Epson V850 work-around for scanning 5x7 negatives.

[Joseph Kashi]

Hi, I'm new to this forum but not to large format photography, which I've done since the 1970s. Recently, I started using 5x7, mostly Ilford Delta 100 BW film in that size, but now scanned rather than wet-printed as previously

Initially, I had some difficulties getting a really in-focus scan of 5x7 film with an Epson V850 using both the included Epson Scan program and VueScan. Mostly, the difficulty arose from the lack of 5x7 film holders that would properly register with the scanner's film holder mode Aside from some wet/dry mounts that involve taping the negative to the ANR glass, I found no easy, commercially available 5x7 solutions for home processing and scanning.

I did find a highly satisfactory work-around and want to pass along the information in the hope that it will save others from similar initial hassles. Basically, I constructed my own 5x7 film holder for the Epson scanner by deconstructing and then revising a dual 4x5 Epson 4990 film holder.

Here's what I did: On Ebay, I bought a new Epson 4990 film holder that was set up to scan two 4x5 negatives at a time. I then carefully cut out the center cross-piece between the two 4x5 sections. I also cut the inner edges of the top-clamps for each 4x5 aperture. Use a very thin hacksaw blade carefully to ensure that only the minimum necessary material is removed in the saw kerf. You'll need that tight fit when you relocate the cut segment to the 7" point and epoxy it into place so that the aperture and edge supports exactly fit a 5x7 negative on the 7" side as well as the 5" sides.

Using a Dremel tool, smooth down any raised areas that interfere with negative seating and with top-clamp locking. The remaining portions of both partially-cut top-clamp pieces should close and lock normally. Epoxy part of one of the removed top-clamp segments to bridge the cut, otherwise unsupported, ends of the two 4x5 clamps, basically so that they form a single longer top-clamp and move simultaneously, supporting each other.

The cut-out section of the second top-clamp is used loose and separately as a friction fit top clamp on the second 5" edge that was part of the center section moved and epoxied at 7 inches. As a means to hold down the remaining 5" edge on the relocated center piece, that removable friction fit piece works well if your initial cuts are smooth and fit snugly.

I now had a 5x7 film holder that registered properly as a film holder of the right height in the Epson V850's central 5" high resolution scanning area and that supported the Ilford Delta 100 5x7 negatives quite flatly without requiring ANR glass on top.

Scanning with both VueScan and Epson scan produced very good to excellent edge to edge sharpness when using the makeshift 5x7 film holder at 2400, 3200, and 6400 resolutions and at most VueScan manual focus ranges from -.36 to +.74.

Using this makeshift 5x7 film holder, I judged that VueScan's manual focus seemed best at +.56. Scans at 2400 dpi and at 3200 dpi resolutions seemed to work best. I judged that the 2400 dpi scan had marginally better tonal separation on Delta 100. 6400 dpi scans had excellent sharpness but low acutance, even after strong post-processing in Lightroom. The test images, made with a Fujinon 210/5.6 NWS lens were sharp at 1:1 even when scanned at 6400 dpi inherent resolution but I did not particularly like the tonal quality at 6400 dpi.

Hence, my judgment was that the best 5x7 BW negative scans on the V850 scanner with VueScan seemed to be at manual focus +.56 and 2400 or 3200 dpi. With Epson Scan, setting the film holder setting, with or without setting 6400 dpi (which activates the high-resolution lens system), resulted in good focus.

On a 5x7 negative, 2400 dpi works out to making 40"x 56" digital prints at full native 300 dpi print resolution without any interpolation needed. Luckily, Delta 100 and the Fujinon NWS 210mm lens had ample native resolution.

[Hugo Zhang]

Is it possible to lay 5x7 negatives on the scanner like prints and simply scan them? I was told to use 24-bit-color and at 300 dpi. I just want to digitize all my 45, 57, WP and 810 and larger negatives and won't do large digital prints.

I struggled for a few hours this afternoon and don't know what to do.

[djdister]

I would avoid setting the negs directly on the glass if possible. In addition to Mr. Kashi's excellent solution, I made a decent 5x7 neg holder out of two pieces of mat board.



Is it possible to lay 5x7 negatives on the scanner like prints and simply scan them? I was told to use 24-bit-color and at 300 dpi. I just want to digitize all my 45, 57, WP and 810 and larger negatives and won't do large digital prints.

I struggled for a few hours this afternoon and don't know what to do.

[Hugo Zhang]

I have just read somewhere that a sheet of undeveloped but fixed 8x10 film could be used as a spacer placed directly on the scanner glass and then the 8x10 negative to be placed shining side down. Does this work? I would find some 5x7, WP and 8x10 blank negatives as spacers.

[Oren Grad]

Is it possible to lay 5x7 negatives on the scanner like prints and simply scan them?

Yes. Emulsion side down. That's what I do.

IMO 300 samples per inch is too low if one intends to print the scans, but with sheet film there are difficult tradeoffs to be made between scan resolution and file size / file manageability. You should make your own decision about what works best for you by making a set of scans at different settings up to the maximum and judging the results and the handling burdens for yourself.

[Pere Casals]

I have just read somewhere that a sheet of undeveloped but fixed 8x10 film could be used as a spacer placed directly on the scanner glass and then the 8x10 negative to be placed shining side down.

Perhaps this would be suitable for the 4990 scanner but not for the V700 - V850 series.

The V850 has two lenses, when you use the area guide you may scan two sheets on glass (with emusion down, as mentioned, to avoid newton rings) as that frame may take a 8x10 negative and the Low Res lens is used. The High Res is used when a holder is detected and it covers only 5.9" maximum width.

The High Res Lens has an effective optical yield of 2900pi in the horizontal axis and 2300dpi in the vertical one, while the Low Res lens has a proportionally lower yield, around 2200 to 1900dpi effective (X-Y) that anyway it's an insane amount of image quality from a 5x7 sheet.

The HighRes lens is focused some mm over the glass, where the holders place the film. The LowRes lens is focused on the glass bed surface.



I have just read somewhere that a sheet of undeveloped but fixed 8x10 film could be used as a spacer placed directly on the scanner glass and then the 8x10 negative to be placed shining side down.

