Pocket Cinema Camera "Blooming Sensor"

[Terry Frechette]

You knew this was coming. Civil discussion please.

This is a technical forum, so please keep the discussion technical.

I'm concerned about this because it isn't something you can fix in post like black sunspots and firmware likely can't fix it either.

I experienced this with my Fuji X10. Fuji had to do a recall and a hardware modification to fix it.

http://www.dpreview.com/news/2012/04/26 ... orb-update

It was termed the 'white orb' issue.

This is exactly the same. Perhaps if a fix is too much to ask for, Blackmagic can suggest ways shooters can avoid it?

@hamid

Please stop writing off topic crap.

I fear that because of this effect you can't capture nice flares and halos at all, not only when it is overexposed.
Just blank 'orbs'...

@hamid

Please stop writing off topic crap.

mate dont read it. move on.

I too am waiting on my pre order but now I am really getting nervous.
can we have Black magic design comment on this issue?
I spent over $1500 on lenses, batteries, sd cards and a cage waiting for this little guy and now I am freaking out ...

Black magic?

I have no experience in this field, but I assumed that the BMPCC has a CMOS sensor. Although I couldn't find any information regarding the sensor type in the camera's specs, you can tell that by the rolling-shutter. According to Wikipedia (http://en.wikipedia.org/wiki/Active_pix ... red_to_CCD), one of the advantages of a CMOS sensor over a CCD sensor, besides the cost, is that the last one dose not suffers from the blooming effect.

So what could have caused this? Maybe it is just a firmware issue…?

As for the car headlights in the above sample, BMCC was probably "Raw" graded footage, with exposure closer to highlights, headlights are Not directly aimed at camera/lens. I saw a similar demo between graded raw in BMCC and a DSLR, which did same "glob" looking blown put highlights, especially on "out of focus" direct lights, which generate "several halos" as a result of "circles of confusion" of out of focus light rays striking either film or a sensor. When a focused light is shown directly at the lens, effect will be more noticeable, and clipping will eliminate halos and turn them into a white circle or glob. More blown out/clipped, worse the effect. Stop lights, car headlights, etc are "focused" bright light sources, and the most difficult to balance with shadow detail ad get it looking right.

I think doing the same test with Christmas glass ball and a direct reflected light, taking exposure from exposing for highlight then opening up to point of blowing out the highlight, in both Film and Video mode to compare results. This should see if there is a "real" bloom issue with sensor.

Anyone with the Pocket Camera, or any camera, could run this test. Would be interesting to see comparison of this test with a DSLR, and BMCC set up together with the Pocket. With all cameras in same controlled situation, shooting ProRes to fair comparison, we can see if there is a real issue, or just a difference in sensor handling clipping.

I have no experience in this field, but I assumed that the BMPCC has a CMOS sensor. Although I couldn't find any information regarding the sensor type in the camera's specs, you can tell that by the rolling-shutter. According to Wikipedia (http://en.wikipedia.org/wiki/Active_pix ... red_to_CCD), one of the advantages of a CMOS sensor over a CCD sensor, besides the cost, is that the last one dose not suffers from the blooming effect.

So what could have caused this? Maybe it is just a firmware issue…?

As I wrote in the first post of this thread:

I've read some articles about the Fujifilm X10 and found out that they had the same problem with their camera, and to get rid of it they had to modify the sensor, becouse the firmware update didn't worked

But I hope that BMD will figure something out without changing the sensor... If not, then they will have a problem... Fuji had to replace any camera with older sensor to the new one with the new sensor if the owner of the camera wasn't happy with the image quality of the old one... If the warranty was not expired.

Perhaps if a fix is too much to ask for, Blackmagic can suggest ways shooters can avoid it?

I think if they don't address it and fix it, that would make me lose a lot of trust in BMD. Fuji fixed it, and so should BMD. If it's just a firmware fix, all the better.

