URSA Mini 4.6 - Repro Case| Sensor Bug -Magenta Edge Shift

Hi Black Magic,

thanks for your great initiative and hard work.

There has been much talk the URSA Mini 4.6 producing edges with Magenta Shifts.
While every camera produce 'some' color shift in where the lens enters into a state of Fall-Off and this may/will result in color-shifts towards those - the URSA Mini 4.6K seems to suffer from a much greater problem which is ultimately amplifying those very fall-off color shifts.

I have been conducting numerous tests and I can encounter and repro this issue 100% of the time.
While I believe that each URSA Mini 4.6K Sensor on all cameras suffer from this issue, I am certain that none of the sensors produce identical artifacts. Thus, some users may encounter shifts in all 4 outer corners while others may only encounter i.e. shifts on the left side of the sensor.

My Camera, suffers from a right-rectangle-edge magenta shift with max-amplification on top-bottom corners.
Please see this following image of an iMac rgb(255, 255, 255) background in a total dark room with Carl Zeiss 50mm Planer situated perpendicularly snug to the screen - shot on 4.6RAW - Aperture 5.6 - ISO 200. Image was white balanced and crush-boosted to fit normalization range between 0 and 1.

imacWhite.jpg (769 KiB) Viewed 15038 times

Next up is a more scientific approach which eliminates any light source and simply reveals sensor noise.

Case

0) Removed Lens and attached URSA Turret Protector placed dark cloth around Turret.
1) No Light In Room nor outside - so totally darkness.
2) ISO 200, Shutter 45deg. WB 6500

Image was crush-boosted to reveal result. The result is highly disturbing and clearly reveals that something is wrong with the sensor. Be that uneven current or mal-placed filters. I don't know. But I do know this, the result should not equal a clear-cut gradient from left to right. It should reveal, noise of course but evenly lit.

Please check this result and please respond to whether this is normal for your sensors. Just in case, I am enclosing the identical scenario from your Black Magic Production Cinema Camera 4K - which is VERY evenly lit just like any of the other cameras I own.

PLEASE NOTE - BY MISTAKE I SWAPPED THE RIGHT/LEFT SIDE LABELS OF SENSOR ON THE IMAGES

URSA_Mini_4.6K_Sensor - Gradient Noise

URSA_Mini_4_6K_Sensor.jpg (699.66 KiB) Viewed 15038 times

BMPCC_4K_Sensor - Linear Noise

BMPCC_4K_Sensor.jpg (783.27 KiB) Viewed 15038 times

In my initial tests, I had not been able to see the errors of the sensor but after a shoot today, it became very evident that something serious is wrong. The Huge Magenta thread on this forum is right. While Magenta is just one of the symptoms there are MANY others - basically this uneven lighting of the sensor causes ALL kinds of color shifts. NOT just magenta but also Cyan and in my case the left part of the image looks different from the right part of the image due to the Gradient !

Before I go so far and call my dealer and void the deal on the URSA Mini, I would like you to respond to this thread. Perhaps you could state whether this is indeed a known issue and MORE importantly - can you fix it with a Firmware Update.

If not, I am going to have to take it back as I cannot use the Images.

Thanks for Reading and Responding

Morten Carlsen

I am enclosing an Image from the same camera where the issue is NOT evident. So as to prove that many may not even have yet found themselves in a situation where the bug becomes important...

Just so no one gets confused... The right side of the wall is intact a pastel lilac. This is not the Magenta Issue

colorChecker.jpg (676.71 KiB) Viewed 14999 times

+1

I'm following this just as much as Im sure everyone else is or SHOULD be anyway.

Morten, can you do the same test with the lens cap with another RAW camera ? Even a stills camera....?

JB

He already did, 3rd picture in his post.

I just took delivery of the 4.6k today and haven't had time to test. Eager to see if BMD responds to what appears to be fairly compelling evidence complete with different behavior from one of their other cameras.

(I know they're looking into it.)

Morten Carlsen, I'd be interested in seeing that no-light noise test made on a 4.6K sensor that had been cooled in the fridge or freezer.

EDIT, sorry misread your post.

JB

Hi Black Magic,

thanks for your great initiative and hard work.

There has been much talk the URSA Mini 4.6 producing edges with Magenta Shifts.
While every camera produce 'some' color shift in where the lens enters into a state of Fall-Off and this may/will result in color-shifts towards those - the URSA Mini 4.6K seems to suffer from a much greater problem which is ultimately amplifying those very fall-off color shifts.

