View Full Version : Geometrical distortions on belt drive scanners
Hi! Finally I want to expose some findings: scanners where the scanning carriage moves with the belt - distort. Scanners where the process is made by lead screw - do not distort. Sad, but true.
I found it accidentally while comparing different scans of the same 35mm negative. I will not show mine, but let's look what he have on the net.
http://www.largeformatphotography.info/forum/attachment.php?attachmentid=172186&d=1511217016
The moving gif is made from two crops from that page:
http://trichromie.free.fr/trichromie/index.php?post/2010/03/02/ScanC41
Two scans were made on IQSmart3. On a gif you can see the difference - it moves. What can be if the face will be scanned in that area? I tell you: nothing good. The face will loose right emotions and will be distorted.
If one wants to explore that, he can visit the page:
http://www.largeformatphotography.info/scan-comparison/
There is a lot to compare. And nearly only drum scanners do not distort. They simply have lead screws. Do not distort: Topaz, Nexscan and Smart 3XX. They also own lead screws.
Having said all that on a large format photography forum I understand that the larger the original the less is the distortion effect. Anyway I think it is good to know that such thing as a geometrical distortion of the belts exists in this not perfect world.
Take care
Al
Evanjoe610
8-Aug-2011, 17:55
Surf,
Now that you got my attention, I will state that you are correct in the sense that after a while belts do tend to stretch and they should be replaced every so often. Yes, the Scitex 340 series do have a lead screw that will require just a wipe with a a bit of dustless grease (sorry, forgot the actual name as it has been a long time!)
As for the Imacon, Hasselblad as part of their servicing, they DO REPLACE the belt on their scanner. Therefore the "chattling" by the lead screw will not "pop" out at you.
As you mentioned in the previous thread, that noise is from a stretched belt that was probably never serviced. I asked about replacing the belt on an Eversmart Pro II scanner, but never received a direct answer. Not sure it was an easy thing to do, but I would definitely like to have it done on my Eversmart Pro II. Can anyone on this forum who had repaced their belt, tell me if was an easy thing to do or was it more technical in nature making it harder? (CORRECT TENSION & BELT POSITION?)
Thank you for bringing this to my attention, I do appreciate.
Hi! Finally I want to expose some findings: scanners where the scanning carriage moves with the belt - distort. Scanners where the process is made by lead screw - do not distort. Sad, but true.
I found it accidentally while comparing different scans of the same 35mm negative. I will not show mine, but let's look what he have on the net.
http://www.largeformatphotography.info/forum/attachment.php?attachmentid=59721&stc=1&d=1312845588
The moving gif is made from two crops from that page:
http://trichromie.free.fr/trichromie/index.php?post/2010/03/02/ScanC41
Two scans were made on IQSmart3. On a gif you can see the difference - it moves. What can be if the face will be scanned in that area? I tell you: nothing good. The face will loose right emotions and will be distorted.
If one wants to explore that, he can visit the page:
http://www.largeformatphotography.info/scan-comparison/
There is a lot to compare. And nearly only drum scanners do not distort. They simply have lead screws. Do not distort: Topaz, Nexscan and Smart 3XX. They also own lead screws.
Having said all that on a large format photography forum I understand that the larger the original the less is the distortion effect. Anyway I think it is good to know that such thing as a geometrical distortion of the belts exists in this not perfect world.
Take care
Al
Thank you for the interest Evan.
Can anyone on this forum who had repaced their belt, tell me if was an easy thing to do or was it more technical in nature making it harder? (CORRECT TENSION & BELT POSITION?)
The first time it is not too easy.
As for the Imacon, Hasselblad as part of their servicing, they DO REPLACE the belt on their scanner. Therefore the "chattling" by the lead screw will not "pop" out at you.
As you mentioned in the previous thread, that noise is from a stretched belt that was probably never serviced.
First thing to start is measuring. You have to learn how to measure the distortion. On belt driven scanners it is always visible, I think due to the nature. You also have to decide when to stop and when to start worrying about it. I must say that it is easier to live without knowing about it. Not alike lead screw the belt can wear differently at different positions on the table and after all there is at least one more critical part inside the scanner: it is reduction gear.
As for the source of new belts. I don't think Scitex made special belts and I believe that best quality industrial belts can be used. I have examined the site of Gates and found the belts that look alike in any aspect.
I think it is a good idea to PM me.
