Scanning Resolution Question

[IanBarber]

Apart from producing a larger file and being able to produce a larger print, will scanning at a higher resolution give you more detail?

Example
If from a 4x5 negative, I want to create an inkjet print which is 16 x 20 inches then in theory in the scanner software, I could set the scan resolution to 1440 which would give me 16x20 inches if I sent the file to the printer at 360ppi

Question
Would I gain and more detail in the scan if I was to say scan it at a resolution of say 2400 and then down size in Photoshop when ready to print ?

[Jim Andrada]

The question is whether or not you'd gain VISIBLE detail in the final print. Overscanning and down-sampling primarily helps by averaging the noise in the scan giving a cleaner image, so It's generally a good thing to do, but it doesn't add any "detail".

Depending on the scanner, you might not get an actual 2400 even if that's what you set into the software - regardless of what the maker claims. What scanner do you have?

[interneg]

Apart from producing a larger file and being able to produce a larger print, will scanning at a higher resolution give you more detail?

Example
If from a 4x5 negative, I want to create an inkjet print which is 16 x 20 inches then in theory in the scanner software, I could set the scan resolution to 1440 which would give me 16x20 inches if I sent the file to the printer at 360ppi

Question
Would I gain and more detail in the scan if I was to say scan it at a resolution of say 2400 and then down size in Photoshop when ready to print ?

2000-2400ppi is arguably all you need from 4x5 - but that depends on you having a scanner that can deliver those pixels in a clean, sharp fashion without halation & other nasty micro-contrast destroying abberations. In other words, a 'higher' resolution scan from an Epson may look far worse than a 'lower' resolution scan from a drum scanner or high end flatbed. Anyway, scanning once at the highest resolution you'll potentially need is generally best practice - the less you handle the negative the better.

[IanBarber]

The question is whether or not you'd gain VISIBLE detail in the final print. Overscanning and down-sampling primarily helps by averaging the noise in the scan giving a cleaner image, so It's generally a good thing to do, but it doesn't add any "detail".

Depending on the scanner, you might not get an actual 2400 even if that's what you set into the software - regardless of what the maker claims. What scanner do you have?

The scanner is an Epson V800. I know people say 2400 is about its maximum from what I have read

[Pere Casals]

The scanner is an Epson V800. I know people say 2400 is about its maximum from what I have read


If scanning beyond 2800 (say 3000) you can get true 2800 Lp/mm from the V800 in one direction, and some 2300 in the other.

Please see http://archivehistory.jeksite.org/chapters/appendixc.htm


Be careful because the sharp look of your image depends a lot on your Photoshop process, not only from the scanner itself.

Edit/scan/sharp always in TIFF 16 bits and with image size larger that needed. Then save a personal copy in TIFF 16 bits. If you are edit again use this source

Then make the image to be posted.

> Resize the image to the final size, Image->Image size . This dialog allows you to selet the resize algorithm, (at the bottom) select "Bicubic, Ideal for reductions", this is critical !!!

> See if unsharp masking again helps for the final look,

[IanBarber]

If scanning beyond 2800 (say 3000) you can get true 2800 Lp/mm from the V800 in one direction, and some 2300 in the other.

Please see http://archivehistory.jeksite.org/chapters/appendixc.htm




Interesting reading, thanks.

[Taija71A]

If scanning beyond 2800 (say 3000) you can get true 2800 Lp/mm from the V800 in one direction, and some 2300 in the other.
Please see http://archivehistory.jeksite.org/chapters/appendixc.htm

? ? ?

http://archivehistory.jeksite.org/chapters/appendixc.htm#_Toc326155573

Table C.2. Epson V750 Scanner: Effective Resolutions for Various Scanner Resolution Settings.

Scanner Resolution Setting: 3000ppi.

Effective Resolution for Vertical Bars (Average of 5 Readings): 2255.
(*Percent Difference from Scanner Resolution Setting: -25%).

Effective Resolution for Horizontal Bars (Average of 5 Readings): 2148.
(*Percent Difference from Scanner Resolution Setting: -28%).

[Pere Casals]

? ? ?

http://archivehistory.jeksite.org/chapters/appendixc.htm#_Toc326155573

Table C.2. Epson V750 Scanner: Effective Resolutions for Various Scanner Resolution Settings.

Scanner Resolution Setting: 3000ppi.

Effective Resolution for Vertical Bars (Average of 5 Readings): 2255.
(*Percent Difference from Scanner Resolution Setting: -25%).

Effective Resolution for Horizontal Bars (Average of 5 Readings): 2148.
(*Percent Difference from Scanner Resolution Setting: -28%).


Sorry, 3000 should be 6400, this is the table:

166181


A 6400 scan can reach 2900 and 2300 , depending on the direction and the particular pass. This are max figures. A you will know in flatbeds film is not flat if not wet mounted. Then you have multipass...

Also this test is a bit subjective as any USAF 1951 test.

Anyway this resolving power is atonishing for 4x5, way, way more what a human eye can see in a full image. We can say that this is an obscene amount of image quality, without discussing much on technical test intrinsics.

But a monitor has 2mpix or 8 mpix, only that, so the way you reduce the image size will be important for a sharp look, much more than the scanner.


This is a 8x10, the bell size is like if it was in a 6m high print. Need more ??? I need way, way less !!!

https://www.flickr.com/photos/125592977@N05/32535835184/in/dateposted-public/

[IanBarber]

A 6400 scan can reach 2900 and 2300

Out of interest Pere, do you scan yours at 6400 knowing that in theory you will be achieving more in the region between 2300 and 2900

[Pere Casals]

Out of interest Pere, do you scan yours at 6400 knowing that in theory you will be achieving more in the region between 2300 and 2900

No... not necessary, only in the case I want an small crop.

Human eyes can see some 60 mpix if moving the eye around without moving the head, a 4k monitor has 8 mpix... so why scanning 500 mpix (from 8x10).

It is funny having such an obscene amount of image quality, I find it only useful to get fun with my digital mates... How many megapixels do you say ? :)


To me LF is great because process purity, movements, long focals... that obscene amount of image quality comes as a bonus. Well, not undesired at all... true, but this is way in excess. The sharp look comes little from that "in excess" image quality, and a lot from excellent subject, execellent light, and excellent photographer.

To me the target is to learn something from AA and Karsh...

I get a lot of fun by measuring Lp/mm from my glasses and scans, but I know very well that this is near the least important thing in LF. The important thing is having artistic ideas to aesthetically exploit the LF look, imho. Problem is that I've a low artistic profile, perhaps that I make too many LP/mm measurements :)

[IanBarber]

I get a lot of fun by measuring Lp/mm from my glasses and scans, but I know very well that this is near the least important thing in LF. The important thing is having artistic ideas to aesthetically exploit the LF look, imho. Problem is that I've a low artistic profile, perhaps that I make too much LP/mm measurements

You havent mentioned what scanner you use Piere, is it one of the Epson V series or something else.

If it's an Epson V series, apart from measuring Lp/mm have you spent any time experimenting with film to platen distance

[Pere Casals]

You havent mentioned what scanner you use Piere, is it one of the Epson V series or something else.

e


I use a V850, of course. IMHO the best choice for a not Pro.

Perhaps a pro may desire another device (or not), if one is scanning 8hours a day one may want something more expensive, but for the price the V800 is an atonishing machine.

See flickr... you will see true artists using it, some use a $3000 Universal Heliar lens, but a V700.

If one invests a lot of money one can get slightly better results with other machines but only in some circumstances, like velvia slides with deep shadows, but a true difference for that is a drum, or a Hassy X1.

For most shots results will depend much more on your post process skills than on scanner, this is including color shots.

If you have a V800/850 you have more than one can desire, really.




film flatness, scanner

This is important for 35mm film, you need to obtain best from that little frame. For LF it is mostly irrelevant, as one has an obscene amount of image quality, anyway. In some cases you may want a 4m high print that can be seen at reading distance with no flaw. In that case you need a good scan, but again postprocess may be more important than hardware (IMHO).

Exception is Velvia/provia with very deep & interesting shadows... there you need a drum.

[faberryman]

I use a V850, of course. IMHO the best choice for a not Pro.

Perhaps a pro may desire another device (or not), if one is scanning 8hours a day one may want something more expensive, but for the price the V800 is an atonishing machine.

See flickr... you will see true artists using it, some use a $3000 Universal Heliar lens, but a V700.

If one invests a lot of money one can get slightly better results with other machines but only in some circumstances, like velvia slides with deep shadows, but a true difference for that is a drum, or a Hassy X1.

For most shots results will depend much more on your post process skills than on scanner, this is including color shots.

If you have a V800/850 you have more than one can desire, really.

This is important for 35mm film, you need to obtain best from that little frame. For LF it is mostly irrelevant, as one has an obscene amount of image quality, anyway. In some cases you may want a 4m high print that can be seen at reading distance with no flaw. In that case you need a good scan, but again postprocess may be more important than hardware (IMHO).

Exception is Velvia/provia with very deep & interesting shadows... there you need a drum.
For 35mm, a scan resolution of 2300 is insufficient for anything but web images and small prints. 2300 will not resolve the grain. You are much better off with a dedicated film scanner at half the cost of the V850. The V850 is a fine scanner for large format.

[IanBarber]

For LF it is mostly irrelevant, as one has an obscene amount of image quality, anyway


For 35mm, a scan resolution of 2300 is insufficient for anything but web images and small prints. 2300 will not resolve the grain. You are much better off with a dedicated film scanner at half the cost of the V850.

What are your thoughts on say 6x6 negatives, would you say 2300 is sufficient for negatives of that size.

[faberryman]

What are your thoughts on say 6x6 negatives, would you say 2300 is sufficient for negatives of that size.
I haven't scanned medium format, so I can't speak from experience.

[Pere Casals]

For 35mm, a scan resolution of 2300 is insufficient for anything but web images and small prints. 2300 will not resolve the grain. You are much better off with a dedicated film scanner at half the cost of the V850. The V850 is a fine scanner for large format.

I agree, for 35mm it is better a dedicated roll film scanner.

Anyway with 2300 dpi optical performance you can get very decent 60x40 cm prints from 35 mm film. For some fllms grain is not an issue, TMX has little grain, an TX grain it is easily resolved.

But it is true that to get a decent scan from 35mm/V850 film flatness is way more important than with 4x5.


Also think that "web images" as a low requirement has changed. Today people starts having even 4k monitors an a fast internet connection. The term "web images" was coined in the 800x600 and 1024x768 era. Today monitors have 8x more pixels and connection speed is no problem.

[Pere Casals]

What are your thoughts on say 6x6 negatives, would you say 2300 is sufficient for negatives of that size.

This is a 35mm scan I made with v850 : https://www.flickr.com/photos/125592977@N05/29604172183/in/dateposted-public/ just click to enlarge to aparent 1m.

These are a 6x6 V700 scans

https://www.flickr.com/photos/55873497@N04/8238049780/in/album-72157631623052598/

https://www.flickr.com/photos/55873497@N04/9338477698/in/album-72157631623052598/

To me, more than enough.

I reiterate, post process is way more important than scanner, in most situations.

[interneg]

What matters is not the headline 'resolution' but the quality of that resolution. In other words, is it in focus, as free as possible from optical flaws, with adequate dmax for the media being scanned, etc? The Epson fails all too easily on those.

This is where the reliance on optical resolution charts falls down. There is a considerable difference between a USAF standard resolution target & a piece of BW or colour neg or transparency film in terms of what needs to be adequately imaged.

If you are considering dropping 3K on a lens, you'd be better off finding a used drum scanner with that cash.

[Pere Casals]

The Epson fails all too easily on those.

This is where the reliance on optical resolution charts falls down. There is a considerable difference between a USAF standard resolution target & a piece of BW or colour neg or transparency film in terms of what needs to be adequately imaged.

