Hasselblad 100C and 35XL

[usm]

I've an idea. Maybe I can fix it with my frame averaging tool, since it has already access to the 16bit raw data of the 3FR RAW files. And since it still outputs a native 3FR file, HNCS is preserved.

Does anyone know if the PDAF map of the Sony sensor in the CFV-100c is documented somewhere?
I can infere it but it would be much better to have an official documentation.

This is not my business, but just a question: Why not implimenting the complete Scene Calibration with the LCC files into your tool? With the integration of your Photoshop workaround for the PDAF and Sensor tilling the output file would be a clean corrected 3FR file. Just dreaming...

 

[mristuccia]

This is not my business, but just a question: Why not implimenting the complete Scene Calibration with the LCC files into your tool? With the integration of your Photoshop workaround for the PDAF and Sensor tilling the output file would be a clean corrected 3FR file. Just dreaming...

Thanks for the suggestion, that is indeed what I'm pondering right now.

I've tested the possibility of correcting the banding without LCC and it seems not possible. They are not clean stripes, they are shaded bands, quite ample in terms of number of rows, so interpolation would not work, it would ruin the image details on such zones. Moreover, the banding is not the same on the whole image, and that cuts out the option for a fixed "blind" correction.

I would need a bit of documentation on how scene calibration works though. Probably it is still a matter of inverting the LCC and using it to counter-balance color and brightness on the real image. But I also think that commercial solutions implement a lot of tricks to even out results and lower the noise in the periphery areas.
Moreover, I suspect that RAW editors don't apply LCCs to the native RAW 16bit data, which is monochromatic by its nature, but rather after demosaiking took place.

Unfortunately, computer imaging is not my aree of expertise, as my focus in IT business is enterprise integration and automation on Telco Cloud Platforms. So, this is really new territory for me, which requires a lot of investment in terms of both time and education.

But I'll try...

 

[Paul Spinnler]

Years ago when the H6 came out I got with my HB rep about testing the H6D100 back on a tech cam. After getting with HB support and buying the necessary accessories and stuff to be able to power the back and trigger it, the rep ghosted me. Since then HB gave me the impression they weren't terribly interested in the tech cam market.

It's disappointing that there are issues with banding on at least some wide angle lenses, but, the tech cam market seems to be an afterthought for HB and not their primary market outside of the X, H, or V ecosystems. I could be wrong but the people trying to use tech lenses with the CFV100 are probably a niche within an already niche market....and it wouldn't surprise me if the market was limited to the people in this thread (especially regarding SK lenses). This issue also seems to only really rear its head in certain settings using gear outside of the HB ecosystem. My gut feeling is that the importance of this issue only extends throughout a very small sample size of their user base that isn't representative of how most people use the CFVII, but of course I could be wrong.

Do these issues pop up on the IQ4 150? I honestly don't know. I used the IQ4150A and the 40HR but didn't really notice anything weird with full movements, but I also wasn't pixel peeping and looking with a magnifying glass or doing extreme edits.

If the IQ4 doesn't share this issue, and if this issue affected my work, and if I didn't rely heavily on the XCD lenses, I'd just save up, grab an IQ4, and move on personally. At this rate, by the time HB does fix the issue, the price delta between used IQ4s and CFV100s will be a lot smaller than it is now.

HB is rumored to have a new camera coming out soon (X2DII is the rumor). Maybe we'll see a CFVII with the issue improved and they were waiting for the new model to put in the work. Or maybe it will be worsened. Who knows.

My own personal feeling is if we're expecting manufacturers to ensure 100% compatibility with legacy, discontinued, and no-longer manufactured items made by third parties, we're bound to be disappointed. I've been tempted by SK lenses but have been sheepish because I worry too much about "what happens if I drop it in the field? can I even get it serviced?" so I've only dabbled with the rodie lenses.

To each their own of course and I understand the disappointment of those affected by the issue. I'm disappointed too, but it doesn't seem like I'm going to be investing in any tech cam glass anytime soon, so it's kind of a moot issue for me for the time being.

