Update - microtek 1800f vs. epson 4990 [Archive] - Large Format Photography Forum

Kirk Gittings

16-Nov-2005, 21:26

Continuing the older thread......After a great deal of testing, (because the first Microtek 1800f that I got developed alot of problems very quickly. It must have been dropped. It was out of alignment etc. so I returned it for a new one), I must say that I am very pleased with this new scanner. It is much closer in quality to an 848 Imacon than to an Epson 4990.

Some of my initial observations comparing Imacon 848 scans (done last summer), my trusty two year old Epson 4990 (with Silverfast AI 6 Studio) and my new Microtek 1800f (with Silverfast AI 6 Studio).

Though I did not have an Air Force target to test things at a more scientific level, I agree with all of Paul Butzi’s observations in his article on his website. My tests did not contardict any of his tests. Read his for more detailed information.

http://www.butzi.net/articles/scannersoft.htm

A summary of my observations to date with B&W scan are:

The sharpest 4x5 scans are with film holders by scanning emulsion up RGB and saving the green channel, in glass carrier emulsion down RGB green channel, but the holder scans were ever so slightly sharper than those from the glass carrier (film taped down).

There is no advantage to using the glass bed vs. the film holder for 4x5 and one big disadvantage. The calibration slot in the glass holder is also glass and collects dust, a big problem here in Albuquerque. The dust particles create holes in the calibration data and streaks in the scan. This is not an issue with the film holder as the calibration slot is open.

Some gain in sharpness by down sampling (3600 to 1800 which is optical res.) but absolutely no less noise. The sharpening is because bicubic sharper does in fact do some sharpening obviously.

The MT is much much faster than 4990 and closer to speed of Imacon.

MT has much less noise than the 4990 and the Imacon (I find the Imacons very noise)

Shadow separation is dramatically better with the MT than I could do with either 4990 or Imacon.

The MT has much less ghosting than 4990 and does not have this odd stretching on a pixel level (gear slop?) that the 4990 does at the edges of film sometimes.

If you scan your B&W RGB and pull the Green channel for B&W, the MT is significantly sharper than 4990 closer to the Imacon.

The noise level is extremely low to begin with and doesn’t need any help from multisampling.

There is no advantage with the MT in terms of sharpness by doing a dummy scan to heat the negative prior to the real scan as in the 4990.

4x5 film Film holder had a slight bow to it requiring it be taped into the main drawer holder. This taping of the 4x5 holder into the main holder slightly increased sharpness.

For me it is ultimately all in the prints. I can tell the difference in my work at 16x20 when I start with a 4990 scan vs. an Imacon scan. I only have free access to an Imacon in the summers so I am trying to approach that Imacon 16x20 level with the MT.

For me the MT 1800F with Silverfast AI Studio does indeed break the 16x20 sharpness limitatation of the 4990 plus it has allot less noise and dimensional accuracy (higher quality gearing or belts). If you don't need Digital Ice (I need it for volume commercial scans so I will keep the 4990 running for that purpose) and you want to print exhibition quality 16x20 I think this scanner is worth the extra expense over an Epson 4990. These are very personal judgements of course.

A friend of mine just bought the new Imacon 949...................

Steve Hamley

16-Nov-2005, 22:40

Thank you Kirk for a very nice and informative post. BTW, I haven't noticed any bow to the 4x5 carrier on my 1800f.

Steve

Paul Butzi

16-Nov-2005, 23:50

Thanks, Kirk, for sharing your results.

Could you write a bit more about noise levels? I've got just the 1800f, with nothing but some Imacon scans done by the local Imacon folks to compare to.

This means I've spent a lot of time trying to find ways to eke better performance out of th 1800f. One of the things I've noticed is that I see pretty dramatic improvements in noise if I do multisampling (at a big cost in time).

Have you done this comparison (single sample scanning versus, say, 8x sampling in Silverfast) to see what sort of noise reduction you get?

Ed Richards

17-Nov-2005, 07:16

What are you doing with sharpening to make the comparison? You once mentioned that you thought the Imacon was still sharpening when you turned sharpening off. I have since seen a technical post somewhere about their needing a special negative setting to tell the machine to really turn off sharpening. Does a little usm to the 1800 wipe out the difference?

Michael Mutmansky

17-Nov-2005, 07:51

Kirk,

Your comments coincide with the results that we have found in the tests we ran for the article back in the june/July issue. The Imacon especially had a high noise floor that suprised me, because If I recall correctly, it appeared to be worse than the consumer Epson and Microtek scanners.

As we discussed privately, I suspect the multisampling will probably only be of benefit for chromes with this scanner, as it's noise floor and DMAX capabilities are both very good on it's own, and well above the normal B&W negative range.

I don't want to encourage any GAS here, but the new i800 from Microtek is available, and my initial impression of it is that it has the absolute best negative carrier that any LF flatbed scanner has ever had. It has some little cams in the carrier mechanism that stretch the negative tight when it is locked into position. I can't speak for performance, but I suspect that it will be at least the equal of the i900 and the 4990. If that's the case, it will be a real nice scanner...

Ted and I discussed whether the carrier from the i800 could be used in the 1800f because it was that nice.

---Michael

Kirk Gittings

17-Nov-2005, 09:03

Paul,

I find very very little noise in the 1800f compared to the 4990 or 848. So mutlisampling has very little effect compared to the noise reduction in the 4990. The 4990 achieves its dmax by interpolation and inherently creates allot of noise and multisampling (8X seems optimum with a dummy scan first to warm up the negative) can have a major effect there. If I go and create allot of shadow noise in the 1800F by applying a very steep curve in the scanner software, the multisampling does some reduction, but technically that appears to be a horse of a different color and it seems less effective in that circumstance with the 1800F.