Perhaps this was good for the preceding 4990 model, it has a single lens that probably is focused in an intermediate position, to work acceptably both for mediums placed on bed and also for mediums raised by the holder, but IMHO placing that extra sheet makes no sense for scan on glass with the V850 area guide. What you may do is wet mounting your film directly on bed glass.



I was told to use 24-bit-color and at 300 dpi.

As Oren pointed, 300 is too low ! If on bed, scan it 2400 dpi at least, better if 3200. You want always a good quality for edition, after edition you donwnsample to the suitable size for the printer (use "bicubic ideal for reductions" setting).

Don't scan 24-bit-color !!! this is 8 bits per channel only, with 256 levels as you modify the curves you soon will obtain a nasty banding.

Scan single channel grayscale at 16 bits per channel, and save it in TIFF format (BMP nd jpg will only save 8bits/ch), you have 65k gray levels and you always have smooth tonal transitions.

[Pere Casals]

Here's what I did: .....

I guess that the scans (and prints) have to be impressive... the V850 shines is 5x7" format if a holder can be provided !!!

That solution (hacking 4x5 holders for 5x7) is not new, but it's really great to know that the solution works nicely, and having the details.

It would be great if you would post a LowRes sample image with a HighRes 100% crop. That scan has to be 230MPix "effective" worth... which is an insane amount in "effective" terms.

[j.e.simmons]

12-15 years ago a user here came up with the solution. Buy two sheets of anti-glare picture frame glass cut to the scanner glass size. Put down the Edson 8x10 mask and put one glass on top, etched side up. Add the negative, then weight it down with the second glass, etched side down. Easy. Very rarely any Newton’s rings. Works up to 8x10. And yes, it uses to lower lens, but I don’t think there’s a way around that.

[Pere Casals]

And yes, it uses to lower lens, but I don’t think there’s a way around that.

By simply Selecting “Film (w/ Film Holder)” in the software it activates the higher resolution (SHR) lens.

https://www.largeformatphotography.info/forum/showthread.php?150931-Epson-V700-High-Resolution-lens&p=1486835&viewfull=1#post1486835

[Jim Andrada]

I wet scanned a lot of 5 x 7 on my 750 and never needed a piece of tape. I used the mounting station and carrier and fluid mounted the neg face down under a piece of the mounting plastic I got from Aztek using their Kami fluid. Average time to mount a 5 X 7 was under 30 seconds. Fastest that I actually timed was 23 seconds.

[Joseph Kashi]

Is it possible to lay 5x7 negatives on the scanner like prints and simply scan them? I was told to use 24-bit-color and at 300 dpi. I just want to digitize all my 45, 57, WP and 810 and larger negatives and won't do large digital prints.

I struggled for a few hours this afternoon and don't know what to do.



Yes, you can lay the 5x7 negatives on the scanner, emulsion side down facing the glass. Then, put Anti-Newton Ring glass on top (mat side down on to top of negative) to hold the negative flat. I found that worked well enough with Epson Scan up to 2400 dpi, but the makeshift 5x7 film holder worked somewhat better with Epson Scan and much better with VueScan. I would urge you to do some consistent, documented scanning tests before deciding upon your final workflow specs. Epson Scan is a good start because it produces decent results without a lot of fiddling.

If you lay your negatives directly on the glass, then use the setting for film guide and try to center negatives on the scanner glass. The Epson Scan software that comes with the scanner is best for this setup. If all you're doing is digitizing old photo negatives or prints, then those minimum quality settings that you mention will work but not very well. You'll likely be better off at 1200 dpi or 2400 dpi and 48-bit color for two reasons:

You extract much more data from the higher resolution and higher bit depth negative scans, and that data may be of value some day to others - if you don't get that data when you can, then you've precluded any higher level use that may be desired in the future, either by you and by those to come; and,

Even if you just print same size, the downsizing and higher bit depth will result in a better print. When digitizing an analog process like a negative or paper print, a higher sampling rate ( higher resolution and bit depth) than the intended final output will minimize the inevitable loss of information in the final digitized result, resulting in better apparent sharpness and less banding.

Disk storage is now very cheap and inexpensive, so there's no practical reason to do tiny scans unless you 1.) don't have the time, or 2. ) store everything "in the cloud".

Scanning an analog item like a negative or a paper print at 300 dpi will result in effective resolution that's lower than 300 dpi. I've forgotten most of the math to calculate the actual loss of effective resolution but higher sampling rates result in much less loss. I've found that 1200 is good enough for a lot of 5x7 negatives on a flat bed scanner and that 2400 dpi high bit rate scans seem optimum. ( As I mentioned in my original post, higher than 2400 dpi scans don't seem to capture any additional real resolution, tend to reduce micro-contrast somewhat and quickly result in excessive file size.)

When I started doing large format photography back at MIT with Minor White back in the early 1970s using the Zone System, we had to go through a lengthy series of camera, lens, and exposure meter calibrations and then calibrations of our film and print exposure and development. These were, and remain, fairly tedious and time-consuming but are unavoidable if you want to do high quality film photography.

Digitizing analog film and paper prints introduces into the final workflow yet another step requiring careful calibration of your monitor, of your scanner settings and technique, and of your printer in addition to the same equipment and film developing calibration tests required for Zone System.

These are just optimized starting points and if you're after the highest quality output, you'll still need to make tweaks for different images and also in your digital post-processing with Lightroom or a similar program. Ideally, a scanned film negative image will have a full and well-differentiated tonal range without being too contrasty. That allows you to optimize afterwards in post.

However, once you have your complete system dialed in, digitized storage and final printing is much easier and more consistent than making a final wet print and you only have to spot the print once, digitally. It's ultimately worth the effort when you're intent upon exhibition quality results.

[Oren Grad]

Then, put Anti-Newton Ring glass on top (mat side down on to top of negative) to hold the negative flat.

I get artifacts in my V700 scans when I have AN glass on top of the negative, I think generated by reflections from the extra glass surface. Often my film stays flat even without; otherwise the negative needs to be held down some other way.

I may try again at some point with antireflection coated glass. It's a pain to find large pieces without subtle flaws, but perhaps that will matter less when the glass is being used to hold down a negative for scanning than it does in a negative carrier or in contact printing.

[Joseph Kashi]

I get artifacts in my V700 scans when I have AN glass on top of the negative, I think generated by reflections from the extra glass surface. Often my film stays flat even without; otherwise the negative needs to be held down some other way.