IQ4 does not have problems with 35 XL - it's the only live view back in the market that works with all Schneider and Rodie lenses - its the ultimate back for tech cams. On top you can go via C1 which has the correction profiles for Rodie glass and great workflow.

Hasselblad backs have awesome color science, and I recently got a H6D-100c, but even with the 40 HR you have color cast and the backs require a more complex workflow to extract the most out of. This said, the colors are sometimes worth it!

 

[mristuccia]

I may be onto something...

Original image (with too much rise for my SK 35XL, this image has been discarded for that reason, but it is good for this test):


LCC applied (on 3FR raw data, still 3FR as output):



Detail (original - 100% crop - note the infamous banding):


Detail (100% crop - LCC applied on 3FR raw data, still 3FR output - note that the banding is automatically fixed):




Second sample detail (original, 100% crop - note the big dust spot on top-center):


Second sample detail (LCC applied on 3FR raw data, still 3FR output - note that the dust spot is automatically fixed):


There is some noise in the areas that were originally darker and got a vigorous vignette correction. But I think that is a natural consequence of the exposure compensation.

I don't want to apply noise correction, it is better to leave it to the sensibility of the person who will post-process the image.

Will apply this feature to my frame-averaging tool first (a single LCC will be applied to the composed 3FR), then maybe I will create a tool specifically for LCC batch processing to a series of images couples (image + LCC). 3FR to 3FR of course.

 

[MGrayson]

Marco,

Excellent examples. I can't imagine using a tech camera without taking LCC's. Vignetting and dust are reasons enough.

Matt

 

[mristuccia]

My work also proves that HB could make their scene calibration algorithm work like mine so that banding and dust spots could be fixed in Phocus.

 

[ruebe]

My work also proves that HB could make their scene calibration algorithm work like mine so that banding and dust spots could be fixed in Phocus.

Exactly, I am shocked (also in a very positive way about your abilities)!

I am not sure what amount of work you put into this, but I suspect HB put more of it into replying to our demands than into actual problem solving.

Marco, if you make this tool work and accessible, from my point of view you need to put a price tag on it, I gladly pay for this and keep Phocus out of my workflow from then on. Only LCC and basic adjustment kept me there (I need to restart the thing several times each session to access files, register LCC's ...)

 

[ruebe]

and one more idea: I almost never need a 100MP file and would love to get a 3FR 50MP File out of a tool like yours. Is that doable, like a sensor+ P1 thing? I'd pay for that as well, especially when it's a 3FR file I can archive

 

[mristuccia]

Exactly, I am shocked (also in a very positive way about your abilities)!

I am not sure what amount of work you put into this, but I suspect HB put more of it into replying to our demands than into actual problem solving.

Marco, if you make this tool work and accessible, from my point of view you need to put a price tag on it, I gladly pay for this and keep Phocus out of my workflow from then on. Only LCC and basic adjustment kept me there (I need to restart the thing several times each session to access files, register LCC's ...)

and one more idea: I almost never need a 100MP file and would love to get a 3FR 50MP File out of a tool like yours. Is that doable, like a sensor+ P1 thing? I'd pay for that as well, especially when it's a 3FR file I can archive

Thanks for your kind words.
This work is taking me a lot of time, but me being a single person, I can imagine that a company can put more resources into projects like these.

I'm considering putting a price tag on the tool, especially if I'll keep adding all new features we are discussing here. It's a lot of extra-time spent for one single person.

As for the downscaling thing (like P1 Sensor+) I think that pixel binning is the best solution, but that would need scaling from 100mpxl to 25mpxl...
Moreover, Sensor+ is an hardware + software solution which cleverly lowers the noise down on both sides. A software-only solution will work fine, but in terms of ISO/noise gain it would not be as good as the real thing of course.

