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!

IQ3 100MP technical camera tests: color cast, mazing artifact, tiling issue, DR etc

modator

Member
I don't know what the purple glow is, but as it's in direct connection to a bright edge on that CCTV I'm thinking that it's some sort of internal reflection, could it be a reflection in the sensor glass or something? Just speculating. If it would have been normal CA of the lens I would have expected it to show more on other edges too, and of course be seen on the IQ380...
From what I see the high contrast area on the camera is affected by a purple fringing like there is a different lens in play, as I'm sure the lens is the same one solution about the problem is the IQ380 don't have micro lenses ... but IQ3100 maybe it have ...
the noise perf. from what I see is in favor of the IQ250.. also there are some color completely different from CMOS to CCD ,, some wall in CCD are greenish,, the CMOS's are yellowish... but maybe there is some difference in white balance point, if not, think about making a good color profile of both camera... the result to me are very strange.

Best regards, Domenico.
 

torger

Active member
All Dalsas since the P65+ have microlenses, that's why you can get the "micro lens ripple" artifact on symmetric wides (which C1 effectively can clean up). They're not at all as destructive to wide angle performance as the microlenses that existed on some Kodak sensors though.

The color difference is indeed strange, I expected it to be smaller. If the profiles are designed to make the cameras look the same they should look virtually the same. You could have some visible differences on high saturation colors or in narrow band light, but in the normal range the difference should not be as large as seen in the examples. So either the profiles are designed differently, there is some white balance issue or some other problem.

I'm not sure how Phase One design their digital back profiles, if they want to differentiate the looks between models, or if they want to make them look the same. Someone who knows?
 

AreBee

Member
torger,

I'm not sure how Phase One design their digital back profiles, if they want to differentiate the looks between models, or if they want to make them look the same. Someone who knows?
As the author of this article, Doug may know. Read from the section: Screw Location - Color. Color. Color.
 

torger

Active member
torger,



As the author of this article, Doug may know. Read from the section: Screw Location - Color. Color. Color.
Thanks for the link. Reading it, it's hard for me to figure out if Doug's trying to provide actual information or just to convince current customers to take the CMOS jump after being fed with CMOS FUD in almost a decade ;)
 

modator

Member
torger,

As the author of this article, Doug may know. Read from the section: Screw Location - Color. Color. Color.
I remember this article, and I read it again, very well done, it's like a romance, of a project made from scratch to the highest levels and color color color with a mythology hero that live in a basement for months cooking the color profiles, a master druid that understand color better and second to none, I'm impressed in how many color research company that sell's their instrument / spectrophotometer and color targets to the entire World to calibrate check and proof the color precision from display to rotogravure print to fine art print to special application are so blind they doesn't understand that their effort only have the result in getting good reproduction of colors of their obsolete targets, well it's only a multi billion dollar color critic mass market...

Honor to the guys that made this magic possible !

But, I already don't know if the background in the image shown above is Yellow or Green, who knows ? for sure there is some human mistake.

Maybe who have made the shot can clarify it to understand which of the two or three is the more color accurate ?

Domenico.
 
It appears that the IQ380 has inherited the corner issues of the IQ260 when doing really long exposure, i.e. for a single long exposure of around 16 minutes and beyond, the Dalsa CCD sensor would suffer from extra noise in the 4 corners of the frame, essentially making it into a crop frame.

Luckily with the new Sony CMOS sensor we are immune to this issue!

24.jpg

LE_darkframe_IQ250_vs_IQ260_vs_P45plus_en.jpg
 
Below shows the test of the 100MP with the 35mm HR (Digaron-S with 70mm IC). We can observe color cast but it cancels nicely with Capture One v9.0.1.

I haven't had time to thoroughly test this lens but my bet is that this lens will be mazing-free if you stay within the 70mm IC.

23.JPG
 
We also did some comparison between the Rodenstock 35HR and the Phase One 35LS. Unfortunately the 35HR we tested had some problems with decentered Copal 0 shutter so the comparison of corner sharpness was rendered invalid. The Phase One 35LS is a really great lens with corner-to-corner sharpness as advertised. I wouldn't hesitate to say it is ready for 100MP. However the 35LS is very huge and bulky when you compare it against its Rodenstock siblings.

So far there is no visible difference of center sharpness as far as we can see if we stop down both lenses to f/8. However there is a visible difference in sharpness at f/4. I would imagine that the technical camera wide angle lenses may still provide some advantages in sharpness when compared against the Phase One lenses.

25.JPG
 
Last edited:
Regarding tiling issue and vertical bandings, yes we confirm that when a wide angle technical camera lens is shifted along the shorter edge of the sensor to the extreme, all three digital backs can suffer from the tiling issue. In addition, the Sony CMOS sensors can suffer from another type of artifact - vertical banding. It would be helpful if future release of Capture One can correct this with LCC.