Ed,

I only have access to the Imacon 848 in the summers so I can't play with it at all right now. I do know this (as I am very proficient with Silverfast), that I am applying no sharpening with the 1800F and it is still in the ball park of the Imacon, what ever the Imacon is really doing. Let me say this, Imacons need regular maintenace to opperate effectively and focus critically. Allot of complaints about Imacons are due to poor maintenance. At the School of the Art Institute of Chicago, where I teach in the summers, they do regular maintenance, and the 848 I use is not one of the machines that is heavily used by all the students. It is fairly new, well maintained, and should represent them well.

With the Imacon I found that it was sharpest scanning in grayscale unlike most other scanners I have tested.

Michael,

Thanks for the feedback. Maybe you guys could bring one of those holders to the VC conference. I would like to check it out.

j.e.simmons

17-Nov-2005, 10:03

FWIW, Clyde Butcher says he uses the 1800f and that he likes it very much. I've seen his latest prints and think they are better than his silver prints were.

juan

tim atherton

17-Nov-2005, 10:18

The Imacon Flexcolor software does a number of things to the data, including sharpening.

True zero for the Unsharp Mask in the settings is actually –60 for example (so if you set it to O it is still sharpening.

And to sharpen, it "automatically converts the RGB colourspace to a LAB equivalent space then sharpens the Lightness channel then converts back to RGB". which you can also prevent by tweakign the settigns

There are a number of tips and hints here (mainy undocumented inthe manual):

http://home.ripway.com/2005-10/480178/Flexcolor4softwaretips.doc

But there is quite a bit going on with the software, which may (or may not..) account for the noise Kirk sees.

Kirk Gittings

17-Nov-2005, 10:23

Juan,

" Clyde Butcher says he uses the 1800f and that he likes it very much. I've seen his latest prints and think they are better than his silver prints were."

A couple of things interest me here. How large is he printing the ink prints from the 1800F and in what way would you say his ink prints are better?

j.e.simmons

17-Nov-2005, 12:32

The Clyde Butcher prints I've seen are along the lines of 16x20 - 20x24 or so - I didn't measure them. They are not the huge, wall sized prints that he has made in the past.

I think he gets better control of the highlights with ink than he does in his silver printing. The shadows are still frequently empty, but they don't bother me so much - perhaps he burns them in Photoshop (I got the impression from our conversation that he uses the burn and dodge tools and doesn't do much with levels and curves). The highlights have some tone in them that the silver prints lack. I shoot in Florida, too, and know that in the swamp there can be a very high SBR. Clyde says he just decides what he wants on Zone V, takes a meter reading and shoots without worrying about the shadows and highlights (TMax-100). He doesn't seem to make any attempt to scale his negatives to paper BTZS style. Perhaps the scanner is able to handle his negatives better than paper does.

Up close, the prints seem as sharp and detailed as his silver prints. I didn't notice anything I would consider a noise problem. I was impressed with the range of tones (I usually print on Azo) and I thought the print color was good. He was not very clear about just what ink he uses. I came away with the impression that he has a technical expert he relies on for equipment recommendations, and that he just uses what is recommended without paying a lot of attention to specifics.

juan

Kirk Gittings

17-Nov-2005, 12:38

Thanks Juan. Interesting.

John Brady

17-Nov-2005, 13:13

Thnaks for getting this thread going again.
This is my first post on the forum. I shoot with an ebony 45su and I currently have been using an epson 4870 to scan tmax 100 and velvia 50/100. I have never been thrilled with the results I have been getting from my epson. The scans tend to be much more soft than I would like. My goal is to print in the 24in x 30in and larger range.
I have had lots of problems with moire also.

For some unexplainable reson I still ordered a 4990 a couple of weeks ago. I just couldnt get myself to open it because I came to the conclusion that it won't be much better than the 4870. It's going back to amazon.

I am going to take a leap of faith and order the microtek 1800f. Thats what I should have done in the first place. Hopefully it will give me more acceptable results.

Juan, where are you at in Florida? I am in Naples. Clyde is having an open house this Saturday at his Venice Gallery. His Dark room will be open all day for tours. Should be interesting.

Thanks jb

Michael Mutmansky

17-Nov-2005, 13:26

Kirk, the bigger question is what size negatives is he starting with...

If he is using 8x10 to start, the 20x24 is only about a 2.5X enlargement, and well within the capabilities of the scanner.

I would suggest that a 4x5 original is going to be pushing the capabilities of the scanner somewhat, but I don't think that Clyde is shooting that small at the moment.

The same goes for all the consumer scanners out there; they all have approximately the same capabilities in terms of resolution, and getting a 20x24 print out os a 4x5 (slightly over 5X enlargement) is beyond their capabilities.

This in particular is a subject that Ted and I will be discussing at the VC conference next summer, and will be dealt with in the scanning workshops that we have planned for next year.

---Michael

Kirk Gittings

17-Nov-2005, 13:57

You are absolutely right Michael and I think he is probably shooting 8x10 from what i remember.
Kirk

John Flavell

17-Nov-2005, 14:04

I really hate to throw another wrench into this process, but does wetmounting change any of the opnions given on this topic?

Ted Harris

17-Nov-2005, 14:36

Update on Michael's comment on the i800 carrier. I spoke with Microtek's product manager for their higher end scanners yesterday. We talked at some length about the i800 holder. The product design folk say they are going to start using that holder design for all their scanners very soon. The question will be what "very soon" means once the desire gets from the new product folks to the engineers. The issue here is that the i800, as it is now donfigured, is slightly thicker than the holders for the glassless scanners; thus, the focus point would be changed, the insert might not fit in the frame right, etc. The engineers are working on it but it may not get to us as fast as the new product folks think it will. I will talk with their chief US engineer sometime in the next two weeks to get further confirmation on this.

robc

17-Nov-2005, 17:36

Thanks for update Kirk...

