Features of a universal Filmscanner and why the Epson 700/800 has or has not.

[rawitz]

From many competent posts we know the highlights and limits of the Epson 700/800 scanner series.
So I dont want here more resolution- , filmholder- or else-tests but anwers to a simple general question:

What features should a universal (all film formats from 135,120 and sheetfilm) filmscanner have?

For me from the today marketed scanners the epon 700/800 series scanner is the only real candidat for that label,
but is not perfect. Why not? I my view a universal filmscanner should have these features:

1.FILMFORMAT from 135, 120 and sheetfilm up to 8x10.
I personally have to scan ULFilms but prefer stitching from 8x10 than going to a 30x40cm/DIN A3 scanner like the Epson 12000XL (which is a perfect device).

2.FILMHOLDERS that can hold the film flat.
The (newer) Epson AN glass-filmholder for me are perfect but should take larger filmstripes and maybe 5x7 film. The Wet-Mounting-Filmholder is a good addition.

3.If 2. is given for me no AUTOFOCUS is necessary.

4.Real net SCANRESULUTION of 3200 dpi for all filmformats out of the device.
I know the highend-filmscanners make up to 4000 dpi for 135 and 120 film, but not for sheet film.
And for sheetfilmers often 2000 dpi is enough.
But I want here an universal filmscanner: and (around) 3200 dpi would be good for all real world (not testcharts) filmformats.
I dont think that 4000 dpi is technical possible for 8x10`/20x30cm devices.

5. ICC DUST-REDUCTION technic YES.
Thats why I gave up my drumscanners for sheetfilm, because manual dustcleaning in post can takes hours if you want i perfect.

6.For SOFTWARE a "professional mode" that
does not cut off highlights and shadows for "crisper" scans, something like rawscans which are "softer and muddy" with much room in gradation and colors for the post.

7.48BIT SCANNING and output therefore would be fine, but does the epson really scan CCD 48bit intern? I dont know.


Second Part:
Why does the Epson 700/800 match or fail your expectations for universal filmscanner?

For me (and many other people) the most critical point is:
Why does the real/net resultion output of about 2400 dpi from filmholder (or 2200 dpi from groundglass) not match the technical given optical resultion of 6400 dpi or 4800 dpi. Is it really only the scan-optic or the whole scan-technic. I dont know ....

And my last question to the technical specialists and gurus:
Is there a real possibility, to technical upgrade the Epson 700/800 series to an universal scanner with the features we want here.
And will the price stay affordable?

regards
Rainer

[Alan Klein]

As an owner of a V850, Epson recently contacted me for a survey of it. I submitted some ideas to them that I thought would improve their scanners. So they are apparently investigating possible future designs and products. You ought to contact Epson and send them an email of some of the features you would find helpful in future scanners. Are you using an Epson now?

[bob carnie]

Hi Rainer- I do a lot of scanning for various purposes , some high and some low resolution depending on my needs at the time, I think the answer or the product will be a hybrid style of scanner that incorporates a very high resolution capture system combined with a backlight frontlight solution. I feel all the work done by many here in that direction is positive. We are now in the timeline where one can get 150mp sensors on cameras, at the moment they are really expensive but as a few years go down the line these prices on the used market will bring the price way down, I think CUSTOM KITS will be available allow one to put on a mirrorless high rez camera and give the worker the option of having a great camera as well a camera that utilizes its best features to do very fast , high quality scans of film. Right now the scanners I have take about 8-15 min to do very high rez scans, with these new systems it will be in the seconds which if one is scanning large groups of older film be quite rewarding. I too would love to see Epson up its game but if you look at it from a marketing perspective the return of their investment would be low as film production is low compared to a scant 25 years ago.
We have seen the 18 year 1mb or 2mb increase by the manufacturers , I now believe they are close to 4 x 5 - 8 x 10 quality with the phase one systems and thus we will see movement at the high end driving each year the price down so us mere mortals can afford one of these cameras.

I actually know of a young man working on this very solution , and I probably will buy my kit from him when he has finished his testing stage. It is only a matter of time now.



Bob

[Alan Klein]

Bob, What's a backlight-frontlight solution? Bedside resolution, the problem with flatbeds like my V600 and V850 is the dMax - you can't get through those darker shadow areas like in Velvia 50.

[rawitz]

Hi Alan,
of course I read your posts and agree with everything. The last years ago I had many discussions with the Epson folks about their scanners at photokina, but photokina is dead. I´m sitting in Germany and have no more contact to Epson. As you have it in US, of course I give you every permission to give them my informations and thoughts to your Epson contact.
And yes, I used drumscanners (Scanmate) since nearly 30 years (!!), I now also have a Nikon 9000 and know many midend-scanners from teaching my photografer customers. Today I use my 15 year old Epson V700 for my sheetfilm scans from 4x5 up to 10x15`as I told upside.
A "universal filmscanner" would be fine not only for me - I´m sure.

Hi Bob,
to the scanning approach with digital camera I did some trials too.
BUT: I nowadays do pics in 8x10`and bigger. My scans on the Epson 700 with 2400 dpi (net) give me 1500 MP RGB and more. I´m not young enough to wait for a digital camera like the PhaseOne 100 or 150 MP to match this from my Epson Scanner.
Not to discuss the expected price for a 500 MP PhaseOne. The Epson can do it much cheaper.

regards
Rainer

[jp]

I think a universal scanner is about as practical as a universal camera.
We need something quick and high quality (a step up from the Epsons, maybe coolscan quality) for scanning rolls of small and medium format film.
Then an upgraded Epson for medium and sheet films. If it must be universal, then a motor feed carrier for the Epson.