I may try again at some point with antireflection coated glass. It's a pain to find large pieces without subtle flaws, but perhaps that will matter less when the glass is being used to hold down a negative for scanning than it does in a negative carrier or in contact printing.

I've found that anti-glare glass from a picture-framing shop also works OK if the mat side is down and in contact with the negative. It's also much less expensive. I bought commercial ANR glass for this purpose but later found that mat glass is OK. It appears that as long as the rougher side is in contact with the top of the negative and the emulsion directly on the scanner glass, then Newton rings are less likely.

[Joseph Kashi]

I would avoid setting the negs directly on the glass if possible. In addition to Mr. Kashi's excellent solution, I made a decent 5x7 neg holder out of two pieces of mat board.

I first tried the mat board approach with corner holders but could not get a scan that was in sharp focus on the V850. That's why I hacked the dual 4x5 4990 holder into a 5x7 holder. There was probably only a small difference in thickness but that was too much for sharp focus.

[Michael Roberts]

Scanning with both VueScan and Epson scan produced very good to excellent edge to edge sharpness when using the makeshift 5x7 film holder at 2400, 3200, and 6400 resolutions and at most VueScan manual focus ranges from -.36 to +.74.

Using this makeshift 5x7 film holder, I judged that VueScan's manual focus seemed best at +.56. Scans at 2400 dpi and at 3200 dpi resolutions seemed to work best. I judged that the 2400 dpi scan had marginally better tonal separation on Delta 100. 6400 dpi scans had excellent sharpness but low acutance, even after strong post-processing in Lightroom. The test images, made with a Fujinon 210/5.6 NWS lens were sharp at 1:1 even when scanned at 6400 dpi inherent resolution but I did not particularly like the tonal quality at 6400 dpi.

Hence, my judgment was that the best 5x7 BW negative scans on the V850 scanner with VueScan seemed to be at manual focus +.56 and 2400 or 3200 dpi. With Epson Scan, setting the film holder setting, with or without setting 6400 dpi (which activates the high-resolution lens system), resulted in good focus.

On a 5x7 negative, 2400 dpi works out to making 40"x 56" digital prints at full native 300 dpi print resolution without any interpolation needed. Luckily, Delta 100 and the Fujinon NWS 210mm lens had ample native resolution.

Joseph,
Thanks for posting this!

Did you apply any post-scanning sharpening in PS/LR? If so, how much and what were your conclusions?

Thanks,
Michael

[Pere Casals]

Did you apply any post-scanning sharpening in PS/LR? If so, how much and what were your conclusions?



Let me say my view.

Some Pro scanners provide a very good digital sharpening, with the V850 you also may activate a (less good) sharpening in the scanning software but I find that doing it manually in Ps is better.

The amount of sharpening you should provide depends on several factors, a main one is the scanned dpi, if scanning with V850 at 6400dpi a good start point would be 2.5pix radius and 50% settings. If dpi is lower then the radius should be lower.

Sharpening is a complex matter, shadows, mids and highlights may require different local settings, in a portrait we may sharpen eyes different than skin... A good sharpening often require a dose of manual adjustmets, anyway today some automatic tools do an amazing job, in a protrait they detect gender, age, race and face features, also they detect overshot to adjust ideal settings locally...

But, "technically" speaking the image provided by the V850 always (IMHO) require some sharpening manual adjustments. Software in Pro scanners required better automation in that to save manpower, and had a number of image enhancing features, a particular (not LF) case are scanners of the digital minilab era (Frontier, Noritsu...), which had an extensive array of very well automated enhacements (color, etc), which included a good automated sharpening.

[Joseph Kashi]

I guess that the scans (and prints) have to be impressive... the V850 shines is 5x7" format if a holder can be provided !!!

That solution (hacking 4x5 holders for 5x7) is not new, but it's really great to know that the solution works nicely, and having the details.

It would be great if you would post a LowRes sample image with a HighRes 100% crop. That scan has to be 230MPix "effective" worth... which is an insane amount in "effective" terms.

A full 16 bit grey scale TIFF scan of the 5x7 negative at 2400 dpi is about 340 megabytes on disk when using both Epson Scan and VueScan. My next set of calibration scans is focused upon tonal range and Zone System development, etc. All of these calibrations are idiosyncratic to each person's technique and systems, so I see those as merely starting points for each person's own tests.

[Joseph Kashi]

Joseph,
Thanks for posting this!

Did you apply any post-scanning sharpening in PS/LR? If so, how much and what were your conclusions?

Thanks,
Michael


Yes, I sharpened quite a bit in Lightroom as part of my calibration tests. I also used Clarity +26 slider. The raw scanner output was not as clean and sharp as one would like for an end result, but responded very well to post-processing, yielding a very clean and sharp final result. LR sharpening, exposure and tonal adjustment, and clarity were all needed for best final result. Neither Epson Scan nor VueScan output was suitable for high-quality final use without post-processing.

I sharpened quite a bit because I used standard high-contrast resolution test targets for the underlying test negatives. I needed a maximally sharp and high acutance/crisp result because the underlying tests were made to determine the comparative and absolute sharpness of a variety of large format lenses.

Thus, I first had to determine optimum scanning setup and post-processing technique for best sharpness so that the lens resolution test results were not seriously affected by sub-optimum scanning technique.

Sharpening the initial scans was needed to determine the end-result maximum resolution of the overall workflow system of lens/film/development/scanning/post processing. Now that I've determined the scanning part of the equation, I can scan and review my 5x7 negatives made with the various lenses.

The scanning technique and post-processing tests were all made with a late-model Fujinon 210mm/5.6 NWS lens, which seems from highly magnified inspection of negatives to be one of my sharpest lenses. Before the scanning tests, I first did tests for camera alignment, initial exposure accuracy and development, and then individual lens resolution, contrast, and shutter accuracy testing.

The next and final set of tests will optimize exposure meter accuracy for better shadow tonality and adjusting development time to avoid Delta 100's tendency toward highlight blocking. Carefully designing the tests and targets can allow several different items to be tested in one shoot.

[Pere Casals]

A full 16 bit grey scale TIFF scan of the 5x7 negative at 2400 dpi is about 340 megabytes on disk when using both Epson Scan and VueScan.