When I asked the question on the previous thread, one thing I was interested in is scanning 6x6 as well as 4x5. Would it be possible to do a comparison of a neg scanned at 4800 dpi on the Epson 4990 verses the same neg scanned on the MT 1800f at 4800 dpi. i.e. the MT scan would be interpolating to 4800dpi. (or scanned at 1800 dpi and upsized in PS).

thanks

Kirk Gittings

17-Nov-2005, 19:24

Rob,
I will try it this weekend, but understand the 4990 is not a 4800 optical resolution scanner either. It accomplishes that with interpolation too.

John,
Wet scanning will work with both and benefit both equally. With the 1800f it should be easier as there is no need for the bottom spacer glass to get it to the focal plane.

robc

17-Nov-2005, 19:39

Kirk,

I had understood that the 4990 was a claimed 4800 optical scanner but it was not actually capable of resolving more than around 2400 (if that). My question was really aimed at finding out if scanning on the 4990 without interpolation would be better than scanning on the 1800f and upsizing/interpolating. If the epson is actually interpolating at 4800, which I assume will degrade the image quality then the test is probably pointless. On the other hand if its not interpolating, but is just not capable of actually resolving 4800 dpi then just possibly the quality could be better than scanning on the 1800f with interpolation to 4800 dpi.

Why oh why do the likes of Epson etc have to claim 4800 optical if its not capable of that....

Roger Richards

17-Nov-2005, 19:43

Hi guys, any comments on how well the Microtek 1800f scans 8x10 negatives? I am concerned if there are problems with Newton rings. Thanks for any help.

Kirk Gittings

17-Nov-2005, 19:58

Does anyone actually know what the optical resolution of the 4990 actually is?

Roger apparently there is no problem with newton rings. I have not seen any with 4x5. You scan on the glass emulsion up. The plane of focus is above the glass.

Walt Calahan

17-Nov-2005, 20:25

Epson 4990

Optical Resolution
4800 dpi

Hardware Resolution
4800 x 9600 dpi maximum with Micro Step Drive technology

Interpolated Resolution
12,800 x 12,800 dpi

Color Depth
48-Bit RGB (internal & external)
16-Bit Grayscale (internal & external)

Dynamic Range
4.0 Dmax

Ted Harris

17-Nov-2005, 20:52

The real optical resolution of the 4990 is in the range of 1800 - 2000 at best. The manufacturer's claims (as well as those for the Canon 9950 and the MT i800) are pure theoretical limits stated for marketing purposes. Epson comes up with this number as a result of having tow smallish 2400 spi sensor arrays stacked one on top of the other ... same with the others. Dmax is also a theoretical number achieved only in an absolute situationthat bears no resemblance to real photography. The actual Dmax of the 4990 is in the range of 2.0 - 2.2. The numbers published in the View Camera article in the May-June issue for the 4870 will be a close approximation of those for the 4990 as the sensors and mechanics are very similar (perhaps even identical).

j.e.simmons

18-Nov-2005, 05:50

My impression is that Clyde Butcher was shooting 8x10 for the exhibit that I saw. The point about negative size is a good one.

John, I'm in Jacksonville. Email if you're coming this way.

juan

Kirk Gittings

18-Nov-2005, 08:25

"If he is using 8x10 to start, the 20x24 is only about a 2.5X enlargement, and well within the capabilities of the scanner."

Michael you must be working with some formula that takes the scanner's optical resolution, the negative size and the quality of the sensor and projects a maximum size for acceptable resolution. is this true?. Is there such a formula or you are making a judgement call based on experience?

Kirk Gittings

18-Nov-2005, 08:31

Walter,

I love this scanner, the 4990, I use it daily. It is capable, with very careful work to produce exhibition quality work up to 11x14. It is a truely amazing piece of technology for the price, but if those claims were correct it would be a miracle at that price. As Ted said:

"The real optical resolution of the 4990 is in the range of 1800 - 2000 at best. The manufacturer's claims (as well as those for the Canon 9950 and the MT i800) are pure theoretical limits stated for marketing purposes."

Roger Richards

18-Nov-2005, 09:51

Thanks, Kirk. This has been a very informative thread. I have been using an Epson 4870 for small prints from 4x5 and 6x7, with half decent results, but now I need a scanner to make 8x10 color prints up to about 20x24. The Microtek seems to fit the bill.

Michael Mutmansky

18-Nov-2005, 10:45

Kirk,

Based on experience. This equates to only a 10x12 from a 4x5, and as you have said above, the consumer scanners can do that well with care.

I somewhat misspoke when I said 'well within' up above, because regardless of the reproduction ratio, a sloppy scan will result in a poor result, so I didn't want to imply that this type of scan is a casual endeavour that can be done without careful attention.

---Michael

Richard Martel

18-Nov-2005, 11:36

From a personal conversation with Clyde Butcher he is using Epson Ultrachrome ink for his inkjet prints. I can not recall the RIP he is using but I think it was Colorbyte.

Kirk Gittings

18-Nov-2005, 12:42

"Thanks, Kirk. This has been a very informative thread. I have been using an Epson 4870 for small prints from 4x5 and 6x7, with half decent results, but now I need a scanner to make 8x10 color prints up to about 20x24. The Microtek seems to fit the bill."

Just to be clear. Almost all my testing has been done on B&W negs. and my recomendations really are limited to that. I would inquire with Ted Harris about color/sizes etc. related to the 1800F.

David Honey

18-Nov-2005, 14:59

To Kirk (or anyone else in the mood to answer!)

