Help w/ Screen Cezanne 5000

[Peter York]

Hi all,

I recently had the opportunity to test a Screen Cezanne 5000 w/ ColorGenius v.2 software that I am considering buying, and the results not what I was expecting.

A Fuji Velvia 100 4x5 chrome was used for the testing. I cannot say if the chrome is critically sharp, but it includes areas that are quite sharp and areas that are out of focus (the shot was at f/32 with a Kodak 100mm wide field ektar). The chrome was placed in the center of the scanning bed (approximately). All tests, wet and dry (w/ an anti-newton overlay), over the entire area of the chrome and with small selected areas of the chrome, at 3000 dpi and 5000 dpi, produced results less sharp than a scan from my Epson 4990 that I used for comparison.

Should I write this scanner off, or are there any adjustments I can make for a re-test to troubleshoot the scanner? The operator had been wet-mounting to an acrylic bed rather than the standard tray. The standard tray looks ok except for some fine scratches, but I can have a glass or acrylic plate made if necessary. Would a sharper chrome, perhaps from a 35mm slide, give any better indication of the scan quality? Can some calibration or diagnostic can be run? Please help! Thank You.

[Pfeiffer Duckett]

'Less sharp' is such a difficult quantification. Do you have the respective images you can post or link us to?

[Peter York]

I'll try to get some images up soon...

[Peter De Smidt]

A scan of a 4x5 inch piece of film with a Cezanne will involve interpolation at 5000 and 3000 dpi. 2100 would give you straight optical resolution, assuming you put the long side of the film parallel to the front of the scanner. (8000 element ccd over 3.75 inches = 2133 spi.)

Make sure that in both cases no sharpening is applied by the scanning software.

[Peter York]

Peter,

My first thought was that with a 3000 dpi scan over the entire chrome, the interpolation might be an issue. The 5000 dpi scans were of a small area on the chrome (they are 3160x3125 pixels). These small sections should (in a properly functioning unit) resolve at 5000 with no interpolation, right?

[Peter De Smidt]

Yep, at less than 1.5" front to back you should get the full optical resolution.

[Joerg Krusche]

Peter,

1. the Cezanne is specified to deliver 5000 dpi actual optical resolution (?) .. provided the scanning area is small enough .. if so has anyone challenged this with for example a scan of an USAF chart 1951 ... and which max number (group and element) of line pairs was obtained ?

2. Which resolution values to enter do you recommend to obtain optical resolution only without interpolation .. does this question make any sense ? ..or does it just not matter ?

The question for actual max. optical resolution does not mean that it is relevant for actual scanning of sheet film .. just would like to know what to expect .. and/or how serious one may take specs ?

best,

joerg

[Peter York]

I'm having image hosting issues... I'll try to have images up later today.

I recall finding the sharpening button accidentially checked in Epson Scan a while ago, and the test scan I used for comparison probably had sharpening applied. I re-scanned the image and the Cezanne is better, but slightly. If my wide field ektar is similar to the sample tested by Christopher Perez/Kerry Thalmann, then at f/32 the lens can only resolve about 2300 dpi in the center. Add in my technique, etc. and the image I tested may be within the capabilities of my Epson 4990.

Since I do not have a USAF 1951 target, I'm planing on re-testing with a 35mm slide and comparing the results against a Nikon Coolscan 5000. This should say much more about the scanner's capability. Any thoughts on if I should shoot some test shots? If so, do you have a recommended procedure?

Thanks all!

[Peter De Smidt]

The Seybold report measured the optical resolution of the a number of professional scanners. I no longer have my copy of the report, but according to them the Cezanne was capable of about 5600 spi.

The Cezanne uses an 8000 element ccd sensor. It is positioned perpendicular to the front of the scanner. Hence, to obtain maximum optical resolution, position the long end of the film parallel to the front of the scanner. To find the optical resolution, measure the image area of the shortest side of the film. Divide that number into 8000. That'll be near the optical resolution that the scanner can produce in one pass. If more resolution is needed, scan slightly overlapping strips.

All of this is complicated since it's positioning system is good enough to give higher resolution in the direction parallel to the front of the scanner. This explains the Seybold Report's findings. The only way to determine what's best with a given film is to do some experimenting. Scanning is a complex interaction between the light source, film, lens, and sensor, and sometimes unexpected results pop up.

With a Cezanne, it's important to mask off non-image areas of the bed. Otherwise flare can cause problems.

