URSA Mini 4.6k The Original™ Magenta Cast Issue

See my post from 4/1 here: https://forum.blackmagicdesign.com/viewtopic.php?f=2&t=46112&p=268490#p268490

I provided DNGs with and without the lens on. The sigma is definitely adding some magenta for some reason but without it, color looks good.

Hey guys, I have taken the lens off the camera and pointed it at a white wall and recorded in 4k RAW.

It does not look magenta.

Here is the link: https://onedrive.live.com/redir?resid=6 ... file%2cdng

Sorry if I missed something but I couldn't to through the whole topic. If you guys want me to try another test let me know.

Regards,
Cristian

Adam, can you post the CinemaDNG Raw ?

Thanks.... This is good news.

Sure thing. Here you go:
https://www.dropbox.com/s/p5w2cj670ldec ... 4.dng?dl=0

ADAM ROBERTS 1: gain balance only
ADAM ROBERTS 2: offset balance only
ADAM ROBERTS 3: gain balance + pushing lift luma down, as in a average grade
Rec 709 gamma and color, ISO 800, 5600 Kelvin, tint 0

Possible causes:
1) Could be the image circle of some lenses does not cover the sensor and the falloff picks up magenta unevenly as per the overall sensor bias. Very improbable as these are known variables.
2) Sensor color shading, as mentioned earlier with telecentric lens design and cross talk between photosites at sensor edge caused by less collimated light. Light leaves a lens and enters RGB micro lenses arranged in a Bayer pattern on a sensor. Sensor color shading happens when the camera assigns the wrong value to a photosite or group of photosites when debayering because the light is entering the microlenses in a scattered way. In other words the sensor is confused as the light is less direct around the edges of frame.
Also not very likely as there are white papers on this going back to early stages of digital sensor design and this is a know issue.
3) In manufacture something is going wrong possibly with the way the sensor is seated ie. It's at a slight angle or some other variable that is offsetting the angle of light from the sensor at the angle it was designed to accept light from a lens. This would cause falloff at the micro lens level. This would explain the differences we see from camera to camera and why this issue wasn't addressed in the design stage.

Two unhappy conclusions... This won't be fixable in post or with firmware.
Neither are capable of solving for a variable that changes with lenses, angle of light, and individuals sensors... Just too complex.

Hi! Here is a DNG direct from my camera UM46K EF.
http://uhlinmedia.se/DNG/URSA_Mini_1_20 ... 000020.dng

Andrew Deme, why on page 12 your screen shot shows +12 under tint in DaVinci Resolve ?

It appears in DaVinci Resolve if you leave the "white balance" setting to "as shot" then Resolve makes default of 12.85 under tint. This is magenta.

Here you go. Shot this morning just for you.

Resolve_RAW_adjustment.jpg

White surface, no lens. Yes there is fall off in the corners but there is not colour shift in the corners. Even my A7s has fall off in the corners.

Download the DNG here:
https://www.dropbox.com/s/2xdymzoh33rm1 ... 9.dng?dl=0

Thanks for the test Adam! Yours is definitely better! I think though that they could do a better calibration and get rid of the dark corners and tint.

So we can assume then it is not a global issue and it will be fixed by BMD in 1-2 months hopefully!
I have postponed my order for now until this is fixed. When I got my BMCC I had to send two units back due to the interference problem. The third one was great, but had an awful noisy fan, luckily they sent me a replacement fan.

I really wanted to believe that their QC would improve with their 5th camera...

Hi! Here is a DNG direct from my camera UM46K EF.
http://uhlinmedia.se/DNG/URSA_Mini_1_20 ... 000020.dng

With white balance set to temp 7150 and tint -18 in Lightroom it looks good. There is shading on the edges but no color cast.

Regarding Adam's last PNG test... I downloaded Adam's png and it looked great in resolve with no coloured corners. It was easy to balance of course.

Having said that because there was no lens applied I don't feel this is a great "one catch all test" to determine if a particular camera has issues with inconsistent color casting because as we've learned (and BMD has confirmed this) the corner casting issue has more to do with the angle of light hitting the sensor which varies in degrees from lens to lens.

