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

[warden]

I keep two Windows Vista machines to run Nikonscan.

Ah, I see. I'm out. It's hard enough for me to keep up with the modern machines!

I only tried Vuescan early on and conducted this test with Kodak 160VC that it has a specific film profile for. I also used the various modifiers from Vuescan compared to a fully automatic scan from Coolscan+Nikonscan.

Yes those presets are pretty useless on Vuescan in my experience. I just aim for a flat scan of the negative and invert for adjustments in Photoshop. I try to have the scanner software do as little as possible.

 

[Les Sarile]

Ah, I see. I'm out. It's hard enough for me to keep up with the modern machines!



Yes those presets are pretty useless on Vuescan in my experience. I just aim for a flat scan of the negative and invert for adjustments in Photoshop. I try to have the scanner software do as little as possible.

Just to be sure, those are with the specific Kodak 160VC film profile and each modifier.
I've scanned over 40K frames of various films with the Coolscan+Nikonscan in ful auto mode and not one goofy result.

Here's one from Coolscan+Nikonscan compared to Noritsu minilab of the same frame of Kodak Gold 100.

 

[138S]

Full res version of Fuji 100 scanned by Epson V700 and Coolscan 5000 http://www.fototime.com/535A4899F477014/orig.jpg

Yes... of course in this case the Epson works worse than the Nikon 5000, because 35mm is the format the Epson makes the worst and slide film is sharp. Instead with Portra and MF result would be mostly equal. The Epson shines the more as the format gets larger.

Probably more important to me is good workflow and color/contrast results in full auto mode. I've scanned tens of thousands of various frames of all types of films with the Coolscan+Nikonscan and only hundreds with the various other scanners. I've never had a bad result from the Coolscan but have encountered many just awful from the others such as this one from Kodak Ektar 100 below.

I also agree, sure that the nikon nails auto exposure. But Frontier and Noritsu have way better Pro image intelligence than the nikons.

Anyway a high quality edition should depart from taking all histogram in 16 bits/channel and manually adjusting curves.

So yes... the Coolscan works better for 35mm than the Epson, specially for slide film, for MF probably we have not much difference, and then the Epson has the advantage that makes 4x5", 5x7" and 8x10" very well while the 9000 only reaches MF.

Not a surprise that the Nikon 5000 is better for 35mm as the scan only covers 24mm when the Epson covers 149.9mm scan width, it's near a miracle that covering x6 more width results are those.



__________

Still older Nikons with too much colimated light were very noisy, one has to select the right Nikon.

 

[warden]

Just to be sure, those are with the specific Kodak 160VC film profile and each modifier.

Yes I understood what you were sharing there. Those presets aren't very useful in my opinion, as you've gone to some trouble to illustrate. Vuescan and other software can make quality color scans, but probably not if you stick to the presets. It does take some practice to achieve an appropriate workflow for your film but the software does allow the user to achieve contrast and appropriate color as a starting point for downstream adjustments.

I don't know what's up with that Noritsu scan, but of course it's awful, contrast and color both. I've never used that scanner or software but I assume it has some control of the output, right? I've used Nikon, Epson, Vuescan and even tried Silverfast once and have never had to settle for a starting point as bad as that Noritsu example. It's weird.

 

[Les Sarile]

I don't know what's up with that Noritsu scan, but of course it's awful, contrast and color both. I've never used that scanner or software but I assume it has some control of the output, right?

I have many examples of very poor results from every other scanner+software from just hundreds of tests with none from the Coolscan+Nikonscan.

 

[warden]

I have many examples of very poor results from every other scanner+software from just hundreds of tests with none from the Coolscan+Nikonscan.

Sounds like you’ve found your favorite. Enjoy.

 

[138S]

I have many examples of very poor results from every other scanner+software from just hundreds of tests with none from the Coolscan+Nikonscan.

Anyway, looking the sample you posted... (http://www.fototime.com/535A4899F477014/orig.jpg)

The 100% crops seen bellow shows in a 24" monitor like if that 35mm frame was enlarged to 1.6m high !!! this is a 44x enlargement.

A remarkable thing is that the raw Epson conversion is better as white walls are neutral in the direct sunlight and a bit blue in the shadows as in the nature happens, but Nikon 5000 left a remarkable cast.