From what I'm reading lately, it looks like quality of output from the Epson 4990 is easily beaten by any of the MicroTeks above the $500 (US) mark.

Even the MT 1800f is currently selling for under $900 (US), and in the scheme of things I don't see this as a whole lot extra to spend, especially on such a crucial bit of hardware.

So, which scanner currently offers best value for money ? And why would someone buy an Epson 4870 or 4990 over say a MicroTek i800 or i900 ?

(I'm primarily interested in scanning B&W 4x5 negatives, especially if it makes a given scanner's job any easier.)

Thanks, and apologies for so many questions!

Dave

David Honey

18-Nov-2005, 15:22

Editing my own post(!) :

> 'From what I'm reading lately, it looks like quality of output from the Epson 4990 is easily beaten by any of the MicroTeks above the $500 (US) mark.' <

I probably shouldn't have said 'easily beaten'. Closer to my understanding is that the 4990 can apparently be beaten in some areas even by the cheaper ($500+) MicroTek scanners.

Is that correct, or are comparisons still close within that price range?

Thanks,
Dave

Ted Harris

18-Nov-2005, 15:35

Dave, in the $500 range the comparisons are fairly close. I prefer the i900 to the 4990 and either to the Canon as msot that have used it (I have not) have found a wide variety of faults. Both Michael and I suspect we will likethe i800 better than the i900 or the 4990 but we have not fully tested it yet.

The 4990 appears to deliver very slightly beteter resolution than the i900.

The i900 absolutely has significantly better dMax than the 4990.

The i900 has somewhat less color fringing than the 4990.

Ther ei sno way fo rus to do any sort of meaningful comparison on the motors or optis on either but they both appear to have goo dbuild quality.

It really is a tossup in this price range BUT,yes IMO, it is worth spending the extra on the 1800f and the next step up is to the 2500f (which is no longer being imported by Microtek USA as there was virtually zero demand after the relese of the 1800f) or an Imacon ... a quantum leap in $$ in either case ... or the used market.

Finally, scanning b&w doesn't necessarily make the scanners job easier as we recommend you scan in RGB and throw away two of the three channels .... you capture much more of the available information that way.

QT Luong

18-Nov-2005, 16:25

And why would someone buy an Epson 4870 or 4990 over say a MicroTek i800 or i900 ?

For the digital ICE.

Kirk Gittings

18-Nov-2005, 18:08

Unfortunately, QT is right. I use a 4990 for volume commercial scans because of the ICE, but I scan my personal 4x5 B&W on the 1800F for better sharpness in 16x20 prints.. To me the best in-house scanner for a commercial/art studio like mine would be a scanner of the quality of the 1800F with Digital Ice or a new Nikon that would handle 4x5.

Ed Richards

18-Nov-2005, 18:39

> Finally, scanning b&w doesn't necessarily make the scanners job easier as we recommend you scan in RGB and throw away two of the three channels .... you capture much more of the available information that way.

From the scanner's perspective, I think they always scan in RGB. The issue is whether you strip out the other two channels in the scanner software or in PS. I use Vuescan and strip it out as it comes out of the scanner so I do not get a HUGE intermediate file.

David Honey

18-Nov-2005, 19:26

OK, thanks. I think I'm getting the picture..

So, apart from not having digi-ICE (fantasizing for a minute you always have clean negs or don't mind fixing it in PS), are the mid-price (i800/i900) MicroTeks (or anything else in the under-$1000 price range) capable of better/sharper results than the 4990 on 4x5" B&W negs?

Or is the 1800f a no-brainer at this point?

Again, apologies for all the questions, but you guys are obviously the best resource as far as LF neg. scanning goes(!)

Thanks again.

Dave

Paul Kierstead

18-Nov-2005, 19:40

On most -- if not all -- scanners ICE does not work on B&W negs (or Kodachrome for that matter...). ICE depends on the IR transmissive properties of film and the B&W does not transmit IR much, defeating it. So if B&W is your game, ICE is not a player.

If there is a scanner out there which does full ICE on B&W, I would love to hear about it...

Kirk Gittings

18-Nov-2005, 21:26

David,

For a couple of years I scanned 4x5 B&W on a 4990. At 11 x14 I honestly could not see much difference between that and a drum scan at that size. For my current retrospective I started pushing the b&w sizes up to 16x20 but found I was having to sharpen too much and creating a very sharpened look at that size. In the summers I teach at the Art institute of Chicago so this summer I did some tests on their Imacon 848. Prints at 16x20 from that were noticeably sharper than scans from the 4990. I rescanned and reworked the 16x20's for the show with scans from the Imacon. After I got that show up and a paired show at a private gallery, I started investigating what I could do in house (I live in New Mexico) to achieve my desired sharpness for a 16x20 b&w. Based on the recommendations of Ted Harris and Michael Mutmansky (two of the most knowledgeable guys around whose opinion I really value) I bought a MT1800F and started testing it against my 4990. The results are posted above.

Kirk Gittings

18-Nov-2005, 22:47

Left is a small section from a 100% 104 MB file of a Tri-x/HC 110 4x5 negative with no sharpening 8X sampling in SF AI Studio. Right is similar negative with a simple levels adjustment to match midtone of the left negative, 1X no sharpening. Now you can see why i needed the 1800F for my 16x20 B&W.

http://www.gittingsphoto.com/Articles/DUO%202.jpg

David Honey

18-Nov-2005, 23:03

Thanks Kirk.

um... I may not have understood correctly -- are they labeled right?

Dave

robc

18-Nov-2005, 23:24

David,

The point here is that the 4990 is giving a softer scan = less resolution. The 1800f is resolving the grain structure = sharper more detailed scan.