I feel like a white paper is going to provide a more even amount of fill for each photo site minimizing what will happen when you actually stick a lens on. ie. creating a CRA of 0 which is not what happens in "real world" environment when you use lenses.

What do you guys think about this?

Regarding Adam's last PNG test... I downloaded Adam's png and it looked great in resolve with no coloured corners. It was easy to balance of course.

Having said that because there was no lens applied I don't feel this is a great "one catch all test" to determine if a particular camera has issues with inconsistent color casting because as we've learned (and BMD has confirmed this) the corner casting issue has more to do with the angle of light hitting the sensor which varies in degrees from lens to lens.

I feel like a white paper is going to provide a more even amount of fill for each photo site minimizing what will happen when you actually stick a lens on. ie. creating a CRA of 0 which is not what happens in "real world" environment when you use lenses.

What do you guys think about this?

Have you viewed the other DNG I posted? Grey wall with XRite chart. That was shot with a Sigma 50mm f1.4

Hi! Here is a DNG direct from my camera UM46K EF.
http://uhlinmedia.se/DNG/URSA_Mini_1_20 ... 000020.dng

With white balance set to temp 7150 and tint -18 in Lightroom it looks good. There is shading on the edges but no color cast.

Thanks Scott for testing my DNG, I thing thats good news for me, right?

Andrew Deme, why on page 12 your screen shot shows +12 under tint in DaVinci Resolve ?

It appears in DaVinci Resolve if you leave the "white balance" setting to "as shot" then Resolve makes default of 12.85 under tint. This is magenta.

This is what I've been trying to say, people see the AS Shot Default Rec 709 debayer and freak out. For whatever reason Resolve always interprets the data with a heavy magenta bias on the tint slider. Also I've noticed the AS SHOT is never a what you set in camera. Its always some derivative of that. If you set to 4800 its 4786 with a +13 magenta bias. Matter fact hold on. Here are some uncorrected and corrected dngs first set based on my eye was around 4000K. Even though the DNG shows 3986 with a +11.20 magenta bias. Bottom one I didnt show the waveform but it was set to 5600K and the AS shot showed 5600K but it also added a 12.5 magenta bias. There is already magenta naturally in that spectrum so why is resolve adding it?

uncorrected1.PNG (327.33 KiB) Viewed 11681 times

corrected1.PNG (222.64 KiB) Viewed 11681 times

5600K 4.6K.PNG (20.18 KiB) Viewed 11681 times

Hi! Here is a DNG direct from my camera UM46K EF.
http://uhlinmedia.se/DNG/URSA_Mini_1_20 ... 000020.dng

With white balance set to temp 7150 and tint -18 in Lightroom it looks good. There is shading on the edges but no color cast.

Thanks Scott for testing my DNG, I thing thats good news for me, right?

Tony,

this is your image neutralized in Photoshop... And after Gamma taken bit down.

The left side image has neutral roll-offs but the right side is def. green. Center is gray.

sensor.jpg (447.57 KiB) Viewed 11650 times

Andrew Deme, why on page 12 your screen shot shows +12 under tint in DaVinci Resolve ?

It appears in DaVinci Resolve if you leave the "white balance" setting to "as shot" then Resolve makes default of 12.85 under tint. This is magenta.

Had me curious as well and is pretty much why I posted it.

A week or so there were a batch of cameras that people had received and posted Adam's 'White Paper' test and they were a breeze to fix with a slight change in gain.

Then I started seeing a 3D cast and I gave up looking as they were a pain to colour correct.

I really have no idea what is going on and when I saw this DNG I just pulled it into Resolve just like anyone else would....showing what the starting baseline is that a user would have to correct from.

You make a good point !

Have you viewed the other DNG I posted? Grey wall with XRite chart. That was shot with a Sigma 50mm f1.4

Hi, I Just took a look at that. WOW ! That looks amazing. If someone could guarantee me that from a mini 4.6k with any lens, I would buy one tomorrow and never complain at all. Adam you got a lucky little camera there.

Can you try with 15mm, 24/25mm, 28mm?

Regarding Adam's last PNG test... I downloaded Adam's png and it looked great in resolve with no coloured corners. It was easy to balance of course.

