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!

IQ160 tiling issue

Pemihan

Well-known member
I have some trouble with lines / tiling on my brand new IQ160.

The attached image (100% crop) are shot with a DF+ / 80mm LS lens
When developing the RAW file in Capture One 7.2 I get the issue as you can see on the left
When delevoping the RAW files in Lightroom there is no issue as you can see on the right.

Also especially when developing the RAWs in Capture One the edges of the image looks somewhat weird as you can see on the second pic.

I have done a clean uninstall and reinstall of Capture One 7.2 which didn't make any difference at all.

I'm not unfamiliar with issues like tiling / lines in the image when using a tech cam and Schneider wides and usually LCC takes care of it, but to see it with the DF+ and 80 LS is new to me and seeing it in C1 but not in Lightroom is strange.

I assume it's some kind of calibration issue and will contact my dealer on Monday, but just want to hear if any on the forum have experienced something similar and if so what was done about it.

I got the back with Value Added Warranty, so don't know if they will simply replace the back.

Thanks
Peter



 

sc_john

Active member
I had the same issue shown in your first image occur with IQ260. In my case, the problem only occurred with shutter speeds greater than 10 seconds. I contacted dealer (Digital Transitions) and sent them a number of images showing the problem. If I recall correctly, the issue was not with the back, but with internal settings in Capture One. They forwarded me a "zone picker" file that when loaded into Capture One eliminated the problem.

John
 

dougpeterson

Workshop Member
Yes, strange. Contact your dealer. I'm sure they will take care of this right away for you. If you upload the raws to a service like Dropbox that will probably save some time on Monday as your dealer will 99% likely need a raw file to complete the appropriate paperwork and resolve this for you.
 

etrump

Well-known member
Since it is Sunday...

Try deleting the preference file and restart c1. I have had extensive problems with this issue. Sometimes it requires sending the back for calibration but most of the time it is a problem with C1. On my Mac it was in my local ~/Library/Preferences file called com.phaseone.captureone7.plist. After deleting this file c1 asked a few initial setup questions and the centerfold issue was fixed.

Thanks to Josh at CI for helping resolve my issue.
 

Pemihan

Well-known member
Thanks John, I hope it is as simple as that.

Dough, I will do that, thanks.

Ed, I actually already did that by doing a clean uninstall but will nevertheless try it again..
Did the issue for you go away on images already in Capture One or only in new images?

It's extremely frustrating, I just got the back this Friday instead of a Leaf Aptus-II 10 that had an issue they couldn't / wouldn't fix.
Had to pay a nice chunk of cash extra as well, so this is not what I wanted..
Maybe I should just stay with my good old Aptus-II 7.

Peter
 

Pemihan

Well-known member
Here's another example, this one is shot with RZ67 and 140mm lens.



Ed's suggestion regarding deleting the Capture One .plist file unfortunately didn't change anything...
 

torger

Active member
Some technical information:

There are two sources of tiling. 1) the sensor is exposed in multiple tiles during manufacturing which causes slight different pixel properties. You can usually see the tiles when looking at the sensor from the side, the IQ160 has eight tiles, split by one horizontal line and three vertical. 2) the back has multiple amplifiers, the IQ160 has four, one for each quadrant.

The amplifiers are not exactly the same so they differ a bit. The tiling caused by amplifiers is much stronger than tiling on the sensor, amplifiers can differ as much as 0.1 stop or so. Some digital back formats write a "cooked" calibrated file, so the raw data is already tuned. The IIQ format does not do that, it writes the raw output from the amplifier channels, and calibration data, ie amplifier and sensor properties measured at the factory. This includes the amplifier responses, native sensor color cast and tiling, and sensor defects (partially dead columns etc). I think some of it it's done per ISO in the newer backs but I'm not sure, so it's lots of calibration information.

With correctly matched and applied calibration data the file should be free from tiling. However, if you increase contrast to an absolute maximum on a plain sky in post-processing some faint tiling will most likely be visible, the eye is very good at detecting tiling so only if there is the tiniest trace left you can see it.

I'm suspecting that Lightroom / DNG converter uses some statistical method in addition to applying the calibration data. It says in the files where the tiles are, so the raw converter could simply analyze this portion of the file, see if it's a smooth surface there and if it is really make sure that the tile does not show up. There may also be differences in how the calibration data is applied. The calibration data does not have 100% coverage, ie you need to interpolate some values, and possibly Adobe's algorithm may be better than Capture One's. It may be possible to get an answer from Adobe in their forums, the Lightroom programmers are sometimes there and are pretty open, ie if they just apply the calibration data or if they smooth the file further.

If you convert to DNG with Adobe, you'll get a "cooked" raw file, ie all calibration data applied. So one alternative is to start working with DNG instead. Capture One is not good at DNG though...