David Honey

18-Nov-2005, 23:45

(duh!) OK, I get it now.

Thanks Rob. Thanks again Kirk.

Ted Harris

19-Nov-2005, 08:58

Dave, unfortunately there is very little that is a 'no brainer' in the LF world. Here, if you are planning on printing 16x20 with frequency the 1800f is a no brainer. If you are seldom going to print larger than 8x10 and then only the infrequent 11x14 then it is no loinger such a no brainer. Yes it will give you a far superior scan with lost more information ... whether or not you need that information is the subject of another,or many other threads.

Don't forget that if you do all yoru printing yourself you CAN'T print 16x20 unless you own one of the Epson professional wide format printers or something similar. For the 'entry level' here from Epson we aere talking ~$2000.

David Honey

19-Nov-2005, 10:41

Thanks Ted.

Small prints are all I can do (and want to do) right now, so a 4990 would probably be good enough get me started.

I guess the good thing about film is that the the negs are the repository of your 'data', not a digital file of finite size. You can always re-scan later if you need to, on a better machine.

Cheers,
Dave

Kirk Gittings

19-Nov-2005, 12:51

Something bothered me about the extreme differences in the previous tests that I posted above. There should have been an evident difference but not that great. So I went back and rescanned the 4990 test with only one pass RGB and kept the blue channel. That is the way to get the absolute sharpest scan with the 4990 that I know of. The new comparison is here. The differences are still evident but this is a fairer test. What is not so evident is the huge amount of noise that is also present in the 4990 scan shadow areas that is not present in the 1800F scan. If you want to download these yourself and look at them on your computer go to: http://www.gittingsphoto.com/Articles/ and pick New DUO.

http://www.gittingsphoto.com/Articles/New%20DUO.jpg

My apologies to Epson. I did not mean to mislead anyone. It was very late.

Ed Richards

19-Nov-2005, 20:46

You may be doing this, but I have lost track in the posts:

Try scanning at 4800 DPI on the 4990, single pass. Then load it into photoshop, dump the extra channels, keeping the blue. Then using PS change image size command to downsample the file to 2400 (divide the pixels in one dimension by 2 and keep the proportions constant). See if that reduces the noise. (It acts as a 4 fold sample.) I do this in one shot with Vuescan, but Silverfast just reduces the scan resolution if you tell it to downsample.

Kirk Gittings

19-Nov-2005, 21:45

Ed,
If that works for you post some examples here. The 1800F has so little noise it is not an issue for me. I am buried with real work and my scanners are tied up for another week trying to meet a big deadline.

Ed Richards

19-Nov-2005, 22:20

Do not worry about it. I was just curious about what resolution you are using when you scan with the 4990 and whether you are downsampling in PS.

Paul B and I am are working on a scan test of the 1800 compared to the 9950f, which we will eventually get done, depending on the vageries of getting mail from Baton Rogue to Wasthington state.:-)

Keith S. Walklet

19-Nov-2005, 23:27

The scan samples are very telling, but I have a couple of questions:

Is it not true that all scanner software introduces a given level of sharpening in the base (no sharpening) mode?

Seems to me that I have heard this to be the case. For example, is the default (unsharpened) Imacon actually run through a USM filter (say of 1.0) as part of the scanning process? If this is the case, how much of the difference in the scan quality can be attributed to the manufacturer's chosen USM and how much to optics?

The reason I ask, is if one can achieve similar (better) results by boosting the sharpening at the scanner level, why not do so? I've asked this question of a number of people who have seen my side by side comparisons from the Epson 4870 to the Nikon 8000 to the TANGO. Mind you, when I began the tests, I intended to illustrate a great disparity in quality. Instead, I find the lower end technology has narrowed the quality gap significantly with the high end counterparts.

The other note I am wondering about is Ted's reference to the 4870/4990 only having an optical resolution of 2000. My own tests seem to contradict this, showing that their is a consistent, marked improvement in the data quality of images I scan on the Epson as I boost the resolution from 1200 to 2400 to 3200 to 4800. Once I move beyond 4800 to 9600, however, their is a significant, very visible drop in quality. This behavior makes it easier to believe the manufacturer's claim of an optical resolution of 4800. Clearly, beyond 4800 dpi, the interpolation (and a poor one at that) kicks in and the image quality suffers.

Paul Butzi

19-Nov-2005, 23:42

Ted's reference to the 4870/4990 only having an optical resolution of 2000. My own tests seem to contradict this, showing that their is a consistent, marked improvement in the data quality of images I scan on the Epson as I boost the resolution from 1200 to 2400 to 3200 to 4800.

Well, it's perfectly possible that the scan quality will improve as you increase the sample frequency beyond the optical resolution. It might, for instance, reduce noise, or even do a better job of approaching the true optical resolution. But the improvement each time you boost the scanning frequency does not mean that you're getting the advertised resolution, nor does it neccessarily mean you're getting higher resolution.

The easy way to tell is for someone with a 4870 or 4990 to scan a 1951 USAF test target, look at the scans, and read the actual resolution from the scan.

Test targets are not cheap but not expensive - $115 for a chrome on glass positive 2" (same size as a mounted 35mm slide) from Edmund Optics www.edmundoptics.com/onlinecatalog/displayproduct.cfm?productID=1790 (http://www.edmundoptics.com/onlinecatalog/displayproduct.cfm?productID=1790)

Ed Richards

20-Nov-2005, 07:20

> It might, for instance, reduce noise, or even do a better job of approaching the true optical resolution.