Lets see how a regular 10x enlargement looks:

Well, at 10x we don't see much a difference, beyond V750 better conversion... Being the Epson an scanner that it also makes MF, 4x5", 5x7" and 8x10", doing 10x from 35mm like the 35mm dedicated Coolscan 5000 is a total success !!!

If you could post an MF sample of Nikon 9000 vs V700 you would see a close match even for 15x.

__________________

Regarding the Noritsu vs Coolscan side by side you posted, let me say that this not tells the truth about what is the general case. In General a Noritsu (or a Frontier "Image Intelligence") have automatic image enhancing features intended for high volume that blows away what the Nikon does. It is true that in some cases like with that shot a "blind" white balance can deliver such a pitfall that the Noritsu operator should correct with a single click, but you overlook that such level of automation found in the Noritsu also makes a superior job for most of the shots, saving a lot of work and delivering amazing prints. From the Coolscan you have a hard job to nail a good RA-4 print, while the Noritsu is able to print automaticly a lot of m2 of well crafted prints per hour.

 

[Les Sarile]

The 100% crops . . .

A remarkable thing is . . .

If you could post . . .

Regarding the Noritsu vs Coolscan side by side . . .

As I said, the Epsons do a good enough job generally speaking and provide a very good value. If you're happy with it then I would certainly not contest that.

The Coolscans were already a hard sell when they were new and many years later - and seemingly higher costs, it's even harder.

You've piqued my interest with the stated advantages of the Noritsu. Please post comparisons and I will as well. TIA.

 

[138S]

Please post comparisons and I will as well. TIA.

You may have many samples at home, all digital minilab RA-4 prints were mostly made by Frontiers or by Noritsus.

In the Digital minilab era most of the competition was in the automatic image enhancing software, this rised the level of general popular photography to levels never seen before.

As I said, the Epsons do a good enough job generally speaking and provide a very good value. If you're happy with it then I would certainly not contest that.

My view is that for 35mm film the best choice is a dedicated 35mm roll film scanner, also an X1 is impressive when wanting to depict grain structure of BW film beyond recorded image. For MF I'm neutral, many solutions will work fine, or better said will work mostly the same.

For 4x5" the V700 delivers 150 effective MPix, this is an insane Image Quality, but it's true that an expensive (4000 or 8000dpi) drum scan will extract more information, if the drum scan has to be 2000dpi then the V700 is better.

It's way cheaper shooting 5x7" and scanning with the Epson than shooting 4x5 and scanning with a drum, and probably the Epson 5x7 image will be better.

Anyway, personally, I'm one of those going back to the darkroom, with a mate, gathering efforts, we are assembling a relatively capable darkroom, aiming up to 47x35" optic prints (what our Drytac flattens) from 5x7" film.

In the future I'll be scanning to produce masks to assist optic printing

 

[Les Sarile]

You may have many samples at home, all digital minilab RA-4 prints were mostly made by Frontiers or by Noritsus.

In the Digital minilab era most of the competition was in the automatic image enhancing software, this rised the level of general popular photography to levels never seen before.

I have many scans from Agfa, Frontier and Noritsus. I acknowledge they can provide volumes of scans but I have no results as good as from the Coolscan - in terms of detail, color & contrast, and that's why I ask for your comparisons.

 

[138S]

I have many scans from Agfa, Frontier and Noritsus. I acknowledge they can provide volumes of scans but I have no results as good as from the Coolscan - in terms of detail, color & contrast, and that's why I ask for your comparisons.

Plase see this this www.sebastian-schlueter.com/blog/2018/3/7/the-magic-fuji-frontier-sp-3000

When I returned to film I got my film scanned in a frontier 2 years.

The Frontier is a high volume scanner taking the right dpi for the printing size, Frontier Image Intelligence feature produces adjustments fron advanced scene interpretation.

Image enhacement may detect gender, age, race, skin type, eye and hair color... Vegetation, flowers, water, sky, snow... Just to apply local adjustments...

Of course there are other commercial software doing that, but what frontier has is also the optimization for RA-4. IIRC not all RA-4 colors can be seen in the monitor.

 

[Marameo]

You may have many samples at home, all digital minilab RA-4 prints were mostly made by Frontiers or by Noritsus.

In the Digital minilab era most of the competition was in the automatic image enhancing software, this rised the level of general popular photography to levels never seen before.

Was Pakon scanner in the game as well?

