Epson v800 8x10 transparency vertical banding

[Sam L]

I've been working on even development of 8x10 negatives and noticed that every negative I scan has distinct vertical bands. After a few rounds of this I noticed it was the same bands on every negative. From a few google searches, I understand that this isn't uncommon and has to do with the way the light source for transparency scanning is made (an array of LEDs). Here's an example of a scan of an unexposed and fixed-out piece of x-ray film. It is perfectly clear, blue tinted:

222708

The vertical banding is not on the negative.

I have never noticed at 4x5 and below.

My question is whether I have a bad scanner or this is just characteristic of V800s at 8x10.

Thanks,
Sam

[_tf_]

It’s a bad scanner, if it’s within warranty, chase that up with Epson. Mine has developed this problem across half of the upper scanning light. The light is made of multiple banks of LEDs arranged in two staggered rows one each side of the diffuser. In correct operation all the banks are at the same intensity, and the diffuser evens it all out. In my case one of the banks is now running at about 1/2 stop dimmer (I can clearly see this when watching the light through a pair of glacier glasses). Mine is out of warranty, Epson quoted me £130 for service, but when I asked them to arrange one, they went quiet, will start chasing it after the holidays again.

[Graeme Hamilton]

Make sure the calibration patches and the top section of glass is clean. If you look at a film holder, that top cut out lets the scanner calibrate itself. That area should be spotlessly clean, even when not using a holder.

[Sam L]

Unfortunately, the calibration area is clean top and bottom. I have had specs of dirt in the calibration area create white streaks down the scan before from calibration area issues but this is something else.

_tf_, good to know that they are not all like this. Mine is just out of warranty but I'll follow up with Epson after the holiday. When I look at the banks of LEDs they look about the same to me but the spacing does seem to correspond to the vertical bands on the scan.

Here's an interesting nuance: when I set Epson Scan 2 to "Transparency Unit" instead of "Transparency Unit (with Film Area Guide)," the scan area is cropped but the bands virtually disappear. Here's the same piece of fixed out film with "Dehaze" set to +70 in Lightroom to bring out the absolute worst in it:

222718

It looks almost perfect. I imagine that's why I never noticed until I started working with 8x10 film. I wonder what the difference is.

[Sam L]

Ah perhaps here is my answer:

https://www.largeformatphotography.info/forum/archive/index.php/t-107027.html

"These have 2 lenses - basically one for glass level scanning and the other (Hi Res) for the elevated film holders.

Unfortunately, the latter has reduced scanning area. It does not cover 8x10. So, by resting the 11x14 glass above the Scanner's glass whilst using the lower Res lens, you will not have the film in focus. Conversely, if you select the High Res lens, you will not cover the film size."

So something may still be wrong with my scanner but it is a different lens that scans the transparency unit, which explains why I don't see the issue with 4x5 and under.

These Epson products are so much trouble. This is my second V800 scanner (first replaced under warranty for a similar reason that did affect 4x5) and I'm also on my second P900 printer (first was impossible to calibrate the print heads).

[Alan Klein]

Are you using the film area guide that came with the scanner?
See pgs 30-33 in the Manual. https://files.support.epson.com/docid/cpd4/cpd41530.pdf

I don't understand why they say not to save as BMP, Tiff or JPEG. Maybe they meant that's what you should save them as. I don;t know the answer to that issue.

Placing Film in the Film Area Guide
You can scan oversized film up to 8 × 10 inch (203 × 254 mm) using the film area guide. Scan using
Epson Scan Professional Mode and select Film (with Film Area Guide) as the Document Type.
Caution: Hold only the edges of the film strips or use gloves, or you may damage the film.
Note: You cannot use DIGITAL ICE Technology when using the film area guide. You cannot use Full
Auto Mode or Home Mode when scanning film with the film area guide. If you scan 8 × 10 inch (203 ×
254 mm) film at 4800 dpi resolution and 24 bit color as the Image Type, do not save it in BMP, Tiff, or
JPEG format.

1. Open the scanner cover.
2. Remove the document mat from the scanner cover, if necessary.
3. Place the film area guide on the scanner glass and align it as shown.
4. Place the film in the film area guide with the shiny side facing down. Make sure the entire image
frame is positioned within the film area guide.
Note: The images and any wording on the film should appear backward. Do not cover the area
shown or colors in your scan may be affected.
5. Close the scanner cover gently to keep the film in place.

[Sam L]

Yes, I am using the film area guide and the calibration area is clear.

The copied instructions are old and/or for PC. On my mac with Epson Scan 2, there is no more professional mode, but some of the concepts still apply. For example, there is "Transparency Unit (with Film Area Guide)" instead of "Film (with Film Area Guide)." Here's what it looks like on a mac with the latest software:

222732

For these development technique experiments, I'm scanning in 16-bit grayscale at 300 DPI and saving as a TIFF, though I think the settings are irrelevant -- I can see the banding in the preview before the actual scan happens.