This is clearly going on with the Canon. I doubt that true resolution ends up being even 2000, but if you have an imperfect optical system, esp. one with random variations as you would get from microslop in the positioning, more samples makes for a better signal.

robc

20-Nov-2005, 07:55

more samples only makes for a better signal if, and only if, each sample is perfectly aligned.
Some scanners will sample x times without moving the scan head. Others do x passes with the head sampling once on each pass. In this instance you have to rely on the scanner being able to put the head in the exact same position on each pass. Thats a tall order for a cheap scanner. So which method does the Microtek, Canon and Epson use?

Ed Richards

20-Nov-2005, 09:49

Rob,

Let me clarify. I am talking about a single pass scan at 4800, not multi-sampling with the scanner. When you downsample that in PS (or directly in Vuescan), each pixel is the average of 4 surrounding pixels That gives you the effect of multi-sampling, but without the scanner having to doing any pecular dance that will just magnify the slop. If you do multi-sampling in the scanner, the Canon leaves the head in one place and takes multiple samples, then moves it. This is less error prone mechanically than rescanning and adding the passes, but the heat distorts the negative and I think it washes out the value. Despite Silverfast's claims about being able to realign multiple scans on a consumer grade scanner, I have not had good luck with their process. As the photo-i site's review of the Canon shows, it may have less noise than the Epson, thus I would see less value in the multi-sampling than would Epson users.

Kirk Gittings

20-Nov-2005, 12:37

"the Canon leaves the head in one place and takes multiple samples, then moves it. This is less error prone mechanically than rescanning and adding the passes, but the heat distorts the negative and I think it washes out the value. "

I tried this scanner (4 of them to be exact) and returned it 4 times for replacement because the sampling process described above left repeated light patterns in middle gray skies scans (subtle). I have never seen one of these Canon scanners that did not do this (I have looked at scans from about 12 of these), but I have seen many photographers who could not see it in their prints. Canon kept telling me that I was being "too picky with a scanner at this price point", but as a large format photographer I am very picky and don't accept patterns in my mid-tone clear skies. I tried and returned 4 of these scanners because of this and finally (after much debate) got Canon to give me a full refund.

"Despite Silverfast's claims about being able to realign multiple scans on a consumer grade scanner, I have not had good luck with their process." It is not always perfect. I think because of environmental conditions like room temperatures. Do not go past 8X and do a "dummy scan" first to heat up the negative etc.

Ed Richards

20-Nov-2005, 13:11

Kirk,

I think that the mottling you saw is caused by the subtle shifts from the selective heating and cooling as the scan head passes over the negative for multi-sampling. I do not use the multi-sampling option in a single pass option because of this.

Kirk Gittings

20-Nov-2005, 16:58

Sorry Ed I did not use multisampling on any of them. I only wanted the scanner for doing volume color commercial work where I don't really care about noise much.

Kirk Gittings

27-Nov-2005, 15:01

Air Force Test Target thanks to the kindness of John Hennesey

http://www.gittingsphoto.com/Articles/USAF%20Compare2.jpg

Kirk Gittings

27-Nov-2005, 15:26

Sorry I forgot the onfo. The 35mm target was placed where it gave the sharpest scan after trying all possiblilities. On the 4990 that was emulsion down (hence the newton rings) on the 1800F it was emulsion up. I think that as the target is on a sheet of glass it actually does not presisely match the plane of focus of either machine. I actually think the earlier test with real film is a better comparison, because you can put the emulsionof real film where it is designed to go:

Silverfast AI 6 Studio

No Sharpening or correction of any kind

1800 DPI (downsampled for internet, oversampling was tried-no gains in noise or sharpness)

displayed at 100%

robc

28-Nov-2005, 10:17

Thanks for the update Kirk,

I'm not sure what it tells me though. What would be really useful would be a scan with 4990 at its native 4800 dpi and a scan from the 1800f at its native 1800dpi. Then crop a small central section of each and save as png24 with no resizing and post to web. That would allow me to upsize the 1800f and compare to the 4990 at 4800 dpi.
I need more dpi than 1800 for my medium format work and am unsure whether upsizing from the 1800f or scanning at 4800 with the 4990 is the better option.

Your efforts on this are much appreciated.

Kirk Gittings

28-Nov-2005, 11:04

The trouble is almost no one but Epson thinks that the 4990 has an optical resolution of 4800. The 4800 scan I did didn't look any better than the 1800 scan. If you downres that 4800 scan, other factors come into play which defeat a proper comparison like bicubic sharpener which sharpens the image slightly. As Sandy King has said about the 4990 being a 4800 OR, "if this is I'll eat my shorts". Yet no one seems to know what it really is.

But the bigger problem with the target since it is mounted on 1/8 inch glass is that it does not match the plane of focus in any way, as film does, on either machine. The plane of focus on both machines is about in the middle of the glass and contrary to popular opinion, I find almost no depth of field on flatbeds.

Because of the plane of focus issue, I am not inclined to waste any more time on the target.

Kirk Gittings

28-Nov-2005, 11:18

Also,

Medium format is a different deal altogether, because it is so hard to get the film to lie flat. Are you shooting 6x9? If not the film holders for the 1800F will not work. If you are not doing high volume MF scans and just need low volume but highest quality, you should probably get the 1800F and wet scan them emulsion up on the glass bed. It is the onlyflatbed scanner that is virtually designed for this. That puts the emulsion at the plane of focus and with the film adhered to the glass it is perfectly flat. You can't do this on the 4990 without a separate piece of glass to get the trans. up to the plane of focus. It doubles the complexity.

Otherwise the 4990 holders are better for multi-size dry MF. Don't cut the film individually. Put the film in a book overnight to get it flat and then scan it.

And no, I do not have the time to do a comparative wet scan test on MF.

