The GetDPI Photography Forum

Great to see you here. Join our insightful photographic forum today and start tapping into a huge wealth of photographic knowledge. Completing our simple registration process will allow you to gain access to exclusive content, add your own topics and posts, share your work and connect with other members through your own private inbox! And don’t forget to say hi!

Tiling..What is it and its causes

aztwang

Member
I see several post's on issues with "Tiling" Can someone enlighten me what Tiling is and what causes it.

Thank You
 

dougpeterson

Workshop Member
Certain combinations of [lens]+[back]+[movement] cause a section of the image to become out of calibration with the rest of the image creating a subtle-but-problematic line to form in the edge of that section.

Depending on the lens, back, and movement, tiling is often removed by an LCC process. But in unsupported combinations the tiling cannot be removed by any automatic process, and could only be removed by manual means in Photoshop.

You can view a basic list of lens support on our tech camera overview page.

Consult with your dealer for more detailed descriptions. At Digital Transitions we have spend dozens of hours shooting various combinations of lenses, backs, and movements to make sure we can provide our clients highly expert advice on the topic.
 

Paul2660

Well-known member
If you look at any of the Dalsa 60Mp or 80Mp chips, you will see a series of lines that appear to divide up the chip. These are readouts to help speed up the delivery of the raw data to the back. When looking the chip, you will see 8 segments, 3 lines up and down and one in the middle.

These readouts can get out of calibration and sometimes never get into calibration, the history of the 80Mp Phase backs has a lot written on the tiling issue.

Phase creates a calibration file for each back, your dealer can help do this I have been told over the phone and it may fix the problem, however sometimes the back has to be sent to Phase One to be re-calibrated.

No two backs are alike and each calibration file is unique to the back.

What does it cause?

A hard line most times running up and down which tends to be more visible in sky portions of an image. If you convert to B&W, the effect will show up even more. Sometimes you can fix the issue in post in Adobe with content aware if the LCC and C1 can't fix it. Also sometimes when you are working the shot in post, and add more clarity or saturation the tiling may become more noticeable.

I believe this is only a problem with Tech cameras and can show up on a center shot or with movements. Movements tend to be worse for it. Often times you will see the tiling lines in the LCC taken during the shoot. As I understand it the LCC and C1 work together to in effect map out the line. But the real key is the back having a good calibration file.

Backs will get out of calibration over time and need to have this process redone. And some backs show this problem worse than others. Also some tech lenses seem to have a greater tendencey to have tiling. The Schneider 35XL and 43XL that I owned both seemed to have issues with very harsh tiling on my IQ160. The 43XL worse than the 35XL. I switched to the Rodenstock 40mm and still see a bit of tiling, but nothing like what I could get with the 43XL. Also it's possible that using a polarizer may make the effect a bit worse, at least in my experience.

To me what tends to cause more issues with movements is mircolens ripple.

Paul Caldwell
 

GrahamWelland

Subscriber & Workshop Member
You'll really notice it if you convert to B&W and start stretching the image in post production. What you'll see is sudden lines between the segments of the CCD imaging chip - and not subtle transitions at all

I can post some examples when I'm next home but I suspect that Paul had plenty in the meantime.
 

torger

Active member
Tiling has two reasons: 1) physical variations on the chip due to multiple exposures during manufacturing "stitching" (still one silicon chip of course), 2) mismatch between multiple readout channels.

The physical aspect is visible to the eye when you look at the chip from an angle, and then you'll notice that it depends on manufacturer. Dalsa is by far the "worst" in this aspect, while Kodak chips don't have any physical tile lines visible at all.

Then there's mismatch of readout channels which also can cause tiling, but I think the physical property is more evident. For example my H4D-50 has a Kodak 50MP chip with two readout channels and you really need to do a scientific deep measurement to detect any difference between the channels. On any Dalsa chip it's easy to measure (I've looked at many), but I think it's more about the physical tiling than channel mismatch, it's either that or Hasselblad is better at calibrating their hardware than Phase One.

The physical tiling is exaggerated if you use tech camera wide angle lenses, as the low angle of the incoming light and increasing crosstalk make the "edges" show more. Capture One has a quite good LCC correction algorithm that helps battle this, but as even slight crosstalk makes the tile line visibility light-dependent an LCC shot cannot cancel it out in full, some filtering is required too, which C1 applies with varying result.

That said most users never have problems with tiling in practice with the Dalsa-based backs. The worst case is if you work with symmetrical wide angles like the SK35 or SK28, and do ultra-high contrast processing which some do in black-and-white and have smooth surfaces in the picture like a sky.

If you use anything else than a Dalsa sensor there are much less issues, practically non-existent. That is the Sony CMOS or the Kodak CCDs (almost) don't have this problem.

There's a related problem called "microlens ripple" which you sometimes can see with tech wides, a wavy horizontal or vertical pattern which is caused by microlens shading when the incoming light angle is too low. Again Dalsa is worst on this (with the old microlens chips being an exception, which cannot be used with tech wides at all), the Sony CMOS also have this problem but to a much lesser extent, while the microlens-free Kodaks don't have it at all of course. Again C1 has a filter for this and quite effectively cleans it up in most cases.