I too am waiting on my pre order but now I am really getting nervous.
can we have Black magic design comment on this issue?

ld make me lose a lot of trust in BMD. Fuji fixed it, and so should BMD. If it's just a firmware fix, all the better.

My concern too... We don't know if BM will actually acknowledge that as an issue.
Now is the right moment to address this, before the cameras actually ship

I'm sorry but I think that you guys are a bit overreacting and even worse over speculating.


Its the same story as always. Based on tree or four random clips people here start going nuts, just wait till you actually receive a camera and have a Camera yourself before complaining about an "Issue" you haven't even experienced yourself.

I don't have the camera but it might be that you could come around this by underexposing the image by e.g. one stop on purpose. (Just with most other cameras the displayed ISO needn't necessarily match the displayed ISO, this might lead to poor exposure by users. A display set to dark or to bright would also do so, especially if zebra is not used while setting the exposure.)
Some sensors are better at retaining details in overexposed Areas (e.G. the ones in Hasselblad Cameras) whilst others are better at recovering Shadows (e.G. Sony sensors as the one in the D800.). On paper both sensors may produce the same dynamic range, however it is very differently distributed.
As a reaction to this you will need to adapt your shooting Style to suit the Camera used.
Beside of that all of the tree clips seemed overexposed to me. (Just look at how little sky detail there is left in all of these shoots.) The Second Clip was clearly very poorly shoot and exposed. Please also remember that Flat ProRes isn't holding the same amount of Dynamic range as Cinema DNG, so it might generally be a smart idea to use a graduated ND Filter in shoots with very bright Sky.
What also concerns me is that it seems that nobody has actually tried to come around this Problem by using different filters, as e.g. Skylight (UV), IR Cut , Polarization etc.
I think that its not to unlikely to dramatically reduce such an issue when using an appropriate filter.

Please also remember that we currently haven't seen any DNG footage till now, I think that it is very well possible that this Issue is nonexistent in DNG.

Finally this discussion reminds me a lot of the discussions that have occurred around the Black Dot Clipping Issues, I think the way this was treated by a lot of People wasn't appropriate in relation to the extent of the Problem.

F.K.,

I have pre-ordered on day one and don't like to have to cancel it because of this.
I'll be happy to be wrong, but there are more than 15 BMPCC clips - all from different users - and fast all exhibit that effect. Did you saw the car lights? It is not normal for any footage to clip that way like erase tool in PS.

Better we share this find with BM as soon as possible, so if it is a sensor issue they address it before everyone get their cameras.

We know issue and I hope after this topic will BMD give an opinion, next week. Sleep well

F.K.,

I have pre-ordered on day one and don't like to have to cancel it because of this.
I'll be happy to be wrong, but there are more than 15 BMPCC clips - all from different users - and fast all exhibit that effect. Did you saw the car lights? It is not normal for any footage to clip that way like erase tool in PS.

Better we share this find with BM as soon as possible, so if it is a sensor issue they address it before everyone get their cameras.

Just to Illustrate what I meant, I attached the Clip I found clipping the worst of all the Clips in this series.
(The Scopes look pretty much the same on all of the Clips exposure wise, just with a little less Clipping.)

Left you can see that I have set an offset so that you can better see the Clipping in the Scopes. Other then that slight offset the Footage wasn't touched.

Now finally coming to the scopes you can see that the footage is clearly exposed way to bright to take full advantage of the Dynamic you get form capturing Prores in the Flat mode. The Shadows start at around 35/35/35 RGB. (Remember that I had -5 Offset)
The Conclusion is easy, on a Scene where the Highlights were Critical the footage retained all of the Shadow details, with a lot of headroom before clipping.
The footage was overexposed by quite a bit so that the highlights where clipping. If it was me recording it I would rather have underexposed the Footage by a fair bit.
Underexposing it would have some clipping in the Shadows as worst case Scenario.
Instead of loosing all information in the Highlights I would have lost a little detail in the far less noticeable shadows.

