanner or in post if I scan flat.
-
I scan MF 120 6x7 on my V600 using 2400 ppi and 48 bit. An image file is about 200mb. Scanning at 6400 would give me a file 7x as large about 1.4gb. At 4800 the file is 4 times as large at about 800gb which I think is the limit for Lightroom. Does anyone recommend something over 2400 which I had heard back when was the most you could get out of the V600? Advice?
Yes.....higher resolution increases file size....no surprise there. File size won't be close to 800 GB though.....maybe you mean 800 MB?
Also, let's say I try different ppi's. What do you look for when comparing to see when there's no more resolution from the scanner? Advice?
I don't know the V600 specifically....but the first thing would be to find the highest *optical* (or native) resolution your scanner is capable of. On my V750, that seems to be 6400. So start at the highest optical resolution and scan in 1/2 increments from there.....6400, 3200, 1600, 800, etc. Don't bother with in between resolutions like 4800 or 2400 as they are likely interpolated.
For testing the detail rendering of your scanner, first off, you must make sure that it's focused properly. If you're using a film holder, you might have to play with "shimming" the holder to get optimum focus. Also try scanning the film emulsion up and down and see if it's makes a difference. On the 4x5 sheet film I was testing, it made a HUGE difference. When testing focus, I would suggest using the highest resolution and scanning just a small section in the center.
After you've got that worked out, time to run the scanning tests at the various resolutions. When picking a film frame or area to scan, I would recommend scanning an area of relatively LOW contrast like fabric texture or something. Scanning areas of high contrast will likely cause "flare" (not sure if that's the correct term) where light coming through clear areas of the film bleed into the denser areas. Scanning a dense area of low contrast detail will give you the best chance of seeing what your scanner can resolve. Oh, and any sharpening in your scanner software must absolutely be disabled. You don't want to get fooled into evaluating *acutance* instead of real detail.
When evaluating the scans themselves, the first thing I do is up-sample all the scans to something like 9600ppi using "Bicubic: Automatic" in PS. After upsampling, use the Window -> Arrange menu in Photoshop and tile them....I use "Tile All Horizontally" to use the full width of my screen and I also hit the TAB key to get rid of tool palettes and such. After tiling them all, click on one of the windows and set the view to either 100% or 200% depending on your personal pixel peep level :) and then move the image around till you find the area of detail you want to view. Once you have one window set to the detail area and view %, use the Window --> Arrange --> Match all option to get them all to exactly the same area.......let the evaluation begin! I find it helpful to slide my chair back to get a good view....then I start closing windows that obviously have poor(er) detail....probably the first couple lower resolutions. That should open up the remaining windows to view a bigger area of the scan(s). Look at them critically.
After you've done that, you could also down-sample them to, say, 3200 and evaluate them again. Quite often the higher-rez scan will still hold up better to down-sampling than the one scanned directly at 3200.
As an aside, the Dmax might be the bigger issue. On the V600 is 3.4 and on the V700 it's 4.0. So shadow areas will scan better the higher the Dmax. When I scan, there always are huge amount of sharpening that are required either by the scanner or in post if I scan flat.
Quotes of certain Dmax values on consumer flat-bed scanners are largely bullshit....or at least meaningless. They are usually calculated from the the native bit depth of the scanner and not the actual *optical* density it's capable of rendering. I was doing some reading on this recently and the best explanation I read was that the bit depth refers to the *size* of the density container....but it says nothing about whether the scanning optics can actually FILL that container.....like having a sack that can hold a million bucks....but that doesn't mean the sack actually HAS a million bucks in it. :)
And those Dmax numbers are kind of meaningless in another context. Most B&W and color neg films won't go beyond about 2.00 Dmax (I've measured them)....and the actual Drange is about a stop less than that. So from a density standpoint, you're not going to find many, if any, neg films that go beyond about 6-7 stops of density (nothing to do with scene dynamic range though). It's true that transparency/slide films have a higher Dmax but I can remember back in my drum scanning days that I would typically never see a Dmax (non-exposed area of the film rebate) over about 3.20......but I've heard that Fuji Velvia can go a bit higher. Point is, Dmax specs beyond about 3.4 are pretty useless since you'll rarely find a film, especially negs, that push those limits.....assuming the scanner's optics are up to the task.
Terry