I have been conducting numerous tests and I can encounter and repro this issue 100% of the time.
While I believe that each URSA Mini 4.6K Sensor on all cameras suffer from this issue, I am certain that none of the sensors produce identical artifacts. Thus, some users may encounter shifts in all 4 outer corners while others may only encounter i.e. shifts on the left side of the sensor.

My Camera, suffers from a right-rectangle-edge magenta shift with max-amplification on top-bottom corners.
Please see this following image of an iMac rgb(255, 255, 255) background in a total dark room with Carl Zeiss 50mm Planer situated perpendicularly snug to the screen - shot on 4.6RAW - Aperture 5.6 - ISO 200. Image was white balanced and crush-boosted to fit normalization range between 0 and 1.

imacWhite.jpg

Next up is a more scientific approach which eliminates any light source and simply reveals sensor noise.

Case

0) Removed Lens and attached URSA Turret Protector placed dark cloth around Turret.
1) No Light In Room nor outside - so totally darkness.
2) ISO 200, Shutter 45deg. WB 6500

Image was crush-boosted to reveal result. The result is highly disturbing and clearly reveals that something is wrong with the sensor. Be that uneven current or mal-placed filters. I don't know. But I do know this, the result should not equal a clear-cut gradient from left to right. It should reveal, noise of course but evenly lit.

Please check this result and please respond to whether this is normal for your sensors. Just in case, I am enclosing the identical scenario from your Black Magic Production Cinema Camera 4K - which is VERY evenly lit just like any of the other cameras I own.

PLEASE NOTE - BY MISTAKE I SWAPPED THE RIGHT/LEFT SIDE LABELS OF SENSOR ON THE IMAGES

URSA_Mini_4_6K_Sensor.jpg

BMPCC_4K_Sensor.jpg

In my initial tests, I had not been able to see the errors of the sensor but after a shoot today, it became very evident that something serious is wrong. The Huge Magenta thread on this forum is right. While Magenta is just one of the symptoms there are MANY others - basically this uneven lighting of the sensor causes ALL kinds of color shifts. NOT just magenta but also Cyan and in my case the left part of the image looks different from the right part of the image due to the Gradient !

Before I go so far and call my dealer and void the deal on the URSA Mini, I would like you to respond to this thread. Perhaps you could state whether this is indeed a known issue and MORE importantly - can you fix it with a Firmware Update.

If not, I am going to have to take it back as I cannot use the Images.

Thanks for Reading and Responding

Morten Carlsen

Morten, This is brilliant!

Thank you for presenting a test that truly digs down deep into the bowels of the sensor! This alongside the open sensor white test, this clearly demonstrates that something is not right with at least some of the cameras. Being that I've seen clean and beautiful examples from the 4.6k and people have reported having NO color issues of any kind, I'm inclined to conclude that this affliction affects only certain cameras.

It really looks like you've uncovered a big piece toward understanding this puzzle! If there was a prize (like a shiny new properly working Ursa Mini 4.6) for getting to the root of this issue, at this moment in time, I would nominate you!

Blackmagic, you're seeing this right??

Blackmagic, you're seeing this right??

Kristian Lam (Blackmagic Design) stated in the other thread that -"This is not a sensor or hardware issue."

EDIT: Attached the full quote below.

The second topic is to do with the magenta corners exhibited in some of the samples posted here. This is not a sensor or hardware issue. Some of you have already correctly pointed out that this has to do with how the light rays are hitting the sensor. This is especially true for lenses that are not telecentric by design. You can find many examples of this happening on many other camera systems and also a myriad of ways to deal with this. The reason there seems to be cameras that exhibit this phenomenon more than other cameras is to do with the combinations of different types of lenses, different aperture or focal length, and even variances between the same model of lenses.

Here is a DNG of the same test on my camera: cap on, same settings as Morten.

https://www.dropbox.com/s/8jeg0dv55f2ge ... 0.dng?dl=0

Screen Shot 2016-04-12 at 10.36.20 PM.jpg (836.58 KiB) Viewed 14709 times

Purely for informational purposes. I have no idea if this is relevant.

Here is a DNG of the same test on my camera: cap on, same settings as Morten.
Purely for informational purposes. I have no idea if this is relevant.

I pulled it into Resolve and pushed it every which way, but no gradient or vignette was visible in the noise pattern that I could see. Though, as you say Matt, who knows if it is a relevant test.