(I would be Thankful for more unedited Prores footage from different Scenes with this blooming "Issue" in it just to make sure that it is always the same way.)

EDIT: Please watch the Image in Full screen otherwise you won't be able to see what I mean. (In case you wonder why the preview on the right looks so bad, I have compressed the JPG quite a bit to avoid unnecessary traffic.)

Just as a small addition. I took a well exposed PMW-F3 S-Log File I found of the Internet. (Just for demonstration purposes, the Image used was not shoot by me.)

You can clearly see how different the RGB-Parade is looking, no clipping in both directions and full use of the colorspace. The right half of it is an the right half of the preview that has been graded, the other half is the ungrades SLOG part. (The Colors might be Slightly off if your Browser can't read the ICC Profile provided in the Image.)

F.K.,

Yes, I agree with you for that example. There is only normal clipping due over-expososure. But that is not what we're looking at here.

The 'orb' effect occurs when a direct light (like flash or very bright reflection) appear as large circles.
Normally you'll get a halo like the bmcc example on the image bellow - not the harsh clipping on the right.

I don't have scientific explanation, but it seems to me that BMPCC sensor can't reproduce the smooth transition between the clipped centre and the aureole around it – it all appears the same!

The example bellow perfectly shows this. I have some hope to the firmware related, as some say the sensors of BMCC and Pocket were very similar if not the same. But still you can find this almost on all footage has been uploaded.

Here is a manipulated image - practically on the Pocket the same scene shot with the vmcc will appear like this:

Screen Shot 2013-08-25 at 4.13.25 PM.png (258.41 KiB) Viewed 24916 times

That's one bug eyed car. How on earth did Blackmagic's test team miss this?

How fortunate that today's cars have small lights.

Can you imagine how big spot could come from these?!?

That's one bug eyed car. How on earth did Blackmagic's test team miss this?

The same way Fujifilm did with the X10, all we can hope now is that BMD will also acknowledge and address the issue.

I tried to recreate the Problem by clipping something out of the BMCC highlights. To me it just seems that the BMPCC prores files clip earlier then the BMCC files, so again for that reason simply exposing the shot lower might do the trick. I am currently also downloading the Prores Files from Phillip Bloom, unfortunately it is still downloading. (Even if this isn't eliminating the Problem it might at least reduce it by a fair amount.)

It might as well be a mistake in the firmware or in the internal signal processing. I think that it is to early to talk about faulty sensors (and all the Bad memories related to those) again.

Ok, two more now from the Philip Bloom clip.

As with the other Clips there was a lot of headroom in the shadows, you can see that this time there is a bit more* (even if not very much) resolution in the highlights, the bit near her nose probably exceeded some kind of internal brightness limit.) I don't know how an to what level the footage was graded so it is a bit difficult to get conclusions out of it but generally the very highlights seem to be recored poorly while the shadows have a lot of data in the.)

I will unfortunately not be able to test this as I have no BMPCC but I will stick with underexpose your shoot and it might fix your problem, as you are pulling the highlights away form the area that seems troubling to the camera.

*I'm not 100% sure about that as it might aswell be caused by the film effect (& grain?) that was used.

I know this might sound odd but might the camera have some internal offset that is boosting the exposure and therefor causing the Files to clip so oddly?

And File No. 2

How fortunate that today's cars have small lights.

Can you imagine how big spot could come from these?!?

< Actors face when shot with blooming sensor

Ok, I just tried out what adding grain did to the Files. I think that the Grain is what is giving the Files form Phillip Bloom the Impression of better highlight resolution, however I doesn't seem to be significantly better in Reality (and from what you see in the Image).

How fortunate that today's cars have small lights.

Can you imagine how big spot could come from these?!?

< Actors face when shot with blooming sensor

We are joking but i have ordered one. Only hope is that BMD and will solve it.