At 2400dpi you don't extract all possible information the V850 is able, see table C.2: https://archivehistory.jeksite.org/chapters/appendixc.htm

For best result you may scan (for a 5x7" sheet ) to a +1 GByte file and later downsampling ("bicubic, special for reductions", choice in Ps) to the point that not much information is lost, depending on if the negative is more or less sharp and on your print size you may notice more or less going well beyond 2400.

The V850 takes 2900 dpi optically effective in the horizontal axis with the HiRes lens, but one has to scan beyond taht (4800 or so) to take that effective resolving power.

In the past PCs had problems with big files, today with an SSD or an M.2 disk it is less a problem !!

[Joseph Kashi]

At 2400dpi you don't extract all possible information the V850 is able, see table C.2: https://archivehistory.jeksite.org/chapters/appendixc.htm

For best result you may scan (for a 5x7" sheet ) to a +1 GByte file and later downsampling ("bicubic, special for reductions", choice in Ps) to the point that not much information is lost, depending on if the negative is more or less sharp and on your print size you may notice more or less going well beyond 2400.

The V850 takes 2900 dpi optically effective in the horizontal axis with the HiRes lens, but one has to scan beyond taht (4800 or so) to take that effective resolving power.

In the past PCs had problems with big files, today with an SSD or an M.2 disk it is less a problem !!


As I noted in my original post, the 2400 dpi was, for me, the best overall compromise.

The amount of additional resolution beyond 2400 dpi is fairly minimal but the file size is much larger and contrast and tonal quality are somewhat (but visibly) reduced.

You may also start bumping into Lightroom's file size limit, and Epson Scan sometimes has difficulty saving the file beyond a 5x7 2400 dpi scan, which results in a 340MB or so file size at 2400 dpi.

Then, of course, effective resolution is greatly affected by your lens, film, and developing choices. If the effective resolution isn't on the film, scanning at a higher dpi makes little difference except perhaps in a totally theoretical as opposed to practical way.

Perhaps that's why I found not much practical benefit going beyond 2400 dpi, using 5x7 Ilford Delta 100 in XTOL 1:2 intermittent agitation for high acutance and a super-sharp Fujinon 210mm NWS lens, which resolves as good as my CM-W lenses and better than the Sironar N lenses, and noticeably better than any of my Dagor and Protar VII optics that I also tested at the same time.

The scanning at 2400 dpi was good enough to see sharply rendered film grain on this very fine grain film, which suggests little real improvement beyond that 2400 dpi point. Beyond that point, one is creating artifacts, not real data.

FWIW, the computer used to make these scans does use an M.2 SSD and an extremely fast Ryzen CPU, 32GB fast DDR memory, and a very fast GPU video card. I built the system for maximum photo post-processing performance.

That's how things worked out for me, but others may find different approaches work better for them, hence the suggestion to make one's own tests, perhaps with these starting points.

[Pere Casals]

As I noted in my original post, the 2400 dpi was, for me, the best overall compromise.

That's how things worked out for me, but others may find different approaches work better for them, hence the suggestion to make one's own tests, perhaps with these starting points.

2400 also works for me most of the times for LF, but if a LF negative is really sharp then I scan it at higher dpi, anyway this is more important in 4x5 than in 5x7, IMHO.

YMMV, of course... it depends on print size, for example.



The amount of additional resolution beyond 2400 dpi is fairly minimal ................ If the effective resolution isn't on the film, scanning at a higher dpi makes little difference except perhaps in a totally theoretical as opposed to practical way.

Of course, it depends on how sharp is your shot and how "microcontrasty" textures are, some shots benefit from a higher dpi and some not. Anyway an image requires a x2 effective pixel count (x1.4 linear resolution) to see a real improvement, Image Quality it's very expensive in image size terms.

At 2400 you get under 2000 effective dpi with the EPSON, but you may reach 2900 in the horizontal axis, if this makes or not a benefit for a particular shot and if that improvement is necessary... it depends...

When I shot ADOX CMS 20 sheets I easily see the difference, TMX resolves more or less depending on microcontrast, datasheet tells very different resolving power depending on contrast.



You may also start bumping into Lightroom's file size limit, and Epson Scan sometimes has difficulty saving the file beyond a 5x7 2400 dpi scan, which results in a 340MB or so file size at 2400 dpi.

With bundled Silverfast, Ps and a SSD disk I've no problem.




The amount of additional resolution beyond 2400 dpi is fairly minimal but the file size is much larger and contrast and tonal quality are somewhat (but visibly) reduced.

I don't agree with that, tonality is exactly the same, scan a crop at 2400 and at 6400, make images of the same size and compare, if you noticed a difference it had to be an edition effect or scanning settings. Why tonal tonality has to change with dpi ?

PD: if scanning 16 bits per channel then you do what you want with tonality in the curve edition, of course a problem is is there if scanning or saving the file in 8bits: banding soon happens.





a super-sharp Fujinon 210mm NWS lens, which resolves as good as my CM-W lenses and better than the Sironar N lenses, and noticeably better than any of my Dagor and Protar VII optics that I also tested at the same time.

In LF lenses there is a sample to sample variation, a Sironar N may be better or worse than a CM-W. Fujis don't usually include shims for the front cell, many Sironars do, if the original shims are lost then the right shimming has to be found.

NWS or CM-W lenses are not better or worse than Sironar N.

An old Dagor/Protar should resolve less in a good shot, of course, another thing is when this makes a difference or not.




The scanning at 2400 dpi was good enough to see sharply rendered film grain on this very fine grain film, which suggests little real improvement beyond that 2400 dpi point. Beyond that point, one is creating artifacts, not real data.

YMMV, but to depict well the film grain structure you need higher efective dpi than the EPSON is able, fortunately grain structure is mostly irrelevant in LF, most of the times. In fact it's really difficult to depict well grain structure in the hybrid processing, nothing like a "through the negative" wet print for a pleasing grain structure (when enlargement vs film nature and processing shows the grain).

[Joseph Kashi]

FWIW, my comments starting this thread resulted from making nearly 150 5x7 test negatives. I tested 34 different LF lenses and chose the sharpest negative of all. In many instances, I retested some of those lenses to be sure that I was getting their best possible on-film performance and discarded any sub-par negatives.