 

[SrMphoto]

Thanks for your kind words.
This work is taking me a lot of time, but me being a single person, I can imagine that a company can put more resources into projects like these.

I'm considering putting a price tag on the tool, especially if I'll keep adding all new features we are discussing here. It's a lot of extra-time spent for one single person.

As for the downscaling thing (like P1 Sensor+) I think that pixel binning is the best solution, but that would need scaling from 100mpxl to 25mpxl...
Moreover, Sensor+ is an hardware + software solution which cleverly lowers the noise down on both sides. A software-only solution will work fine, but in terms of ISO/noise gain it would not be as good as the real thing of course.

I doubt that Sensor+ lowers the noise except at pixel level.

Leica has implemented multi-resolution modes on their latest cameras. While PR initially claimed higher DRs, engineering clarified that the same DR effect can be accomplished by downscaling in Photoshop.

+1 on your LCC tool. It would be nice if it could be invoked from LrC so that the user can select images in the catalog.

 

[cly]

Marco, if you make this tool work and accessible, from my point of view you need to put a price tag on it, I gladly pay for this

I'm still undecided whether to invest in a CFV100 (SK28, 35, 43, 60, used with an IQ250), mainly because of the banding issue, but if I do, I'd be in too.

 

[MGrayson]

I'd buy it, and I won't start demanding support and complaining about slow releases and bug fixes. I promise!

Matt

 

[alexfth]

The implication is to test this properly you'd need to use a lens that you can shift the same amount on both, with the same magnification. Does the APO-Digitar 47/5.6 have the problem on the 100 CFV 100C? If you use the L-bracket you can shift it the same distance on an F-Universalis with GFX as you can with a 100 CFV 100C. @John Leathwick has this arrangement (L-Bracket and F-universalis with the 47mm) and has not mentioned having issues with PDAF banding.

I tried the Schneider Apo 47XL with the X2D on the F-Universalis and it indeed showed banding on vertical shifts…

 

[rdeloe]

I tried the Schneider Apo 47XL with the X2D on the F-Universalis and it indeed showed banding on vertical shifts…

Whoa. That's not good. So this is not just a CFV 100C problem.

 

[Paul Spinnler]

I'm still undecided whether to invest in a CFV100 (SK28, 35, 43, 60, used with an IQ250), mainly because of the banding issue, but if I do, I'd be in too.

I'd be cautious with a full line up of SK ... better IQ4

 

[harshdreams]

Agree with Paul, I have a similar SK line of lenses 28,35,60 & 90 and am glad I stuck to the IQ4 ... with the 28XL I can shift upto 10mm, anything more though is possible but I get sensor tiling which I am trying to sort out... but the IQ4 works great for the rest of the lenses....

 

[mristuccia]

I'm curious. Once we found a way to fix the banding, what would the advantage of an IQ4 be to, say, a CFV-100c, X2D or a GFX 100 when using the SK lineup?
Does it allow more shifting and/or less CA?

 

[anyone]

I'm curious. Once we found a way to fix the banding, what would the advantage of an IQ4 be to, say, a CFV-100c, X2D or a GFX 100 when using the SK lineup?
Does it allow more shifting and/or less CA?

Capture One as raw converter and the larger sensor (apart from the resolution difference). Both a matter of preference arguably.

 

[SrMphoto]

I'm curious. Once we found a way to fix the banding, what would the advantage of an IQ4 be to, say, a CFV-100c, X2D or a GFX 100 when using the SK lineup?
Does it allow more shifting and/or less CA?

Larger sensor, higher resolution, in-camera frame averaging, dual exposure, raw histogram, wider wide angles. Additionally, no banding fix is required, but if the fix is combined with the necessary LCC, there is no additional overhead. If Marco succeeds in launching his tool (LCC with dust and banding correction), it will be a significant expansion of CFV-100c's usage.

 

[mristuccia]

But none of the above mentioned features have something to do specifically with the SK lineup. They are mostly valid for any technical lens brand, aren't they?