From Youtube: Drum Scan vs DSLR vs Epson (Via Nick Carver)

Les Sarile · Feb 21, 2020

Helge said:
It doesn't at all look like JPEG artifacts (why would you compress the hell out of a scanned image anyway?) and should be the same for the Coolscan image, unless you used very different compression ratios.

Filesize for web. Besides, it's "real world" not pixel peeping evaluation.

138S · Feb 21, 2020

Helge said:
You can't discount the poor optics either, or use them as a proxy for a low pass filter.

Why do you say that the V700 has a poor optics ?

The "higher resolution" lens resolves 2900 dpi effective horizontally over a 5.9" scan width, this is 17.110 pixels. You find not many lenses around having the capability to resolve 17000 pix in a row !!!

People stating that the Epson has bad lenses don't know much about glass or about scanners... 170002 is 300MPix in a 5.9" square frame, I can tell you that to resolve that one should use really fine glass working at optimal magnification for the glass.

Helge said:
It's an artifact in the same family as interference patterns.

Sorry, but you are a bit lost... the noise you see in the V700 scans at 6400dpi is mostly from vibrations in the vertical displacement of the carriage, combined with the dual row arrangement in the sensor. Any expert in scanning identifies that at first sight.

No aliasing: zero aliasing in the Epson.

Helge said:
Discounting field curvature

What field curvature?

The V700 lens is optimized for extreme flat field, curvature can be in the film, but not in the lens field. The new ANR holders do ensure film flatness if being careful.

A problem in the Epson is that it covers 5.9" with the lens, even when scanning a single 35mm strip the lens+sensor cover the 4 strips, so lens is x4 times better that what you see.

For this reason the Epson stars for MF and up, being decent for 35mm.

Other scanners have a "zoom system", the Epson avoids the zoom, but the lens works pretty well at it's optimal magnification.

Lenses in the Epson are extremly good, it delivers not more effective dpi performance for 35mm because covering 150mm when scanning a 24mm wide frame, not because the lens has any flaw.

Helge · Feb 21, 2020

Les Sarile said:
Filesize for web. Besides, it's "real world" not pixel peeping evaluation.

Well, you are doing quite a bit of pixel peeping in your test already. And there is quite bit if difference between the Coolscan and the Epson.
No one is saying anything about webuse where you're at most going to really ever "need" 8 MP.
It's always about print, and printing medium or large at that.

Les Sarile · Feb 21, 2020

Helge said:
Well, you are doing quite a bit of pixel peeping in your test already. And there is quite bit if difference between the Coolscan and the Epson.
No one is saying anything about webuse where you're at most going to really ever "need" 8 MP.
It's always about print, and printing medium or large at that.

Again apologies but I'm sure I have some more appropriate comparisons with regards to aliasing that I'll have to look for if there is more interest.

Marameo · Feb 21, 2020

138S said:
[...] in the past a Pro scanner required small files containing no dummy pixels, this is a mission the Pakon, the Nikon and Flextights did very well.

I wonder how scanners such as the Pakons/Frontiers color correct the new Portra 400 since the they are older than the film.

If I target the Pakon should I be better picking the Fuji 400H?

Thanks

138S · Feb 21, 2020

Marameo said:
I wonder how scanners such as the Pakons/Frontiers color correct the new Portra 400 since the they are older than the film.

If I target the Pakon should I be better picking the Fuji 400H?

Thanks

Today we have new tools: 3D LUT Creator, or equivalent.

Let me make you a recommendation, just totally control your inversions (and many other things) by mastering powerful color management tools. If you are concerned about color management and you have a refined aesthetic criterion this will empower you a lot.

You may use this technique for what you point:

Helge · Feb 21, 2020

138S said:
Why do you say that the V700 has a poor optics ?

The "higher resolution" lens resolves 2900 dpi effective horizontally over a 5.9" scan width, this is 17.110 pixels. You find not many lenses around having the capability to resolve 17000 pix in a row !!!