I'll detail those tests and calibrations not to be argumentative, but rather in the thought that doing individualized systematic testing, given a practical starting point from which to deviate and further test one's own systems, may be useful to some readers on this forum.

I then made about 100 test scans of that sharpest negative at all resolutions and focus points from 1600 to 6400 dpi. Included in my test targets were three large 21 step wedges plus specific micro-contrast/tonal gradient finely-detailed BW images along with the usual resolution test charts so that I could compare resolution, dynamic range, and tonal gradation in each negative.

Development was in intermittent agitation XTOL 1:2 on Delta 100 - this provided higher acutance/sharpness negatives. I first tested and found the best negative time and agitation technique for optimum sharpness prior to scanning. Tmax 100 is no longer available in 5x7 over the counter, hence Delta 100 in diluted XTOL as recommended for Delta 100.

As a practical matter, slight tonal quality improvement was discernible under magnification in the highlight areas of the 2400 dpi finely-detailed BW tests, at least to me after repeated inspection comparing 2400dpi and 3200dpi scans of the same super-sharp negatives.

Very little if any discernible resolution improvement was noted when scanning the same test negative at 3200 dpi nor 4800 and 6400 dpi after comparable post-processing in Lightroom and secondarily DXO. If there was any additional useful resolution, it was too slight for me to reliably detect under a high-quality 10X magnifier or 1:1 on the calibrated monitor.

Hence, my conclusion that a 340MB file scanned at 2400 dpi was probably the most practical overall approach balancing all of the various factors. Scanning beyond the effective physical limit of a device and its focus capabilities primarily introduces artifacts. The 2400 dpi scan still gives a 300 dpi native resolution print with crisp detail at a 40"x56" size for 5x7 full-frame printing of 5x7 negatives and 32"x40" for a scanned 4x5 negative. That's basically as large as even a 44" wide digital printer can print. I'm not sure that there is any practical benefit beyond that point for virtually any subject. Diffraction limits at f/32 and smaller, and limited depth of field/focus issues, are likely to be more significant determinants of discernible on-print resolution at very large print sizes.

While an undergraduate and then graduate student at MIT way back when, I studied some with Minor White, who headed that program until his death. He authored the "New Zone System Manual" with Zakia and emphasized that practical tests and individual calibration, not theory or what others said, were always the ultimate determinant of best photographic practice for any one person and their systems. Theory does help us understand and hopefully can guide us toward a good starting point for our own tests and toward constant improvement in our technique, but in the end individualized calibrations and practical tests are the foundation of Zone System and any similar approaches to careful photographic technique.

[Pere Casals]

First, let me reiterate that a 5x7" at 2400dpi should deliver very good results.

Anyway, IMHO, a 3200dpi scan may deliver a discernible improvement if the negative is sharp.

There are several see tests around:

see here C.2 table: https://archivehistory.jeksite.org/chapters/appendixc.htm

https://www.largeformatphotography.info/forum/showthread.php?150020-Scanner-Comparison-2019-Epson-Flatbed-Eversmart-Flatbed-Drum-Scanners&highlight=pali





. I tested 34 different LF lenses. In many instances, their best possible on-film performance and discarded any sub-par negatives.

It would be interesting you post what lp/mm you found for the lenses you tested.

It can be compared with other practical tests made

http://www.arnecroell.com/lenstests.pdf
http://www.hevanet.com/cperez/testing.html




Development was in intermittent agitation XTOL 1:2 on Delta 100


Reportedly (good source) a Delta 100 4x5" negative has 461Mpix effective inside, a 8x10 one 627Mpix, a 5x7 should be beyond 500MPix effective with not much effort. See table https://www.onlandscape.co.uk/2011/12/big-camera-comparison/

197645



I then made about 100 test scans of that sharpest negative at all resolutions and focus points from 1600 to 6400 dpi.


You may know that the V850 delivers a different resolving power for the horizontal axis than for the vertical one, many scanners have the same behaviour, including the Hasselblads.

At 2400 dpi you take from a 5x7" sheet around 150 MPix effective (2000*7*2100*5) from the 500MPix effective available. If scanning 6400 then you would take 233MPix efective (2900*5*2300*7). Don't think that 233 vs 150 is a big improvement in the Image Quality...

That 6400 scan (16 bits) would be 2.8GBytes, but most of the benefit would be reached by 3200dpi, so a 3200 dpi scan may be worth if the negative is sharp enough and we want to print really big.




Tmax 100 is no longer available in 5x7


There is an easy solution, just cut two 5x7 sheets from a 8x10 sheet. You may use an IR night vision device for convenience, but it is also easily done in the dark, one may practice first with lights open using a discarted 8x10 negative.






As a practical matter, slight tonal quality improvement was discernible under magnification in the highlight areas of the 2400 dpi finely-detailed BW tests, at least to me after repeated inspection comparing 2400dpi and 3200dpi scans of the same super-sharp negatives.


Sorry, but I've scanned quite a lot of T2115 Stouffer wedges and I've never noticed an effect from the dpi on the tonal scale, in fact I developed a C++ crafted tool to speed up film and paper calibrations, you may repeat your tests, if you found a tonal problem then you may have a flaw in the test, perhaps you did not take all the histogram, saved 8 bits, you had some banding... I can tell you that I checked that very well and I'm pretty sure of it. A lower dpi may have lower noise because pixel binning averages several pixels, but if you properly downsample the higher dpi image you have the same noise, no problem in the tonality, for sure.


https://www.largeformatphotography.info/forum/showthread.php?150045-New-darkroom-calibration-software


https://c1.staticflickr.com/5/4872/39776268813_073b8f6f3e_o.jpg

What makes an improvement is using Multi-Exposure in Silverfast, as it overcomes some of the limitations from the CMOS range, compared with PMTs. This is mostly important for recovering Provia/Velvia extreme shadows.





Scanning beyond the effective physical limit (2400?) of a device and its focus capabilities primarily introduces artifacts


No! what artifacts?

The V850 will never introduce artifacts because resolving power is limited by the optics, and not by the sampling. Nyquist-Shannon explains very well when you introduce artifacts, in the V850 case sampling is x2 the optical performance (6400 vs 2900), with the optics working like an strong LPF that prevents any artifact. Other scanners (in what limitation comes from sampling density rather than from optics) may generate artifacts, a V850 not.