[Peter De Smidt]

Peter,

You want an area with fine detail (to test resolving power), an area with smooth tones (to see how the scanner handles grain), and a representative dynamic range (to check noise.) If you shoot both fine-grained film and large grained film. Check both. The Coolscan excels with fine-grained film, but in my experience doesn't do as good a job with grainier film.

[Joerg Krusche]

Hi,

the Seybold reports ... hmmmm ... much better would be if someone of us had checked with a USAF 1951,

best,

Joerg

[Peter De Smidt]

Hi,

the Seybold reports ... hmmmm ... much better would be if someone of us had checked with a USAF 1951,

best,

Joerg

Yes, of course. If you'd like to volunteer a USAF 1951 slide, I'd be happy to scan it.

-Peter

[Bob McCarthy]

Berfore going very far I would put a strong loupe to the original transparency that didn't look sharp.

On more than one occasion, any softness I saw in a cezanne scan was due to softness in the original.

just a suggestion!

bob

[Bob McCarthy]

I used to own a 4990, still have a Nikon 5000 and neither hold a candle to my Cezanne. Could the bed be poorly done, I have the proper original bed, it sounded like you have a modified or reworked unit??

bob

[Joerg Krusche]

Peter and Bob,


1. The USAF 1951 chart on film can be purchased here for around 50 USD .. I am playing with it at the moment .. and it may give quite some interesting insight.. Peter .. if still interested I may send you the the chart in a few weeks .. would then need your ship to address. I would be very interested to have your assessment then.

2. I may have access to a Cezanne in about two months .. and there is no question .. that is a great scanner .. just like the Lanovia, Eversmart etc.

3. With regard to resolution of scanners .. I have the feeling that this is an area where specifications may be way off from what you may expect in real life .. a seemingly good example may be the often discussed Epsons .. with specified 4800/6400 dpi .. and actual may be 2000 dpi .. if at all .. and only after some strong sharpening.

Those specified 4800/6400 dpi may be defended by the manufacturer using whichever model .. it is hypothetical resolution .. in private life many might just call it lxxx .. the friendliest word might be "unrealistic".

As a customer .. even if buying used/seconhand I just want to know what is real.. and the USAF 1951 test is a favourable one for the scanner due to max. contrast of the test patterns.

4. Personally I believe that a scanner that delivers true 1800 to 2100 dpi without sharpening is all one may need for MF to LF and if scanner does deliver true 2500 dpi .. that is a superb flatbed scanner .. and a joy to work with.

It may be a nice exercise to try camera/lens on Norman Koren's test chart and then scanner or enlarger on the mentioned above chart.. but then return to photographing .. appeal of most images is not a function of resolution..


Best regards,

Joerg

[Bob McCarthy]

I agree with you that specifications and especially resolution are not the measure of the worth of the scanner.

But I have to disagree regarding the Cezanne being comparable to the Epson. Epson gives you the resolution of the stepper motor and the theoretical resolution of the sensor based upon the number of sensing pixels in its line array. But it forgets about the lens being somehow in play in the number it offers to its prospective customers.

With the high end prepress flatbed scanners it is the opposite, the end customer is knowledgable and wants the real specs.

The screen stepper is light years more accurate than a few hundred dollar consumer scanner. These things were $30,000 new.

The lense is a high dollar proper one.

The lense can zoom in to the specific area being scanned with great accuracy and autofocus on the subject matter. No fixed focus plastic lens here.

If it used Epsons way of claiming, it would be an 8000 spi scanner btw.

5000+ is the optical resolution with the Cezanne being properly adjusted and in alignment.

bob

[Peter De Smidt]

Where can it be bought for $50? Edmund lists it at $117:
http://www.edmundoptics.com/onlinecatalog/displayproduct.cfm?productID=1790&PageNum=1&StartRow=1&itemid=9125

[Peter De Smidt]

And that's for the low resolution version. How high does the standard version allow one to test to?

[Joerg Krusche]

Peter,

you can buy it here in Germany .. I correct myself .. gave the price from memory .. it is 59 euros.. the test chart goes > 11000 dpi ..

My point is that it may be informative to check the imaging chain .. camera/lens/film .. scanner or darkroom .. to have a better feel how elements interact and where spending money will will improve the final result .. there is a first class article by Tim Vitale (2007) in the net .. he deals with all elements that are interlinked in the imaging chain. He comments also on scanning.