Note that if the lens is not shifted heavily, then the tiling issue is hardly observed on the Sony CMOS sensors. I will need to find another chance to further test the safe movement range to avoid tiling issues on these sensors.

For the vertical bandings, we always get it when a lens is shifted on these Sony CMOS sensors. So far this is true for almost all Rodenstock wide angle lenses (regardless of image circle) I have tested specifically for this case. It could become problematic for the sky.

26.JPG
 

tjv

Active member
While I appreciate these test results, is it not fair to say that the results will be rendered mostly invalid after the new patch for C1 is released tomorrow (if delivered as promised)? I.e. the 100 will improve somewhat?
 
While I appreciate these test results, is it not fair to say that the results will be rendered mostly invalid after the new patch for C1 is released tomorrow (if delivered as promised)? I.e. the 100 will improve somewhat?
I still have these RAW files so I will give an update if I find anything has changed at all (other than the LCC bug of v9.0.2).
 

AreBee

Member
voidshatter,

Regarding tiling issue and vertical bandings, yes we confirm that when a wide angle technical camera lens is shifted along the shorter edge of the sensor to the extreme, all three digital backs can suffer from the tiling issue...
I thought that tiling was an adverse effect related only to CCD sensors. :(
 
voidshatter,



I thought that tiling was an adverse effect related only to CCD sensors. :(
If you don't shift that much then it's hardly visible on the CMOS sensors. For my real world usage I have only encountered this once and that was when I was too greedy to shift it to the extreme.

On the other hand, it is very easy to see that on the CCD sensors even unshifted, and that could be 3 vertical lines in addition. The 8 partitions are clearly visible on a fullframe CCD sensor by eye, but when you stare at the IQ3 100MP surface it looks like a whole (without any partitions at all).

IQ260+23HR unshifted:

lithography_partition_1.jpg

IQ260+40HR unshifted but at ISO 200:

lithography_partition_2.jpg
 

torger

Active member
If it is like it has historically been the suppression banding and tiling can be improved, but the loss of saturation clearly seen in the long exposure sky (crosstalk effects) is not.

It is interesting though that the bug caused a green cast, maybe it is because they're trying to do something with the saturation... it shall be interesting to see what the update can do.
 

torger

Active member
voidshatter,



I thought that tiling was an adverse effect related only to CCD sensors. :(
It's related mostly to Dalsa CCD sensors, Kodak does not have the same issue. But all large(?) sensors have it to some extent. With the CCDs there are two reasons, one during manufacturing when the sensor is exposed in several steps leading to a uniformity variation (tiles is highly visible on Dalsa sensors if you just look at it), and the other is when you have multiple readout channels with ADC that doesn't match up perfectly. I think Dalsa sensor have the most visible tile lines of all when visually inspecting the sensor itself, and there's also been some considerable challenges to match the ADC channels.

In an extremely processed dark frame I can get one tile line on my 50MP Kodak (which has two readout channels) but I've never had any real world problems.

With the CMOS there's no problem with ADC channel matching (I think), but there can still be uniformity variation tiles from manufacturing.

Normally the tiles are 100% neutralized with the LCC shot, but if you have some crosstalk (which you have on the wides, especially when shifted) the tile lines appear with different strength depending on what actual light/color there is at the line which means that the LCC shot cannot cancel it out fully. To remove the tile line you then need to make some filtering after the LCC has been applied, which Capture One does with varying success.

If you actually get problems with tile lines depends a lot of what post-processing style you have. If you do low contrast stuff you rarely get any issues. If you do black and white high contrast processing you're almost guaranteed to get at least some issues. With the Dalsa CCDs that is... I don't know how many issues there are with the Sonys, I have seen tile lines at least one time but only at crazy crazy extreme processing just to test if I could get a line visible.

Here's an image showing Dalsa tile lines:
 
Here's an image showing Dalsa tile lines:
I'm surprised that you still kept my image :)

I do not see such tiles on the IQ3100 or the IQ250, but I did not have a decent camera to shoot the surface of the IQ3100 when I tested it. I would expect the tiling issues to be far less with the CMOS.
 
RawDigger version 1.2.4.437 supports both IQ3 100MPix formats, 14-bit and 16-bit; available from our site at http://www.rawdigger.com/download
List of supported cameras is available at http://www.rawdigger.com/usermanual/cameralist
Thanks for the update! I confirm that RawDigger 1.2.4 now reads the IQ3 100MP IIQ file correctly and is consistent with the DNG files converted by Capture One v9.0.2.

Unfortunately the IQ250 is still not fully supported (at least not cooked as Capture One). I guess no one cares about that crop sensor :D

44.JPG
 
Top