If you have artifacts this should be from compression, etc.




MIT way back when

I've spent several thousand hours in the OpenCourseWare, mostly in MV and ML.

[Joseph Kashi]

I may have not been adequately clear in my prior posting, and hence may have inadvertently caused some confusion, leading the discussion off the OP topic. My apologies. I will attempt to clarify.

The very slight apparent highlight degradation that I discussed earlier was not in step wedges where, you are of course correct, there would not be any variation when scanning a solid tone step wedge. I should have been clear that I was discussing tests that included my own composite test target that included a printed high contrast BW landscape image with delicate highlight and deep shadow detail, three large standard step wedges to allow accurate densitometer readings, as well as standard high-contrast ISO 12233-style resolution test targets.

Where there seemed to be very slight but discernible contrast loss was in the extreme highlight shoulder areas of that BW enlargement portion of my composite test target. 2400 dpi seemed slightly better at retaining extreme highlight detail there than the 3200 dpi images and 3200 dpi did not provide resolution that was discernibly better under a 10X Wista ground glass loupe.

There may well be very slightly better contrast and highlight tonality at one ultimate scan resolution and arguably slightly better resolution at a different scan rate. This is not so strange when one considers that contrast tends to drop at higher frequencies/resolutions ( MTF curves, etc.). But, really, in these discussions, LF has such a reserve of image quality that these minimal differences, when they occur, are of no practical significance in real-world photography. From a 2400 dpi scan of a 5x7 negative, we’re at a native 300 dpi in a 40"x56" print - that’s literally laser printer quality sharpness and crispness.

While I generally do not put much credibility in Internet postings, I did find two on-point V850 reviews by respected publications that seem pertinent to others possibly reading this thread later.

The UK’s respected journal Amateur Photographer’s review of the Epson V850 did a series of tests starting at 3200 dpi through 12800 dpi and reported that each additional increase in apparent dpi slightly but discernibly reduced contrast and crispness.

ScanDig’s extensive V850 review, which was a very favorable review by the way, noted that their tests indicated that the V850's actual resolution limit was in the approximately 2300 dpi to 2600 dpi range.

There followed discussion indicating that the ultimate resolution of Epson’s scanners is affected by the scanning software used. Silverfast SE 8 Plus seemed to do about 13% better in their tests than the Epson Scan and VueScan that I used. I may need to get a copy of SE 8.5 Plus.

Also, it’s worth noting that the film in my tests was scanned using the 5x7 film holder whose construction I detailed in starting this thread. Using the V850's film holder mode automatically engages the V850's high resolution fixed focus lens and thus physically scanned the test films in the center channel at the 6400dpi previously suggested.

Hence, any lower dpi final output in fact would have been initially sampled at the higher sampling rate you suggest in at least one dimension and then downsized to whatever resolution was set as the output dpi. The other dimension would have the scanner’s step rate as the limiting factor, again as you noted.

It may well be that the algorithms of different scanning software affect sharpness and contrast in at different rates in the final output. ScanSoft, VueScan, and Epson Scan all work differently and that may well account for the variation.

With effective resolutions that vary along different axes and that differ from the set output dpi, interpolation is occurring in at least some of those various output dpi modes along with other “artifacts” as I would term them, such as aliasing. It may be that one scanning program does a better job at interpolation than others and that may account for any differences noticed.

One other point of interest: an uncompressed 16-bit greyscale TIFF made at 2400 dpi from a full-frame 5x7 negative results in a consistent on-disk file size of 335MB to340 MB with both Epson Scan and VueScan. That would seem to be the actual final TIFF file size. Was the smaller size file quoted earlier in this thread the multiplication sum of the total dots on each side rather than the actual file size stored on-disk? Was it compressed or a JPEG, low bit-range, etc.?

[Pere Casals]

I may have not been adequately clear in my prior posting, and hence may have inadvertently caused some confusion, leading the discussion off the OP topic. My apologies. I will attempt to clarify.

Don't worry, we are here to debate, and in my case to learn, every day I learn something in this forum.




Was the smaller size file quoted earlier in this thread the multiplication sum of the total dots on each side rather than the actual file size stored on-disk? Was it compressed or a JPEG, low bit-range, etc.?

TIFF format can compress or not, depending on settings, compression is without loss, so the original image is exactly restored, like with ZIP files. JPG compression has an information loss, the restored image has not the same exact values: Wavelets method elimiates information that is less relevant for our eye.

A non compressed TIFF has the size of the pixel count multiplied by "bytes per pixel", plus a header of variable size that may include many things, including ICC profiles.

See in Ps Image Size dialog how many pixels you have.






Also, it’s worth noting that the film in my tests was scanned using the 5x7 film holder whose construction I detailed in starting this thread. Using the V850's film holder mode automatically engages the V850's high resolution fixed focus lens and thus physically scanned the test films in the center channel at the 6400dpi previously suggested.


:)

Of course !!!! You won't find a 3200 vs 2400dpi enhacement with the low res lens !!!!!!!

We had a bad communication...

When you try your new 5x7" holder you will find an slight improvement at 3200, if the shot is sharp, as you will find with 4x5".




ScanDig’s extensive V850 review, which was a very favorable review by the way, noted that their tests indicated that the V850's actual resolution limit was in the approximately 2300 dpi to 2600 dpi range.

"2300 dpi to 2600" is a right value for a "general rating" for the highres lens, but an deep evaluation should consider Hor vs Vert vs Slanted resolution, height precision, MTF... We know for sure that that the V700-V850 products resolve 2900 Hor and 2300 Vert, or at least 2800-2200, the average is 2500 or 2600, and the minimum is 2300 for a vertical patern.

For an effective calculation I use X*Y*2900*2300, which is paroximately the same than X*Y*2600*2600


ScanDig should say 2900H, 2300V and 2600 average. Many tests around find similar results, and it nails what I personally found with the glass slide target:

https://archivehistory.jeksite.org/chapters/appendixc.htm
https://www.largeformatphotography.info/forum/showthread.php?150020-Scanner-Comparison-2019-Epson-Flatbed-Eversmart-Flatbed-Drum-Scanners&p=1477975&viewfull=1#post1477975


Here there is a height vs perrformace study for both axis:

197711




2400 dpi seemed slightly better at retaining extreme highlight detail there than the 3200 dpi images and 3200 dpi did not provide resolution that was discernibly better under a 10X Wista ground glass loupe.