Having said that because there was no lens applied I don't feel this is a great "one catch all test" to determine if a particular camera has issues with inconsistent color casting because as we've learned (and BMD has confirmed this) the corner casting issue has more to do with the angle of light hitting the sensor which varies in degrees from lens to lens.

I feel like a white paper is going to provide a more even amount of fill for each photo site minimizing what will happen when you actually stick a lens on. ie. creating a CRA of 0 which is not what happens in "real world" environment when you use lenses.

What do you guys think about this?

Have you viewed the other DNG I posted? Grey wall with XRite chart. That was shot with a Sigma 50mm f1.4

I concur! Adam's DNG of a grey wall with XRite chart looks absolutely great and normal! I pushed the saturation and contrast hard to TRY to make Mr. Magenta rear it's head and all the colors behaved nicely!

Adams test was scientific and so was mine. And it reveals the non-uniform behavior of the camera.

To be "scientific" these tests must have a control. To know for certain whether the results you're getting from pushing the image to unrealistic extremes mean anything, there's got to be a known standard of comparison. Otherwise, no actual conclusion can be drawn from any of this.

In this case, it would require a known sensor that everyone agrees is standard that is then put through the same test as the Ursa Mini 4.6k sensor and given an identical grade in Resolve.

Do you have access to a camera sensor that you feel could function as a control to test against your 4.6K?

Adams test was scientific and so was mine. And it reveals the non-uniform behavior of the camera.

To be "scientific" these tests must have a control. To know for certain whether the results you're getting from pushing the image to unrealistic extremes mean anything, there's got to be a known standard of comparison. Otherwise, no actual conclusion can be drawn from any of this.

In this case, it would require a known sensor that everyone agrees is standard that is then put through the same test as the Ursa Mini 4.6k sensor and given an identical grade in Resolve.

Do you have access to a camera sensor that you feel could function as a control to test against your 4.6K?

Totally correct.

My test is far from scientific. I was simply trying to create a test that was repeatable so that others could have a simple test to follow that they could then compare to others. Shooting a white wall has to many variables to be repeatable.

This test gives people something simple to do to identify a sensor that might have an issue. It's not perfect but a reasonable starting point. They can then take it up with BM Support.

There were many people panicking and posting images that had a cast but not a shift or had done the test without the light being flat. We don't need mass panic. That helps no one. We need a simple test that'll help people check for the issue. They can then do "scientific" tests with BM Support if they have an issue.

Adams test was scientific and so was mine. And it reveals the non-uniform behavior of the camera.

To be "scientific" these tests must have a control. To know for certain whether the results you're getting from pushing the image to unrealistic extremes mean anything, there's got to be a known standard of comparison. Otherwise, no actual conclusion can be drawn from any of this.

In this case, it would require a known sensor that everyone agrees is standard that is then put through the same test as the Ursa Mini 4.6k sensor and given an identical grade in Resolve.

Do you have access to a camera sensor that you feel could function as a control to test against your 4.6K?

Totally correct.

My test is far from scientific. I was simply trying to create a test that was repeatable so that others could have a simple test to follow that they could then compare to others. Shooting a white wall has to many variables to be repeatable.

This test gives people something simple to do to identify a sensor that might have an issue. It's not perfect but a reasonable starting point. They can then take it up with BM Support.

There were many people panicking and posting images that had a cast but not a shift or had done the test without the light being flat. We don't need mass panic. That helps no one. We need a simple test that'll help people check for the issue. They can then do "scientific" tests with BM Support if they have an issue.

I think your test is a great.
I only meant to say we should extend it by using it on other cameras (an Alexa, for example) as a point of comparison. That might temper the panic.

I think your test is a great.
I only meant to say we should extend it by using it on other cameras (an Alexa, for example) as a point of comparison. That might temper the panic.

Yeah. I did post an example shot on the A7s alongside my first test. It too shows fall off toward the edges.

Worth doing the test on other camera.

Adams test was scientific and so was mine. And it reveals the non-uniform behavior of the camera.

To be "scientific" these tests must have a control. To know for certain whether the results you're getting from pushing the image to unrealistic extremes mean anything, there's got to be a known standard of comparison. Otherwise, no actual conclusion can be drawn from any of this.