If you work with technical wides tiling can appear because pixels vary a bit in wide angle response between the sensor tiles, ie how much pixel vignetting you get. The calibration data in the file has no information on what the wide angle response is so to avoid tiling with lenses with very short flange distance (like the SK35) you'll need LCC and a good LCC algorithm that in addition to fixing color cast can supress tiling too.
 

torger

Active member
I would guess that if you work in black-and-white and apply color filters in post-processing, especially supressing green, there's a larger risk that tiling becomes visible.

When doing black and white with color backs one really do not want to supress green (especially not at shooting time with a filter), as a major portion of the captured information will then disappear. With green channel intact color backs work quite well for black-and-white though.
 

Pemihan

Well-known member
Thanks for the writeup Anders..

I do indeed make quite a lot of B&W but this issue is present when just making a basic B&W conversion and can also be seen in the color images with very little adjustments albeit to a smaller degree.

The case is handed over to my dealer and hopefully I'll have an answer soon.

Peter
 

torger

Active member
I've looked at Abobe's conversion a bit more and experimented with zeroing some tags etc. It does look like Adobe does additional things to eliminate tiling, ie it applies the calibration data first, and then probably by statistical analysis along the edges it does further refinements. Haven't studied C1 in detail but based on the examples posted above I'd guess that C1 only applies the calibration data and that's it, meaning that if the calibration data is not 100% on the mark, tiling may be visible.

It depends on who you ask which approach is better. Surely by doing analysis along the edges of the actual image you can always eliminate tiling in full, but you also risk disturbing an edge that is truly there, eg if you happen to line up a vertical feature exactly on the centerfold. I think that with frequency analysis near the edge there should be no risk of harming any real subject though, and probably it's very difficult to make calibration data that is 100% accurate, so additional refinement after applied calibration data may be the way to go.

Disclaimer: analyzing what really happens inside the converter is not so easy, so there may be some tag magic I've missed, eg maybe Adobe is just applying the tags and no more but does it better than C1. I don't think so though, I'm quite sure that Adobe's better result in this case is due to some Adobe-specific refinement step.
 

Paul2660

Well-known member
Bit off the topic, but Torger, you wrote:

"There are two sources of tiling. 1) the sensor is exposed in multiple tiles during manufacturing which causes slight different pixel properties. You can usually see the tiles when looking at the sensor from the side, the IQ160 has eight tiles, split by one horizontal line and three vertical. 2) the back has multiple amplifiers, the IQ160 has four, one for each quadrant."

When you say "exposed" in the manufacturing process what does that mean?

This issue of just how the larger chips are manufactured comes up a lot. I have been told that it's still one single chip i.e baked all at once, but the line are the readouts. You have to have the readouts since they speed up the process of pulling the data off the larger chips.

I have also read that the actual chip 60 and 80MP takes the image in a series over these readouts and this can also help to cause the tilting as the light can actually vary during the exposure.

All this is done at the speed of light or near so it's way over my head. Just curious.

As you and Ed pointed out, the calibration process is very key to the whole image looking good and I believe that each back has a separate calibration but again I don't know for sure.

Thanks
Paul
 

dougpeterson

Workshop Member
One silicon wafer. Exposed in several windows of exposure. With readouts between each window of exposure.

Each back is individually characterized and calibrated, not just for readout parity.
 

Pemihan

Well-known member
I got a RAW file from another IQ160 from my dealer and tried converting it to B&W like I normally do. No tiling or centerfold or anything, even with some rather extreme adjustments. So it points towards my back being completely out of whack.
Still haven't heard anything though..

As a side note I can tell that the calibration file in my Aptus-II 7 out of the blue got corrupted. When opening the RAW in Capture One or Leaf Capture it looked horrible, but when doing the same in DCRAW which doesn't read the calibration file it looked fine.
After a new calibration file was loaded all was good again.

So yes the calibration process is key to the final image.

Peter
 

padams

New member
Had the Exact same problem with my IQ140. Sent it in for repair - which did not fix the issue. Sent in more files and PhaseOne replaced the back with a new one.
 

GrahamWelland

Subscriber & Workshop Member
Call me greedy but I'd love to see a production photography version of the 48x84mm chip. A 6x12 ratio chip would be awesome! Heck, I'd even take a crop down to 6x17. :grin:
 

Pemihan

Well-known member
Thank you very much everyone for your comments.

As said I got a RAW from another IQ160 which doesn't have this issue so I was thinking it could be interesting for me to get some more examples from different IQ160 backs.

So if some of you have one or two RAW files you can send me via for example Dropbox I would be much appreciated. Preferably images with a lot of sky in it since that's where I mostly have the issue.

My email is pm (at) petermisfeldt . com

Peter
 
Last edited:
Top