Here is a DNG of the same test on my camera: cap on, same settings as Morten.

https://www.dropbox.com/s/8jeg0dv55f2ge ... 0.dng?dl=0

Screen Shot 2016-04-12 at 10.36.20 PM.jpg

Purely for informational purposes. I have no idea if this is relevant.

Mine looks the same:
https://www.dropbox.com/s/9tcz1892qv2xu ... 7.dng?dl=0

Morten: I want to make sure I'm boosting the footage the same way you are in post; What software/setting were you using?

I don't think there's a right way to do a test like this.

But i think a lens cap test isn't relevant. Noise or a boosted image will always introduce other variables.

I'm also not so sure about a computer screen as a light source. Firstly, I think you want a LARGER source of light relative to the sensor. And the colour of a screen is never what we consider white, no matter what the value is of the image generated (RGB wise)

I can see a MASSIVE chunk of dirt on the sensor or back of the lens there as well...top right...That would possibly affect the results of this version of a test.

And to be honest, the same test with and without lens tells us more.

JB

Blackmagic, you're seeing this right??

Kristian Lam (Blackmagic Design) stated in the other thread that -"This is not a sensor or hardware issue."

EDIT: Attached the full quote below.

The second topic is to do with the magenta corners exhibited in some of the samples posted here. This is not a sensor or hardware issue. Some of you have already correctly pointed out that this has to do with how the light rays are hitting the sensor. This is especially true for lenses that are not telecentric by design. You can find many examples of this happening on many other camera systems and also a myriad of ways to deal with this. The reason there seems to be cameras that exhibit this phenomenon more than other cameras is to do with the combinations of different types of lenses, different aperture or focal length, and even variances between the same model of lenses.

JoeyL, I'm fully aware of what Kristian said. Kristian also mentions nothing about why open sensor tests, shot in a variety of ways (using a blank white wall, covered with white paper, covered with a PhaseOne issued white translucent plexiglass specifically designed for mapping lens cast on sensors, in direct sun, in open shade and using an industry standard color correct transparency viewer) all still show an uneven cross color magenta cast. This would not be the case if it was purely a lens/angle of light ray created cast. This is not to say that lens cast is not also at play here, which is fairly common, especially on large medium format sensors, but the open sensor test tells us that something else is contributing and exacerbating the magenta problem in certain cameras. I also believe that Morten's capped sensor noise floor test also reveals an underlying hardware/manufacturing problem. One potential manufacturing problem could include, but is not limited to, not calibrating each individual sensor with it's own custom "sensor shading" adjustment that John Brawley had spoken of many posts ago in another thread. At that time, he suggested that final production "sensor shading" (which had not been done yet) could have an effect on reducing or eliminating the magenta cast that was being seen at that time in many beta test clips that had been posted by different beta testers. (John, please correct me if I'm paraphrasing you incorrectly)

I don't think there's a right way to do a test like this.

I think there is !

If images exhibit artifacts with Filter on Lens -> Remove Filter and reshoot. If artifact persist without the filter -> Exclude next-in-inline possible culprit -> Lens. Try with a different Lens. If artifacts remain identical with different lens, move down a step on the ladder -> Remove Lens and check sensor. If sensor reveals identical artifact - Culprit equals Sensor.

But i think a lens cap test isn't relevant. Noise or a boosted image will always introduce other variables.

1 - It was not a lens cap test. It was turret cap test.
2 - You are right on the money that other variables will be introduced - but one thing that should definitely not happen is that the noise should not produce a consistent gradient clearly identifiable from left to right or from one direction to another.

I'm also not so sure about a computer screen as a light source. Firstly, I think you want a LARGER source of light relative to the sensor. And the colour of a screen is never what we consider white, no matter what the value is of the image generated (RGB wise)

screenRec.jpg (62.17 KiB) Viewed 14374 times

True - but if you align the lens perpendicularly-snug against the screen covering an area of 57mm in diameter and the result corresponds to the artifacts induced when shooting a white wall 3 Meters away - THEN - one can safely assume that the problem is in fact identical. And when THAT result corresponds 100% with the result of boosting noise on a turret-cap black image, then the evidence is no longer circumstantial but hard fact.

I can see a MASSIVE chunk of dirt on the sensor or back of the lens there as well...top right...That would possibly affect the results of this version of a test.

The dirt stains on the lens aren't playing a major factor here. If they would then where those are situated that would be the only place where Magenta shifts would be visible. Clearly that is not the case.

And to be honest, the same test with and without lens tells us more.