In this case, it would require a known sensor that everyone agrees is standard that is then put through the same test as the Ursa Mini 4.6k sensor and given an identical grade in Resolve.

Do you have access to a camera sensor that you feel could function as a control to test against your 4.6K?

The only control or standard that's needed in this case is the target ideal of uniform color and brightness. While virtually no sensor will achieve 100% absolute perfection, it is not unreasonable to expect an acceptable degree of artifact free normalcy in what is routinely seen in modern professional sensors. Global color casts, tint bias and vignetting are all normal variations that are part of what is reasonable. Within this thread alone there have been examples of both clean and affected sensors, so while it hasn't been 100% scientifically rigorous because the tests have all been conducted by different people, I believe it has come close enough to say something is amiss with certain sensors.

I'll post my cap test to this thread as well Original 4K Ursa red channel bias and the 4.6K blue channel bias. I saturated both to 100 and upped the color boost to 100. No anomalies both sensors have uniformity.

4.6KCapTest1.jpg (890.86 KiB) Viewed 11406 times

UrsaCapTest.jpg (986.33 KiB) Viewed 11406 times

To be "scientific" these tests must have a control. To know for certain whether the results you're getting from pushing the image to unrealistic extremes mean anything, there's got to be a known standard of comparison. Otherwise, no actual conclusion can be drawn from any of this.

Science is theory and observation. A study. It has to start somewhere. And before any standards can be derived from science there is a lot of trial and error. If there were a standard to measure by and the issue had already been quantified, there would be little need for science.

Taking pictures with turret caps on and without lenses to reveal possible flaws of a camera , is, when employed on a Camera using Lenses to capture real-life images, scientific by nature

In this case, it would require a known sensor that everyone agrees is standard that is then put through the same test as the Ursa Mini 4.6k sensor and given an identical grade in Resolve.

What if the sensor is working and performing just fine in other camera bodies ? But the camera body in question has an off-center cooling unit blowing cold air from the right side of the sensor thus generating non-uniform temperatures of the sensor. In this case the sensor would be fine but its cooling would induce its flaws. What if the the color filters on the sensor were unevenly applied by one manufacturer and evenly by the other ? In this case the sensor would perform perfectly well in the latter condition but exhibit major flaws in the other.

We don't even know if the sensor is the problem. How could we then agree on any one sensor ?
The sensor is an important but very small part in all that could skew the quality of an image.

Given the nature that Black Magic would not intentionally ship nor kidney-test each camera prior to shipping, they only way for them to learn that something went wrong in production is when the camera operators report back...

In this case of the UM46. When first getting one's camera and paying a premium for it, no one would expect it to be faulty. And thus (me at least) not begin by conducting weird and extreme scientific tests. One would unpack his camera (like I did) be happy as a baby and start shooting some test footage and look at it...

If the person in question has had past experience in buying cameras he would know what to expect from one.
If that same person had been shooting 1000 of images for the past decade or more - it would be safe to assume that he/she would have some idea of how images look captured from a wide variety of scenes.

So when he eagerly and with great passion takes his first look at the test photos he would not sit there with a grim look on his face expecting to see artifacts, he would naturally assume that the camera which he has been waiting for the past 12 months (like the cameras he bought in the past did) is going to blow him away and thus justify his 9K investment !

If most the test photos reveal artifacts that the user has never before been confronted with by any camera not even a $100 cheap-cam on an iPhone, naturally he is going to think that this is fixable in his favorite software or that he did something wrong. So he double checks. Triple checks. Quadruple checks. If all his experience neutralizing and grading images the past decades won't allow him to correct for the artifacts then and only then he adventures on to the scientific stage of shooting white backgrounds and other extremities like naked sensor shots or with turret/lens caps on. Who in their right mind would begin shooting images that aren't images because the lens is detached, of evenly lit white backgrounds - as the first shot of a brand new camera which they just bought ?

I did exactly what I described above, shot images in my apartment of my daughter (we happen to have white painting on the walls) - on our balcony which has white stone walls, in our kitchen with lots of natural gray by nature caused by stainless steel pottery.