It's a good idea to use the Rodenstock Digaron wides for your Dalsa sensor rather than the symmetrical Schneider Digitar, but it won't solve the problem 100%. So if you get a Dalsa CCD and is really into high contrast post-processing you should be prepared to now and then do some hand-editing to clean up.
 

AreBee

Member
torger,

If you use anything else than a Dalsa sensor there are much less issues, practically non-existent. That is the Sony CMOS or the Kodak CCDs (almost) don't have this problem.
Just so that I am clear, are you saying that Sony CMOS sensors will not have a tiling issue and Kodak CCD sensors almost will not, or that Sony CMOS and Kodak CCD sensors almost will not?
 

Paul2660

Well-known member
When looking at the 50Mp Sony chip you don't see the readout lines or visible segmentation.

So I would say no tiling issues with the Sony. However I found the C1 does not correct the micro lens ripple on the Sony as it does on Dalsa chips. The micro lens ripple on the Sony is much less defined and presents as more of a series of vertical smears. I found that Phase One agreed that they are there but I don't feel they have tried to correct for them as Phase does not consider the 50mp chip highly recommended for movements. And to see the issue you need a solid background like a sky with no clouds and be shifting 10mm or more.

Paul
 
torger,

Just so that I am clear, are you saying that Sony CMOS sensors will not have a tiling issue and Kodak CCD sensors almost will not, or that Sony CMOS and Kodak CCD sensors almost will not?
I only know that the Sony 33x44 sensor is a single tile but not really, it's tiled during the actual etching process like a jenga tower, so that it works seamlessly. I have done some crazy exposure pushes on my 645Z and never noticed any lines or exposure shifts, so if it does exist, Sony did a bang-up job on hiding it.
 

torger

Active member
torger,



Just so that I am clear, are you saying that Sony CMOS sensors will not have a tiling issue and Kodak CCD sensors almost will not, or that Sony CMOS and Kodak CCD sensors almost will not?
Actually I'm not 100% sure so I'm deliberately a bit hazy on that point :). The more recent Kodaks have more than one readout channel so mismatch can occur, but it's very well hidden. I assume that also the Kodaks and the Sonys are "stitched" in some way during manufacturing, however much leaving tiles virtually invisible unlike in Dalsa's manufacturing process.

I have a faint memory that I managed to pull out a tile line from an IQ250 raw, but not in real processing but in a crazy test to just see if it's there. I can't reproduce that though now, so I may have mixed up files or something. I think it's safe to say that any tile issues if they are there are at such a low level that even the most extreme real processing should not experience problems, and this counts for both microlens-free Kodaks and the Sony CMOS. With Sony CMOS you can get a sort of microlens ripple though.

With any Dalsa-based back, even the older 33 megapixel ones, it's fairly easy to force a tile line so there's no doubt that it's there, and as said it's very visible to the naked eye when looking at the sensor too. There's also the added issue with matching the CCD readout channels. I have no idea why Phase One seems to have more troubles doing that than Hasselblad. I haven't had the opportunity to test a H5D-60 (Dalsa) for tiling, comparing that to an IQ160 (same sensor) would give some indication if the channel matching aspect is a Phase or a Dalsa problem. Today I don't know.
 
Actually I just remembered that I was once testing out an IQ180 and wanted to see how it did in available light, so I turned it up to around ISO800, no sensor+, and one of the shots I was inspecting in post has significantly more noise in the lower-right quadrant than in an other, almost as if it were shot a stop higher. Anyone else run into something like that?
 

AreBee

Member
Sounds like CMOS all but eliminates the issue then.

Hopefully BSI will in due course eliminate ripple/mazing/colour cast.
 

torger

Active member
Sounds like CMOS all but eliminates the issue then.

Hopefully BSI will in due course eliminate ripple/mazing/colour cast.
I don't think it's likely that color cast will disappear. As far as I understand one component in color cast is that the tiny R G B filters in front of each pixel changes response slightly depending on incoming light angle, and that issue won't disappear with BSI.

Mazing is a crosstalk artifact and that has a good chance to dissappear with BSI.

Ripple is a microlens artifact and while BSI is excellent base for making microlens-free sensors (as the photo diode can cover larger surface than otherwise AFAIK) I think we will not see that from Sony. Even with BSI there are still gains to make in sensitivity and aliasing reduction when adding microlenses. Ripple in the current Sonys are less bad than in the current Dalsas though, and I see it as quite likely that ripple is further reduced with the microlens type used with BSI so if we're lucky it will be negligible. Otherwise a purpose-made algorithm will be able to clean it up. Due to color cast most likely will continue to be there we will be making LCC shots anyway, which provides input to the algorithm that can clean up microlens ripple as well.

LCC is indeed a bit of a pain, but the fact that we tech cam users can accept the hassle of making LCC shots gives us unique optical designs that are impossible for more user-friendly" workflows. So in a way the natural slow workflows that makes LCC acceptable is what keeps tech cams on top in terms of image quality :)
 
Top