 

[Helge]

"Real world" test using Fuji 100 color negative and scanned by Epson V700 and Coolscan 5000.

Full res version of Fuji 100 scanned by Epson V700 and Coolscan 5000 http://www.fototime.com/535A4899F477014/orig.jpg

This image perfectly illustrates one of my main gripes with the Epsons (and most consumer scanners really, to a lesser degree though).
Not only is the resolution much lower, than the actual resolving power of the film, which is expectable and something you could learn to live with.
But what is unacceptable, is the false grain structure known as grain aliasing, especially apparent in the sky on this image.

In the rest of the image, it masquerades as false detail., while less noticeable, still contributes very negatively to the overall feeling of the image.

It's a function of on one hand the sensor being too low resolution, and imposing that low resolution regular matrix, on the irregular grain of the film.
But also on the other hand, the lumps can be attributed to a lesser but still significant degree, to the optics being low resolution, and of generally bad quality.
That often results in detail at the edge of resolving power of some lens types, getting what can be described at boost in contrast. Not in a good way though.

This of course also transfers perfectly to any size of film you throw at the scanner since the grain structure and size is much the same.
So all that wonderful detail and smoothness in your medium format and large format shots is going to get tainted with excessive grain, long before the actual resolution limit of the film is reached.



On top of all of the above, I can't be the only one who see clear artefacts of the scanning action of the sensor in the Epsons in the final image. There are clear, what I can only describe as drag marks on the image, noise and other artifacting seems to get smeared across the frame.

 

[138S]

Was Pakon scanner in the game as well?

There is this interesting review:_ http://www.sebastian-schlueter.com/...-go-printing-scans-from-a-kodak-pakon-scanner

There are the normal annd the Plus version... the Plus scans higher dpi and it's faster for the same job.

I don't know all the Pakon details, but's also a high volume Pro scanner, making 200 to 500 frames per hour (depending on resolution) with ICE enabled. The Packon would deliver high MPix files, but every pixel in the image was meaningful.

The Frontier (etc) was not optimized to extract every bit of detail from the negative, but it was designed for taking from the negative the detail that could be printed by the minilab, but information in the image had little overhead, this is delivered information than was of high quality without dummy pixels.

Today we have computing performance in excess, but in the digital minilab era that was a bit scarce , so for high volume image had to contain the maximum information with the lower pixel count.

The Frontier has no linear sensor like flatbeds have... it has an area sensor like DSLRs. In those times the Fuji Frontier had "Pixel Shift", it sported a Piezo actuator for the area CCD displacement... In those kind of machines all was optimized to print decently dozens of m2 per hour and processing hundreds of frames with the least manpower possible, having to print hard all day long to pay the $250,000 it could cost (apply correction to today's $), one had to sell several $ millions in prints to earn the profit that would pay the machine and keep the company alive. So there was no room for dummy pixels in the image.

Today we can scan with a V700 at 6400dpi, 16 bits/channel, this ends in a 500MByte image for a 35mm frame (6400*6400*1.5*4*2) and later we can do pixel peeping in Ps with a PC sporting 16GB RAM and a M.2 disk... this is a game changer for the V700, that benefits from a processing massive files to extract every bit of useful information, 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.

 

[Lachlan Young]

This image perfectly illustrates one of my main gripes with the Epsons (and most consumer scanners really, to a lesser degree though).
Not only is the resolution much lower, than the actual resolving power of the film, which is expectable and something you could learn to live with.
But what is unacceptable, is the false grain structure known as grain aliasing, especially apparent in the sky on this image.

In the rest of the image, it masquerades as false detail., while less noticeable, still contributes very negatively to the overall feeling of the image.

It's a function of on one hand the sensor being too low resolution, and imposing that low resolution regular matrix, on the irregular grain of the film.
But also on the other hand, the lumps can be attributed to a lesser but still significant degree, to the optics being low resolution, and of generally bad quality.
That often results in detail at the edge of resolving power of some lens types, getting what can be described at boost in contrast. Not in a good way though.

This of course also transfers perfectly to any size of film you throw at the scanner since the grain structure and size is much the same.
So all that wonderful detail and smoothness in your medium format and large format shots is going to get tainted with excessive grain, long before the actual resolution limit of the film is reached.



On top of all of the above, I can't be the only one who see clear artefacts of the scanning action of the sensor in the Epsons in the final image. There are clear, what I can only describe as drag marks on the image, noise and other artifacting seems to get smeared across the frame.