Yes - the test with the turret cap on in a dark room are much more important than those with Lens on as they reveal that the sensor has a major issue...

-- Best
Morten

Hi There,

I went ahead and downloaded the CinemaDNGs from Fanon and Matt and compared with mine.

1) I opened all in Photoshop
2) Made them Monochrome
3) Applied a 5px Gaussian Blur
4) On Left side of image -> grabbed the color of a median gray
5) On Right side of image -> grabbed the color of a median gray
6) Drew an interpolated gradient from right to left.

I did the same to turret-cap black image from my BMPCC

The whiter parts of the image are indicative of a Magenta Color-shift.
Notice the strong blobs in Matts image which are basically not there on Fahnons or mine.

While Fahnons Sensor according to my tests, should reveal the least color shift, there is still a very clear gradient going from left to right which would make it very difficult for him to shoot a centered talent sitting on a white background. My Sensor is the one with the strongest gradient which would make it impossible for me to shoot what I described in Fahnons case. Matt's Sensor has random blobs combined with a slighter gradient. That would explain 'the Porsche in the parking lot' image that one user uploaded with random blobs of magenta on the asphalt.

While the BMPCC also reveals slight irregularities - they are far from the severity of the URSA Mini 4.6k. I have never noticed any issues of this magnitude in the year having operated that camera.

That said, there are always going to be irregularities whether RED, Canon, Sony or BMD - it is just that for some reason those irregularities are exaggerated on the URSA Mini 4.6K.

I am no expert in Sensors. So perhaps someone else who is could chime in. Could it be that either the current is off or that the color filters on the sensor have been unevenly glued on (Are they glued on ??? ) or how do they get attached ?

Anyway - here are the results - The BMPCC Image Result is in the following Post. I could only attach 3 images to this post !

FahnonBennet.jpg (713.72 KiB) Viewed 14356 times

MattWhite.jpg (657.58 KiB) Viewed 14356 times

MortenCarlsen.jpg (651.47 KiB) Viewed 14356 times

For some reason I could only add 3 images to the last main post

So here is BMPCC

BMPCC.jpg (487.77 KiB) Viewed 14353 times

Here is another test done simulating what would happen if I were to shoot a talent on white background..

It would be 100% impossible to correctly grade the image in that scenario.

Check the gray card in the center. It is ALSO revealing (within it) a slight magenta offset from left to right.
Almost as if the sensor would be divided into two halves - with the problem getting more prominent the further the off center.

This is definitely something that I have never seen on any other camera and even though I have been doing color grading the past 20 years - I would NOT want to endeavor into grading this one. Although possible with A LOT of gradient layers in Photoshop - I would not want to spend THAT amount of time to fix this issue on EVERY shot That in the end would be have to be charged to the client.

WhiteBGTest.jpg (742.73 KiB) Viewed 14281 times

What disturbs me the most is that the beautiful imagery from John's Film featured on the BM Website does not seem to suffer from this.

1 - Could it be that each sensor is not QC'd ?
2 - Could it be that the ColorFilters attached to the sensors are influenced by Temperature and humidity which will vary greatly from country to country and during flight @ 33,000 feet ?

I don't believe that the sensor used in the prototype cameras where affected from this issue at all. Otherwise they would never have shipped them and as we all know BM took their time to make sure they got it right.

So what could cause this weird issue occur ?

What disturbs me the most is that the beautiful imagery from John's Film featured on the BM Website does not seem to suffer from this.

Morten, while the BM website film may not suffer with the magenta issue, several test footage seemed to have them. So, in a way, it was a known issue, at least in some prototype cameras.

Check this thread, when some people actually pointed out the problem. But most refused to believe at that point of time.
https://forum.blackmagicdesign.com/viewtopic.php?f=2&t=46112&p=245516&hilit=magenta#p245516

What disturbs me the most is that the beautiful imagery from John's Film featured on the BM Website does not seem to suffer from this.

Morten, while the BM website film may not suffer with the magenta issue, several test footage seemed to have them. So, in a way, it was a known issue, at least in some prototype cameras.

Check this thread, when some people actually pointed out the problem. But most refused to believe at that point of time.
https://forum.blackmagicdesign.com/viewtopic.php?f=2&t=46112&p=245516&hilit=magenta#p245516

Thx Subrata,

1) I too believe that anything beta should not be judged until release. Hence the coined term Beta
2) I am beta testing for a lot of companies (software mostly) and have been doing so for the past 20 years.
We get a beta (sometimes even and Alpha) software find the bugs which in return are being fixed and then the Golden Master is released. Of course, if the beta testers aren't qualified the Golden Master is going to suffer from their incompetence.