Most those images revealed strong magenta shifts on the right side of the camera and towards the left side, cyan shifts to go to strong magenta on the outer left side (not as strong as the right side). The center was gray. I would have to choose to either balance for edges or center. Only none of the result were uniform.

So then and only then did I venture into to the scientific state of testing the camera.

All those cap and wall tests when boosted to the extreme revealed identical patterns recognizable from the normal images I had shot. And if that is the case it is safe to assume that the artifacts present in my normal images were not caused by cam-operator-error nor by weird lighting. So the tests were merely revealing what I had already seen in the test images. And thus used for comparison only.

Remember - I did not start off by taking weird scientific images and boosting their flaws with the camera I had been waiting for so long and paid 9kEURO for - I was forced into that corner. As matter of fact, if you scroll back in this here thread you will find that I took the stance that folks with the issue should 'just' properly balance their imagery. I was refusing to believe that it could be the camera. I was very wrong and as you can read here, I was sorely corrected

I did not start this topic of Magenta - it was there long before I began writing. Dating back to the days of beta testing even. I am not sitting in a dark room downloading other folks' CinemaDNGs judging from afar - I am judging first-handedly after putting down 9kEURO prior to waiting for 12 months AND kidney-testing my camera.

Shutting up about it, pretending it is not there JUST because it is not convenient, isn't going to make it go away, you know !

Here the DNG shot on studio with controlled light
http://adelica.it/test1.dng

Attached the snap (snap from camera dng, untouched)

Recap of camera settings:

Zeiss CP.2 50mm T2.1 EF mount @ T11
ISO 800 - 180° - 5200K - Film range - Full area sensor

Have you viewed the other DNG I posted? Grey wall with XRite chart. That was shot with a Sigma 50mm f1.4

Hi, I Just took a look at that. WOW ! That looks amazing. If someone could guarantee me that from a mini 4.6k with any lens, I would buy one tomorrow and never complain at all. Adam you got a lucky little camera there.

Can you try with 15mm, 24/25mm, 28mm?

Here are a few lenses shot this morning. Light is not flat. Its entering a window on the left of frame so it falls off from left to right.

Sigma ART 18-35mm f/1.8 at 18mm

UM46_Lens_Test_Sigma18.jpeg (417.4 KiB) Viewed 11339 times

Here are 2 more:

Carl Zeiss ZF.2 85mm f/1.4

UM46_Lens_Test_Zeiss85.jpeg (441.85 KiB) Viewed 11331 times

Helios-44-2 58mm f/2

UM46_Lens_Test_Helios58.jpeg (404.9 KiB) Viewed 11331 times

All of these have been had global colour corrections done to them and as you can see the colour is pretty consistent across all the lenses.

There are many cameras out there without issues. If you have a camera that has issues you need to contact BM Support so that can address it. This also helps them find the fault on their end so it can be sorted out for all future cameras.

Download the DNGs here:
https://www.dropbox.com/s/kxdfmbvc0gh1a ... t.zip?dl=0

Here the DNG shot on studio with controlled light
http://adelica.it/test1.dng

Attached the snap (snap from camera dng, untouched)

Recap of camera settings:

Zeiss CP.2 50mm T2.1 EF mount @ T11
ISO 800 - 180° - 5200K - Film range - Full area sensor

Looked at it. I'd say the shot is under exposed by over a stop. It does show some magenta shift in the corners. Difficult to tell how bad tho as the light has fall off too. Might be worth running the white paper test.

Alessandro_Resolve.jpg (940.1 KiB) Viewed 11323 times

Alessandro_Correction.jpg (719.85 KiB) Viewed 11323 times

Looks like we have a bit of it with ours as well - doing more tests this morning, and contact both Adorama for a possible return and Blackmagic for a possible repair/replacement, but here's our first taste:

Quick update - the cast is there (I'll post test shots later), but in addition, my sensor has a single-pixel wide red band in the right side of the frame. Barely perceptible in many circumstances, but in ProRes LT it becomes more prominent (which is how I first noticed it).