People stating that the Epson has bad lenses don't know much about glass or about scanners... 170002 is 300MPix in a 5.9" square frame, I can tell you that to resolve that one should use really fine glass working at optimal magnification for the glass.

LOL, Sorry but you are a bit lost... the noise you see in the V700 scans at 6400dpi is mostly from vibrations in the vertical displacement of the carriage, combined with the dual row arrangement in the sensor. Any expert in scanning identifies that at first sight.

No aliasing: zero aliasing in the Epson.

What field curvature?

The V700 lens is optimized for extreme flat field, curvature can be in the film, but not in the lens field. The new ANR holders do ensure film flatness if being careful.

A problem in the Epson is that it covers 5.9" with the lens, even when scanning a single 35mm strip the lens+sensor cover the 4 strips, so lens is x4 times better that what you see.

For this reason the Epson stars for MF and up, being decent for 35mm.

Other scanners have a "zoom system", the Epson avoids the zoom, but the lens works pretty well at it's optimal magnification.

Lenses in the Epson are extremly good, it delivers not more effective dpi performance for 35mm because covering 150mm when scanning a 24mm wide frame, not because the lens has any flaw.

Extremely good is very subjective. Can we agree that it's at least worse than the one in most dedicated scanners? Even with perfect focus, it still has a lot of space to cover up to the nearest Nikon or Minolta scanners lenses. And those scanners aren't even that good.

You can say that as much as you want, but there is still amble, and in my eyes irrefutable, evidence of GA in all Epson scans.
Whether it is by vibration and displacement of he carrier, or some other cause (which it looks like to me) makes little difference if that is an unavoidable disadvantage of scanning on the Epson.

57 lppm is pretty piss poor for a lens made to do one thing at a limited aperture with known backlight.
AFAIR they are even cheapening out on coating on the "budget version" of the scanner.

I was talking about my hypothetical phone scanning example.
Why should they avoid zoom, when it is clearly needed and very beneficial for the type of film ninety percent (probably more really) of their customer base is going to use‽
Tells you something about the aspirations and ambitions of the people behind it, I think.

Helge · Feb 21, 2020

Les Sarile said:
Again apologies but I'm sure I have some more appropriate comparisons with regards to aliasing that I'll have to look for if there is more interest.

Sure, if it's not too much bother.
But it's not like there is not a lot of examples out there.

138S · Feb 21, 2020

Helge said:
Extremely good is very subjective. Can we agree that they are at least worse than the one in most dedicated scanners? Even with perfect focus it. still has a lot of space to cover up to the nearest Nikon or Minolta scanners lenses. And those scanners aren't even that good.

17000 pixels effective resolved in a 5.9" row of the V700 is not subjective, this IS objective technical information. The nikon 5000 resolves only 3650 efective pixels in the scanned row. Still the 5000 haS a superior dpi effective because the row is 1/6 of the Epson row size.

At 17000 pixels per row the Epson lens can only be very extremly good, if you don't understand that...

Helge said:
You can say that as much as you want, but there is still amble, and in my eyes irrefutable, evidence of GA in all Epson scans.
Whether it is by vibration and displacement of he carrier, or some other cause (which it looks like to me) makes little difference if that is an unavoidable disadvantage of scanning on the Epson.

The question is at what enlargement you see that noise in a print, for sure at 10x you don't see that noise at all. Of couse the Epson is worse for 35mm than a Coolscan for 30x but absolutely not worse at x10.

Anyway we cannot speak about aliasing when there is no aliasing, but another thing, because such a coarse mistake would discredit what we say.

Helge said:
57 lppm is pretty piss poor for a lens made to do one thing at a limited aperture with known backlight.

Do you thing film resolves more ?

TMX (a sharp film) resolves 200lp/mm in 1000:1 contrast but around 60 lp/mm only at low contrast.