3) If this issue with Magenta Cast, which John Brawley as a beta tester acknowledges in the thread you linked to,
was obviously known and identified prior to Final Release, then according to that statement of his, the beta testers came thru' with due diligence pointing out the troublesome sensor issue of Magenta casts.

So what on earth caused Black Magic to knowingly ship a BetaTester-Certified Malfunctioning Camera ?

I personally do not think they would do that as it would introduce irreparable damage to their name. Hence I believe in being part of the Solution as opposed to being part of the problem. What could have cause all those malfunctioning cameras to be shipped around the world?

Perhaps they were NOT faulty at the day of Shipment ?
What happens to a Camera's internals in the cruising altitude of a 747 at negative 60° Celsius ?
Did the cameras perform fine when leaving the plant ? Where they assembled in humidity A and did the components distort in Humidity B ?

Who knows - all I care about is that the Mini 4.6 gets up and running both for me personally and for Black Magic Design which reputation in the Camera Business is depending much on this camera to be successful and perform great so that they can further keep developing awesome cameras which most can afford.

I agree with Morten 100%

I have yet to place my order, do you think its worth rolling the dice? Hope for a good one? And if not get it RMA?

Honestly I dont really like any other camera in this market and the Raven is JUST out of my budget for a camera at this time, and I know NAB is next week, but then add another 6-8 months to that... I guess just some advice?

Thanks guys. I hope we all get it worked out and keep doing cool **** with our BMD cameras!

I agree with Morten 100%

I have yet to place my order, do you think its worth rolling the dice? Hope for a good one? And if not get it RMA?

Honestly I dont really like any other camera in this market and the Raven is JUST out of my budget for a camera at this time, and I know NAB is next week, but then add another 6-8 months to that... I guess just some advice?

Thanks guys. I hope we all get it worked out and keep doing cool **** with our BMD cameras!

Hey Alberto,

look at it like this:

1) If you place your order, it will take at least 1-3 months until you get it depending on where you are located. I believe waiting lists are long
2) If you are on no dire need to get a functioning camera the next 2 days for a shoot then why not place your order ?

If the camera's sensor is malfunctioning like mine, you will get a replacement.
If they will never get the sensor to work as expected and as worst case scenario quit manufacturing the sensor you will get your money back from your dealer.

So according to my risk assessment | You don't have anything to loose unless you need a perfectly functioning camera by tomorrow evening or 3 or next week.

I am personally in the scenario which you could be facing. This past Friday I dropped 8k EURO on the UM46 incl. the VF etc. and two fast cards. Heck I even ordered the CFAST to SSD for $599 from the US. So in total I am looking at about 9k EURO but I have a camera where I cannot use any of the images due to a malfunctioning sensor.

My dealer is aware of the issue after reading this thread and immediately offered me the money back for the entire package. I told him to hold his horses and wait for Black Magic to figure this one out. I want my camera and what it can do to succeed and I am willing help getting there too !

So - my advise to you is genuine as I am sitting in the boat of unfortunates

I hope they get this issue worked out before they start shipping the turret for my BIG ursa!
Light hitting at a weird angle? not whet the lens is capped.

my biggest concern is will BM acknowledge this as a fault OR just a characteristic of the way the sensor responds to light through different glass. Depending on that answer could leave me with a camera with an unacceptable colour shift that I am told to live with.

my biggest concern is will BM acknowledge this as a fault OR just a characteristic of the way the sensor responds to light through different glass. Depending on that answer could leave me with a camera with an unacceptable colour shift that I am told to live with.

Tell you customers this is what they will get. Like it or not.
I'm sure they will understand.

Guys - I am in the boat having paid for this Camera and having the actual problem.

You might not even have that issue with your, once it arrive

This will be dealt with and the more we are part of the solution the faster this will happen.
That means - lets keep it as positive as possible so that others don't have to experience what I and others just did.

No need beholding the Pale Horse before it gallops onto the scene !

Morten, Thanks! Sound advice!
Lets roll the dice lol

Ok,
I am positive they will understand.