Also hope not like that:

You could have removed the "compressed RAW" in favor of the "Blooming"

So strange as on the original BMCC it was better to slightly over expose the image for noise and DR. Does trying to salvage your footage from blobby highlights sacrifice these other things too? I love how some people are trying to defend this issue or calling it a non issue. Its a huge issue, it ruins footage and this is why we are freaking out. For BMD not to comment on this that they are aware and its nothing or that they are fixing it is a big mistake.

You could discount this camera also by 50% with this issue and I still would not buy it. I was going to take this with a small fuji XE1 up to the Black tusk this weekend to get some great footage but now im not so sure...its too much of a hike to waste on blooby footage.

Its a huge issue, it ruins footage and this is why we are freaking out. For BMD not to comment on this that they are aware and its nothing or that they are fixing it is a big mistake.

The first commentary I saw on this was probably late Friday Australia time. It's gained momentum over the weekend on several forums.
It's still not 9am Monday morning Melbourne time as i write.
To start ragging on BMD for not commenting, when the whole issue has cropped up out of business hours, and even if they admit there is a problem, might need to discuss and agree on a strategy to fix it going forward before announcing anything, is the 'big mistake' IMO

Just downloaded Phillips footage and looked into it - I saw those "white blob" anomalies before, but it was material from my FS100, and I think I might have an idea what it is.

It's way to soon to draw my conclusion, since I haven't talked to Phillip yet (he is probably sleeping at this time). I go out on a limb here, but if I'm right, it has nothing to do with the camera per se.

I try to reach Philip tomorrow maybe it's a much easier fix, than we all think.

I agree it is too soon, Im mainly suspecting we hear nothing from them as usual so I may be jumping the gun. I would like to be surprised with an up front approach to this one is all so i can be comfortable with purchasing all my accessories like others so I can be up and running.

Frank I hope you are correct it would be really great if they just had to tweak something to fix this issue really fast.

Perhaps these orbs are our friends from another world and they just want to be noticed?

Frank,

I just looked through the raw clips John Brawley has put online for download as well as from Ravens and the problem is definitely there. Look at any clip John has posted that has deep DOF. Specifically if you look at highlight glints from shiny ( often metallic ) objects. Once you see it, you really can't un-see it. I'm assuming that John is exposing properly for at least the majority of his shots as he is a professional.

Check out some of John's clips that a user graded:



This effect is definitely occurring at 0:23, 0:35, 1:18

I checked the original prores footage, its there even before grading, just more obvious after.

Very disconcerting!

That's one bug eyed car. How on earth did Blackmagic's test team miss this?

i found this clip. shot with Prores Film, BMCC. not BMPCC.

1:12

ss.png (600.29 KiB) Viewed 24668 times

there's same white orbs, so...i think(hope) BMPCC will be ok with RAW too.
i'm feeling a little relieved.

...
however, it's not solved.
just assuming white orb issue occurs only in Prores.
there is an issue. and if firmware update can solve it, i hope BMD do it.
now i'm waiting for RAW update.

Nothing on the original Blackmagic Camera looks the issue that's being discussed here. At this point, people are becoming hysterical...

I just watched the 9th episode of House of Cards which is shot on the Red Epic, and I saw what looks to be the same orbs on the street lights of a couple of scenes. Perhaps its just me and I'm starting to see orbs everywhere, but it sure looked like the same anomaly.

Which may prove Kholi's point.

i found this clip. shot with Prores Film, BMCC. not BMPCC.


there's same white orbs, so...i think(hope) BMPCC will be ok with RAW too.
i'm feeling a little relieved.

Black cabs actually have round headlights like that. Doesn't look like blooming to me at all.

I've been reviewing my footage from last week's test. The only time I am seeing this issue is with blown-out areas. I have seen similar "orbs" from other cameras' blown out areas as has been mentioned here already.

I'm putting up two frame grabs shot at Abel Cine, NY. Lens, by the way, was Optex 5.5mm Super16 on a M43 to PL adapter. The vignetting you are seeing is due to my holding a matte box and Mitch Gross shooting through the matte box. I think the overall image was overexposed in the first place and you can see the clipping.