You've hit the nail on the head.

 

[138S]

with the Epsons .....
But what is unacceptable, is the false grain structure known as grain aliasing, especially apparent in the sky on this image.

Helge, this is totally wrong, you may find other defects in the Epson images but you won't never see aliasing.

First, please review what aliasing is: https://en.wikipedia.org/wiki/Aliasing

Then consider that the Epson is an optically limited system, as it is a 40k pixels sensor covering 5.9" the sampling outresolves the optics by a 2.2x factor, and in that situation aliasing simply is not possible.

Consider what hapens a in a DSLR incorporating a LPOF, in the V700 case the optics works like a ultra strong LPOF cancelling any aliasing.

if you want we may review the math suggested in the wiki article.

especially apparent in the sky on this image.

Please review the #132.

At 10x the noise in the Epson scan is not seen at all, and beyond 10x the image is soft anyway because film/optics/etc so that noise is only seen in when pixel peeping at not useful image enlargements, so that noise is totally acceptable.

One question... why do you hate the Epson in that way ?

 

[Lachlan Young]

It may not be aliasing, rather it may be compression artefacting from on-board compression in the scanner hardware or shortcomings in the bit depth of the Analogue Digital Converter - which is then all then dumped into a 16-bit tiff container. You have to remember that the usual test threshold that mattered in the mid-90's was making a 250% enlargement at 175lpi - ie 2.5*350ppi = 875ppi, normally from a well exposed 4x5 transparency of very fine grain. At those specifications, compression artefacting etc would be less noticeable because the file at those settings has very little visible granularity - this is the same reason why Netflix etc can claim to stream 4K, but struggles so badly with grainy film and intensely busy action sequences.

 

[Helge]

Helge, this is totally wrong, you may find other defects in the Epson images but you won't never see aliasing.

First, please review what aliasing is: https://en.wikipedia.org/wiki/Aliasing

Then consider that the Epson is an optically limited system, as it is a 40k pixels sensor covering 5.9" the sampling outresolves the optics by a 2.2x factor, and in that situation aliasing simply is not possible.

Consider what hapens a in a DSLR incorporating a LPOF, in the V700 case the optics works like a ultra strong LPOF cancelling any aliasing.

if you want we may review the math suggested in the wiki article.




Please review the #132.

At 10x the noise in the Epson scan is not seen at all, and beyond 10x the image is soft anyway because film/optics/etc so that noise is only seen in when pixel peeping at not useful image enlargements, so that noise is totally acceptable.

One question... why do you hate the Epson in that way ?

I'm perfectly aware of what aliasing is and it's definition.
You might want to look up grain aliasing (<click).
It's an artifact in the same family as interference patterns.

You can't discount the poor optics either, or use them as a proxy for a low pass filter.
Bad optics often come at a price higher than just a lack of resolution and digitally correctable apparitions.
Some of the high frequency information still makes it through the lens. If you are lucky in attenuated/low contrast form (still enough to cause GA), at other frequencies in accentuated untrue form.
You just have to look at the image in the example above, or the many examples in the grain aliasing link, to see this perfectly illustrated.

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.

 

[Helge]

It may not be aliasing, rather it may be compression artefacting from on-board compression in the scanner hardware or shortcomings in the bit depth of the Analogue Digital Converter - which is then all then dumped into a 16-bit tiff container. You have to remember that the usual test threshold that mattered in the mid-90's was making a 250% enlargement at 175lpi - ie 2.5*350ppi = 875ppi, normally from a well exposed 4x5 transparency of very fine grain. At those specifications, compression artefacting etc would be less noticeable because the file at those settings has very little visible granularity - this is the same reason why Netflix etc can claim to stream 4K, but struggles so badly with grainy film and intensely busy action sequences.

Well Epson claims 48 bit RGB AFAIR. This is probably true, though I'm almost certain it's way overkill for the sensors actual range.
If you can get companding artifacts from that bit depth, you are FUBAR as an engineer.
It doesn't look like that kind of thing either.
As said, it's striations and lumps.

 

[Lachlan Young]

Well Epson claims 48 bit RGB AFAIR. This is probably true, though I'm almost certain it's way overkill for the sensors actual range.
If you can get companding artifacts from that bit depth, you are FUBAR as an engineer.
It doesn't look like that kind of thing either.
As said, it's striations and lumps.