If you are considering dropping 3K on a lens, you'd be better off finding a used drum scanner with that cash.

People that had a drum and a V750 made 95% of the work with the Epson. For most shots it is simply not worth the drum required wet mounting for the results.

Density is not near a problem with BW and color negative film.

For slides one may use Multiexposure feature for deep shadows, but beyond 3.0D (and this is a lot) no flatbed do it well, real difference is a drum.


Drum vs V750 crops

http://www.simonkennedy.net/wp-content/uploads/2012/01/drum-scan-vs-epson-v750.jpg

At the level one can see a difference no enlargement is good...

[Jim Andrada]

I was using an Epson 750 and kept debating what to upgrade to. I looked into Drum scanners and finally wound up getting an IQsmart 2 flatbed. So far I'm completely satisfied with it. I use it for everything from Minox to 8 x 10. The workflow is vastly simpler and it can do a very credible job of scanning prints as well.

I also have a Coolscan 8000. It was quite nice for MF, but unfortunately it broke a plastic part and I haven't been able to find a replacement. Fortunately it failed shortly after I installed the IQsmart. However, the workflow with the IQsmart is easier that with the Nikon since you can put a whole roll of film on the bed, select the photos you want to scan, and let it run all night on its own.

[Pere Casals]

I was using an Epson 750 and kept debating what to upgrade to. I looked into Drum scanners and finally wound up getting an IQsmart 2 flatbed. So far I'm completely satisfied with it. I use it for everything from Minox to 8 x 10. The workflow is vastly simpler and it can do a very credible job of scanning prints as well.

I also have a Coolscan 8000. It was quite nice for MF, but unfortunately it broke a plastic part and I haven't been able to find a replacement. Fortunately it failed shortly after I installed the IQsmart. However, the workflow with the IQsmart is easier that with the Nikon since you can put a whole roll of film on the bed, select the photos you want to scan, and let it run all night on its own.

If I was a Pro I also would prefer the 70 lbs IQsmart 2, a tank, and for certain LF conditions it delivers an slightly better image than the cheap V850. But I don't think I would be able at all to know if a 1m BW print form 4x5 negative is from a V850 or IQ2. Probably I would instantly recognize if that print was processed by the skilled man or not.

[interneg]

People that had a drum and a V750 made 95% of the work with the Epson. For most shots it is simply not worth the drum required wet mounting for the results.

Density is not near a problem with BW and color negative film.

For slides one may use Multiexposure feature for deep shadows, but beyond 3.0D (and this is a lot) no flatbed do it well, real difference is a drum.


Drum vs V750 crops


At the level one can see a difference no enlargement is good...

You'll see the difference very clearly at a 3x (or less) enlargement in my experience - a 1m across print would be like night and day. Especially in the highlights that will not be horribly aliased/ burnt out & the non-mushy critical details. Inkjet or chromogenic output reveal these differences in a way that even quite high resolving monitors don't. There's some noticeable spherical aberration in that 35mm scan you posted earlier which I am fairly sure is not coming from the camera lens. Sharpening will not really solve those problems - it will however increase the obviousness of the aliased grain.

Realistically, Epsons struggle to reach a useful transmissive dmax & you can see the ugly effects in the highlights. In fact, the Epsons are bad enough that pretty much any 24x36 sensor DSLR from the last 5-10 years with a macro lens & a suitable light table will be a vastly better scanner - & with a bit of care/ stitching will get you comfortably into drum scan territory.

Also, I think you'd be surprised how cheap older high end pre-press flatbeds can go for today - sometimes barely even V850 money.

[Pere Casals]

Realistically, Epsons struggle to reach a useful transmissive dmax & you can see the ugly effects in the highlights.

Especially in the highlights that will not be horribly aliased/ burnt out & the non-mushy critical details.


Not at all, if one makes an automatic exposure with V850 it will clip the extremes to have good midtones, but you can take all histogram and then you can compress the range like you want.

Here you can see real densities (very dificult to print in a darkroom), and how well it worked the V750 I used then.

https://www.flickr.com/photos/125592977@N05/28693688313/in/dateposted-public/

Densities from BW or color negatives are not a challenge for the Epsons, at all !!! Do you make BW negatives beyond 3.0D ?

Even in that case you can use Multiexposure feature, in some bundled software you have to purchase the S.F. SE Plus Upgrade ($50) for it.



Inkjet or chromogenic output reveal these differences in a way that even quite high resolving monitors don't. .

This was in the past, today's monitors (and tablets, etc) have better dynamic range than paper, and with monitor you can enlarge a crop.





There's some noticeable spherical aberration in that 35mm scan you posted earlier which I am fairly sure is not coming from the camera lens.

It can come from the lens because it was shot wide open f/1.8 (nikon 50mm AFD). Poor f/1.8 Lp/mm performance (vs f/6.5 aprox peak performance) is due espherical aberration.

Here you have another 35mm from CMS 20 film. https://www.flickr.com/photos/125592977@N05/16482785085/in/dateposted-public/

Just enlarge and see the zipper of the coat, where it has glare from sun...



Also, I think you'd be surprised how cheap older high end pre-press flatbeds can go for today - sometimes barely even V850 money.

Yes, it can be a very good option, but you have obsolete drivers and software, and no warranty. A V850 is new, Windows 10, and 2 years warranty, and it is not a 70lbs tank. So no trouble.





You'll see the difference very clearly at a 3x (or less) enlargement in my experience


Not true at all, this is 8x10 scanned over the glass, so with the LowRes lens of the V750 (now I use 850...)

The bell has the size like if it was in a 6m high ultra monster print. So at the end you'll have to bin pixels to send it to a lambda to get prints of common sizes.

The detail you see in the crop (see the bolts over the bell, the grains in the stones) tells about the exceding amount of resolving power.

Some people belive than a a 20mpix image downsized from 800mpix is better than one donwsized from 400mpix, but this is not true. What it is true is that the sharp look depends on the downsizing algorithm you use, in PS one has to use "bicubic, ideal for reductions".


So see what obscene amount of information is in excess from a V750 scan !!!


https://c1.staticflickr.com/1/669/32535835184_2a10b880f7_o.jpg

[Jim Andrada]

Yes, I agree 100%. The IQsmart does a vastly better job retaining smooth tones in the highlights compared to the Epson. So far I haven't felt the need to wet mount with the IQsmart but even if I did, wet mounting on a flatbed is quick and easy compared to a drum scanner - it's the part of the workflow that tipped me the most toward the big flatbed.

Negative - well, it's an old machine and it has a slow processor by modern standards so 4k dpi scans take some time. But being able to simply mount a lot of film on the scanner at once and set up scan windows means that I can mount a bunch of negatives and fire it up and let it run all night on its own. Also need an old Mac to run it but I had an old Macbook that's perfectly adequate. (There IS a Win version of the software but I was advised to avoid it.) It's built like a tank and I have no doubt it will outlast me, and maybe the next couple of owners as well.

[interneg]

So see what obscene amount of information is in excess from a V750 scan !!!

It's bereft of what matters - real sharpness. Fine detail is resolved to one extent or another, but it is not really sharp - ie there's some resolution but it's aberrated to the point of being unrecoverable in a convincing way. Fundamentally, it'll make a print look 'soft' & no amount of sharpening will help that - especially if it was sat alongside a print originated from a better scanner. It will show up in print, even at very small enlargements. BTDT. That image should be razor sharp at the size you've posted it at. Furthermore, those broken up, aliased highlights are highly characteristic of Epson scans & are symptomatic of wildly insufficient dmax that begins to fail in the very low 2.0's - if that. The wall below the clockface sticks out like a sore thumb because of this problem with the highlights.

More to the point, those high pixel density screens you vaunt so much barely reach 2/3 of the resolution of an average inkjet print & a little over 50% of the resolution of a Lambda or similar. Just to make it clear, a 4K television is not a substitute for a decent graphics monitor & that's even more the case if we're comparing with a 4K+ graphics monitor. Again, BTDT.

Finally, I've scanned plenty of stuff shot with a Nikkor 50/1.8 - the level of aberration in your images seems atypical (when it appears in a known good example, it's mostly out towards the edges) & I've scanned CMS20 - yes it will show up some spherical aberration in optics, but at a much lesser scale (& I mean much less). The fact that it's all over the image in that way is not characteristic of a 50/1.8 in good working order. Most of that aberration is more likely scanner lens artefacts, or dirty flatbed glass covered in out-gassing from the plastic shell of the scanner. Note that it 'blooms' around the highlights - which is usually indicative of a cheap (or dirty) scanner optic.

[Pere Casals]

Yes, I agree 100%. The IQsmart does a vastly better job retaining smooth tones in the highlights compared to the Epson. So far I haven't felt the need to wet mount with the IQsmart but even if I did, wet mounting on a flatbed is quick and easy compared to a drum scanner - it's the part of the workflow that tipped me the most toward the big flatbed.

Negative - well, it's an old machine and it has a slow processor by modern standards so 4k dpi scans take some time. But being able to simply mount a lot of film on the scanner at once and set up scan windows means that I can mount a bunch of negatives and fire it up and let it run all night on its own. Also need an old Mac to run it but I had an old Macbook that's perfectly adequate. (There IS a Win version of the software but I was advised to avoid it.) It's built like a tank and I have no doubt it will outlast me, and maybe the next couple of owners as well.

I don't agree.

What it is true is that you may need mutiexposure for extreme highlights, and V700/750 could not have the silverfast SE PLUS version boundled, if coming with bare SE version not suporting multiexposure, could need the PLUS upgrade, some $50

Extreme higlights can have high densities, as shadows in slides, beyond 3.0D you clearly need multiexposure.

IMHO Epsons can do a perfect job with highlights of negative film, both color and BW. For extreme velvia shadows one may need a drum, as no flatbed is perfect for that.

[Pere Casals]

It's bereft of what matters - real sharpness. Fine detail is resolved to one extent or another, but it is not really sharp - ie there's some resolution but it's aberrated to the point of being unrecoverable in a convincing way. Fundamentally, it'll make a print look 'soft' & no amount of sharpening will help that - especially if it was sat alongside a print originated from a better scanner. It will show up in print, even at very small enlargements. BTDT. That image should be razor sharp at the size you've posted it at. Furthermore, those broken up, aliased highlights are highly characteristic of Epson scans & are symptomatic of wildly insufficient dmax that begins to fail in the very low 2.0's - if that. The wall below the clockface sticks out like a sore thumb because of this problem with the highlights.

More to the point, those high pixel density screens you vaunt so much barely reach 2/3 of the resolution of an average inkjet print & a little over 50% of the resolution of a Lambda or similar. Just to make it clear, a 4K television is not a substitute for a decent graphics monitor & that's even more the case if we're comparing with a 4K+ graphics monitor. Again, BTDT.

Finally, I've scanned plenty of stuff shot with a Nikkor 50/1.8 - the level of aberration in your images seems atypical (when it appears in a known good example, it's mostly out towards the edges) & I've scanned CMS20 - yes it will show up some spherical aberration in optics, but at a much lesser scale (& I mean much less). The fact that it's all over the image in that way is not characteristic of a 50/1.8 in good working order. Most of that aberration is more likely scanner lens artefacts, or dirty flatbed glass covered in out-gassing from the plastic shell of the scanner. Note that it 'blooms' around the highlights - which is usually indicative of a cheap (or dirty) scanner optic.


interneg.... i've posted an lowress image with a crop to show the amazing detail level...

if you don't understand that the detail level you see in the bell crop is much more than enough to get a razor sharp, sharp, sharp big print in a lambda... well, not worth to continue the discussion.

[interneg]

interneg.... i've posted an lowress image with a crop to show the amazing detail level...

if you don't understand that the detail level you see in the bell crop is much more than enough to get a razor sharp, sharp, sharp big print in a lambda... well, not worth to continue the discussion.