On the graded shot, I lowered the Offset and dropped highlights. Did no color adjustments. And applied the Blackmagic Cinema Camera 3D output LUT. All in one node. Those lights in the background definitely clipped and those are the orbs. Bulbs themselves are spherical.

So I think it is more a function of exposure and range rather than a fault with the camera per se. I'm not seeing it in properly exposed footage with no extreme highlights.

Ned Soltz

Optex55_Graded.png (950.31 KiB) Viewed 24607 times

Which may prove Kholi's point.

I think it's a legitimate concern, but so was the CCD smear issue that plagued nearly every CCD camera that one wanted to use, right on up to the Genesis, which shot many major motion pictures and yea... you can spot CCD smear in them.

Can't remember if it was fixed in the F35.

If anyones' lost on what I'm talking about, you point a CCD camera like the HVX200 or Panavisions' Genesis at a bright enough light (talking street lamps) and a line shoots up and down the frame, stays there.

well, i think there's almost same issue with bmcc('s ProRes).
here's my guess.

bmcc has 2.5k resolution.
and...bmcc's ProRes, resized from 2.5k to 1920x1080, probably resizing white orbs too.
so they appears smaller than bmpcc...as not seem so weird.

well, i think there's almost same issue with bmcc('s ProRes).
here's my guess.

bmcc has 2.5k resolution.
and...bmcc's ProRes, resized from 2.5k to 1920x1080, probably resizing white orbs too.
so they appears smaller than bmpcc...as not seem so weird.

There isn't an issue with the 2.5K Camera, not this issue.

Please do not create another problem out of thin air, there's already enough noise.

call me crazy, but I'm pretty sure I've had the subtle "blooming sensor" (if that's what it's called) issues with overexposed shot on the 2.5k. This could also depend the lens and filters? When I get a second I will try and find an example.

Please do not create another problem out of thin air, there's already enough noise.

apologize for my comments.
i never wanted raise a new issue.
i did not mean negative...
i'm not good at english, it's so hard to communicate smoothly.

i have no negative with bmd&bmpcc&bmcc.
i believe this issue(with bmpcc) will be solved with firmware update.
(RAW available, or changes of ProRes compression, or, whatever)
so i'll quit with this topic.
once again, i apologize for my comments.

Please do not create another problem out of thin air, there's already enough noise.

apologize for my comments.
i never wanted raise a new issue.
i did not mean negative...
i'm not good at english, it's so hard to communicate smoothly.

i have no negative with bmd&bmpcc&bmcc.
i believe this issue(with bmpcc) will be solved with firmware update.
(RAW available, or changes of ProRes compression, or, whatever)
so i'll quit with this topic.
once again, i apologize for my comments.

I think the Pocket Camera "circles" are a real thing to deal with.

But, it is not on the 2.5K camera. I've blown highlights out all over the place on 2.5K with all types of glass in various situations, this does not happen. That's what I mean, it is not a problem with that camera.

For my 2 cents worth, I have the camera on pre-order and this concerns me. It seems that highlights are rounded off harshly to create blobby areas. This eats into what evers around it. You can see this quite clearly with the "pinocchio" nose shot from Philip Blooms latest. The extreme highlight on the roof behind blooms outward and eats into the models nose. One would expect this sort of extreme backlight to wrap around the subject in a far more elegant manner surely. By any standards this is troubling. I can understand if you point a camera at a bright light you are going to get overexposure especially at night, but the characteristic of the bloom is quite unappealing once you notice it, on the water highlights and on chrome details in some of the pictures submitted to this thread. Even more worrying is "Black Hole Sun".

My biggest concern is not the blooming but that, despite a lot of confusion and panic, no one from Blackmagic has issued a statement. I would assume they are aware of this and maybe looking into it but it would calm a lot of people if they at least gave us the official line.

http://www.flickr.com/groups/1773207@N2 ... 252110028/ for same issues with other sensors