There's very much some nasty aliasing going on - and what you say about the use of slightly unsharp optics (intentional use of diffraction?) as a de facto low pass filter with the idea of re-sharpening later does make sense.

 

[warden]

This image perfectly illustrates one of my main gripes with the Epsons (and most consumer scanners really, to a lesser degree though).
Not only is the resolution much lower, than the actual resolving power of the film, which is expectable and something you could learn to live with.
But what is unacceptable, is the false grain structure known as grain aliasing, especially apparent in the sky on this image.

In the rest of the image, it masquerades as false detail., while less noticeable, still contributes very negatively to the overall feeling of the image.
.....

The image you're referencing shows two things that I'd never try with an Epson: scanning 35mm and scanning with the Epson at 6400dpi, both bad ideas with the Epson as far as I have read, so if the point of the image is to show using probably the wrong tool for the job and using it badly, well job done. (No offense intended to Les, I think he's just maxing out the machines and showing us what it looks like)

I agree with your point about most consumer scanners showing this behavior and other problems too eventually if you stress them enough.

My question for the scanner experts follows. Take a medium format image, say 6x6 Provia. Scan it like an adult would, after reading the instructions for both the scanner and software and using appropriate settings for the best quality and resolution for each scanner. (In other words don't hobble your scanner by setting it for 10,000dpi if it can only deliver quality at 2,500.) Print the results using the best printer on the best honest to goodness paper. At what size will the quality gap appear on paper between scanners? Specifically I'm referring to normal viewing distances, with the naked eye. No magnifying, No loupes, no friggin' microscopes, just be normal and look at the print.

My gut tells me that up to 15X15" it doesn't matter what scanner you use for medium format because the prints will all be grainless, sharp, and very nice indeed. But maybe I'm wrong. What do the experts think? Could you see the difference at 5x5"? 10x10? 20x20?

 

[Les Sarile]

Regarding the artifacting in my shot of the Excalibur building, it is likely jpeg/compression artifacting and not intended 1X magnification viewing only. Apologies for not posting this disclaimer.

 

[Helge]

Regarding the artifacting in my shot of the Excalibur building, it is likely jpeg/compression artifacting and not intended 1X magnification viewing only. Apologies for not posting this disclaimer.

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

 

[Helge]

The image you're referencing shows two things that I'd never try with an Epson: scanning 35mm and scanning with the Epson at 6400dpi, both bad ideas with the Epson as far as I have read, so if the point of the image is to show using probably the wrong tool for the job and using it badly, well job done. (No offense intended to Les, I think he's just maxing out the machines and showing us what it looks like)

I agree with your point about most consumer scanners showing this behavior and other problems too eventually if you stress them enough.

My question for the scanner experts follows. Take a medium format image, say 6x6 Provia. Scan it like an adult would, after reading the instructions for both the scanner and software and using appropriate settings for the best quality and resolution for each scanner. (In other words don't hobble your scanner by setting it for 10,000dpi if it can only deliver quality at 2,500.) Print the results using the best printer on the best honest to goodness paper. At what size will the quality gap appear on paper between scanners? Specifically I'm referring to normal viewing distances, with the naked eye. No magnifying, No loupes, no friggin' microscopes, just be normal and look at the print.

My gut tells me that up to 15X15" it doesn't matter what scanner you use for medium format because the prints will all be grainless, sharp, and very nice indeed. But maybe I'm wrong. What do the experts think? Could you see the difference at 5x5"? 10x10? 20x20?

The lower spec 500 shows the same artifacts. The grain aliasing would be the same for any format.
It's problematic because it makes medium format look like it has more grain than 35mm. It also makes sharpening and other image editing very troublesome.

 

[Lachlan Young]

My gut tells me that up to 15X15" it doesn't matter what scanner you use for medium format because the prints will all be grainless, sharp, and very nice indeed. But maybe I'm wrong. What do the experts think? Could you see the difference at 5x5"? 10x10? 20x20?

Bitter experience tells me that you might get what looks like a nice enough 10x10" - until you look at it beside one from a higher end scanner (not a coolscan etc). It's most frustrating when you have a long printing job ahead anyway & you have to add in re-scanning negatives rather than getting on with running what were supposedly supplied as print-ready files.

It also makes sharpening and other image editing very troublesome.

Yes, very much so!