1000:1 are 10 stops difference between white an black bars, a situation you never find in real photography at 200cycles/mm on film, at the end system resolution (film+lens+shake+DOF+etc) usually resolves no more than the Epson, only special shots in the optic lab may resolve more in the negative.

Helge said:
AFAIR they are even cheapening out on coating on the "budget version" of the scanner.

This is about product segmentation, but general impression is that the coating of the V750/V850 does nothing most of the times.

Helge said:
I was talking about my hypothetical phone scanning example.
Why should they avoid zoom, when it is clearly needed and very beneficial for the type of film ninety percent of their customers are going to use‽
Tells you something about the aspirations and ambitions of the people behind it, I think.

Not hypotetical, me I've tested that 9000dpi effective can be obtained with a DSLR and a reversed lens, the question is how many shots would benefit from (say) better than 3000 effective dpi scanning perfomance. Not many in 35mm ! and almost none MF and up.

Single reason I find to go that high is grain structure depiction...

Helge · Feb 21, 2020

It's well known fact that you can "get away" with a worse lens the larger the format is. For example a stopped down triplet in front of a 4x5 piece of film, will still kill a Double Gauss of the highest quality on a 135 camera WRT resolution.
Nothing new in that.
You seem caught in a loop of repetition and truisms.
Less resolution is less resolution.
If you want resolution, the Epson is not where to get it.
And stop comparing to the Flextight. That scanner brand was never optimal for either 135 and esp, not for large format.
Not terrible, but not optimal either.

The fifty percent MTF limit is for lenses and was something agreed upon with the advent of serious digital cameras.
Have you ever seen the MTF curve for a sensor?
Film was always meant to have contrast and microcontrast controlled in printing and now when scanned too. This will bring the the low contrast detail up.
Somewhat the way big hexerei is performed to the raw Bayer image off the sensor.

The way we as humans perceive contrast and detail depends very much on how large the print is too.

ABX tests have show that in large prints, practically everyone prefer the subtle degradation of detail with film, rather that the sudden cliffside falloff into mush or pixels with digital.

And there clearly is grain aliasing. What else would you attribute the pseudo grain to?

If 135 benefits from DSLR scanning medium format does too. Simple as that.
You might not get quite a resolution increase precisely equivalent to larger area of film, but it's damn close. And the grain is the same, so to avoid GA you need the resolution anyway.
Many experiments have shown that there is new detail to be had from a 135 piece film up to about 80 MP, and with stuff like ADOX CMS 20 even higher.
It all depends on the lens, the precision in focus and if a tripod was used of course.
The same goes for digital of course.
My guess would be even with VR/IBIS and autofocus, that ninety percent or more of all digital shots are not resolving to the full potential of the sensor.
Difference is that with film, you are punished for not scanning at the highest possible resolution.

Lachlan Young · Feb 21, 2020

138S said:
The question is at what enlargement you see that noise in a print, for sure at 10x you don't see that noise at all.

Only if you've sandpapered your corneas. It's horrendously obvious if even the slightest sharpening is applied.

Helge said:
Have you ever seen the MTF curve for a sensor?

For your amusement - see pg. 123 specifically. A collapse to 50% MTF response at only 650ppi is not good by any metric.

Shows why the Epson fails so badly. And why the Coolscans aren't great either.

Helge · Feb 21, 2020

Lachlan Young said:
For your amusement - see pg. 123 specifically. A collapse to 50% MTF response at only 650ppi is not good by any metric.

Shows why the Epson fails so badly. And why the Coolscans aren't great either.

What a brilliant find! This paper seems like it was made to shoot down 138S (sorry dude, I hope you know this isn't personal. I like your other posts).
Practically everything discussed in this thread is beautifully illustrated here, plus more.
On top of that, they actually took the damn thing apart and exposed some of the "specmanship", as they call it, right down to measuring the sensor and the coverage by the lenses.

I sure as hell hope Harvard College Observatory didn't throw away their astronomical plates and film back then, because they thought they "had captured all the data that was there".
Not if the papers authors had their way, anyhow.