Regards
Al
Evanjoe610
9-Aug-2011, 10:06
Al,
Thank you for your advise. I have PM you to carry our conversation off line.
8x10 user
9-Aug-2011, 11:35
I don't think there are issues with distortion on the Eversmart or IQsmart line of scanners. They were made in part for copydot work and distortion would really mess things up there.
One thing that you may see is a difference from an XY stitch machine like the Eversmart and a xy zoom (stitch free) machine like the Aztek plateau. The stitched scan should be more accurate due to the final scan being made by multiple "telephoto" shots while the plateau uses one scan. You might see natural differences due to the perspective of the lens in the scanner.
Creo Scitex scanners go through a long calibration process where geometric distortion and alignment is checked.
As far as belts, for the creo scanners there is newer and older belt. The newer one is a little better and it is the white belt.
Which belt driven scanners have you seen distortion in? I could imagine it in a cheap one... But a Creo Scitex?
I don't think there are issues with distortion on the Eversmart or IQsmart line of scanners. They were made in part for copydot work and distortion would really mess things up there.
Problems with distortions do not influence stitch or copydot process.
Creo Scitex scanners go through a long calibration process where geometric distortion and alignment is checked.
We are talking about tens of microns here and it has no influence on calibration too.
As far as belts, for the creo scanners there is newer and older belt. The newer one is a little better and it is the white belt.
Which belt driven scanners have you seen distortion in? I could imagine it in a cheap one... But a Creo Scitex?
All I know: the newer - the cheaper everything goes. Kinda joke.
Yes. Creo Scitex. The gif was made from IQSmart3 scans and I see no big difference with that also on Eversmarts, Screens and Imacons.
OK. Time to make homework.
Scan the original at 3200 dpi. Scan area: 10mm by 10mm. Then rotate original 90 degrees on the table and repeat scan. Save both at different names.
If you were precise in 90 degrees you can only crop the area. Start from upper left corner and crop it close to one pixel accuracy. Then crop the bottom right corner. See the sizes in pixels and scale one scan to have both of exact size. Then you can compare the two by fast switch. What moves - is the distortion. The step of it in Eversmarts is 5 mm. You can also try to measure the absolute size of the distortion. Average old scanner has about 3 pixels. Nothing good, nothing too bad. The scanner from the first post was measured like 4 pixels. No good.
How can you be certain that your image was perfectly alligned along on the glass in the first place and how one could be certain about rotating it by 90 degrees precisely ?
What about the calibration ? Was it recently done?
Then for Eversmarts there are certain factors such as Head Tilt and Rotation. When the last time were they checked and adjusted on the scanners produced the images for analysis?
How can you be certain that your image was perfectly alligned along on the glass in the first place and how one could be certain about rotating it by 90 degrees precisely ?
I will answer only one question. You have to rotate one image back 90 degrees in fotoshop afterwards. I have fogotten to mention it. It's a good idea to have some lines on the original, so you can fine tune rotate in ps. In other words it is absolutely not necessary to be precise in the original placements. You can fine tune it in the program.
Other questions have no connection with the problem.
Now I want to post scanner comparison based on the pictures from well known page:
http://www.largeformatphotography.info/scan-comparison/
Have a look how Eversmart Supreme distorts geometrically compared to Tango drum scanner. And what is interesting: all Tangos do not have any distortion when compared to each other. There are several Tangos and two Eversmart Supremes. Both Supremes heavily distort.
http://www.largeformatphotography.info/forum/attachment.php?attachmentid=172187&d=1511217812
Supremes are often used in the museums for backing up the images. I believe that most of that backups are good only for trash folder cause they are distorted. What do you think?
8x10 user
12-Aug-2011, 16:55
I don't know. You have to be careful about generalizations with this type of tech. One scanner could be off in some way or another while another might be perfect. Also I don't know about how the scans were made, if all the scans were fluid mounted, if they were scanned at another resolution then interpolated or what, ect. There are just too many factors to for me to take this evidence of geometrical distortion in all Eversmart scanners. Maybe when my scanner comes back from service I will run some tests of my own.
The supreme's are also used for scientific research.
I don't know. You have to be careful about generalizations with this type of tech. One scanner could be off in some way or another while another might be perfect.