It's going to look rubbish alongside a print from the same negative scanned on a better scanner. Multiple people have gone over this multiple times with you. Either get a drum scan & see for yourself or stop trying to pretend the Epson is something it isn't. It's OK for basic proof scans, but for top quality negative scans that do justice to the negatives, you need something much better. The quality of the resolution is what matters, not the supposed quantity.

[Bill Burk]

Not true at all, this is 8x10 scanned over the glass, ... So see what obscene amount of information is in excess from a V750 scan !!!

This speaks to me about the value of shooting 8x10 in the first place. Although such detail might never be revealed in a print, someone tasked with restoring the bell tower one day... could use the information in your negative to cut a new headstock.

[Pere Casals]

This speaks to me about the value of shooting 8x10 in the first place. Although such detail might never be revealed in a print, someone tasked with restoring the bell tower one day... could use the information in your negative to cut a new headstock.

Yes... but 4x5 is not very far from 8x10 in terms of technical IQ, in practice. In theory a 8x10 should give "4x more pixels" total, but with common situations you obtain 1.5x to 2x more.

Also as a 4x5 can have much more resolution than what a human eye can see, any practical improvement can only be seen with monster prints seen ar reading distance.

[Pere Casals]

It's going to look rubbish alongside a print from the same negative scanned on a better scanner. Multiple people have gone over this multiple times with you. Either get a drum scan & see for yourself or stop trying to pretend the Epson is something it isn't. It's OK for basic proof scans, but for top quality negative scans that do justice to the negatives, you need something much better. The quality of the resolution is what matters, not the supposed quantity.


Interneg... the quality of the resolution is LP/mm


Tell me how 30 Lp/mm can be better than 40 Lp/mm because that 30 can be of better quality than 40...

Resolution Quality? in what terms? DR? Acutance? Color accuracy? what numbers?


Look, I've been comparing a lot of drum scans (and flextight scans) to v750 scans, and I saw it is very clear when drum matters a lot and when not at all. Also I've been technically evaluating what in hardware (or in the driver) digital image enhancing resources are used or not, and how that can be adressed in post process. One has to know what a human eye can see or not at all.

See here how and edited V750 scan will look like one from a drum. You can take the images from the collaborative LF test and try on your own...

https://c1.staticflickr.com/9/8561/28420386682_d481942db8_o.jpg


IMHO, most difference is in the man that scans and edits with PS, with the exception of velvia very deep shadows, where a drum rocks, and all (good) flatbeds have same amount of problems with scanning stray light (depending on scene, a very dark overall slide has lower stray light for a flatbed).

http://www.largeformatphotography.info/scan-comparison/scanimages/prem-4x5-fullframe-u.jpg


Also you will know that real photographic conditions are not lab conditions. If camera has a shake of 0.03mm you won't have 40Lp/mm, Camera planes are not exactly aligned, there is tolerated DOF... etc, etc, etc...

So it is difficult to find negatives that will need extreme scanner performance to shine. It is true that expensive scanners cook internally more the image by default, but you always can do that with Photoshop.

At the end what counts is photographer.

[interneg]

the quality of the resolution is LP/mm


You cannot tell USAF guys that their 1951 chart has a flaw !!! ...because 30 Lp/mm can be better than 40 Lp/mm because that 30 can be of better quality than 40...

Resolution Quality? in what terms? DR? Acutance? Color accuracy? what numbers?

MTF tests will tell far more than a single target designed to be imaged at high contrast. Edge definition, accutance, aberrations all have their own tests too. Accutance is vitally important to the appearance of sharpness & the Epson has very low accutance.


I've been comparing a lot of drum scans (and flextight scans) to v750 scans, and I saw it is very clear when drum matters a lot and when not at all. Also I've been technically evaluating what in hardware (or in the driver) digital image enhancing resources are used or not, and how that can be adressed in post process.

See here how and edited V750 scan will look like one from a drum. You can take the images from the collaborative LF test and try on your own...

It's pretty obvious that the Epson operates nowhere near the performance of any of the high end scanners. That you omitted to compare it with the Heidelberg Tango in your selection is very telling.

That test has its own small flaws - ie using interpolation on the Imacons instead of looking at everything at their actual resolutions & no real indication of maintenance/ adjustment on the scanners. But even the up-scaled Hasselblad/ Imacon scan blows away the Epson for sharp resolution & that's supposed to be lower in resolving power than the Epson. The colour fringing on the Epson is horrific.

Your touchingly naive belief in the USM as a cure all is hilarious. It might enhance apparent 'sharpness' but it also increases grain and other noise too. A truly crisp, in focus, non aberrated scan may not even need sharpened at all.

Believing you can magically 'fix it in post' is just that - magical thinking. If the data isn't there, you can't make it appear unless you indulge in CGI. Instead of 'comparing' online resources and producing voluminous pseudotheory, how about getting a drum scan made of that negative? Or are you afraid that it'll tell the truth about the Epson?


At the end what counts is photographer.

Most photographers operating at the top of their game are incredibly fussy about the qualities & handling of colour. High end scanners enable you get access to far more of the relevant information than an Epson. The End.

[Pere Casals]

MTF tests will tell far more than a single target designed to be imaged at high contrast. Edge definition, accutance, aberrations all have their own tests too. Accutance is vitally important to the appearance of sharpness & the Epson has very low accutance.

How much acutance V750 vs IQ2 ??? None, zero, nothing: the same in practice

Better USAF Lp/mm is always from better MTF graph.








It's pretty obvious that the Epson operates nowhere near the performance of any of the high end scanners. That you omitted to compare it with the Heidelberg Tango in your selection is very telling.

Go to LFPF Collaborative scan test and instead Howtek take Tango images, cook in PS to the optimum and compare. Also compare to other flatbeds.

I don't doubt drums have better performance, that sometimes help, sometimes not at all. What I say is that V750 compare very well with other expensive flatbeds, and even is near from certain drums in certain conditions, not always of course.






Instead of 'comparing' online resources and producing voluminous pseudotheory, how about getting a drum scan made of that negative?

Sorry, but you are not evaluating well the thing.

"LFPF Collaborative scan test" is not a lab MTF test, but a practical one with practical photographic results, and it is a neutral test.

So you have in front of your nose very enlarged crops from matching 45 negatives (made at the same time, and sent to very skilled people), scanned with most prominent scanners.

If you can't judge that in practical photographic terms it is not worth to continue the discussion, so the end.

My recommendation is that you take "LFPF Collaborative scan test" crops and try to make it match in PS, you'll have the opportunity to abandon silly prejudices that are around, and you'll learn what matters and what not in a digital image.

(In the LFPF Collaborative some shadow detail of the red box it is much worse in the V750, I guess this is because ME was not used)

[interneg]

I don't doubt drums have better performance, that sometimes help, sometimes not at all. What I say is that V750 compare very well with other expensive flatbeds, and even is near from certain drums in certain conditions, not always of course.

I read you all right, you are desperate to elevate the V750 and are producing voluminous clouds of obfuscation and disinformation. Unlike you, I know how higher end scanners behave compared to the Epson because I have used them. You are increasingly becoming a textbook illustration of Dunning-Kruger.

[bob carnie]

I read you all right, you are desperate to elevate the V750 and are producing voluminous clouds of obfuscation and disinformation. Unlike you, I know how higher end scanners behave compared to the Epson because I have used them. You are increasingly becoming a textbook illustration of Dunning-Kruger.

I had to look that one up.... I think I have found my nameskeep.


Pere - you are becoming a PIA to say the least

[Corran]

+1 to everything interneg is saying. This is far afield from the original question. With regard to the Epson Vxxx scanners what I have observed is that the highest available resolution is achieved by scanning at 4800 or possibly 6400 and then downsampling to 2400 DPI. My observation with most consumer scanners is that from the point of "maximum resolution" the scan should be further reduced another 50% to look its best. This brings the Epson to about 1200 DPI which gives roughly a 4x enlargement ratio, or about 16x20 max from 4x5, which matches what most users have reported. I used an Epson V700 for a while before buying various other scanners. Actually I had a sample size of 4 Epsons at the local college and they all performed about the same. Finding the right height of the holders is essential for getting the best usable resolution.

[alanbutler57]

I don't know about detail comparisons etc. But I usually scan 4x5 negatives at 3,000 -3200 dpi on the Epson V750.

Then, I reduce the image size to 10,000 by 8,000 pixels in PS using "Sharper, Best of Reduction" setting. This does a very good job of sharpening the image without halos etc. I usually then apply and edge mask and hit with a healthy dose of unsharp mask.

The resulting image is big enough for a good sized print if I choose to make one and the bigger files look to take higher levels of unsharp masking better to my eye. Of course it takes a looooong time to scan.

[bob carnie]

Today I am printing files from the new Phase One XF - the 100mb beauty.

44 x 60 inch prints on Silk Bayrta paper. I can say that this is the first time since making large murals on Cibachrome from 8 x10 trans that the image quality I am seeing is equally impressive.

There is a significant difference at this magnification from the clients old camera NikonD 800 to this new camera.

I think with these new systems we are reaching very high quality output and once this quality is out there the bar has been set.

[Corran]

Bob, the native pixel measurements on the XF shows a bit less than 200 DPI for that size print. I assume you are upsampling and processing appropriately for that size print? As I've said before, digital files can be uprezzed pretty significantly and still make very sharp prints. I don't doubt for a second you have more than enough resolution for that size despite the "apparent" lack of pixels at native resolution. Many have argued when I've said I could easily print my D800 files to 36x24 and hey guess what that is the exact same roughly 200 DPI situation for the native pixel sizes.

[bob carnie]

Bob, the native pixel measurements on the XF shows a bit less than 200 DPI for that size print. I assume you are upsampling and processing appropriately for that size print? As I've said before, digital files can be uprezzed pretty significantly and still make very sharp prints. I don't doubt for a second you have more than enough resolution for that size despite the "apparent" lack of pixels at native resolution. Many have argued when I've said I could easily print my D800 files to 36x24 and hey guess what that is the exact same roughly 200 DPI situation for the native pixel sizes.

Hi Corran - the files were prepared in Capture One by my client to the size I printed.... my observations is that from a medium format anything ... this new device reminds me of the good old days of 8 x 10 trans direct print to super size Cibachrome.
I have made 4ft x 72 inch cibas in my past life.


I have been waiting a long time with digital capture to see this , my point is I think the quality issue is for me at least put to bed between 8 x10 trans and digital capture... Looking at these large prints say it all.

[Pere Casals]

I read you all right, you are desperate to elevate the V750 and are producing voluminous clouds of obfuscation and disinformation. Unlike you, I know how higher end scanners behave compared to the Epson because I have used them. You are increasingly becoming a textbook illustration of Dunning-Kruger.

False.

I just stick in what "LFPF Collaborative scan test" shows, that's consistent with what I found when comparing drum scans to the cheap Epson. What Epson is shown in post #33. Not more, not less.

About the Dunning-Kruger... you see the mote in the other's eye.... Well, post #33. A Fact.

[faberryman]

I just stick in what "LFPF Collaborative scan test" shows, that's consistent with what I found when comparing drum scans to the cheap Epson. What Epson is shown in post #33. Not more, not less.
In comparing drum scans to the Epson, have you actually had one of you negatives drum scanned and compared it to the Epson, or are you relying on internet reports and images? If you had a drum scan done of one of you negatives, which drum scanner was used and at what resolution was it scanned?

[interneg]

Today I am printing files from the new Phase One XF - the 100mb beauty.

44 x 60 inch prints on Silk Bayrta paper. I can say that this is the first time since making large murals on Cibachrome from 8 x10 trans that the image quality I am seeing is equally impressive.

There is a significant difference at this magnification from the clients old camera NikonD 800 to this new camera.

I think with these new systems we are reaching very high quality output and once this quality is out there the bar has been set.