[Alan Klein]

Try 2400. 300 is too low. That could be causing the banding.

[Sam L]

An update: I contacted epson support and they confirmed it's a hardware failure. The bands correspond to the LED array in the transparency unit. Since the scanner is just over a year old, it is out of warranty. While doing this, I was reminded that I had previously returned the scanner to Epson for the same problem but more severe and affecting the high-res camera that works for 4x5 and smaller negatives. The replacement did fix that but apparently I never tested 8x10. Epson quality control ls Bad.

I also remember that if you contact them with a new scanner that has an issue, they will send you a refurbished scanner as a replacement.

Out of options with the V800, so I ordered a new V850 Pro from Amazon (easy to return). It is a lot better but still not perfect. Here is a scan of the same 8x10 transparency using the film area guide. As always, the calibration area is clean. I have boosted "dehaze" +50 in Lightroom to see the unevenness:
222908

Same transparency rotated 90 degrees in the scanner (notice my finger prints have moved to the top):
222910

In both there are dark bands throughout and lighter areas on the sides. I asked amazon for a refund and ordered another V850 Pro.

Same transparency, scanned with the second V850 Pro:
222911

This one is actually a little worse than the first V850. It has more pronounced banding and a distinct stripe down the center. The dark stripe looks like dust in the calibration area but the area is spotless and it persists after wiping top and bottom. I don't see any dust under the glass there (however there is a speck on the lower glass near the middle).

Four scanners (including the original), and none can scan 8x10 transparency evenly. I strongly suspect that none of the V800s or V850s can do this. Anyone willing to run the test on their scanner?

[Kiwi7475]

How much variability is there really in the banding? Like if you go to PS and hover the mouse over, how much variability is there in the K values (or RGB values)? I understand you’re emphasizing “artificially” the effect to show it here, but I was wondering how much is really the impact. If it’s like 2 or 3 grey levels it probably won’t ever show up in practical terms. I can’t imagine it’s that high and be a systematic issue, a lot of people would have reported it.

[Sam L]

For the old V800 scanner, using the eyedropper with a 5x5 average box on the unadjusted scan, I see about 9% difference between the light and dark areas. As you can see, it is really obvious in unadjusted photos. You can't have your sky looking like this:

222927

For the new V850 scanner, it is a lot better. I see 4% difference between light and dark areas when looking at the unadjusted original, which is tolerable as long as I don't use dehaze or clarity:

222925

By scanning the maximum available area of the 8x10 with the higher-res camera that crops the edges out, the variance drops to 2%, which is nearly undetectable. This explains why I never noticed before: within the area of a 4x5 negative, the variance is just 1% and likely below that when you just look at the sky area of a photo.

222926

The thing is that I do often boost the clouds with "dehaze" and sharpen with "clarity." I assume that's pretty normal in a hybrid film-digital workflow? Maybe people are doing DSLR scans for 8x10? I also can't quite fit the film borders in the scan area and I see a lot of film borders in other people's 8x10 scans. Maybe those are scans of contact prints?

[Kiwi7475]

For the old V800 scanner, using the eyedropper with a 5x5 average box on the unadjusted scan, I see about 9% difference between the light and dark areas. As you can see, it is really obvious in unadjusted photos. You can't have your sky looking like this:

222924

For the new V850 scanner, it is a lot better. I see 4% difference between light and dark areas when looking at the unadjusted original, which is tolerable as long as I don't use dehaze or clarity:

222925

By scanning the maximum available area of the 8x10 with the higher-res camera that crops the edges out, the variance drops to 2%, which is nearly undetectable. This explains why I never noticed before: within the area of a 4x5 negative, the variance is just 1% and likely below that when you just look at the sky area of a photo.

222926

The thing is that I do often boost the clouds with "dehaze" and sharpen with "clarity." I assume that's pretty normal in a hybrid film-digital workflow? Maybe people are doing DSLR scans for 8x10? I also can't quite fit the film borders in the scan area and I see a lot of film borders in other people's 8x10 scans. Maybe those are scans of contact prints?

I have to confess I never use the dehaze or the clarity. I feel they create halos and other artifacts that look unnatural to me. I mostly use curve adjustment and gradients and other minor adjustments such as contrast, etc. and local doge and burn. But I’m not saying you shouldn’t use them, only that I don’t.

I don’t have a fixed, unexposed 8x10 sheet to try it out on my v850 though, but I’ve never noticed something like this, perhaps I don’t do large enough adjustments to trigger them.

I mostly scan my 8x10’s with the “lower res” optics (just a ~15% loss), right on the scanner glass (with an ANR glass on top) using the area guide. You can definitely scan a bit more than the 8x10 sheet so you can get the holder’s cut-outs if you center the sheet precisely.