My Camera, suffers from a right-rectangle-edge magenta shift with max-amplification on top-bottom corners.
Please see this following image of an iMac rgb(255, 255, 255) background in a total dark room with Carl Zeiss 50mm Planer situated perpendicularly snug to the screen - shot on 4.6RAW - Aperture 5.6 - ISO 200. Image was white balanced and crush-boosted to fit normalization range between 0 and 1.

imacWhite.jpg

Morten Carlsen

Hi Morten Carlsen, Can you please make available your imacWhite image in a DNG file? I'd like to bring it into Capture One Pro and see if a successful lens cast correction file can be created. Of all the tools I know of, the lens cast tool in Capture One is the only one capable of dealing with lens cast. I had mentioned this before and showed an example of it working, but I'd like to try it on your file too. If this type of tool existed in Resolve, I believe most of these issues could be dealt with, but for now it remains in a stills only program unless there's a way to export the correction file and plug it into Resolve?

I'd like to bring it into Capture One Pro and see if a successful lens cast correction file can be created. Of all the tools I know of, the lens cast tool in Capture One is the only one capable of dealing with lens cast... If this type of tool existed in Resolve, I believe most of these issues could be dealt with...

Just a heads up but a LCC is very shot dependent, and to be used correctly needs to be redone as focal length change, focal point change, and movements are adjusted. Wether it can be used for multiple is dependent on what one views as acceptable. Also it's a pain to an LCC all the time.

I'd like to bring it into Capture One Pro and see if a successful lens cast correction file can be created. Of all the tools I know of, the lens cast tool in Capture One is the only one capable of dealing with lens cast... If this type of tool existed in Resolve, I believe most of these issues could be dealt with...

Just a heads up but a LCC is very shot dependent, and to be used correctly needs to be redone as focal length change, focal point change, and movements are adjusted. Wether it can be used for multiple is dependent on what one views as acceptable. Also it's a pain to an LCC all the time.

With medium format technical cameras with swings and tilts, LCC is indeed very specific and limited to just that one setup and is as you mentioned, very shot dependent. But with with non-tilting, center mounted lenses (like most other cameras), a LCC has a bit more effective "wiggle room" and I believe could be created for each lens and basic range of t stops in advance, ready to drop in a color correction program (if that feature could exist). The LCC itself, is not dependent on any scene or shot, it maps out the lens cast created by the sensor from each specific lens using a white translucent plexiglas in front of the lens. But no doubt, that varying the focus and aperture would change the "perfect fit" of a LCC to some degree, but I'd be willing to bet it would offer a very good start toward taming this problem.

[Kristian Lam]

Hi Morten,

I've emailed you to get hold of the DNGs.

Morten and company,

I was in line to but mini 4.6K but when the dealer called me and said I was next, at same time I heard about these issues and got very skeptical so I stepped out of line and said no thanks. This is not $995 pocketcam (which I own by the way) - as morten pointed out here this is nearly $10k or more.

I've been grading some raw images from mini 4.6k available online and notice in Resolve if you do a color boost and lower exposure just a little you can easily get magenta cornering to appear, and also around the frame edges in some cases. If you increase exposure the colored corners tend to go away.

Has anyone tried lens wacking to see if magenta increases/decreases?

For what it's worth, my invoice from my dealer was about $15K Canadian. I'm starting with nothing , and $15K is just for the camera, shoulder mount, battery plate, BM viewfinder, battery, charger, 2x Wise 256GB CFast 2 cards, Wise card reader. No lights, no audio, no rails, no matte box, no follow focus, no lenses, no filters, no tripod, no fluid head. I've added a B4 Cine zoom yesterday. Top of the line iMac arrives this week. Pegasus2 R6 arrived already. I will add a set of SLR Magic APO lenses and SLR Magic Vari-ND, Schneider 750 nm IR Cut, and 5 NiSi ND filters. The missing pieces will take a year or more! The camera needs quite a supporting cast.


Sent from my iPad using Tapatalk

Let's summarise.

It's perfectly normal to have the edges and corners of a sensor have fall off with lenses. This is why LCC, peripheral illumination and adobe's lens profiles have corrections for exactly this well known and established issue.

There are also known casts with many cameras. You can't expect a camera to have perfect WB on a shot out of the box. No camera will. Check out this little doozy on the Alexa, known to generally be green....have a look at the uncorrected shots that HAVE the IR filter. Look at the UNGRADED stills. They're very very green.

http://www.thehurlblog.com/what-is-ir-p ... it-tiffen/

And a search of REDuser gets you this....

http://www.reduser.net/forum/showthread ... gon-Sensor

http://www.reduser.net/forum/showthread ... reen-Shift

http://www.reduser.net/forum/showthread ... -the-image

So there's two ways that you could explain a GENERAL cast or dark corners or BOTH.