Interesting to hear about this - was making some 50x65" prints recently for a client from stitched digital files to get 300ppi of resolution & spent most of the project feeling that scanned 8x10 would've made the whole job much easier...

Have you tried the newish Fine Art Baryta Satin from Hahnemuhle? Even better than the Silk Baryta (which I like a lot) I feel - the colours are amazingly good & it has even more of a darkroom FB paper feel than I've seen before - even compared to Canson's papers etc.

[Pere Casals]

Today I am printing files from the new Phase One XF - the 100mb beauty.

44 x 60 inch prints on Silk Bayrta paper. I can say that this is the first time since making large murals on Cibachrome from 8 x10 trans that the image quality I am seeing is equally impressive.

There is a significant difference at this magnification from the clients old camera NikonD 800 to this new camera.

I think with these new systems we are reaching very high quality output and once this quality is out there the bar has been set.



It has to be noted that D810 and D800E can deliver 50% more resolving power than the D800 you mention, in area terms, because D810 and D800E lack that lowpass optical filter. Of course that +50% happens if lens (and shot) is good enough to not be a limiting factor.

Still the IQ3 has 2.5x the FF surface, so difference has to be seen, anyway the D800E shortens that difference.

What I don't know is when 60, 80 or 100 Mpix on the IQ3 makes a real difference, as for the lens it is clearly difficult working in practice at "up to" 100Lp/mm performance glass manufacturer speaks about.

[bob carnie]

Interesting to hear about this - was making some 50x65" prints recently for a client from stitched digital files to get 300ppi of resolution & spent most of the project feeling that scanned 8x10 would've made the whole job much easier...

Have you tried the newish Fine Art Baryta Satin from Hahnemuhle? Even better than the Silk Baryta (which I like a lot) I feel - the colours are amazingly good & it has even more of a darkroom FB paper feel than I've seen before - even compared to Canson's papers etc.

I love the silk baryta - as well their photo baryta which has a bit of texture.... Both are really nice, I am not a big fan of luster or satin but some of my clients like it.

There are clients of mine using this 100mb camera and stitching making monster files, I am worried my measly PS printing platform will need to be upgraded to handle these monster files, keeping and printing in 16 bit (though not critical over 8bit) means files over 1 gb which starts slowing down things.

I have a 60 inch printer but for sanity's sake have made all my offerings based on 44 inch( i feature 5 Hannamuhle papers) ... I can see the day where I will be making 60 inch x 10 ft prints for clients using this advanced imaging methods and using stitching to make huge, huge, prints.

Hanamuhle has made good efforts to provide printers like me , help when I need it and marketing help as well...

[bob carnie]

It has to be noted that D810 and D800E can deliver 50% more resolving power than the D800 you mention, in area terms, because D810 and D800E lack that lowpass optical filter. Of course that +50% happens if lens (and shot) is good enough to not be a limiting factor.

Still the IQ3 has 2.5x the FF surface, so difference has to be seen, anyway the D800E shortens that difference.

What I don't know is when 60, 80 or 100 Mpix on the IQ3 makes a real difference, as for the lens it is clearly difficult working in practice at "up to" 100Lp/mm performance glass manufacturer speaks about.

I let my eyes do the talking and not charts... those days are gone for me... kinda like testing film ... been there done it.

[Pere Casals]

In comparing drum scans to the Epson, have you actually had one of you negatives drum scanned and compared it to the Epson, or are you relying on internet reports and images? If you had a drum scan done of one of you negatives, which drum scanner was used and at what resolution was it scanned?

Yes, I've drum scanned (hired service) with Imacon 848, and X5, and I scanned V750 the same again. Mostly it was 35mm Velvia 50 shot in Finland with very special winter light by a mate that's a remarkable artist, winner for example of the Zoom Photo Festival, Saguenay, Canada.

Resolution was this one (raw from drum before editing):

166296

There was an underexposure bracketing, the best images came from pretty underexposed slides that had highlight plenitude for all that Polar Circle light beauty. One can stay hours viewing those slides on a boosted light table with a loupe. It is amazing to see how Velvia records deep shadows. All digital is pure crap compared.

Target size for the exhibitions was only 60x40, but all could be very well enlarged to 1m without flaw.

The X5 did an amazing job, really. Something where a V750 would come pretty short.

In the same way I see no advantage beyond V750 for a BW 4x5 negative with 1.8 densities. What I see is that a good postprocess and understanding printers is critical.

It was Bob Carnie that months ago stated that not all Lambda shops were obtaining same results with FB paper, and this is not only because he said that, it's a lot like that. The printer skills !!!

Let me add that in real world photography a lot of shots have more shake (or other optical flaws) than lab testings, and a common situation is that the on negative resolved information is the limiting factor, not the scanner. Then what counts is to apply the right sharpening algorithm to the right area.

And about color management, again what counts is operator and software, IMHO.

[Pere Casals]

I let my eyes do the talking and not charts... those days are gone for me... kinda like testing film ... been there done it.

This is ok... anyway for learners like me it is interesting to understand the scientific basis of thinks. This leads to understand when specs matter or not. Of course nothing substitutes artistic push and creativity. If a true artist only has a hammer he makes something like La Pietà.

[faberryman]

In comparing drum scans to the Epson, have you actually had one of you negatives drum scanned and compared it to the Epson, or are you relying on internet reports and images? If you had a drum scan done of one of you negatives, which drum scanner was used and at what resolution was it scanned?

Pere,

You have responded to several other posts, but not to mine. Do I take it then that you have not compared one of your own negatives drum scanned against Epson scanned, and are only relying on internet posts and images for your opinions? In evaluating you opinions, it is important to know whether you speak from actual experience.

[Pali K]

Yikes not again. I will start and end my contribution to this thread with this post. PM me if you want to discuss anything privately.

I have the following scanners with multiple years of use with each.

Epson v700
Eversmart Pro
Scanmate 5000 Drum Scanner
Scanmate 11000 Drum Scanner
Heidelberg Tango Drum Scanner

Eversmart Pro is a lower model of a pro flatbed that will outshine Epson V series scanners in a heartbeat. There is simply no debate when you see the performance side by side over 100s of same negative scans. No technical comparisons are necessary when your eye will tell you the obvious.

Drum Scanners including "cheap" Scanmate 5000 will outdo most flatbetds. Eversmart Pro and higher models will actually outperform entry level drum scanners in some tests but not in difficult slides or overexposed negatives. This is where the purity of the PMTs really can't be beaten.

Tango is the Holy Grail for Slides and Scanmate 11000 is the Holy Grail for negatives if you have the modified board and upgraded IR/UV filters.

Epson V series is a fantastic scanner for it's price. It's also the best unless you use pro scanners regularly and realize that it's not the best :)

Pali

[faberryman]

Epson V series is a fantastic scanners for it's price. It's also the best unless you use pro scanners regularly and realize that it's not the best :)

Your post, based on actual experience, is extremely helpful.

[Corran]

Epson V series is a fantastic scanners for it's price. It's also the best unless you use pro scanners regularly and realize that it's not the best :)

:D

[Peter De Smidt]

And no amount of test scans will convince any of the True Believers on either side. As long as you're happy with what you get, why worry that someone else might think it's sub-optimal?

[Pere Casals]

Pere,

You have responded to several other posts, but not to mine. Do I take it then that you have not compared one of your own negatives drum scanned against Epson scanned, and are only relying on internet posts and images for your opinions? In evaluating you opinions, it is important to know whether you speak from actual experience.

I've answered you in post #47, regards

[faberryman]

My apologies. It appears that your drum scan experience is limited to 35mm for which the Epson is wholly inadequate for prints of any but the smallest size.

[Pere Casals]

My apologies. It appears that your drum scan experience is limited to 35mm for which the Epson is wholly inadequate for prints of any but the smallest size.

This is, my posts #19 and #31 explains why (for me) the V750/850 is more than I need for 120 and for LF.

For 35mm I understand that V850 is also more than enough, as handheld shots (if not 1/30000 flash) normally have more than 0.02mm camera shake on film, so the V850 is not the limiting factor, but the shot itself.

If on film blur (because lens, shake) is more than 0.02mm then a better scanner only scans more blur, so not advantage over V850.

Also I'm pretty sure that another scanner won't deliver better BW tonality.

A 35mm Adox CMS 20 shot, with 50mm stopped to f/6.5 on a tripod or VR lens would benefit from a drum scan. A TMX 35mm film with 50mm stopped to f/2, or handheld, won't benefit from better than 2500 dpi optical resolving power. The point is having more than 0.02mm on film blur or less.


Do your shots have much less than 0.02mm on film blur ??? Then a drum may be better for big enlargements.

[faberryman]

I agree, for 35mm it is better a dedicated roll film scanner.


For 35mm I understand that V850 is also more than enough, as handheld shots (if not 1/30000 flash) normally have more than 0.02mm camera shake on film, so the V850 is not the limiting factor, but the shot itself.

Feeling conflicted?

[Pere Casals]

Feeling conflicted?

No conflict, a Nikon Coolscan if a better scanner for 35mm than the V850, by a large margin, no doubt, 3x "more optical pixels". But if you shot handheld with a Nikon F5 like me there is no advantage, because if you shot handheld it is very difficult that a shot has less than 0.02mm shake blur, so better scanner performance won't deliver a better image for PS edition.


One reason to use the coolscan is to depict film grain, but you also may prefer to decrease grain, like when using diffusion enlarger instead condenser. PS edition also can make grain more or less sharp.


PD: Let me add that at one point negative color films (Portra, Ektar, 160NS, and all consumer) were reengineered to have larger color clouds to scan optimally well in digital minilabs (frontiers, noritsus), so prefect for the V series. It is not the case of Velvia/Provia.

[faberryman]

If you are happy with your Epson scans of 35mm, more power to you.

[Pere Casals]

If you are happy with your Epson scans of 35mm, more power to you.

My 35mm shots are handheld and dynamic, usually I shot totally wide open. A better 35mm scanner simply won't make a difference at all. And I prefer investing in DIY emulsion making than in "pixel peeping".

What's for LF with the V850 I obtain way more resolving power than I need.

Rather than spending in an expensive Flatbed I prefer to spare some money for drum service for the case a Velvia shot deserves it, as no Flatbed will do it.

[interneg]

And it should be noted that Hasselblad/ Imacon scanners are not under any circumstances a 'drum' scanner - no PMT's, no fluid mount on perspex drum. They are 'virtual drum' with flexible holders, but are really just a very good CCD scanner. To refer to it as a drum scanner is fundamentally dishonest. I'm extremely familiar with them & while they don't necessarily have the learning curve of drums, they are outperformed by the better drum scanners.

As for 'I don't need a better scanner for 35mm because I shoot handheld'... let's just say that's one of the more laughably bizarre statements I've heard in a while.

[interneg]

There are clients of mine using this 100mb camera and stitching making monster files, I am worried my measly PS printing platform will need to be upgraded to handle these monster files, keeping and printing in 16 bit (though not critical over 8bit) means files over 1 gb which starts slowing down things.

Going above the 4GB mark is where things start to get really entertaining... One of those 50x65s hit 12GB in 16 bit I recall - bringing a pretty powerful computer to a crawl. Looking at building a new machine here too, with a decently serious GPU in order to handle those sort of files.

[Pere Casals]

And it should be noted that Hasselblad/ Imacon scanners are not under any circumstances a 'drum' scanner - no PMT's, no fluid mount on perspex drum. They are 'virtual drum' with flexible holders, but are really just a very good CCD scanner.


"are not under any circumstances a 'drum' scanner" yes, but better than most drum scanners, 4.8D for the X5. Hassy is Hassy.



As for 'I don't need a better scanner for 35mm because I shoot handheld'... let's just say that's one of the more laughably bizarre statements I've heard in a while.

It is not bizarre at all. You can test that for shure very easy. Just take a good DLSR and shot a resolution target handheld, with both VR ON and OFF, also with tripod. Then calculate the blur radius you have on sensor/film, it is a plain calculation, I think you should be able to perform, but on any doubt I can explain it to you.