[Kiwi7475]

So I just took a new 8x10 x ray sheet and scanned it. Did it without fixing it but the emulsion is a “flat” middle grey (well, purple-ish but in b&w a nice mid gray), so it shouldn’t make a difference wrt assessing the banding issue.

I scanned it with Epson Scan making sure all adjustments were off, at 1200 dpi in 16 bit greyscale, as tiff.

There is vertical banding but it’s very small, the K values go from 47 to 50, or effectively a +/-1.5 %. It’s almost invisible on my screen, and with any information overlaid it wouldn’t show unless you’re just shooting against a perfectly white background (or grossly over-blown skies).

[Kiwi7475]

I also took the image and applied a 30 clarity and a 30 dehaze in camera raw filter in PS. It shifted the overall level but the banding remained quite the same, a +/-2 % from 56 to 60.

Of course these filters are probably non linear so one can’t make generalized statements, but it’s probably likely that with moderate clarity and dehaze adjustments it won’t also exaggerate the effect.

[Sam L]

Thanks Kiwi!

I had assumed that the x-ray film was more opaque than it is with the purple emulsion.

The 3% total variation of K with your scanner is 25% better than V850 I tested (4%). I'm surprised you didn't see the banding as more pronounced at +30 dehaze. I've been looking at +50 dehaze with no clarity adjustment, which is admittedly more than I would apply to a real photo. I imagine the V800 had real problems (9%) while the new V850s are just like this. 3% or 4% falls below the threshold that most people notice or care about.

I'll have to test some real photos with sky to see if I care.

[Gary Beasley]

I know astrophotographers do a darkframe shot used to subtract uneven pixel values from the images they shoot. Is this something that can be done with a scanner as well?

[domaz]

I know astrophotographers do a darkframe shot used to subtract uneven pixel values from the images they shoot. Is this something that can be done with a scanner as well?

In theory.. scan a piece of clear film than scan the actual film. It could work..

[Kiwi7475]

I know astrophotographers do a darkframe shot used to subtract uneven pixel values from the images they shoot. Is this something that can be done with a scanner as well?

You would need to never change the selection of the scan area, not by one pixel, or it would not cancel it by subtraction. It’s different from a background picture which always covers the entire sensor.

Or would would have to do a background scan prior to every single scan— gets a bit tedious (and you need to keep dust away from your reference background sheet…).

Also, we would need to make sure it’s perfectly repeatable between (even consecutive) scans. Small fluctuations in the mechanical alignment could shift the banding response.

[Gary Beasley]

This sounds like a feature that someone should build into the scanner driver, kind of like the IR scan to remove dust on many machines.

[Sam L]

I suspect it is a feature of the scanner -- specifically what it's doing with its calibration area. I just think it isn't very good at it over large areas. I'm sure it's a hard problem.

If you look at the light source in the transparency unit, you can see it's a double row of discrete LEDs. Since the light source is extremely uneven, the scanner must be doing some magic to adjust for the dark and light areas. Furthermore, LED intensity will change over time, so the scanner must be periodically adjusting. I bet this was less of an issue with the older scanners that used a tube for a light source.

One bit of corroborating evidence is that a speck of dust in the calibration area this will result in a white vertical line down the scan (if you are scanning a positive transparency). I imagine this to be akin to the scanner detecting a dimmer LED in that area and boosting the values illuminated by it as it travels down the length of the scan area.

There is some instruction on removing scanner artifact banding in photoshop on the web. I tried it but it does not work satisfactorily and it is really a hack to cover up bad hardware. It works by scanning a clear (but tinted) transparency and then scanning a negative without adjusting the scanning area. As above, the clear transparency will show the scanner's bias as dark and light bands. You can place the bias scan over the negative as a layer in photoshop. It will be aligned as long as you don't change the scanning area. Rather than subtracting it, it works better to invert it and use it as a semi-transparent layer with one of the various layering methods (e.g. multiply). Sounds great but it doesn't work very well. I didn't work on it very long but I could imagine heavily blurring the bias scan first and using it as a mask for an exposure adjustment layer might work better.

[Alan Klein]

I'm curious. If the 1-3% of banding occurs normally with this scanner, where else might we see this problem on normal shots?

[Sam L]

I'm curious. If the 1-3% of banding occurs normally with this scanner, where else might we see this problem on normal shots?

Well, technically it affects every scan of every negative. It's not a good choice for, say, archival preservation of large format negatives.

However, it is rare that you will see the unevenness in a normal photo. The exceptions seem to be truly empty sky or perfect light-toned untextured backgrounds. I first noticed this in the sky of a landscape on an overcast day and I see it in the test shots I use to evaluate development evenness. The latter are photos of a blank white evenly-lit stretched canvas that is out-of-focus. So perhaps you could not scan Avedon style portraits well with these scanners.

Some other options are sourcing an older scanner (speculatively), medium format DSLR "scans" (maybe good for 100MP), and drum scanning services for $$$$.