I still think many of the posted examples are falling into the above two categories. It's expected behavior.

There are also some examples where it seems to be more than that, where the colour cast isn't uniform or is tied to exposure....as it goes darker it goes more magenta. That is going to be harder to fix. That's the kind of issue that you should be taking back to BM and doing it in a way that's reproducible with your examples.

JB

For what it's worth, my invoice from my dealer was about $15K Canadian. I'm starting with nothing , and $15K is just for the camera, shoulder mount, battery plate, BM viewfinder, battery, charger, 2x Wise 256GB CFast 2 cards, Wise card reader. No lights, no audio, no rails, no matte box, no follow focus, no lenses, no filters, no tripod, no fluid head. I've added a B4 Cine zoom yesterday. Top of the line iMac arrives this week. Pegasus2 R6 arrived already. I will add a set of SLR Magic APO lenses and SLR Magic Vari-ND, Schneider 750 nm IR Cut, and 5 NiSi ND filters. The missing pieces will take a year or more! The camera needs quite a supporting cast.


Sent from my iPad using Tapatalk

That's a very nice clean slate you're starting with Rick! I'm curious, where did you determine that the Schneider 750nm IR cut was the one to get for the 4.6k over the 680nm IR cut?

At this point it's speculation buttered my rationale:
The 680nm wavelength is in the visible light spectrum. I think it's deep red or far red. Infrared begins beyond 700nm. The Schneider IF489 filter begins a very gradual cut around 650nm and continues a gentle slope to 750nm at which point it eliminates true infrared. I think that gentle slope will give a better rendition of skin than the modern CUT filters that are like an abrupt curtain closing which takes some of the colour from skin. Here's what I posted on the BMCuser forum:

Yesterday 01:02 PM
The Schneider IR CUT IF489 has a spectrum response curve that I like in theory. I haven't found the spectrum analysis for their True Cut 750 filters but I hope it's very similar. What do you think?

http://www.schneiderkreuznach.com/en...ir-cut-filter/


Sent from my iPad using Tapatalk

You want the 680 IR cut


Sent from my iPhone using Tapatalk

There are also known casts with many cameras. You can't expect a camera to have perfect WB on a shot out of the box. No camera will. Check out this little doozy on the Alexa, known to generally be green....have a look at the uncorrected shots that HAVE the IR filter. Look at the UNGRADED stills. They're very very green.

http://www.thehurlblog.com/what-is-ir-p ... it-tiffen/

And a search of REDuser gets you this....

http://www.reduser.net/forum/showthread ... gon-Sensor

http://www.reduser.net/forum/showthread ... reen-Shift

http://www.reduser.net/forum/showthread ... -the-image

So there's two ways that you could explain a GENERAL cast or dark corners or BOTH.

I still think many of the posted examples are falling into the above two categories. It's expected behavior.


JB

John, the issue with Red is nothing but choosing the wrong OLPF setting or the wrong white balance. Going through the threads makes that very clear. That's very easily correctible in post, specially when people are shooting RAW.

Lens fall-off is also a known issue, we have lived with that. Certain lenses create a kind of vignette in the image (Zeiss 16 mm on an Red sensor for example). However, I've never seen this inconsistent colour shift in any of those cameras.

What Morten has posted here is his sensor image in a dark environment without any lens attached to the camera. So the question of lens falloff does not arise in his tests. Also, he's tested the BMPC4K alongside in same condition and you can easily say that his UM46 definitely has a sensor which is really problematic. No IR cut filter can solve his problem.

Now, it seems not all UM46 cameras are having this problem. At least it's not perceivable. That's good. I am not sure, maybe they too have this problem which is not as severe and is not quite noticeable in normal scenario.

In any case, this problem needs to be addressed. Maybe a LCC type of correction can be added to DaVinci Resolve for people having this issue? BMPC4K has a major Black Sun issue, which they fixed in Resolve (upto a point) and maybe something can be done now? I don't really know though.

I had been a fence-sitter with this UM46 and was planning to get one. As an extensive user of BMCC and BMPC I simply love the colour that the BM cameras can produce. With UM46 I was elated as this was the camera that would have actually propelled BMD into real camera professional arena. I have a feeling that Red understood this and that resulted in their Red Raven.