I do that calculation often for industrial video systems that are on a machine with vibrations, sometimes a LED illuination delivers blur but a 1uS strobe freezes it very well. I use that for contactless metrology.

Once you have your on film blur radius you can guess how many scanning dpi you need. Anyway you can see it by eye without calculations, technical calulations are no need at all for art, but anyway doing it once in a lifetime is a good learning to know matters or not.

Regards

[faberryman]

PS edition also can make grain more or less sharp.
You can't make the grain more or less sharp in PS if the grain itself is not resolved by the scan, as is the case with the Epson flatbed.

With respect to camera shake, when I scan I would rather have sharp blur than blurry blur so I use a film scanner. But to each his own.

[Pere Casals]

You can't make the grain more or less sharp in PS if the grain itself is not resolved by the scan, as is the case with the Epson flatbed.

With respect to camera shake, when I scan I would rather have sharp blur than blurry blur so I use a film scanner. But to each his own.

Tabular grain (TMX, Delta) is not evident even in 35mm ...for LF not seen at all.

The particular case of big grain films, (HP5, Silvermax) are well seen with V700/850, and a sharpening (or softening) action with very small radius will adjust well grain depiction.

This an example with silvermax vs TMX with V750

https://www.flickr.com/photos/125592977@N05/21478354193/in/dateposted-public/

https://www.flickr.com/photos/125592977@N05/29604172183/in/dateposted-public/



You can find really good photographers like this one (that I like a lot and inspires me), working with grain and using V700

https://www.flickr.com/photos/55873497@N04/albums/72157633306014832

https://www.flickr.com/photos/55873497@N04/albums/with/72157633306014832

Of course with LF... no grain !!!




With respect to camera shake, when I scan I would rather have sharp blur than blurry blur so I use a film scanner. But to each his own.

"sharp blur" and blurry blur to me it's the same in terms of image quality. But "Sharp blur" (strange concept :)) may show more grain.



I would like to add that grain depiction can be important in digital process: Amazonas team that edited Salgado's Genesis struggled a lot with it. Half of the shots were digital but they introduced artificial grain (with DXO filmpack, IIRC) to have consistence with TXP 645 shots, then printed it on Delta 100 8x10 sheets with a Kodak LVT Rhino, 4 images per sheet, and enlarged optically.

What I mean is that grain can be a digital postprocess concern, of great aesthetical complexity. And this also concerns how grain is scanned. I've been exploring all that with a lot of personal interest... but still I've a lot to learn...

[IanBarber]

Pere, how much testing have you done with Epson V series with regards to the required distance between the film and the scanner glass to achieve optimal sharpness from the scanner on Medium Format films or any size film come to that

[faberryman]

Tabular grain (TMX, Delta) is not evident even in 35mm ...for LF not seen at all.
I see grain in my 10"x15" (10x) darkroom enlargements from 35mm Delta 400, particularly in areas like the sky. The Epson V700 cannot resolve that grain, so similar size prints from those scans are not sharp; they look mushy. Scans at 3900 ppi clearly show the grain structure on screen and in the prints, and I don't use sharpening to enhance that grain. These are my actual results; they are not theoretical.

I would post comparative scans from an Epson flatbed and a 3900 ppi film scanner, but after doing my initial 35mm film scans on an Epson V700, I saw what they looked like, deleted them, and bought a film scanner. Again, if you are scanning large format, an Epson flatbed may be just the ticket, though it is no where near as good as an X5 or a drum scanner.

[Corran]

I think Pere is seeing grain aliasing. And worse his overly-aggressive unsharp masking is accentuating this grain. A smoother overall image can be achieved with higher-resolution scans - and yes, this applies to any format, because the film is roughly the same regardless of how large a piece it is (yes, enlargement ratio will play a part in what it looks like though). I have seen plenty of very grainy scans from 120 and 4x5, even with slow film (also important here is manipulations in contrast and highlights). "Resolving grain" is not as simple as seeing grain in your scans.

From what I've seen, grain aliasing looks more like random noise and less like film grain to me. Pere's 35mm silvermax shots clearly show this fuzzy noise in the scan that to me is a perfect example of aliasing, especially when exacerbated by USM.

[faberryman]

I think Pere is seeing grain aliasing. And worse his overly-aggressive unsharp masking is accentuating this grain. A smoother overall image can be achieved with higher-resolution scans - and yes, this applies to any format, because the film is roughly the same regardless of how large a piece it is (yes, enlargement ratio will play a part in what it looks like though). I have seen plenty of very grainy scans from 120 and 4x5, even with slow film (also important here is manipulations in contrast and highlights). "Resolving grain" is not as simple as seeing grain in your scans.

From what I've seen, grain aliasing looks more like random noise and less like film grain to me. Pere's 35mm silvermax shots clearly show this fuzzy noise in the scan that to me is a perfect example of aliasing, especially when exacerbated by USM.

Here's an example (not mine) of grain aliasing. It isn't pretty:

166310

What you see on the left is multi-colored pixels; on the right, grain, or more accurately, dye clouds.

[Pere Casals]

I see grain in my 10"x15" (10x) darkroom enlargements from 35mm Delta 400, particularly in areas like the sky. The Epson V700 cannot resolve that grain, so similar size prints from those scans are not sharp. Scans at 3900 clearly show the grain structure on screen and in the prints, and I don't use sharpening to enhance that grain. These are my actual results; they are not theoretical.

Well... I'm not very interested in Delta or TMY grain, to the those grain structures are lacking character, to me it looks DSLR noise, while I love TX/TXP/HP5/FP4/Silvermax/APX and P3200 grains. For Delta and TMX I prefer to hide grain as much I can, so I can't tell.

I'm pretty sure that a Coolscan would deliver a better grain, but don't think that 2800 to 3900 dpi (say effective optical) is a big change for grain depiction.

[Peter De Smidt]

I agree completely with Corran.

Pere, you like to avalanche technical information in your posts. Some of it is fine, but some is not. For example: "...yes, but better than most drum scanners, 4.8D for the X5. Hassy is Hassy." That dmax number, like the ideal world theoretic numberd from other manufacturers, does not reflect real world use. It is marketing bull. You've been told this many times, and yet you fail to understand this. You check real world dmax by scanning a calibrated grayscale and plotting the results. You don't check it by looking at manufacturer specs. Your persistent failure to recognize this undermines your credibility.

[Pere Casals]

Here's an example (not mine) of grain aliasing. It isn't pretty:

166310

So this is not grain aliasing:

https://c1.staticflickr.com/3/2947/15473025262_7034af3422_b.jpg

[Pere Casals]

I agree completely with Corran.

Pere, you like to avalanche technical information in your posts. Some of it is fine, but some is not. For example: "...yes, but better than most drum scanners, 4.8D for the X5. Hassy is Hassy." That dmax number, like the ideal world theoretic numberd from other manufacturers, does not reflect real world use. It is marketing bull. You've been told this many times, and yet you fail to understand this. You check real world dmax by scanning a calibrated grayscale and plotting the results. You don't check it by looking at manufacturer specs. Your persistent failure to recognize this undermines your credibility.

Peter, I was wrong, sorry, not 4.8D... but 4.9D is the specs. This is X5 maximum readable density in optimal conditions, this means than it will read very well Velvia DMax, (under 4.0D).

We all know very well that DMax of any scanner is not usable, but mentoning the 4.9 specs it was only way to say to Interneg that X5 it is not a joke, he said about X5 "They are 'virtual drum' with flexible holders, but are really just a very good CCD scanner", one can understand that it was a second rate machine, but it is a first rate device.


For the rest, if you find wrong technical information from me, please tell me. I like to learn, specially in what I'm wrong.

[Pere Casals]

Pere, how much testing have you done with Epson V series with regards to the required distance between the film and the scanner glass to achieve optimal sharpness from the scanner on Medium Format films or any size film come to that

To me it has some importance for scans from very good (tripod) 35mm film. Much less importance for 120, and little importance for LF. I you want to know exactly you should scan an USAF 1951 glass slide.

[faberryman]

Tabular grain (TMX, Delta) is not evident even in 35mm ...for LF not seen at all.


Well... I'm not very interested in Delta or TMY grain, to the those grain structures are lacking character, to me it looks DSLR noise, while I love TX/TXP/HP5/FP4/Silvermax/APX and P3200 grains. For Delta and TMX I prefer to hide grain as much I can, so I can't tell.

If you think tabular grain is not evident even in 35mm, why do you dislike its character and try to hide it as much as you can?


I'm pretty sure that a Coolscan would deliver a better grain, but don't think that 2800 to 3900 dpi (say effective optical) is a big change for grain depiction.

In all likelihood, the reason you don't think so is that you have never actually done the comparison for yourself using your own negatives. Sometimes theory can only take you so far.

[Pere Casals]

If you think tabular grain is not evident even in 35mm, why do you dislike its character and try to hide it as much as you can?

TMX has a very fine grain, with no evident structure (to me). So I don't find that grain aesthetically useful. So I've been using a good fine grain developer, Xtol. I like other TMX features.

From TX I like dramatic structure: grain is larger in the dark greys than in the lighter ones, TX has an strong aesthetical subculture around it that I love. HP5 has larger grains in the mids so also a classic with a well developed imaging trend around it.

I just try to learn from others (that inspire me) how can it be aesthetically exploited... sometimes I feel I've too low artistic profile to get advantage from it...




In all likelihood, the reason you don't think so is that you have never actually done the comparison for yourself using your own negatives. Sometimes theory can only take you so far.

Never used a coolscan, but I've been observing how photographers that I admire get results from it, these are some coolscan shots I collected, because grain depiction (not only because the image itself), some with washed areas showing well grain:

https://www.flickr.com/photos/12205295@N03/14314453828/in/photolist-nNVjq5-pkazcB-hmAYLo-oTZqCt-dH232s-raGX85-q3QmYt-pmq6sr-nYFg4G-qTEhGA-n4YQRn-nH48Y9-nNmwvZ-rufrPR-npXVZN-fUeYJr-qpebR2-hgKyui-ne8ozN-bnhS7H

https://www.flickr.com/photos/12205295@N03/15313086781/in/photolist-nNVjq5-pkazcB-hmAYLo-oTZqCt-dH232s-raGX85-q3QmYt-pmq6sr-nYFg4G-qTEhGA-n4YQRn-nH48Y9-nNmwvZ-rufrPR-npXVZN-fUeYJr-qpebR2-hgKyui-ne8ozN-bnhS7H

https://www.flickr.com/photos/12205295@N03/8340459476/in/photolist-nNVjq5-pkazcB-hmAYLo-oTZqCt-dH232s-raGX85-q3QmYt-pmq6sr-nYFg4G-qTEhGA-n4YQRn-nH48Y9-nNmwvZ-rufrPR-npXVZN-fUeYJr-qpebR2-hgKyui-ne8ozN-bnhS7H

https://www.flickr.com/photos/12205295@N03/14308055843/in/photolist-nNmwvZ-URWEzu-rufrPR-oTZqCt-UB1LAZ-pCuF4c-jPGPAP-oKdoqQ-gLHscD-gPhEsT-nCYmsS-bmE4yr-RFXNHj-fBqW3z-67XuJz-76Q6kL-bMGmzr-bn6YdX-989Gbb-6xC5Hw-bExdqY-5vdhUL-nmKNyN-noCP4C-sG5jUn-SftT83-nDLPvj-8QSF5w-TPrKFu-fHG8Xh-6sjDPh-jSqwxq-g3Vdnb-ne62kj-Rb4YXE-4Y8dkX-npXVZN-nmK4jR-6aqa4U-ijX2iG-dH232s-86xLcq-8BFhFH-9gKyyE-raGX85-8cKz3x-U32X2X-QJcCJr-nH3nHL-3kHubZ

Also I've processed Plustek scans for a mate, and compared to V850.