Full image:


Red line (right in the middle of the image):


These two factors combined, I'm just returning the camera and putting in a Scarlet-W order (I get $2,500 off as a Scarlet Dragon owner, and as much as I love Blackmagic for getting me into the RAW video game and prize my BMPCCs, I have had better experience and reliability with RED products).

Kudos to everyone sharing here. This has been a helpful, detailed, and civil discussion.

Thanks. This helped us make a decision and move forward.

I'll post my cap test to this thread as well Original 4K Ursa red channel bias and the 4.6K blue channel bias. I saturated both to 100 and upped the color boost to 100. No anomalies both sensors have uniformity.

4.6KCapTest1.jpg

UrsaCapTest.jpg

Brandon, your capped sensor test looks great and wildly different from what Morten is getting from his camera. This is also backed up by your other real world shots that show even color for the most part and where there is a touch of magenta, it is easily fixed. This to me confirms that some and hopefully most cameras are fine and some are not. This also shows that the capped sensor test is relevant, reliable and repeatable. As soon as I get my replacement camera, I will do the capped test as well as the open sensor test. If both of these show no irregularities, I would bet anything that real world shots with a lens will be fine.

So this weekend I'm doing more scientific tests, but I just wanted to see how it looks in a more real world scenario that would reveal the issue using the same workflow I have been using for years with any number of cameras successfully. I pointed my UM46 and a pocket camera at a wall that was lit from the top by a household overhead light and from the side by a window. The pocket and the UM46 should look close as I shot both simultaneously with 50mm and 28mm Contax Zeiss lenses to compensate for sensor size differences. Both shot at 3400k @800 ISO in raw 3:1. DNGs at the bottom. I'm not happy...

Note that I only ballparked the grade here and wasn't trying to get them to match exactly; remember my point was do just do things the way I normally would and see if there's a practical issue ahead of further testing. They're both a bit green, but the point is the uniformity and the differences. Here's the pocket:

Pocket

Pocket_opt.jpg (675.32 KiB) Viewed 11082 times

And the UM46:

UM46_opt.jpg (686.92 KiB) Viewed 11082 times

And the DNGs:

https://www.dropbox.com/sh/vctk0gnmkx6o ... nwaXa?dl=0

I pointed my UM46 and a pocket camera at a wall that was lit from the top by a household overhead light and from the side by a window.

Oh wow. We should send you a modern light switch from German manufacturers Busch & Jaeger or Berker.

I pointed my UM46 and a pocket camera at a wall that was lit from the top by a household overhead light and from the side by a window.

Oh wow. We should send you a modern light switch from German manufacturers Busch & Jaeger or Berker.

Hehe, yeah. Pre-war Brooklyn building with a super that does the bare minimum. At least the 4.6k is resolving a lot more detail! The upside-down "On" looks super sharp and terrifying...

So this weekend I'm doing more scientific tests, but I just wanted to see how it looks in a more real world scenario that would reveal the issue using the same workflow I have been using for years with any number of cameras successfully. I pointed my UM46 and a pocket camera at a wall that was lit from the top by a household overhead light and from the side by a window. The pocket and the UM46 should look close as I shot both simultaneously with 50mm and 28mm Contax Zeiss lenses to compensate for sensor size differences. Both shot at 3400k @800 ISO in raw 3:1. DNGs at the bottom. I'm not happy...

Note that I only ballparked the grade here and wasn't trying to get them to match exactly; remember my point was do just do things the way I normally would and see if there's a practical issue ahead of further testing. They're both a bit green, but the point is the uniformity and the differences. Here's the pocket:

Pocket_opt.jpg

And the UM46:

UM46_opt.jpg

And the DNGs:

https://www.dropbox.com/sh/vctk0gnmkx6o ... nwaXa?dl=0

...the pocket looks very even and nice...the Mini ist very uneven. green in the middle, magenta in the corners....where is some information from BM??

...the pocket looks very even and nice...the Mini ist very uneven. green in the middle, magenta in the corners....where is some information from BM??

They already answered a couple of days ago. The sensor has no issues. Problem are lenses. All of them.

...the pocket looks very even and nice...the Mini ist very uneven. green in the middle, magenta in the corners....where is some information from BM??

They already answered a couple of days ago. The sensor has no issues. Problem are lenses. All of them.