Unfortunately, BMD goofed up this time too. Let's recollect:

1. The camera was announced at NAB 2015, with promises of delivery in July 2015. Final product came in March 2016, a day before Red started shipping heir Ravens, which was announced in September.

2. It was advertised and people were bragging about the switchable Global and Rolling shutter. That was a dream. Finally, the Global Shutter thing was dropped.

3. Now this magenta/ colour shift issue. No possible solutions I can find from BMD, excepting similar answers like yours. But well, this may be "expected behaviour".

Frustrating enough, no?

At this point it's speculation buttered my rationale:
The 680nm wavelength is in the visible light spectrum. I think it's deep red or far red. Infrared begins beyond 700nm. The Schneider IF489 filter begins a very gradual cut around 650nm and continues a gentle slope to 750nm at which point it eliminates true infrared. I think that gentle slope will give a better rendition of skin than the modern CUT filters that are like an abrupt curtain closing which takes some of the colour from skin. Here's what I posted on the BMCuser forum:

Yesterday 01:02 PM
The Schneider IR CUT IF489 has a spectrum response curve that I like in theory. I haven't found the spectrum analysis for their True Cut 750 filters but I hope it's very similar. What do you think?

http://www.schneiderkreuznach.com/en...ir-cut-filter/

Sent from my iPad using Tapatalk

A very reasonable theory for sure, but the proof is always in the pudding. Are you able to exchange the 750 if it turns out you need to dig deeper toward visible light to cut any red pollution? Maybe they could let you test the 680, 715 and 750 and return the ones that didn't work as well? I recall the tests I saw for the original BMCC and it clearly needed the 680 to look it's best and the new 4.6k has been reported to have much of the 2.5k's DNA. BM cameras have always been very sensitive to IR and in past experience, need a pretty heavy IR hammer.

John, the issue with Red is nothing but choosing the wrong OLPF setting or the wrong white balance.

Or.....it's "known" to have a green tint when used with the right OLPF, it's just that it has a green tint ????

JB

John, the issue with Red is nothing but choosing the wrong OLPF setting or the wrong white balance.

Or.....it's "known" to have a green tint when used with the right OLPF, it's just that it has a green tint ????

JB

Actually not. Can you go through the full thread you posted? The answer lies there.

More than that, I wouldn't mind any tint in UM46 - CMY/RGB (or any colour you choose) if that is consistent throughout the sensor. Not only at the corners.

Actually not. Can you go through the full thread you posted? The answer lies there.

Perhaps I speak from personal experience and knowledge and use a cursory search of reduser to point out what is common knowledge to my peers and friends in post. Just like we also know Alexa's are green.

JB

Actually not. Can you go through the full thread you posted? The answer lies there.

Perhaps I speak from personal experience and knowledge and use a cursory search of reduser to point out what is common knowledge to my peers and friends in post. Just like we also know Alexa's are green.

JB

John, yes. I have personal experience too. Not always in shooting, but tons in post. Our post house does about one film per month, most of them are Red. Just don't forget that our country produces the largest number of feature films in the world, so we have the problem of seeing things in bulk.

Btw, I do find BMPC4K a bit greenish too, much like the Alexa. *grin*

John, yes. I have personal experience too. Not always in shooting, but tons in post. Our post house does about one film per month, most of them are Red. Just don't forget that our country produces the largest number of feature films in the world, so we have the problem of seeing things in bulk.

Btw, I do find BMPC4K a bit greenish too, much like the Alexa. *grin*

GREAT !

So the actual point...

Every camera has a cast. It's not abnormal to have a cast.

JB

John, yes. I have personal experience too. Not always in shooting, but tons in post. Our post house does about one film per month, most of them are Red. Just don't forget that our country produces the largest number of feature films in the world, so we have the problem of seeing things in bulk.

Btw, I do find BMPC4K a bit greenish too, much like the Alexa. *grin*

GREAT !

So the actual point...

Every camera has a cast. It's not abnormal to have a cast.

JB

Yes. It's not abnormal to have a consistent overall colour cast. No one is bothered about that.

The problem is when the colour cast is inconsistent. That's the problem that is being discussed here.

Yes. It's not abnormal to have a consistent overall colour cast. No one is bothered about that.

The problem is when the colour cast is inconsistent. That's the problem that is being discussed here.

Umm. Yes. Isn't that what I said !!!?????

JB

John,
The images I see from Benton and Morten are not quite examples of consistent colour cast. These are problems and very serious ones. I am trying to figure out if this can be handled in Resolve, but I doubt.