Lachlan Young · Feb 21, 2020

Helge said:
What a brilliant find! This paper seems like it was made to shoot down 138S (sorry dude, I hope you know this isn't personal. I like your other posts).
Practically everything discussed in this thread is beautifully illustrated here, plus more.
On top of that, they actually took the damn thing apart and exposed some of the "specmanship", as they call it, right down to measuring the sensor and the coverage by the lenses.

I sure as hell hope they didn't throw away their astronomical plates and film back then, because they thought they "had captured all the data that was there".

A while ago, I ended up manually effectively replicating the Epson sensor effect by taking two slightly offset native 1200ppi scans of a frame of 120 on a Flextight, then essentially auto-aligning & doing an up-res on them & then blending, much as the Epson appears to do. I then made a native 2400ppi resolution scan on the Flextight and compared the up-res scan to it. The upshot was that the 'modelled Epson scan' was potentially quite high resolving in high contrast areas, and a bit higher resolving overall than the 1200ppi scan, but also noticeably less 'sharp' than the 2400ppi scan & had acquired some odd artefacts along the way. I do still have the files, but the question is which hard drive they're on...

The major headache on the Flextights is the software, especially the onboard inversions. Solve that, & you find that most of the high end CCD scanners are all pretty much in the same ballpark, apart from absolute maximum resolutions and which ones cover better than 4x5. A bit of stitching in PS lets you get the maximum resolution of the Flextight on 120 as well - and with a bit of care, potentially 3200ppi on 4x5...

Helge · Feb 21, 2020

Lachlan Young said:
A while ago, I ended up manually effectively replicating the Epson sensor effect by taking two slightly offset native 1200ppi scans of a frame of 120 on a Flextight, then essentially auto-aligning & doing an up-res on them & then blending, much as the Epson appears to do. I then made a native 2400ppi resolution scan on the Flextight and compared the up-res scan to it. The upshot was that the 'modelled Epson scan' was potentially quite high resolving in high contrast areas, and a bit higher resolving overall than the 1200ppi scan, but also noticeably less 'sharp' than the 2400ppi scan & had acquired some odd artefacts along the way. I do still have the files, but the question is which hard drive they're on...

The major headache on the Flextights is the software, especially the onboard inversions. Solve that, & you find that most of the high end CCD scanners are all pretty much in the same ballpark, apart from absolute maximum resolutions and which ones cover better than 4x5. A bit of stitching in PS lets you get the maximum resolution of the Flextight on 120 as well - and with a bit of care, potentially 3200ppi on 4x5...

Interesting story. Strange things can happen when you start to "get clever" in the frequency domain.
Do you have a theory about exactly what happens?
The Flextight I once had access to, was definitely best with 120.
Seems like the whole basic principle of the machine was designed to support pro photo studios with need for "rapid" access to fast, high quality 120 scans.

Les Sarile · Feb 21, 2020

Helge said:
Sure, if it's not too much bother.
But it's not like there is not a lot of examples out there.

I prefer to work with examples made by the participants to answer any questions.

warden · Feb 21, 2020

Helge said:
On top of that, they actually took the damn thing apart and exposed some of the "specmanship", as they call it,

I caught that term too and chuckled. It's the word of the day for sure.

Lachlan Young · Feb 21, 2020

Helge said:
Interesting story. Strange things can happen when you start to "get clever" in the frequency domain.
Do you have a theory about exactly what happens?

I might hypothesise that fractionally imprecise alignment has something to do with it not assembling adequately - if you consider that the Heidelberg Tango has a minimum aperture of only 10 microns (ie 2540ppi), yet delivers on average about twice that in useable resolution and has about 10000spi of 'resolution' available in stepping along its leadscrew, it has to be oversampling & assembling in a very controlled and precise manner - the Epson and similar very manual methods are attempting to do this, but not accurately enough at the pixel level. Perhaps. I don't really know why, all I know is it happens.