I found it on february. Now it's august. At first I didn't believe my eyes cause nobody never in my life thought that it was possible. Finally I am sure in what I am saying. While scan the computer think that scanner mechanics is perfect and if it is not we get distortions. I believe that those scanners were not made by the fools and while new they scanned more or less OK. Also those scanners were made for printing industry where nobody ever cared about such things. I know what I am talking about. While the size of the scan is OK then no problems. And the sizes are always OK, but what happens inside those sizes... You can see now.
Another nice picture: Screen Cezanne_vs_Tango. You see - no good for screen too.
http://www.largeformatphotography.info/forum/attachment.php?attachmentid=172188&d=1511217986
SergeyT
12-Aug-2011, 18:39
Exactly my point. There are so many unanswered questions and variables...
I wish that "distortion" (even if it present indeed) was my biggest concern with producing quality scans ;)
SergeyT.
SergeyT
12-Aug-2011, 18:42
Another nice picture: Screen Cezanne_vs_Tango. You see - no good for screen too.
http://www.largeformatphotography.info/forum/attachment.php?attachmentid=59888&stc=1&d=1313199072
Have you found two scanners which produce identical scans on pixel level?
Thanks,
SergeyT.
Have you found two scanners which produce identical scans on pixel level?
Yes. Tango, Scitex Smart 3XX, Topaz, Nexscan. They are perfect in that terms. May be Nikons with glass holders, but I had no chance to test that setup. Without glass holders they distort and tend to be unsharp. Anyway Nikon distortions feel better because it is close to lens distortion. They own lead screw. Not sure do Nikons have problems in size proportions or not. I have not tested them.
And it happens not only on transparencies. Look at the reflectives scan. The picture is from the summer of 1945 btw. Nice print, not from the family album though. I see the face changes. I do not like the faces to change. That's why I scan my family pictures only on the lead screw driven scanner. I feel better when I am sure that proportions of dear faces remain intact.
Smart vs Eversmart (eversmart that is in average state, before I made it better)
http://www.largeformatphotography.info/forum/attachment.php?attachmentid=59889&stc=1&d=1313201692
Valdecus
13-Aug-2011, 01:41
The issue became apparent to me for the first time some years ago when I tried to scan a single large format film with different exposures to be used as input for HDR processing. None of the scanners I tested (from the Epson V750 flatbed to Imacon's 848, Hasselblad's X5, and various drum scanners like Howteks) were able to scan *the same* image twice with absolute pixel accuracy. Even with drum scans, certain pixels along the long axis of a file 20,000 pixels wide were off by as much as 10 pixels when compared to the first scan.
For HDR processing, this definitely wasn't accurate enough as the scanned images need to be aligned perfectly. Distortion of the scanned results made this impossible, and even Photoshop wasn't able to properly align the results with its advanced alignment algorithms.
Cheers,
Andreas
Even with drum scans, certain pixels along the long axis of a file 20,000 pixels wide were off by as much as 10 pixels when compared to the first scan.
It's a good remark, Andreas. And we have to mention repeatability now. It's a different animal which also lives in the precision land. The scanner can have a perfect repeatability of the distortions. That's why Eversmarts stitch periodically distorted images on the fly and can pass all the calibrations.
Tango vs Imacon949:
http://www.largeformatphotography.info/forum/attachment.php?attachmentid=172189&d=1511218178
Well... I can add that Screen 1045AI and 1030 also distort. These guys managed to make drum scanners without lead screw. OMG.
The Deer Gunter
29-Mar-2015, 13:16
Well, alomst 5 years later ... can I still throw in a couple of questions? I will begin to try and master an iQsmart3 and its oXYgen scan software somewhere in the next couple of weeks, and stumbled upon this page while Googling. It's my first post on LFP by the way :-)
I already have some experience with the iQsmart, but it is only now I will be able to study it properly. An example:
https://www.flickr.com/photos/thedeergunter/6250812458/
1) Does the geometrical distortion affect image quality in any other way? I'm particularly concerned about loss of sharpness.
2) Is it inevitably inherent to each and every belt driven scanner, even when you're talking about a brand new out of the box device?
3) Does recalibration have an influence on the degree of distortion?
Thanks for your time! Cheers.
Well, alomst 5 years later ... can I still throw in a couple of questions? I will begin to try and master an iQsmart3 and its oXYgen scan software somewhere in the next couple of weeks, and stumbled upon this page while Googling. It's my first post on LFP by the way :-)
I already have some experience with the iQsmart, but it is only now I will be able to study it properly. An example:
https://www.flickr.com/photos/thedeergunter/6250812458/
1) Does the geometrical distortion affect image quality in any other way? I'm particularly concerned about loss of sharpness.