If you prefer that I send you the test chart .. let me know,

Best ragards,

Joerg

[Joerg Krusche]

Peter,

the chart goes up to group 7 with element 6 .. i.e. up to 11500 dpi or > 220 line pairs/mm .. it does cover all we might need,

best,

Joerg

[Peter De Smidt]

Hi Joerg,

Thank you for the info and the generous offer of letting me borrow the slide for a bit. That sounds like a good mid- to late January project.

Regards,
Peter

[Joerg Krusche]

Bob,

I certainly did not want say that the Epsons and the Cezannes play in the same league .. quite the opposite .. it would be unfair to compare since they are also in very different price regions .. I said that there is an inflation of dpi's with Epson et al. .. not Cezanne..until further I believe that Screen's specs are ok .. just to use an analogy ... you buy a car specified at 100 miles/h .. and you cannot get above 30 .. if at all .. would you be satisfied .. and let the seller get away with it ?

I do not question that the Epson's are useful devices .. it is just about their specs .. my question is about the old high-price/high performance professional flat beds .. and their actual performance .. what you can get out of them ..and there is one way .. a simple test scan.

Bet,

Joerg

[Bob McCarthy]

Joerg,

A few years back Scott Rosenberg and I did a scanner test with quite a few scanners. We had use of a Cezanne, a Creo Supreme, a Leaf, and a Sprintscan 45. There were others including Nikon scanners for smaller formats and other Creo models. We compared these to a consumer scanner (my 4990). Ted helped us and the outcome was at 1200 dpi, we were challenged to see a difference. The Epson did just fine with B&W and was only challenged when scanning some of the slide films, Velvia comes to mind. As we moved the resolution up the lesser scanners fell behind. While the improvement was real with the epson is was not the equal of the others as we approached 2000 spi. Where the consumer scanners fell behind was not in comparing zone X vs. Zone 0 (your resolution test) but in comparing Zone 3 vs. Zone 7 (low contrast subjects). It was obvious that lpm was a poor measure.

Out conclusion was the larger the piece of film the lesser quality of scanner was required. 5x7 or 8x10 on an Epson at 1200 spi was great on the Epson especially for B&W. Up to 3 or 4X enlargements on a consumer scanner were for the most part every bit as good as using a high end scanner. Medium format and 4x5 enlarged significantly required a superior scanner.

The Creo Supreme and Cezanne were the top scanners in our test. No drum scanners were tested.

No scanners were harmed in testing.

bob

[Peter York]

Well,

To make a long story short, I now own a Screen Cezanne.

My original transparency was not sharp enough to see a big difference in resolution, yet while I was looking over the scans at home, the color rendition was far more accurate (I have to embarrassingly disclose here that I have never profiled the Epson for color). Areas of smooth gradation were much better with the Cezanne, and whenever I adjusted the Epson scan to roughly match colors in one area, another area would mismatch. Overall, my subjective impression of the Cezanne scan was that it was much more vibrant.

The re-tests we performed today showed a BIG improvement with a 35mm slide. The Cezanne was pulling out as much detail as the Coolscan 5000, and apparently with less noise.

This scanner will force me to re-think my entire setup, technique, etc. to maximize its potential. It will also be a very useful instrument to test lenses, etc.

I will post more to the "Cezanne Users Unite" thread when I have run some tests/scans. After lugging that beast across town, I might need a nap. However, I can hear my film and scanner calling me!

Thanks to all who responded! I greatly appreciate your help.

[Bob McCarthy]

Let me offer, the Cezanne software (ColorGenius) has built in profiling. It will take your hit rate way up if you go to the trouble.

Oh yea, congrats'. Best value in pro level scanning.

bob

[Peter De Smidt]

Hi Peter,

Congratulations!

I have some tips on scanning with a Cezanne at: http://peterdesmidt.com/blog/?p=361

Peter D.

[sanking]

Joerg,

A few years back Scott Rosenberg and I did a scanner test with quite a few scanners. We had use of a Cezanne, a Creo Supreme, a Leaf, and a Sprintscan 45. There were others including Nikon scanners for smaller formats and other Creo models. We compared these to a consumer scanner (my 4990).
.......

The Creo Supreme and Cezanne were the top scanners in our test. No drum scanners were tested.

bob

In comparing Scitex/Creo Eversmart scanners with Cezanne or any other flatbed on the market, it should be mentioned that the optical resolution of the Eversmart is over the entire bed, which is 12X17", with no interpolation. In other words, if the optical resolution is 5600 ppi you can scan any size negative or transparency, up to 12X17", with that resolution. This is because the Eversmart scanning software use XY stitching, which means that the scan is made in rows and then stitched by the scanner software.