But only when on the 4.6K sensor. Logical. We have not even heard BMD comment on their 4.6K in terms of the micro lenses over the sensor (apologies if I missed their comment).

Are there micro lenses over the sensor photosites?
Are the micro lenses all in the identical orientation so that the sensor will look it's best only when using telecentric lenses?
Are the micro lens designs different than those for the smaller sensor cameras (BMCC, BMPCC, BMMCC)?

Simple answers to a few questions but perhaps BMD considers that information to be proprietary.

Adam the color chart needs to fill the frame or shift to the corner, as we are testing the sensor and not using it to balance the image. We can then look at the numbers assigned to each patch and see if there is a color shift at the corners using the eyedropper tool. Also the dsc cam align or the one shot would be great here as we would be able to see where the patches lined up on the vectorscope.

I pointed my UM46 and a pocket camera at a wall that was lit from the top by a household overhead light and from the side by a window.

Oh wow. We should send you a modern light switch from German manufacturers Busch & Jaeger or Berker.

Hehe, yeah. Pre-war Brooklyn building with a super that does the bare minimum. At least the 4.6k is resolving a lot more detail! The upside-down "On" looks super sharp and terrifying...

Maybe it's really saying "No"? As is Nooo.. Is that magenta?

Since I'm a novice it's easy for me to get sucked in by the thought that magenta in the corners on shots without a lens is not (easily) correctable.

But as some have pointed out, it's very much fixable in RAW. Of all the DNG:s I've downloaded from this thread only one had a very slight problem.

The key is that the global (fixable) color cast paired with a slight global vignetting causes different "slopes" of the vignetting on the three color channels. Especially the blue channel on the examples I've seen. This is probably common knowledge, but it means that you can't just play around with channel levels because the vignetting won't align. Unless you use the tint slider.

Then again maybe I'm just not that sensitive to the issue.

Untitled_1.1.4.jpg (310.86 KiB) Viewed 10948 times

Oh, and for some reason the cat is mesmerized by this image.

IMG_6366.jpg (121.93 KiB) Viewed 10934 times

Adam the color chart needs to fill the frame or shift to the corner, as we are testing the sensor and not using it to balance the image. We can then look at the numbers assigned to each patch and see if there is a color shift at the corners using the eyedropper tool. Also the dsc cam align or the one shot would be great here as we would be able to see where the patches lined up on the vectorscope.

These were not intended as a sensor test. They were to show Morten Carlsen that a grey surface shot with my camera does not turn into a "piñata-colorer gradient". See this post: viewtopic.php?f=2&t=46112&p=271310#p271321

My sensor does not have colour shift issues. As you can see from various other test shots I've posted. I've simply been sharing mine so other can see that there are cameras out there that are fine. I shot the same grey wall and chart with different lenses to show that there is not magenta issue from different lenses.

Double post...

Maybe it's really saying "No"? As is Nooo.. Is that magenta?

Ha!

Since I'm a novice it's easy for me to get sucked in by the thought that magenta in the corners on shots without a lens is not (easily) correctable.

But as some have pointed out, it's very much fixable in RAW. Of all the DNG:s I've downloaded from this thread only one had a very slight problem.

The key is that the global (fixable) color cast paired with a slight global vignetting causes different "slopes" of the vignetting on the three color channels. Especially the blue channel on the examples I've seen. This is probably common knowledge, but it means that you can't just play around with channel levels because the vignetting won't align. Unless you use the tint slider.

Then again maybe I'm just not that sensitive to the issue.

Untitled_1.1.4.jpg

Oh, and for some reason the cat is mesmerized by this image.

IMG_6366.jpg

I don't know about you, but it's a flaw I've never seen actually visible in footage from any other camera. Do you want the camera that requires "fixing" in a way that no others do? I just want a sensor that performs to spec and doesn't have flaws that I'd have to, or worse a client would have to, always grade out. Like I said, the pocket was right next to it under the same circumstances and the grade was the same and the results were different as you can see.


PS - Did you pump up the saturation? Yours looks considerably flatter than my grade.

PPS - Your cat is hilarious.