2) Is it inevitably inherent to each and every belt driven scanner, even when you're talking about a brand new out of the box device?
3) Does recalibration have an influence on the degree of distortion?
Thanks for your time! Cheers.
What scanner is not belt driven?
The Deer Gunter
29-Mar-2015, 23:53
What scanner is not belt driven?
SURF mentioned a couple of them earlier in this thread ... "Tango, Scitex Smart 3XX, Topaz, Nexscan", and likely some of the Nikon scanners as well.
SURF mentioned a couple of them earlier in this thread ... "Tango, Scitex Smart 3XX, Topaz, Nexscan", and likely some of the Nikon scanners as well.
I have not yet seen a seitex or nikon scanner not belt driven. The Tango and others are drum scanners.
Scanners with step motors are prone to the same questions you raise BTW.
Even an imacon has 2 drive belts. They are steel reinforced belts. When they fail, the machine does not work, but it does not happen over time. It either works or it does not, as is the case with all others.
The Deer Gunter
30-Mar-2015, 07:13
Scanners with step motors are prone to the same questions you raise BTW.
Hi,
I raised the questions because someone (SURF) raised the issue, being 'there is a fundamental difference between belt driven scanners opposite lead screw driven scanners causing the former to generate geometrical distortions'. Apart from the fact which scanner uses which system, do you doubt this statement, or not? I'm just interested knowing the limitations of each system.
Hi,
I raised the questions because someone (SURF) raised the issue, being 'there is a fundamental difference between belt driven scanners opposite lead screw driven scanners causing the former to generate geometrical distortions'. Apart from the fact which scanner uses which system, do you doubt this statement, or not? I'm just interested knowing the limitations of each system.
Instead of breaking your noodle over this theoretical maybe, you should tell us what scanners you are considering.
The Deer Gunter
30-Mar-2015, 11:51
Instead of breaking your noodle over this theoretical maybe, you should tell us what scanners you are considering.
I agree. In the next couple of weeks I'll have the privilege to have permanent access to a Creo iQsmart3 scanner at home. I came to this thread while Googling. The claimed distortions are illustrated at the very beginning of this thread using ... indeed, the iQsmart3.
Thanks for your opinion.
The Deer Gunter
4-Apr-2015, 13:52
Instead of breaking your noodle over this theoretical maybe, you should tell us what scanners you are considering.
And then all became quiet ...
Allow me to answer my own questions then, and that of many others I suppose.
I installed a Creo iQsmart3 from scratch: yesterday the hardware part, today the software part. All went well. I also went through a couple of (6x7) test scans. And I can tell you this:
There is not one sign of geometrical distortion on my samples. You can perfectly overlay two consecutive scans without moving one pixel! My faith in belt drive scanners is restored completely.
So what went wrong with the scans posted on the Collaborative Large Format Scanner Comparison?
http://www.largeformatphotography.info/scan-comparison/#
I can only guess someone is either ...
- using the wrong Mac (iMac, Mac mini, MacBooks of any kind are not compatible with iQsmart scanners).
- using a messed up Mac OS X. Updating the OS over the internet causes the entire scanner installation to become invalid, and requires a reconfiguration from zero, including recalibrating the device!
- using the scanner while actively connected to a network.
- using a scanner that needs recalibration.
... or a combination of the above.
I have used a G5 MacBook Pro (ppc) before with the iQsmart3 and also experienced deformations. I was sceptical at first, but the difference using a G4 PowerMac (tower) is clear. They make a much better couple now.
Peter De Smidt
4-Apr-2015, 14:18
Well, that's good news. Happy scanning!
The Deer Gunter
4-Apr-2015, 14:37
Well, that's good news. Happy scanning!
Happy scanning to you too :-)
I also went through a couple of (6x7) test scans. And I can tell you this:
There is not one sign of geometrical distortion on my samples. You can perfectly overlay two consecutive scans without moving one pixel! My faith in belt drive scanners is restored completely.
If you compared "two consecutive scans" that's you compared repeatability. It's often very high and it's another animal. But faith is a good thing to have.
PS. I have restored the images from 2011. Some years passed and they disappeared.
Powered by vBulletin® Version 4.2.5 Copyright © 2024 vBulletin Solutions Inc. All rights reserved.