Helge said:
The Flextight I once had access to, was definitely best with 120.
Seems like the whole basic principle of the machine was designed to support pro photo studios with need for "rapid" access to fast, high quality 120 scans.

I think it was really aimed at publications wanting to take pre-press in-house - and those who needed to scan up to A2-A1 size print formats without the costs and mess of drum scanning. Not necessarily intended for high industrial throughput in a pre-press house (unlike a drum scanner), but capable of delivering similar quality for 4-colour or various digital outputs. I've found them pretty consistent across formats, though a lot depends on model and how well looked after they are. In general 120 is a good compromise format - the image is big enough for the designed-in edge effects in the film emulsions to kick in, but the lenses are often as good and sharp and contrasty as some of the better 35mm glass, and the sheer surface area of film has a granularity benefit without the sharpness robbing unflatness of bigger formats - and I think these all play together to make a decent scan from 120 seem subjectively 'better' than bigger or smaller formats. The Flextights can be pretty ruthless in showing up optical/ sharpness flaws in camera lenses to a greater extent than you might see in a moderately sized darkroom print.

138S · Feb 22, 2020

Helge said:
It's well known fact that you can "get away" with a worse lens the larger the format is.

Of course, but a lens resolving 300MPix effective in a 5.9" frame, in macro conditions, this is a fine lens working at optimal magnification, no doubt. I'm used to check performance of MF and LF lenses in macro conditions, belive me, a 300MPix in 5.9" projection is a really fine lens.

A lens of this quality would cost a lot if sold as retail photographic grear, try to buy two new lenses like those in a V700 and you will see that cost can be several times the price of a V850, but you know... industrial components mass produced have a 1/15 of the retail price or less.

Several false staments and lies have been thrown by Epson haters about the V700 glass, some were saying that lenses were bad plastic... At first I believed this lie, but when I saw that it optically resolved 2900pix over a 150mm field then I found that it was a coarse lie.

138S · Feb 22, 2020

Lachlan Young said:
Only if you've sandpapered your corneas. It's horrendously obvious if even the slightest sharpening is applied.

Shows why the Epson fails so badly. And why the Coolscans aren't great either.

It looks that you may need some teaching about sharpening

(please take this with some humor)

Let me show you an easy recipe, for Portra scanned in the Epson at 6400dpi just apply a 2.5pix radius 50% sharpening and you get the same than a Drum scan or Creo high end flatbeds:

Before:
https://www.largeformatphotography....rum-Scanners&p=1479176&viewfull=1#post1479176

After (2.5pix radius 50%):
https://www.largeformatphotography....rum-Scanners&p=1479178&viewfull=1#post1479178

No overshot, no problem. Matching results. Note that the other scans came from two Creo and a Scanmate 11000 drum

Don't tell me that you are not able to do that in Ps !!!!! (also please take this with some humor)

Let me add that at lower dpi radius has to be lower, and other films (E-6, BW) require other settings and may be a bit more difficult to optimize, but please ask if you are not able, I'd be happy to show you how it can be done and what limitations the V850 has, those limitations are not a concern most of the times.

Lachlan Young said:
And why the Coolscans aren't great either.

Coolscans are very good, but you should learn that there are several generations, with different collimations (Callier) in the illumination, the ED version is fine.

______________

Disclaimer:

1) If I was professionally scanning for others all day long I would not use a V850, but a Pro scanner.

2) A home I obtain mostly Pro results with the V850, with some effort in the manual image optimization that Pro scanners do well automaticly but with the Epson you have to work in the straight image.

3) A good scanning Pro service with a good Professional operator makes wonders if he wants, he has an excellent machine and he masters powerful edition tools, and probably he has a refined aesthetic criterion.

albireo · Feb 22, 2020

Helge said:
Discounting field curvature and having an even, clean backlight and spot on focus, you could in theory get better results by just taking snaps of the film with your phone and stitching them.

Absolutely hilarious. This thread is the gift that keeps on giving.