I am satisfied that on 4x5 b&w that the 18ooF is a superior product and that is the sole reason I bought it. I continue to use the 4990 for batch scanning 6x9 color for my business.

robc

28-Nov-2005, 12:35

OK end of topic...

Jeremy Holmes

5-Dec-2005, 15:40

I think the key thing to remember here is the price point of each unit. The Artixscan 1800f is priced at $999 MSRP while the 4990 is priced at roughly half that. I would compare the 4990 more towards the i8oo or i900 scanner if you wanted to do a fair comparisson. With that being said I think that the Microtek offerings are quite good. I may be biased as I've used them in the past, but overall I think that the quality of the scans is phenomenal. The best thing to do though would be to try all the scanners if you had the opportunity to. I agree with Ted that the stated resolution is one thing, but is more of a marketing tool rather than a real life tool. I personally start at the price point of all the units that I'm looking at and then look at the background of each unit and its potential for scans. Having done that I personally believe that the optics and outcome of the 1800f and i900 put it in the top tier of scanners.

Kirk Gittings

5-Dec-2005, 18:29

Jeremy,

" I would compare the 4990 more towards the i8oo or i900 scanner if you wanted to do a fair comparisson. "

This thread started because someone specifically wanted to know how an Epson 4990 and a Microtek 1800F stacked up against each other. Since I own both I offered to do a test. Unfortunately I am a working photographer and do not have time or interest to buy or borrow every scanner out there and test them.

Greg Bowl

7-Dec-2005, 17:49

I've just recently tested an 1800F and I900 with 8x10 chromes, looking for noise levels and

shadow depth. The 1800 had the least noise, the 900 had horizontal noise / lines from the scan head, as well as more green channel noise in general..

I was going to try an epson 4990, but I probably will not at this point, I don't think it will come to the level of the 1800 based on the comments I'ver read so far. I have about 25 8x10 chromes to scan, some are old E3 process film that is seriously color shifting as it ages. Has anyone made a holder to keep 8x10 film flat without taping, or devised some method with the 1800 (or 900) holders ?

I'm eventally planning on making some pretty large prints from the 8x10 images, so the noise in dmax areas is an issue. I didn't have the opportunity to try multi-sampling with either scanner, I'm not sure that would deal with horizontal line noise anyway. Any thoughts?

Kirk Gittings

10-Dec-2005, 14:22

Here are some excellent tests of the 1800f that mirror some of my own conclusions by Scott Rosenberg.

Kirk Gittings

10-Dec-2005, 14:27

Sorry my paste finger got disoriented.

http://www.srosenberg.com/Website7/1800f.html

Kirk Gittings

10-Dec-2005, 14:29

Again:

www.srosenberg.com/Website7/1800f.html (http://www.srosenberg.com/Website7/1800f.html)

Ed Richards

10-Dec-2005, 19:31

> I was going to try an epson 4990, but I probably will not at this point,

If money matters, you should try the Canon 9950. Paul and I found it to be comparable with the 1800, with perhaps lower noise with the 1800 and better resolution with the 9950. Since then, Vuescan has been tweaked to improve the performance of the 9950, and I will be running the test again to see if there is a difference. Seems to be, but might be wishful thinking. One big change for someone interested in 8x10 is that Vuewscan's latest version can downsample on the fly, so the full file does not need to go into memory, allowing higher rez scanning without using up all the memory.

Kirk Gittings

10-Dec-2005, 21:25

My last 4990 after just 8 months took a dive this week. Of course I scan around 50 images a week. Which is hard on these plastic beasts. I only use it for 120 commercial images so I replaced it with a Nikon 8000. In less than three years I have gone thru 4 Canon 9950's, 1 Epson 3200, 2 4870's, 3 4990's, two Microtek 1800f's and now a used Nikon 8000. So my current stable is the Microtek for 4x5 B&w and the Nikon for the commercial volume color. Clearly these under $1000 scanners are not made for serious volume professional use. I'm not sure what the solution is here. I would love to find a reasonably priced scanner that would do it all. A drum scanner is simply too slow for the volume we need. I can't afford an Imacon, I could almost by a car with that and I need a new SUV, my current one has 200,000 miles on it and still have one kid in college. It is starting to make some sense to shoot the commercial work digitally.

QT Luong

11-Dec-2005, 21:34

I am a bit surprised that your scanners fail so much. Obviously, it is not a LF scanner and therefore has smaller moving parts, but my Nikon 4000 has scanned more than 10,000 images in a few years, and is still going strong.

QT Luong

11-Dec-2005, 21:37

Kirk, don't bang on your scanner when the scan doesn't come out as clean as you hoped :-)

Ed Richards

12-Dec-2005, 06:44

Do I remember correctly from a previous post that you were turning it on it's side? If so, that would contribute to its short life. You did not wear out the 9950s, you did not like them.:-) (Canon offers a 3 year swap out warranty for a nominal sum for 9950.)

It is hard to imagine any commerical work done in 120 that could not be done as well with digital, esp. if a scan from a 4990 is good eough. The Canon 5 D is now an affordable way in to a full frame camera, which is critical to doing wide angle. If you are doing 50 images a week, film costs have to be adding up, plus all the scan time and post processing to get to where you would start with a digital image. The time you saved could translate into more jobs and income, or more time for LF.

Frank Petronio

12-Dec-2005, 06:54

I haven't found much commercial work that can't be done with a D2X or Canon type of camera. I'd save the hassle of scanning for your image archive and new large format work.

At the rate you are going through scanners a DSLR would save you a lot of money.