Sandy King

[Bob McCarthy]

Sandy, the Cezanne works identically (ie, full resolution over the entire bed) to the Creo, with one single exception, that is the stitches have to be assembled with an external program, ColorGenius passes that step downstream. I use CS3 to stitch the panels.

I wonder if there was a patent or licensing issue at play. Its pretty trivial to incorporate flat stitching into ColorGenius. The geometry is pretty simple with a scanner.

bob

[sanking]

Sandy, the Cezanne works identically (ie, full resolution over the entire bed) to the Creo, with one single exception, that is the stitches have to be assembled with an external program, ColorGenius passes that step downstream. I use CS3 to stitch the panels.

I wonder if there was a patent or licensing issue at play. Its pretty trivial to incorporate flat stitching into ColorGenius. The geometry is pretty simple with a scanner.

bob

Bob,

I understand how the Cezanne works and while I do a lot of stitching
myself I must say that having the passes stitched with the scanner
software into one file with the original scan is a great convenience
compared to stitching with an external program.

There must be some sort of patent or licensing issue related to hardware
or software that prevented Fuji from using XY stitching. It is not as
simple as it may seem because the scanner glass of the Eversmart has
special bars and a coding system that makes it impossible to use another glass
bed of the same size. Some people have inadvertently wiped away the bar
design on the plate while cleaning, which basically makes the plate
useless unless you replicate the pattern because while you can scan the stitching program does not work.

I have the Edmund high resolution target and have tested the Eversmart Pro with it. The optical resolution of the Eversmart Pro is 3175 spi, and as you can see from the attachment the scanner is very close to 100% efficient in optical versus real resolution in that it actually delivers about 64 lp/mm. Also, if you look closely at the target you will see that in one direction the resolution is over 80 lp/mm. The actual scan was made at 6400 spi so anything beyond 3175 spi is only in one direction.


Sandy King

[Bob McCarthy]

Sandy,

In our test was both a Pro and a Supreme. As I recall the Supreme was the superior performer. I don't know the Creo line other than superficially, therefore I am at a loss to explain the difference.

My purpose in participating in the test was to evaluate the Cezanne I was intent on purchasing. The cezanne appears to be priced more favorably in comparison to the Creo line.

Also I found the Cezanne was marginally outperforming the Nikon 5000 I own for 35 mm scanning.

In any case for large format any of these pro level scanners is a joy to own compared to a consumer level scanner I was using at the time.

I have to agree, having the scanner software do the stitching is a convenience. I wonder if the bars are a location pointer to minimize or nullify the amount of overlap required to do the stitch!! A small amount of overlap is required to do it the Screen way.

bob

[sanking]

Bob,

Briefly, the first Eversmart has maximum optical resolution of 2540 spi, the Eversmart Pro has optical resolution of 3175 spi, and the Supreme and Select have optical resolution of 5600 spi. The IQ smart series uses the XY stitching software developed by Scitex but are considered by many experts to be a step below the EverSmart in terms of performance. All use the same Kodak 8000 ppi tri-linear CCD but the greater resolution of the Supreme and Select is made possible by placing the CCD closer to the material to be scanned and scanning a narrower row.

There is no question but that the Cezanne is priced more favorably than the Eversmart scanners,and it is certainly a fine scanner, much superior to any of the Epson or Microtek consummer scanners.

I suspect you are right in that the bar pattern serves to minimize overlap, of which there is very little needed when scanning with the Eversmart. Photoshop likes about 20-30% of overlap and I find that it is often necessary to manually stitch the MF negatives that I scan with the Leafscan 45 at 5080 spi.

Sandy King



Sandy,

In our test was both a Pro and a Supreme. As I recall the Supreme was the superior performer. I don't know the Creo line other than superficially, therefore I am at a loss to explain the difference.

My purpose in participating in the test was to evaluate the Cezanne I was intent on purchasing. The cezanne appears to be priced more favorably in comparison to the Creo line.

Also I found the Cezanne was marginally outperforming the Nikon 5000 I own for 35 mm scanning.

In any case for large format any of these pro level scanners is a joy to own compared to a consumer level scanner I was using at the time.

I have to agree, having the scanner software do the stitching is a convenience. I wonder if the bars are a location pointer to minimize or nullify the amount of overlap required to do the stitch!! A small amount of overlap is required to do it the Screen way.

bob