The Epsons could be improved by less plastic. A sheet metal construction would reduce static electricity greatly (and thus dust). I manage by keeping the scanner in a dust free darkroom, something most people can't do.

I don't have any problem with the resolution of the Epson for large format film. It would be nice to have more resolution AND speed for smaller film.

It would be nice to have a brighter light source either for more depth of field of focus, or more dynamic range in overly dense negatives.

[interneg]

For me (and many other people) the most critical point is:
Why does the real/net resultion output of about 2400 dpi from filmholder (or 2200 dpi from groundglass) not match the technical given optical resultion of 6400 dpi or 4800 dpi. Is it really only the scan-optic or the whole scan-technic. I dont know ....

It's nowhere near precise enough in construction - and the way it tries to squeeze resolution from the 1/2 pixel overlapping sensors & interpolating them together (not very well) will deliver a misleadingly high resolution (from an ultra high contrast chart) at very poor MTF. So, the 2200-2400ppi across the sensor is really only about 2x1200ppi + questionable interpolation. High MTF is critical to good scan results - and that overall transfer function is sensitive to both mechanical and optical components/ precision.

And my last question to the technical specialists and gurus:
Is there a real possibility, to technical upgrade the Epson 700/800 series to an universal scanner with the features we want here.
And will the price stay affordable?

If it was built to much higher standards and used leadscrew rather than belt drive systems - and an optical path that is designed for high MTF over absolute resolution, then it might be a potentially good machine. The available length of linear scan sensors is why most of the high end flatbeds resorted to XY scanning & stitching.

Could a better machine be built? Unquestionably, yes. Could it be built for less than USD 10,000 selling price, I don't know - maybe if you could sell a lot of them & leverage newer manufacturing techniques. If it could deliver 1500ppi across the bed & 3-4000ppi down a narrower strip (6000ppi could probably be done, if the thing was well enough designed) at USB 3 speed, then it's potentially very good. Better yet if it automatically complies with how colour neg is supposed to be 'seen' by the paper and how colour transparency is supposed to be viewed (easily done with appropriate RGB LED and engineering). Autofocus and reliable focus calibration should not be difficult to implement either - if it's designed to quality, not price.

[Joshua Dunn]

There will never be a one size fits all solution for high quality film scanning. However I do think we need more options than the market is willing to provide on it's own.

I am working on building a better solution for a scanner, although I don't see it being any faster, just better in resolution and bit depth. I think the real problem is what do you buy if you don't buy the Epson V850? If you buy the V850 it's about $1100 and that comes with Silverfast. But for those of us that want a step up above the V850, there are not a lot of options. I have looked into buying a drum scanner but that is even more impractical than the V850, however with a lot of effort, will give you better results. Realistically they are a lot of money on the used market for such old (although still viable) technology. I would like to see a option that is an easier to use than a drum scanner, but better quality than the Epson V850. The price point that I feel most prosumers would be comfortable with something like that could be around $2000.

-Joshua

[Bernard_L]

For me (and many other people) the most critical point is:
Why does the real/net resultion output of about 2400 dpi from filmholder (or 2200 dpi from groundglass) not match the technical given optical resultion of 6400 dpi or 4800 dpi. Is it really only the scan-optic or the whole scan-technic. I dont know ....

6400 dpi is not optical resolution; just marketing. That is obtained with high sampling (small steps) but because the optical resolution is 2200-2400, at 6400 successive samples are essentially repeats of the same data. In techspeak that is called oversampling.

Leaving marketing aside, the 2400dpi figure is obtained by scanning a resolution chart (there are other means, but let's keep it simple) and looking for the last (finest) patch with discernible lines. At that point, the contrast is almost zero. So even the technically correct figure of 2200-2400dpi does not match very well the perceived resolution. In techspeak one says that the modulation transfer function (MTF) drops to zero at the limit of resolution.

One way to improve the perceived resolution is sharpening (in techspeak MTF restoration). Meaning boost the response at the spatial frequencies below the resolution limit, where the contrast has dropped but not fallen to zero. Best done with unsharp masking (USM). Two parameters must be chosen properly: radius and amount. If done properly, a significant improvement in perceived shaprness is obtained. If overdone, it can be ugly: an image that "hurts the eyes", and, in extreme cases, bright rims around dark areas, and vice versa.

[rawitz]

Hi Joshua,
I fully agree! A better than Epson V850 scanner is not an idea of a small minority of highenders, but the wish of many people and for a reasonable price (about 2000$) it would be a market runner. I had two drumscanners and gave them away for nothing. SCSI bus is antique and runs on antique computers with antique windows/mac systems. Even my Nikon 9000 with firewire bus only runs on my old "highend" notebook. We are practising photographers and not spleeny antique-cars-collectors.

Hi Berhard,
We know that the 4800/6400 dpi optical resolution is sampled. The Epson technical specs say, the scan unit consists of 2x3 (RGB) scanlines with each of 20400 pixel. The second linetriple has an offset of 1/2 pixel to the first triple and with this "optical sampling" the 4800 dpi (groundglass) and 6400 dpi (filmholder) resolution is reached.
But even with only one scanline-triple without oversampling there is a resolution of 2400 and 3200 dpi (half of oversampled). Why dont we get it, for me 3200 dpi net output would be enough. Is it the inferior "high quality" lens or what else?

regards
Rainer