Probably you may have this effect in a situation and the counter in another situation, let me point what IMHO it that can happen.

Linear sensors have "analog binning" capability, the V850 sensor is 40K RGB pixels (120,000 total) in 6 rgb rows, two for each color. A monster, it takes 5.9" at 2900 effective !

"analog binning" capability is featured to speed up scanning at lower dpi, in principle you have to make one analog digital conversion for each photosite, and later you have to average several values to downsample to the user requested dpi. "analog binning" allows to average "in sensor" the electrical voltages or several contiguous pixels so for alower dpi count you require only a fraction og the A/D conversions.

It is undisclosed (until I know) if the V850 used that, but we can bet for it.

Because of digitizing effects, it is not impossible that find a better texture ina a 2400 native dpi for a particular texture, but also it can happen the counter, it depends on patern vs pixels alignment. A similar effect happens when you scan a USAF 1951 target several times and each pass have an slightly different result.

Extreme highlights in BW are no SNR challenge for the V850, you have to go to very underexposed Provia/Velvia to have problems.




The UK’s respected journal Amateur Photographer’s review of the Epson V850 did a series of tests starting at 3200 dpi through 12800 dpi and reported that each additional increase in apparent dpi slightly but discernibly reduced contrast and crispness


Many people ignore that Image Size dialog in Ps have a Combo control in the bottom, to select resizing algorithm, "Bicubic, ideal for reductions" ha to be selected for crispness.

Probably that respected journal Amateur P is not aware that each acquisition kind requires a particular post processing, I'm dayly working in MV, many times raw coding convolutions in C++, and my experience is the counter. The better the IQ you have the better you may get a good downsampled image.




Silverfast SE 8 Plus seemed to do about 13% better in their tests than the Epson Scan and VueScan that I used. I may need to get a copy of SE 8.5 Plus.

You can upgrade your SF license for a fraction of the cost.

I only use SF SE Plus when I need Multi-Exposure, I get the most raw possible image, and I sharpen manually in Ps or other. Probably SF has better digital sharpenings than ES, but I don't want those. Digital sharpening is important, and those included in scanner software are convenient but not optimal, and they degradate information for a following advanced processing.




From a 2400 dpi scan of a 5x7 negative, we’re at a native 300 dpi in a 40"x56" print - that’s literally laser printer quality sharpness and crispness.


Of course ! The V850 is a great LF scanner, and 5x7" it's format it makes better, because you still can use the HR lens that takes 5.9", and it exploits most of the sensor width for singkle shot. For 35mm it scans 4 frames in the scan width plus the plastic separators.

[Joseph Kashi]

This thread has been an interesting discussion spanning real world results and theory. Rereading it, I have a few summary thoughts about using the V850 to scan LF film:

1. Even with an advertised native 6400 dpi resolution, the V850's effective, or at least optimum, resolution setting is lower, likely at 2400 to 3200 dpi range with a 16-bit grey scale TIFF file. A 16-bit TIFF would have an effective 2400dpi resolution of 201.6 +/- MP and an on-disk file size of about 345 MB including TIFF file overhead.

2. A 16-bit grey scale TIFF should be able to encompass the complete tonal range of a negative with careful exposure and development.

3. Optimum results are a combination of resolution, proper tonality, and acutance/microcontrast.

4. Aside from the scanner’s own mechanical stepping limitations, there are multiple analog and analog to digital processes that occur between initially taking the photo and digitally printing the final result. At least some of these steps can degrade the potential image quality inherent in a particular negative but the impact can be minimized at each step by using very sharp optics and very fine grain film like Delta 100 to produce a very high resolution negative, developing carefully in a high acutance developer, and scanning at optimum settings for that image.

5. Even a 2400 dpi effective resolution is more than adequate for 300 dpi native resolution in print sizes of 32" x 40" for 4x5 negatives and 40"x56" for 5x7 negatives. This is laser-printer sharp images with 4x5 and 5x7 negatives, assuming top-tier film, lenses, and technique.

6. At this quality level, careful technique when taking photos, including focusing accuracy, physical diffraction limits at very small apertures, and limited depth of field with 3D subjects is probably more determinant of effective resolution and image quality on the final negative and scanned file than marginal differences in scanning resolution and micro-contrast.

7. Scanning software makes a difference. Although Epson Scan and VueScan are capable of high quality results, Silverfast seems to seems capable of slightly higher ultimate effective resolution, although I cannot comment about the microcontrast and acutance.

8. An acceptable sharp scan of a 5x7 negative can be made using Epson Scan by laying the negative emulsion side down on the scanning glass with ANR glass properly positioned on top of the negative using the film guide mode.

9. Somewhat sharper scans are possible using the “film holder mode” and the 5x7 film holder whose construction started this thread.

[Pere Casals]

This thread has been an interesting discussion spanning real world results and theory. Rereading it, I have a few summary thoughts about using the V850 to scan LF film:

1. Even with an advertised native 6400 dpi resolution, the V850's effective, or at least optimum, resolution setting is lower, likely at 2400 to 3200 dpi range with a 16-bit grey scale TIFF file.

The Epson does not reach 3200, it makes 2300 in the vert axis and 2900 in the Horizontal, 2600 effective averaged. This is if scanning at high dpi, 4800 or the like.



7. Scanning software makes a difference. Although Epson Scan and VueScan are capable of high quality results, Silverfast seems to seems capable of slightly higher ultimate effective resolution, although I cannot comment about the microcontrast and acutance.

Regarding resolution, Epson Scan, Vue and SF both do the same, Silverfast SE Plus has multi-exposure for deep velvia shadows. Epson Scan and the "non Plus" SF version lack multi-exposure

[Gary L. Quay]

I'm getting lines in my 5x7 and 8x10 scans that look like they are from the LED's that are used in the scanner. Is this due to laying the film directly onto the scanner glass, and would a film holder that holds the neg away from the glass help? I saw that there was some discussion about film holders earlier, but none mentioned this issue.

Thanks!

[Steven Ruttenberg]

My take. I did focus and resolution tests and at 6000dpi the scans were much sharper than at 2400dpi which was soft. There was a bit more at 6400. What I do is s a at 6400 and do a 2x2 bin to 3200 using viewscan. I don’t worry about charts, graphs, apparent optical quality vs actual optical quality, etc. I scan at whatever gives me the best file.