Greg Bowl

12-Dec-2005, 07:57

I decided to buy a demo I900 from Calumet and let it multisample the dmax noise away. If my project prints do what I'd like, then I can justify having the 8x10 chromes drum scanned before they do finally disintigtrate, color wise, anyway. I did find that Silverfast was pretty good, I hadn't used it before, and the extreme red cast overtaking the old E3 process chromes was easy to correct out. I hadn't bought a flatbed since my last Umax Powerlook 1200, which I had Binuscan for, and it didn't handle the old film dmax areas, and the E3 film itself seemed to give the scanner fits. The 1800 is a less noisy scanner for certain, but not sure the $1000 was smart at this point. Ultimately I don't know where the market for large format scan technology is going. I believe the 1800 is considered "discontinued" by Microtek, there isn't enough demand to continue producing machines that can handle this format. I think the newer I700 has a thinner film path than the I900, which is probably going to be the trend. I actually don't shoot film at all anymore, I'm just trying to digitize old images before they go away, to get what I consider "final" scans. It's too bad, there is nothing like an 11x14 original chrome, nothing at all...

Paul Butzi

12-Dec-2005, 08:18

re: medium format and scanner versus an EOS-5d.

For some years I've been toying with the idea of some medium format system, mostly after going out and working on the beach with a guy who uses a Pentax 6x7. Watching him work inspires great envy because it's so darn fast - and I'm not particularly slow with the 4x5. And for prints up to 16x20, the prints from carefully done 6x7 look darn good.

But some time ago, a friend sent me some images from an EOS-1d, and I printed them pretty large and was very impressed. I thought, actually, that they were competitive with 6x7.

So when the EOS-5d came out, I snapped one up, and I've been using it fairly heavily.

Working carefully (that is, off a solid tripod, mirror locked up, lens well shaded, careful selection of f-stop, etc.) the images are awesome at 16x24. Shot at ISO 100, they're essentially noise free, which has an incredible impact on the perceived quality of the image at that size. By 22x33, the prints are lacking in resolution and the lack of noise no longer compensates. On the other hand, at that size you'd be seeing degradation from a 6x7 scan, too, if it's off a scanner like the 1800f or similar.

This has totally killed my desire to buy anything medium format with the expectation that I'd run film through it.

To be honest, I've been stunned by the image quality of the 5d. It's simply way, way better than even my most optimistic dreams for full frame 35mm digital.

The question, of course, is what the cost would be to switch to a 5d and a complement of lenses from whatever MF system you're using now.

But if you're doing a LOT of MF work, and scanning costs (cost of scanners and time to scan negs) are eating you up, you might be able to recover those costs very quickly.

Kirk Gittings

12-Dec-2005, 08:57

Ed,
Yes I do do a small amount of scanning with the 4990 on its side as explained in earlier threads. Many people have taken to this practice with no apparent difficulties but that is less than about 5% of my scans done that way on the last 4990.

Don't be foolish enough to think that I am doing something wrong here. The really frightening thing is that 7 of the above mentioned scanners were bad almost OUT OF THE BOX! They didn't last a week. Now I really check these things out thoroughly when I get them. I really put them through their paces and look at the scans in depth looking for banding, bad alignment etc. But some of them simply had major failures in the first week. These things are simply not made well. Pull one apart and look at the tracks and belts. As a former industrial mechanic, the mechanical parts of these look like toy parts. Nor can they be made much better at this price point. They are simply not made for heavy critical professional use.

Paul, That has been my thinking too. Full chip Canons with PC lenese are the standard now of digital architectural photography.

Ed Richards

12-Dec-2005, 16:48

> They are simply not made for heavy critical professional use.

I agree - if nothing else, they are much slower doing high rez scans and that matters if you are doing this for money.

> As a former industrial mechanic, the mechanical parts of these look like toy parts. Nor can they be made much better at this price point.

I have learned that this part does not always follow - really good design means nothing is stronger than it needs to be, and we often do not know what real weak link is. I remember having the same feeling when I saw my first overhead cam driven by a belt instead of a chain, but it turns out belts work pretty well.

Lee Hamiel

13-Dec-2005, 12:49

Hello to this forum as this is my first time here.

Great information & I look forward to more.

I find this thread of great interest as I am in the process of getting a scanner to do 4x5 & have been trying to come to a conclusion as to what will work best for me.

A year or so ago my son & I had visited Clyde Butcher at his Venice, Florida studio & were very lucky to have a long chat with him in his office where he was doing LF printing on an Epson LF printer - at this point in time I was not shooting any LF & Clyde was very pleasant & talkative as my son explained his desire to become a professional photographer.

At the time Clyde said he was using a Microtek scanner - don't know the model but from other posts here it seems that it was the 1800f - I also don't know the model of the Epson printer but it was one of the larger one - 9600? Anyhow - the print coming out of the printer was in the range of 3'x5' & the quality was very impressive. I have done a lot of B&W darkroom printing & I was blown away by the quality. I don't believe Clyde has an expert advising him as he admitted that he has taught himself by trial & error.

Regards

Kirk Gittings

13-Dec-2005, 13:39

Welcome Lee,

I think you will find that by learning from the experiences posted here that you can shortcut some of that trial and error time.

Frank Petronio

13-Dec-2005, 13:41

I keep think back to the 30x40 Iris I made from am 8-bit greyscale Kurzweil flatbed ($3600) circa 1991. It is still hanging in my garage, hardly faded with a nice full range and detail.

For double page spreads from 4x5 for commercial work, my old Epson 3200 did just fine. And the 1640 before that was not bad. Even the old Agfa Arcus did a lot of publication work.

Since most of your requirements are way over what most commercial artists need (11x17 @ 300 dpi versus 30x40 @ 300 dpi) I bet you'll never be satisfied with any of the sub $1000 flatbeds. Wouldn't a Scitex-Creo-Kodak Smart Scanner series flatbed be the real solution for high quality, speedy production, and build quality?