What F stop is everyone using for their tests? I've noticed that my sensor is largely clean as long as i stay faster than F.8. If i stop down past that then the image starts to shift a bit.

My current work-around is to put the camera into windowed uhd, gives the same sensor coverage as the 4k ursa (roughly) and helps avoid the issues when they come along.

My test was at f/2.8.

Got bored and decided to do another test and had the bright idea to throw on a Hoya IR Cut just to see if it made a difference. I think it did. It would explain why on my first day of shooting I didn't see anything odd (I had IR Cut on the entire time that day). Again, not a scientific test (will do that Sunday), just a quick real world test to see differences between two images.

Raw 3:1, 800 ISO, 3400k, f/1.7 on a Contax Zeiss 50mm.

No IR Cut:
https://www.dropbox.com/s/m30vw6mbp319i ... t.dng?dl=0

IR Cut:
https://www.dropbox.com/s/epcfoy5rs73ov ... t.dng?dl=0

**Sorry if someone else already tried this; this thread is 13 pages long and I must have missed it...

At the moment I see a lot of people putting their opinions forward in regard to customers questions around the quality and performance of their camera, especially in regard to the 'Magenta' issue.

Right now we are seeing a lot of 'My Camera is OK' and so is yours, versus 'My Camera Isn't OK'...and we all need to analyze it to death.

Then there are a long line of opinions in relation to how the various forms of 'Magenta' happen to appear, be it an even or uneven cast, lens created or enhanced or simply it doesn't matter as all other cameras show some sort of anomaly anyway.

I for one will happily admit to being keen to see BMD succeed, with the primary reason being that I am an Aussie and secondly I kinda like the underdog.

Having said that, I have no idea why BMD seem to want to cower in the corner and hide...for goodness sake, as employees start standing up for yourselves and step up and be counted and take the issue in your stride....speak up, the forums will love you for it.

The last thing any Aussie (Kiwi's included) wants to see us doing is offshoring courage, conviction and mateship !

And as for the Magenta issue...faark.....how hard is it to tweak the way Tint works in Resolve and give customers the ability to remap the cast with a slider or two....geez, even if you create a new slider...who cares. Worst case, give em the ability to map their sensor against a color chart.

Anyone that has setup a particular brand of Flightsim with a joystick, pedals, throttle, buttons, pedals and 'Top Hat' know exactly what I mean. It's all hand carved and tuned to support the particular user, their equipment and circumstances....kids can do it, surely the adults can as well.

Seems most people here don't have perfect color vision, so go the 80/20 rule and get on with it as the 4.6K obviously has an image to die for...Brawley proves it again and again and again.

I don't know about you, but it's a flaw I've never seen actually visible in footage from any other camera. Do you want the camera that requires "fixing" in a way that no others do? I just want a sensor that performs to spec and doesn't have flaws that I'd have to, or worse a client would have to, always grade out. Like I said, the pocket was right next to it under the same circumstances and the grade was the same and the results were different as you can see.


PS - Did you pump up the saturation? Yours looks considerably flatter than my grade.

PPS - Your cat is hilarious.

True, it shouldn't need that kind of fixing. But since it is fixable, maybe it's just the metadata that is incorrect.

I don't think I added any saturation to that particular version, but it's not horrible even when I do.

I don't know about you, but it's a flaw I've never seen actually visible in footage from any other camera. Do you want the camera that requires "fixing" in a way that no others do? I just want a sensor that performs to spec and doesn't have flaws that I'd have to, or worse a client would have to, always grade out. Like I said, the pocket was right next to it under the same circumstances and the grade was the same and the results were different as you can see.


PS - Did you pump up the saturation? Yours looks considerably flatter than my grade.

PPS - Your cat is hilarious.

True, it shouldn't need that kind of fixing. But since it is fixable, maybe it's just the metadata that is incorrect.

I don't think I added any saturation to that particular version, but it's not horrible even when I do.

If Fahnon's camera suffers from irregular magenta corners (which hasn't been fully shown yet) like some cameras have been getting, it is NOT fixable in post. If you can bear to read though all 13 pages of this thread, you will see that there are cameras that exhibit NO issues under any circumstances and cameras that show irregular magenta corners under just about all conditions with or without a lens.