Understand me, I try to learn (not to teach) to use grain aesthetics, I feel that's a very complex discipline. Something I yearn when doing LF.

[Peter De Smidt]

Peter, I was wrong, sorry, not 4.8D... but 4.9D is the specs. This is X5 maximum readable density in optimal conditions, this means than it will read very well Velvia DMax, (under 4.0D).

We all know very well that DMax of any scanner is not usable, but mentoning the 4.9 specs it was only way to say to Interneg that X5 it is not a joke, he said about X5 "They are 'virtual drum' with flexible holders, but are really just a very good CCD scanner", one can understand that it was a second rate machine, but it is a first rate device.


For the rest, if you find wrong technical information from me, please tell me. I like to learn, specially in what I'm wrong.

I just did. Moreover, if "We all know very well that DMax of any scanner is not usable", then "mentioning the 4.9 specs...was [the] only way to say to Interneg that the X5 is not a joke...." didn't do a thing. Interneg didn't say the X5 is a joke. He said, as you quote yourself!, "...but [the X5] are really just very good ccd scanner", which is perfectly accurate. It's not a slight against the X5. So, you misunderstood his position, and your evidence that he was wrong was not credible. You don't take care with what people tell you or with what you write. Instead, you try to steamroller fellow conversationalists with a ton of material. That doesn't convince people, assuming that's your goal. It makes your positions more questionable.

[interneg]

So this is not grain aliasing:

It's certainly an illustration of hideous over-sharpening. TX in D-76 should never look that ugly.

The Coolscan seems to aliase/ emphasise grain horribly. I know this because I have routinely had to rescan (on the X5) negatives that were turned into aliased, seemingly over-sharpened messes on Coolscans. No idea why this is - is there some built in sharpening bug in the software?

Also, the X5 is a first rate machine (and a scanner I use heavily) - but it's a CCD scanner with a feed path that curves the film around a virtual drum, not an actual PMT drum scanner where the original is fluid mounted on to a perspex drum. Misquoting others to suit your purposes is a very stupid idea.

[Pere Casals]

I just did. Again. Moreover, if "We all know very well that DMax of any scanner is not usable", then "mentioning the 4.9 specs...was [the] only way to say to Interneg that the X5 is not a joke...." didn't do a thing. Interneg didn't say the X5 is a joke. He said, as you quote yourself!, "...but [the X5] are really just very good ccd scanner", which is perfectly accurate. It's not a slight against the X5. So, you misunderstood his position, and your evidence that he was wrong was not credible. You don't take care with what people tell you or with what you write. Instead, you try to steamroller fellow conversationalists with a ton of material. That doesn't convince people, assuming that's your goal. It makes your positions more questionable.

Peter, thanks for saying what you think.

If you read all the post you'll see what is the interneg intention with that comment.

Please let me put my thoughts in order, about the V850 the OP has, and asks about:


> If your scan (from a LF negative) way outresolves (optically ) monitor and lambda print, then you don't need a sharper scanner.

> After developing, BW tonality is about curves/edition, and image size reduction algorithms.

> Color is IT8 calibrated, important thing is edition skills and software

> V850 works very, very well for densities under 3.0D. Have higher interesting densities ? ...go drum or flextight.

> For 120 you can enlarge a lot from V850 scans

> For 35mm it is not a very good scanner, but perfect for Portra.

> It is difficult to find 35mm handheld negatives that have inside resolving power beyond 2000 dpi, because it is easy having on negative 0.02mm blur, (shake, lens, film flatness). So a very good scanner will shine in some shots only, mostly from tripod.

> A coolscan may deliver slightly better grain, from 2800 (or 2500 ) to 3900 grain improves a bit. You can work grain mood in PS, but scanning well grains is good.


IMHO digital photography world has a lot of pixel peeping, you know. IMHO this also happens a bit with scanners. IMHO ultimate scanner performace is secundary in front of having the good criterion to bring an image to a sound artistic output.

[Pere Casals]

It's certainly an illustration of hideous oversharpening. TX in D-76 should never look that ugly.

Also, the X5 is a first rate machine (and a scanner I use heavily) - but it's a CCD scanner with a feed path that curves the film around a virtual drum, not an actual PMT drum scanner where the original is fluid mounted on to a perspex drum. Misquoting others to suit your purposes is a very stupid idea.

no sharpening used at all in that ugly shot... All of we know what flextight is.

[Peter De Smidt]

The Coolscan seems to aliase/ emphasise grain horribly. I know this because I have routinely had to rescan (on the X5) negatives that were turned into aliased, seemingly over-sharpened messes on Coolscans. No idea why this is - is there some built in sharpening bug in the software?



That was my experience using a Coolscan V for a couple of years. It was very good with fine-grained bw film, when the grain wouldn't be visible in the final print, but it was very poor with grainy film, such as HIE, as noise went through the roof. My flatbed at the time, a Canon 9950, did a much better job than the Coolscan with grainy BW film.

[Bill Burk]

I didn't like that ugly shot's texture either.

This is the kind of look I try to achieve in my prints, I print to 11x14 from all my negatives whether 4x5 (on the left) or 35mm (on the right). (These are scans from enlarger prints on Ilford Galerie - some scan/jpg herringbone artifacts appear in this image that of course do not exist on the print).

If you can choose scanner settings that approach this kind of appearance on output prints at 11x14 inches, then I would say you have found a scanner that is good enough (for me).

http://beefalobill.com/imgs/45y35.jpg

[Pere Casals]

That was my experience using a Coolscan V for a couple of years. It was very good with fine-grained bw film, when the grain wouldn't be visible in the final print, but it was very poor with grainy film, such as HIE, as noise went through the roof. My flatbed at the time, a Canon 9950, did a much better job than the Coolscan with grainy BW film.

I suspect, not sure, that when the digital sampling is of similar size than grains or color clouds a lot of noise is generated. The digital sample of an spot can even have 0% or 100% of the signal.

Long ago (15 years) I was told by a Noritsu service man (digital minilabs) that film manufacturers made color clouds larger (color negative film) to match the best performance with sensors of the era that were in the Frontiers, Noritsus, etc. Films became less sharp, worse for optical enlargers, but perfect for minilabs that were adding some digital sharpening to compensate for the larger clouds. In this way digital minilabs performed very well, consumers saw a lot of improvement as image enhancing software started making wonders. So they sold a lot of expensive gear around the world.


What's about BW grain, I love it, (not for all shots...) so I've been observing contact copies with 60x loupe and comparing with scans, and trying different sharpening radious/strengths to get the natural depiction. Then what I saw is when downsizing the image to the target output everything was destroyed again, then there is jpeg, and Web Browsers resizing...

When I was exporing how Amazonas worked grain for the Genesis digital shots I understood how difficult it had to be. Taschen doesn't like botched jobs... IMHO Amazonas soon nailed the TXP tonality, but grain had to be a nightmare.

They concluded that best way was to print the artificial DXO grain (on digital Canon shots) on sheets with a Rhino, and enlarging those grains, to get consistence with the TXP shots.

[Corran]

This is a 3000 DPI scan of 35mm Tri-X developed in Acufine and shot at an EI of 800. The inset is a 100% view (30x45 print if viewing on a 96 DPI monitor). The grain is visible even at a small size when I develop with Acufine in areas of smooth tone, as seen on the back wall. I like the look of this in many situations where I would shoot Tri-X so that's not a problem, plus it enhances the feeling of sharpness at smaller sizes due to the texture.

http://www.garrisaudiovisual.com/photosharing/epicon-2230grain.jpg

In contrast, here is a landscape shot that I scanned at 5000 DPI of Tri-X developed in XTOL, shot at box speed. The 100% view has some grain visible but you can't see it in normal sizes (100% view with a 96 DPI monitor equates to a 50x75 print). I don't shoot landscapes with Tri-X but I was finishing up a roll and just scanned this yesterday specifically at a higher resolution to compare. The higher-rez scans also smooths out the grain.

http://www.garrisaudiovisual.com/photosharing/doncarter-7316grain.jpg

Neither looks like the shot posted by Pere with what I could only describe as digital-esque noise. That image looks like what my D800 produces when shot at 6400-12800 ISO and viewed at high magnification and sharpened.

This is far afield of the OP but since we are talking about grain and scanning I thought I would post this. I don't think there is anything wrong with wanting to emphasize grain in an image but the point is that the scanner, scanner resolution, and grain aliasing are important considerations with how the resultant digital image looks.

[chassis]

Recent experience is giving me the idea that scanning at maximum effective resolution for your hardware+software system is generally desired. e.g. for the Epson Vxxx scanners, the max effective resolution is generally accepted to be a bit more than 2000 dpi, with a scan resolution set at 3000 dpi. Image resizing can be done after cropping, color/tone management and de-dusting. I have found color (tone) and shadow detail can be lost if less than max effective resolution is used. Just my experience and self learning conclusions.

[Peter De Smidt]

With the Cezanne, grain keeps getting better, i.e. finer and more natural, all the way up to 6000 spi.

[interneg]

In contrast, here is a landscape shot that I scanned at 5000 DPI of Tri-X developed in XTOL, shot at box speed. The 100% view has some grain visible but you can't see it in normal sizes (100% view with a 96 DPI monitor equates to a 50x75 print). I don't shoot landscapes with Tri-X but I was finishing up a roll and just scanned this yesterday specifically at a higher resolution to compare. The higher-rez scans also smooths out the grain.

That essentially tallies with my experience with the Hasselblad.




When I was exporing how Amazonas worked grain for the Genesis digital shots I understood how difficult it had to be. Taschen doesn't like botched jobs... IMHO Amazonas soon nailed the TXP tonality, but grain had to be a nightmare.

They concluded that best way was to print the artificial DXO grain (on digital Canon shots) on sheets with a Rhino, and enlarging those grains, to get consistence with the TXP shots.

The Salgado 'Genesis' work had copious amounts of awful HDR amongst numerous other issues. Not a great look. The LVT films were only made for exhibition prints. No idea why you'd hold it up as a great example of TXP's tonality.

[IanBarber]

Recent experience is giving me the idea that scanning at maximum effective resolution for your hardware+software system is generally desired. e.g. for the Epson Vxxx scanners, the max effective resolution is generally accepted to be a bit more than 2000 dpi, with a scan resolution set at 3000 dpi

No one has mentioned software from what I can see. Is there a census that one scanning software out performs another or are they more or less producing the same results

[faberryman]

no sharpening used at all in that ugly shot... All of we know what flextight is.
Are you suggesting that the portrait you posted was scanned by an X5. I would be very surprised. First, the image data says it was scanned at 3201 ppi, and, more importantly, when you enlarge the image, you see pixels before you see the grain structure. It appears to be, and certainly looks like, a medium resolution scan with aliasing. If it was done on an XF, there was operator error.

[Pere Casals]

Neither looks like the shot posted by Pere with what I could only describe as digital-esque noise.

Anyway this is not a V750 scan, but a V500 one, sorry I accidentally deleted part of the text explaining that

https://www.flickr.com/photos/125592977@N05/15473025262/in/dateposted-public/

I wanted to show that even a low end V500 is able to show TX grain.



Are you suggesting that the portrait you posted was scanned by an X5. I would be very surprised. First, the image data says it was scanned at 3201 ppi, and, more importantly, when you enlarge the image, you see pixels before you see the grain structure. It appears to be, and certainly looks like, a medium resolution scan with aliasing. If it was done on an XF, there was operator error.

No... this was from my first scanner I had, a V500. Sorry for the mess...