If I worried constantly about charts, graphs, sample images, then nothing gets done. Find out what gives you the file you want and don’t look back.

[Pere Casals]

I'm getting lines in my 5x7 and 8x10 scans that look like they are from the LED's that are used in the scanner. Is this due to laying the film directly onto the scanner glass, and would a film holder that holds the neg away from the glass help? I saw that there was some discussion about film holders earlier, but none mentioned this issue.

Thanks!

Please post the flawed image and I'll may be able tell you.

Download last software/driver versions, I had several issues (lines) with software that came with the unit. Some incompatible compression algorithms... etc

[Gary L. Quay]

201767

I'm not sure if this will be very good. The forum limits pictures to 250K unless they're on a server. You can see the lines, though. Thanks!

[Steven Ruttenberg]

can you put it on Flickr?

[archphotofisher]

[QUOTE=Joseph Kashi;1524868

I now had a 5x7 film holder that registered properly as a film holder of the right height in the Epson V850's central 5" high resolution scanning area and that supported the Ilford Delta 100 5x7 negatives quite flatly without requiring ANR glass on top.



Do you have an image of this holder and it's modifications, love to see it?

[Steven Ruttenberg]

I tried dry scanning and it sucked. Once I started wet scanning my scans quality went way up.

[Gary L. Quay]

can you put it on Flickr? I can, but I don't want anyone on Flickr to see it. I'll try making it private, and see if it will post here.

[Gary L. Quay]

can you put it on Flickr?

https://www.flickr.com/photos/foolscape_imagery/49674998642/in/dateposted

It won't display here, but I right-clicked on the little box with the question mark, and it took me to the image. I may be doing something wrong. I've been off of this forum since 2014, and the way images post may now be different even though I followed the instructions on the FAQ page.

[Gary L. Quay]

can you put it on Flickr?

Can you see it on this reply?

https://live.staticflickr.com/65535/49674998642_6f756d48d0_k.jpg (https://flic.kr/p/2iFBmrQ)Crown Point, June 2013 (https://flic.kr/p/2iFBmrQ) by Gary L. Quay (https://www.flickr.com/photos/foolscape_imagery/), on Flickr

[Pere Casals]

....

I think it is not the scanner, but flaws in your film.

Scan it rotated 90º and see if the lines also rotate.


________

You have a lot of scratches and dust.

Handle film with care specially when wet, final rinse with some distilled water, squeeze with care if emulsion is delicate (Foma, etc).

Use a cheap HEPA air purifier to get rid of dust in the drying and when scanning, I use a Honeywell HAP-16200E I found in very cheap offer, cheaper choices would also work.

Regards

[Steven Ruttenberg]

I was thinking the same thing. I have had troubles in sky areas similar to that in the past.

[Gary L. Quay]

I think it is not the scanner, but flaws in your film.

Scan it rotated 90º and see if the lines also rotate.


________

You have a lot of scratches and dust.

Handle film with care specially when wet, final rinse with some distilled water, squeeze with care if emulsion is delicate (Foma, etc).

Use a cheap HEPA air purifier to get rid of dust in the drying and when scanning, I use a Honeywell HAP-16200E I found in very cheap offer, cheaper choices would also work.

Regards

Thanks for your help!

I was having some issues at the time with dust (it was 2013), but most of it on this scan was due to running an experiment to get rid of the lines. I placed it on a piece of glass kept a couple millimeters over the scanner glass to see if distance would help. I didn't spend time cleaning it because I didn't know if would work. The scratches weren't on the one that was done on just the scanner glass.

So, I rotated the negative, and the lines rotated. I have had this issue with multiple films, both color and black and white, negative and slide, processed by me and processed by a lab. I scanned a 4x5 slide yesterday, and the sky area showed the same lines. I notice it most on areas where there is uniform tone, like skies, but the lines are throughout the negative or slide.

I've been thinking about trying Silverfast ai 8 to see it that would help. I'm using Epson Scan 2. I went through my older scans the V800 before I had to stop using the original Epson Scan (it was 32 bit, and Apple went to 64 bit, and it stopped working), and the ones before I used Epson Scan 2 did not have the issue. I'll try to post one of those. Again, thank you for the help so far.

[Gary L. Quay]

Here's one done with the original Epson Scan program.

https://live.staticflickr.com/65535/39657769693_b94eca7914_k.jpg (https://flic.kr/p/23qqrgT)Hay Rolls near Silverton, Oregon (https://flic.kr/p/23qqrgT) by Gary L. Quay (https://www.flickr.com/photos/foolscape_imagery/), on Flickr

[Pere Casals]

Thanks for your help!

I was having some issues at the time with dust (it was 2013), but most of it on this scan was due to running an experiment to get rid of the lines. I placed it on a piece of glass kept a couple millimeters over the scanner glass to see if distance would help. I didn't spend time cleaning it because I didn't know if would work. The scratches weren't on the one that was done on just the scanner glass.

So, I rotated the negative, and the lines rotated. I have had this issue with multiple films, both color and black and white, negative and slide, processed by me and processed by a lab. I scanned a 4x5 slide yesterday, and the sky area showed the same lines. I notice it most on areas where there is uniform tone, like skies, but the lines are throughout the negative or slide.

I've been thinking about trying Silverfast ai 8 to see it that would help. I'm using Epson Scan 2. I went through my older scans the V800 before I had to stop using the original Epson Scan (it was 32 bit, and Apple went to 64 bit, and it stopped working), and the ones before I used Epson Scan 2 did not have the issue. I'll try to post one of those. Again, thank you for the help so far.


My pleasure. Feel free to also contact me by PM if you have any issue in the future.

Remember (case you don't do it): to always save files in TIFF 16 bits/channel, try to get all the histogram with manual exposure, the image may look dull, but then in Photoshop (or etc) you adjust contrast and by deformating the cuve in S you adjust toe/should, if images are important better doing in that way as you keek your choices for the edition.


Regards

[Gary L. Quay]

The final resolution to the lines in the scans issue I was having was that the whole lid / transparency scanner needed to be replaced. It works fine now. I still haven't bought Silverfast, but I intend to in the near future.