Lachlan Young · Feb 22, 2020

138S said:
It looks that you may need some teaching about sharpening (please take this with some humor)

Let me show you an easy recipe, for Portra scanned in the Epson at 6400dpi just apply a 2.5pix radius 50% sharpening and you get the same than a Drum scan or Creo high end flatbeds:

Before:
https://www.largeformatphotography....rum-Scanners&p=1479176&viewfull=1#post1479176

After (2.5pix radius 50%):
https://www.largeformatphotography....rum-Scanners&p=1479178&viewfull=1#post1479178

No overshot, no problem. Matching results. Note that the other scans came from two Creo and a Scanmate 11000 drum

Don't tell me that you are not able to do that in Ps !!!!! (also please take this with some humor)

Let me add that at lower dpi radius has to be lower, and other films (E-6, BW) require other settings and may be a bit more difficult to optimize, but please ask if you are not able, I'd be happy to show you how it can be done and what limitations the V850 has, those limitations are not a concern most of the times.

Coolscans are very good, but you should learn that there are several generations, with different collimations (Callier) in the illumination, the ED version is fine.

______________

Disclaimer:

1) If I was professionally scanning for others all day long I would not use a V850, but a Pro scanner.

2) A home I obtain mostly Pro results with the V850, with some effort in the manual image optimization that Pro scanners do well automaticly but with the Epson you have to work in the straight image.

3) A good scanning Pro service with a good Professional operator makes wonders if he wants, he has an excellent machine and he masters powerful edition tools, and probably he has a refined aesthetic criterion.

Rather than trying to hide behind the most low level of unsharp masking techniques, how about reading that article I linked to & considering its implications for your claims?

138S · Feb 22, 2020

Lachlan Young said:
Rather than trying to hide behind the most low level of unsharp masking techniques, how about reading that article I linked to & considering its implications for your claims?

Lachlan:

Before:
https://www.largeformatphotography....rum-Scanners&p=1479176&viewfull=1#post1479176

After (2.5pix radius 50%):
https://www.largeformatphotography....rum-Scanners&p=1479178&viewfull=1#post1479178

These are facts, totally trustworthy evidence. No words necessary, just view images.

Any clue about why those image crops do match ?

Lachlan Young · Feb 22, 2020

138S said:
Lachlan:

Before:
https://www.largeformatphotography....rum-Scanners&p=1479176&viewfull=1#post1479176

After (2.5pix radius 50%):
https://www.largeformatphotography....rum-Scanners&p=1479178&viewfull=1#post1479178

These are facts, totally trustworthy evidence. No words necessary, just view images.

Any clue about why those image crops do match ?

So to be clear, you took Pali's screenshot of those 4 comparisons and sharpened it? You didn't download the original files and look at them?

138S · Feb 22, 2020

Lachlan Young said:
So to be clear, you took Pali's screenshot of those 4 comparisons and sharpened it? You didn't download the original files and look at them?

Have I to guess that you don't understand that side by side ?

Ok, see this one, perhaps there it's better explained: https://petapixel.com/2017/05/01/16000-photo-scanner-vs-500-scanner/

This is the full image

This is a crop of a 1m enlargement, so a 15x enlargement as you see it in the monitor from 6x6:

Yes... the X5 image is a bit better in that 15x enlargement side by side, but a mild sharpening in the Epson image made by me allows a Perfect match:

Try 0.1pix / 56% in the Epson image of the posted crops and you have the same, even at 15x. (Download crops from Petapixel web site)

Lachlan Young · Feb 22, 2020

138S said:
Have I to guess that you don't understand that side by side ?

I understand it perfectly. And the implications it makes about your technical abilities, honesty and understanding. Do you understand this?

Once again, Did you perform that operation on the screenshot or the original files? Do you understand what the implications are of not trying the sharpening operation on the original files?

From Youtube: Drum Scan vs DSLR vs Epson (Via Nick Carver)

Marameo

Privacy & Transparency

Privacy & Transparency