Here you have V850 + TX samples I collected.

https://www.flickr.com/photos/idandog/27866147816/in/photolist-JsrhUq-qLDqmc-G9Pe83-UXLRDv-qvYrQv-CZU8WM-q48wMp-rhVR4s-Rit2Pr-SnzQuj-NJzcSJ-UanFeW-PaYULT-GUDWTf-QpJD8h-SpitDm-TSQNah-SeaFAu-FMmEth-DptYiK-UdXJFt-q7JFZR-Mj22Uy-PAcDcu-DnaMkG-CzmREi-TSQGT5-DxsKqK-D5CCeG-DBo9gq-CzmJWt-SSuFZX-D8esVt-CJkpPD-EUcKtu-DC1uDX-FhQ6fM-wmpJgX-Smf38k-SyxEFZ-qPNMUa-qDjhxa-D4vVAe-F3uLMX-Rit5Xa-qgbXvN-pTbWAr-Gvn4YB-tV5k2X-D7aVYe

https://www.flickr.com/photos/red_eyes_man/24146107552/in/photolist-CMH7xb-D5cVDV-DT8Xg7-Rik4uN-CPY4D2-D8Th9B-Dw5JDj-GbZPwE-CES8QV-BYbn9v-CLuXUP-D9dMGy-E3WWc9-D8xSTh-CZUtmZ-Dv5s8T-D7hoZo-CJfttw-CLvaZk-DX2x84-Dr8HWg-DR9Qpb-Cn4sw6-DthwV9-CTNr84-CwozNi-D2QQz6-DzGaGY-DDGFMH-EW1ZBm-PqMVQr-DRa97f-Dt6LiR-uooQ6v-PntPZh-EEHdTq-Ds4PQ2-D6cB3B-PntSmG-DczR5W-E2hiNz-EtxN4A-FAk8fU-CVyRUz-EcvbEp-C8nYDo-E24Qxs-EAnjHJ-DLkvaj-DeBDFs

https://www.flickr.com/photos/red_eyes_man/25209533183/in/photolist-EAnjHJ-DLkvaj-DeBDFs-FPDyma-EUcNUN-ENWHEk-DRyfd8-Eaz4Wx-ExAo2j-EtiZMC-Ey5bcy-EtiMGq-FHQCEL-EHexSp-EpFrVR-PMd2uV-EcKU1w-FCC22e-vNMErC-DWHigU-EncEjn-CwoFaF-FAqicp-Eajm8q-DrAGjf-FqVYxk-E7PrEN-F9AhhY-EEH3os-Pf2pVq-PBbEfM-Fywq1q-EzTdDd-Pub7KS-CEBvbH-EcKyCd-GrJE54-F31G1n-GtYYEe-EtycHu-GbZFwb-GtZ1ur-GrJB1Z-ESDRrY-EpQNcF-FAfjyU-wWsrCZ-DYTVWz-F8ajFG-G4Szcd

[faberryman]

Anyway this is not a V750 scan, but a V500 one, sorry I accidentally deleted part of the text explaining that

https://www.flickr.com/photos/125592977@N05/15473025262/in/dateposted-public/

I wanted to show that even a low end V500 is able to show TX grain.
Sorry Pere, the portrait scan you posted does not show grain. If you enlarge it you see pixelation before you see grain structure, as you would expect from a low-end flatbed scanner like the V500.

[Pere Casals]

Sorry Pere, the portrait scan you posted does not show grain. If you enlarge it you see pixelation before you see grain structure, as you would expect from a low-end flatbed scanner like the V500.

Yes, pixels are larger than grains. Anyway if you take a very good scan and you reduce the image size to the 1080 lines (vert direcction) a common monitor has... then you will see someting similar. Even with a 4k monitor. This is not enlarging a crop, but showing the full image, of course.

For a lambda/lightjet print there is also a pixel count target, so it will depend on the print size.

[Pere Casals]

With the Cezanne, grain keeps getting better, i.e. finer and more natural, all the way up to 6000 spi.

Yes, this is for 35mm. But Cezanne has 2000 hardware DPI for 4x5 (1325 dpi optical), this is if placing negative oriented in the best direction as pixel density decreases with larger negative. So it is very good for 35mm but you won't be able to get good grain for 120 and LF, if you are to make big prints.

The way to get very good results with Cezanne is scanning 1" strips and stitching the strips in PS, as Seybolt report did, if I'm not mistaken.

http://www.kar.fi/Skannaus/pixelperfect1_part2_seybold_1999_vol28_nro11.pdf


In normal conditions Cezzane delivers 5300 optical points from the hardware 8000, so max efective 4x5 resolving power is 5300/4 = 1325 dpi, if scanning all at one time, rather stitching strips with PS. This is equivalent to what V500 performance obtains with rolls.

Correct me if I'm mistaken.

Still Cezanne is an amazing pre-press war horse, and 1325dpi from a 4x5 may be more than enough. And then you have amazing resolving power for 35mm...

Stitching strips it's not convenient, anyway the amount of 4x5 quality it can deliver it's amazing.

[Corran]

Blah blah blah let's not get into this again, the thread is not about Epson vs. Cezanne. You continue to state that the Cezanne actually has less resolution than an Epson scan in one go which is simply stupid and demonstrably false.

[Peter De Smidt]

Scanning in strips is not hard at all. All of the strips can be setup at once. Press "scan" and walk away. It's child's play to merge the strips manually in photoshop.

No, the Cezanne delivers more than 5300, as tested with a chrome on glass high resolution target, and as supported by the Seybold report and other users.

[Pere Casals]

Blah blah blah let's not get into this again, the thread is not about Epson vs. Cezanne. You continue to state that the Cezanne actually has less resolution than an Epson scan in one go which is simply stupid and demonstrably false.

Bryan, the Cezanne delivers more resolution than V850 for 35mm film, but much less than V850 for LF sheets. If you scan 1" strips from 4x5 sheets and stitch that in Photoshop then you get Amazing results, but if you scan the sheet in one pass then you get 8000/N dpi (hardware), where N is the negative width in inches. And then divide by 1.5 (from Seybold results) to get optical dpi.

Ask Mr De Smidt, he knows very well that machine and he can lead you to get very good results with your machine, with the stitching technique.

[Pere Casals]

Scanning in strips is not hard at all. All of the strips can be setup at once. Press "scan" and walk away. It's child's play to merge the strips manually in photoshop.

No, the Cezanne delivers more than 5300, as tested with a chrome on glass high resolution target, and as supported by the Seybold report and other users.

Hello Peter,

Seybold report stated 5300 dpi optical performance for all the image width, so for 4" this is 1325dpi , for 1" this is amazing 5300 dpi.

166371

http://www.kar.fi/Skannaus/pixelperf...ol28_nro11.pdf , page 16.


This is very consistent with the 8000 pix sensor and having a zoom optical system to have a continous variable medium scan width for the sensor. V850 has two fixed focal lenses, the highress one delivers 6400dpi (hardware) that translates from optical 2800 to 2300 depending on the axis and other factors, it works until 5.9" IIRC. The Lowress lens delivers 4800, so some practical optical 1900 to 1700 dpi for 8x10 sheets.


Cezanne is a $36k (year 1999 dollars) beast, so this gear it is not a joke: a war machine.

I'm thinking that perhaps Photoshop stitching may take advantage of BW grains to make a perfect stitch, if image has detail enough to allow a good pre-align of strips, then perhaps single grains can nail the stitching. I'm just speculating.

It would we interesting to see how stitching works in the limits of the strips, if result is seamless that would give top performance with BW sheets, but a bit it depends (I guess) on the stitching result.

[Corran]

Pere, no one cares what you think based on what you read on the internet.

[Pere Casals]

Pere, no one cares what you think based on what you read on the internet.

Look, it's not what I read in the internet, it's the Seybold report.

Bryan, it's up to you. It's you that have a Cezanne, and were stating that a D800 has similar resolving power than a 4x5 sheet, not long ago. And saying you would not give a s#it for what the Joe Cornish test (made with the Phase One sales representative UK) concluded.

We are here to learn one from the other and share knowledge and art, not to see who knows the more, IMHO.

[interneg]

Pere, no one cares what you think based on what you read on the internet.

Nor does he care to read anything that does not agree with his bizarre assertions - from the Seybold report: "The Cézanne’s result comes very close to the figures stated in the specifications. In this case, however, the interpolated result is even better, with lines visible at 120 lp/mm - beyond the manufacturer’s claimed resolution. This is surprising, and it differs from the results with the other scanners."

aka 6000ppi of real resolution... And he's still not able to understand why a Screen or a Hasselblad or a Heidelberg Tango will produce a vastly better 2000ppi scan from 4x5 than the Epson's nominal 2400, preferring to hide behind obfuscatory nonsense about sharpening & wrongheaded application of metrological techniques to regular photography.

[Corran]

...preferring to hide behind obfuscatory nonsense about sharpening & wrongheaded application of metrological techniques to regular photography.

Right.


...not to see who knows the more, IMHO.

Your behavior belies this.

I don't know why I even bother. Since this negative was still on the scanner I did a quick test. Scanned a tiny area at 3000 DPI, less than half an inch in width, and then scanned it with the boundaries set to the max width as if it was a full sheet of 4x5. Difference is minimal. Finally I scanned the small piece at 6000 DPI and resized down to 3000, which showed a small improvement. Note that this is 35mm on a tripod with a lens at f/11 - so an actual sheet of 4x5 will show less improvement due to diffraction at typical apertures of f/22 or smaller.

http://www.garrisaudiovisual.com/photosharing/scannertest-3000dpi2.jpg

To be clear, yes scanning at a higher resolution in strips is better - I've done this many times when preparing files for printing very large. In practice, the resolution achieved even at 3000 DPI and with a full 4x5 sheet is far and beyond your claimed 1350 DPI and certainly better than an Epson. There's simply no argument here. I assume some sort of software/hardware trickery that you you or I are unaware of. Let's move on.

[Pere Casals]

Nor does he care to read anything that does not agree with his bizarre assertions - from the Seybold report: "The Cézanne’s result comes very close to the figures stated in the specifications. In this case, however, the interpolated result is even better, with lines visible at 120 lp/mm - beyond the manufacturer’s claimed resolution. This is surprising, and it differs from the results with the other scanners."

aka 6000ppi of real resolution... And he's still not able to understand why a Screen or a Hasselblad or a Heidelberg Tango will produce a vastly better 2000ppi scan from 4x5 than the Epson's nominal 2400, preferring to hide behind obfuscatory nonsense about sharpening & wrongheaded application of metrological techniques to regular photography.

Please quote a bizarre assetion of those.


Hasselblad or Tango will deliver a clearly better image than a V750 if dealing with velvia deep shadows. But I don't understand what difference is there for a 1.8D BW sheet. Really I don't see the way...


If scanning 8x10 then the V750 will deliver a pretty better image than with the hassy. Note that with X5 you will need scissors for 8x10 :)

Then a Tango 8x10 scan is (or was) for several hundred $.


About Cezanne... you'll have nice results if stitching and stitching resulted ok.


Beyond this you can like more the Cezanne results and I can like more the V850.... for the rest MTF math is MTF math, so I see no discussion there.

[Pere Casals]

claimed 1350 DPI and certainly better than an Epson. There's simply no argument here. Let's move on.

Bryan, the Cezanne sensor has 8000 pix, the scanner lens zooms to see the 4" of the negative, so you have 2000 digital samples for each inch your 4x5 negative. this is a fact. Then you have a resolving power loss from the zoom lens and from digital sampling. Seybolt says that there a 1.5 loss factor, from 8000 to 5300.

This ends in around 1325dpi, optical, if not stitching strips, or course. Then the scanner firmware/driver makes a sharpening that delivers an improved look from the 1325 peformance, but this also limits the amount of sharpening you can later do with PS.

This is technical exact information, go and find a flaw in it.

If you are happy with you Cezanne I'm also completely happy with V750 results, do you think one may need more resolving power ?

https://c1.staticflickr.com/1/669/32535835184_2a10b880f7_o.jpg

[Corran]

You were just shown that it's not quite that simple. I am done debating this with you, there is no benefit.

[interneg]

Bryan, the Cezanne sensor has 8000 pix, the scanner lens zooms to see the 4" of the negative, so you have 2000 digital samples for each inch your 4x5 negative. this is a fact. Then you have a resolving power loss from the zoom lens and from digital sampling. Seybolt says that there a 1.5 loss factor, from 8000 to 5300.

This ends in around 1325dpi, optical, if not stitching strips, or course. Then the scanner firmware/driver makes a sharpening that delivers an improved look from the 1325 peformance, but this also limits the amount of sharpening you can later do with PS.

This is technical exact information, go and find a flaw in it.

6000ppi is 120 lp/mm, 5300ppi is about 105lp/mm. That is likely to be the design specification of the optical system - ie aimed to deliver 120lp/mm, but to understate the specifications in order that, no matter the sample variation, the specs are maintained. Now, when that lens zooms to cover 4x5 at 2000ppi, it needs to resolve approx 40lp/mm. See where this is going? In other words your linear scaling is wrong because you failed to consider lens design limitations at the higher end of resolution. Indeed, given your penchant for nonsensical waffle, if such a high resolving lens was sold for other uses, you'd be drooling inane phrases to praise its resolution.

[Pere Casals]

6000ppi is 120 lp/mm, 5300ppi is about 105lp/mm. That is likely to be the design specification of the optical system - ie aimed to deliver 120lp/mm, but to understate the specifications in order that, no matter the sample variation, the specs are maintained. Now, when that lens zooms to cover 4x5 at 2000ppi, it needs to resolve approx 40lp/mm. See where this is going? In other words your linear scaling is wrong because you failed to consider lens design limitations at the higher end of resolution. Indeed, given your penchant for nonsensical waffle, if such a high resolving lens was sold for other uses, you'd be drooling inane phrases to praise its resolution.

interneg, Lets make the math accurately:

2000/25.4 * 5300 / 8000 / 2 = 26 Lp/mm

Then if you try to resolve on film 26 Lp/mm detail with a 26Lp/mm system you'll have aliases, as always with any scanner.

But, gentlemen, on film 26 Lp/mm. This can be more than enough for our output... true, but 26. Then there is the stitching way...


Note that it is impossible to retain the 2000 value through the zoom lens, a 5300/8000 (Seybold) is a fair degradation factor. This is in the same way the 6400 V850 DPI ends in 2300-2800, even a worse factor.

[Pere Casals]

You were just shown that it's not quite that simple. I am done debating this with you, there is no benefit.

Bryan, there is no benefit, true. It is impossible to make you understand why you see similar Image Quality from a D800 than from a 4x5 sheet.

After you discredited the Joe Cornish's test, the Seybold report ...and Cezanne datasheet itself, I'm also not able to go further.


I'm happy with the V850, you are happy with the Cezanne, so this is a happy world.

In the other side, I've to say that I like your photograhs, I'm not able to match that by a large margin.

[Pali K]

Apples are not oranges but some oranges are apples :)

I said I was done but it's hard to stay out of another entertaining debate about scanners.

Bryan, I am pretty sure your scans are from an Epson because it clearly shows more DPI than 1325 :)

Gotta love Pere's love for Epson.

Pali

[Pere Casals]

Apples are not oranges but some oranges are apples :)

I said I was done but it's hard to stay out of another entertaining debate about scanners.

Bryan, I am pretty sure your scans are from an Epson because it clearly shows more DPI than 1325 :)

Gotta love Pere's love for Epson.

Pali

Hello Pali,

I'm happy to see you joining the pixel party. Entretainment is for sure :).

Still we should debate more about 11x14" wood gear !

Pali, did you remember these posts:

http://www.largeformatphotography.info/forum/showthread.php?132325-Changing-from-v700-to-IQSmart2-for-8x10&p=1342349&viewfull=1#post1342349

http://www.largeformatphotography.info/forum/showthread.php?132325-Changing-from-v700-to-IQSmart2-for-8x10&p=1342352&viewfull=1#post1342352

Have you tried to apply best sharpening to both V750 and expensive flatbeds before comparing?

IMHO different scanners may apply different in hardware sharpening, so to make justice a way is to sharpen all samples to it's optimum before comparing...

Regards.

[Corran]

Apples are not oranges but some oranges are apples :)

I said I was done but it's hard to stay out of another entertaining debate about scanners.

Bryan, I am pretty sure your scans are from an Epson because it clearly shows more DPI than 1325 :)

Gotta love Pere's love for Epson.

Pali

:rolleyes:
Well don't forget your popcorn. Or perhaps it's me who needs popcorn.

I just scanned a 6x9 image from yesterday @ 4500 DPI (roughly the max it can go in one go) which is a retake of my 35mm image from in the creek on Monday. Even looking at what equates to a 15x enlargement it looks fantastic. I am tempted to print this one humongous. I probably don't have the wallspace for a 60x40. On the other hand I am closing on my house in about 3 weeks and that would be a nice addition...

Anyway, [/blog]

[Pere Casals]

:rolleyes:
Well don't forget your popcorn. Or perhaps it's me who needs popcorn.

I just scanned a 6x9 image from yesterday @ 4500 DPI (roughly the max it can go in one go) which is a retake of my 35mm image from in the creek on Monday. Even looking at what equates to a 15x enlargement it looks fantastic. I am tempted to print this one humongous. I probably don't have the wallspace for a 60x40. On the other hand I am closing on my house in about 3 weeks and that would be a nice addition...

Anyway, [/blog]

Bryan, at least you have to admit we get some fun !!!!

Popcorn is always a good idea for all :)

[Peter De Smidt]

Hello Peter,

Seybold report stated 5300 dpi optical performance for all the image width, so for 4" this is 1325dpi , for 1" this is amazing 5300 dpi.

166371

http://www.kar.fi/Skannaus/pixelperf...ol28_nro11.pdf , page 16.



And if you read more of the report, you'll see that the resolution went up in a measurably objective way from the purely optical resolution by scanning at higher values due to the quality of the positioning system. [But I see that Interneg already pointed that out pages ago. Time to find some coffee.]

[Pere Casals]

And if you read more of the report, you'll see that the resolution went up in a measurably objective way from the purely optical resolution by scanning at higher values do to the quality of the positioning system. [But I see the Interneg already pointed that out pages ago. Time to find some coffee.]

I've read very well that good report. You are referencing page 27 and 28 (http://www.kar.fi/Skannaus/pixelperfect1_part2_seybold_1999_vol28_nro11.pdf),

so you speak about this table:

166381


This says that if you oversample (interpolation, with presumably further internal sharpening) then USAF 1951 resolution improves by a 15% linear, this also happens with ScanView ScanMate F8 Plus (it improves 6% with interpolation).

So this says that for 4x5 with interpolation you get 1325 dpi * 1.15 = 1524dpi. For 8x10 this would be 762 dpi.

Of course, this is without scanning strips and stitching in PS.

I think I'm not wrong, please correct me if I made any mistake in my reasoning.

[interneg]

I've read very well that good report. You are referencing page 27 and 28 (http://www.kar.fi/Skannaus/pixelperfect1_part2_seybold_1999_vol28_nro11.pdf),

so you speak about this table:

166381


This says that if you oversample (interpolation, with presumably further internal sharpening) then USAF 1951 resolution improves by a 15% linear, this also happens with ScanView ScanMate F8 Plus (it improves 6% with interpolation).

So this says that for 4x5 with interpolation you get 1325 dpi * 1.15 = 1524dpi. For 8x10 this would be 762 dpi.

Of course, this is without scanning strips and stitching in PS.

I think I'm not wrong, please correct me if I made any mistake in my reasoning.

You are wrong because you are assuming a linear scale that is not the case in reality. Just because resolution is limited at 6000ppi for 35mm, it does not linearly scale to 4x5. The Screen can capably resolve a sharp 2000ppi for 4x5 in one pass & more in multiple pass. If you were correct, the Imacon/ Hasselblads without the 8000ppi mode would not be able to resolve 2048ppi on 4x5 - yet somehow they can.

You continue to refuse to acknowledge the obviously poor performance of the Epson in comparison with regard to aberrations & other crucial sharpness robbing characteristics.

[Pere Casals]

You are wrong because you are assuming a linear scale that is not the case in reality. Just because resolution is limited at 6000ppi for 35mm, it does not linearly scale to 4x5. The Screen can capably resolve a sharp 2000ppi for 4x5 in one pass & more in multiple pass. If you were correct, the Imacon/ Hasselblads without the 8000ppi mode would not be able to resolve 2048ppi on 4x5 - yet somehow they can.

You continue to refuse to acknowledge the obviously poor performance of the Epson in comparison with regard to aberrations & other crucial sharpness robbing characteristics.


Interneg, Flextight X5 also has a loss from the 8000 hardware pixels to 6900 optical. The loss is lower than the Cezanne because the hasselblad has a way better zoom lens.

Here you have a 1951 scan form an X5:

http://www.filmscanner.info/Bilder/UsafHasselbladFlextightX1.gif

So see group and element, and check here https://en.wikipedia.org/wiki/1951_USAF_resolution_test_chart this is 7-1 -> 128 (Perhaps 6-6 -> 114) Note (I see 128*25.4*2 = 6502 Lp/mm), they say 6900. So around that.


... while the X1 delivers 6150 for 35mm.


Source: http://www.filmscanner.info/en/HasselbladFlextightX1.html


Of course when zomming Cezanne to 4x5 (or 8x10) the resolving power for the 4" (or 8") will be lower than 5300/4 (or 5300/8) optical dpi this is an easy guess as zoom optics works worse in the long side of the focal range, as you may know.

But I don't know at all how worse it will be, so I don't say it.


Cezanne is a sound scanner for 35mm and 120, for 4x5 it has severe limitations, and for 8x10 we are talking of clear sub 1000 dpi. This comes from it's pre-press nature. A perfect gear for pre-press...


Hasselblad was focused on 35mm and 120 format. They are 120 format people, as you know very well, so 4x5 capability comes as a bonus, not as the main target of the machine.

About V750, this bell it would be from a 6m high monster print http://www.largeformatphotography.info/forum/showthread.php?139643-Scanning-Resolution-Question&p=1395781&viewfull=1#post1395781

poor performance ? not for me. Belive me, the Cezanne is way worse than the cheap V750 for 8x10, and way better for 35mm. (Not considering stitching).

[Peter De Smidt]

[COLOR="#0000FF"][B] Belive me, the Cezanne is way worse than the cheap V750 for 8x10, and way better for 35mm. (Not considering stitching).

Only if you use it poorly. If you use the Epson poorly, you'll also get subpar results. It's as if you were to evaluate the Epson, but not be willing to find the ideal scan height.

I scan 8x10 on a Cezanne. Have you?

[Pere Casals]

Only if you use it poorly. If you use the Epson poorly, you'll also get subpar results.

Peter, I'm aware that if stitching strips, and if this results in a seamless image, then the result has to be awesome. But also there is no doubt that information I posted is fair, about those 5300 optical dots spreaded in 8 inches, if talking about 8x10.

[Jim Andrada]

Thank ... for the Ignore button - it does wonders for some of the more contentious among us.

[Pere Casals]

I scan 8x10 on a Cezanne. Have you?

No, I've never scanned a 8x10 with a Cezanne, but as you did that, it would be great if you post an image to show how it does it, with and without stitching strips.

I'm curious to see the practical difference from scanning straight or by joining strips with Cezanne.

I mean a sample like this one, with a crop of the 1/20 height: http://www.largeformatphotography.info/forum/showthread.php?139643-Scanning-Resolution-Question&p=1395781&viewfull=1#post1395781

I know Cezanne does a very good job for 35mm and 120, but I'd like to know how it does with 8x10 without joining strips.



Thank ...

Jim, I don't care. If you can point a single mistake in what I posted I'd be happy to learn.