Blackmagic Forum

I received my Blackmagic Pocket a few days ago and have been out a few times to do some extreme low light tests before I shoot a short film. I've come across what might be an issue with my camera but not sure exactly what it is. I shot about 20 clips and it was in the latter half when it happened. My battery was probably at 15-5% when this 'grid' like effect came up. I'm uploading the original clip (taking some time and will post link when it's done. I've put some pictures below in the meantime.. Has anyone else come across this? Down scaling the clip (etc - putting in 720p timeline) - seems to amplify the issue. Note that these are extreme low light conditions (pushing the camera massively!) - so I did expect to see square artefacts but not quite like this. At 720p it is quite noticeable so you should see it clearly in images below.

Setting at the time:
1600 ASA
24p
172.8 shutter
F1.8 (which is wide open on lens - so a little soft)

Open the images in new window - full res to see it.

This is downscaled 720p example. You'll probably have to download rather than just play to see it..
https://www.dropbox.com/s/7b633n9t6y7h2 ... e_720p.mov

That's not good, I think you should get in touch with BMD support, even in low light you shouldn't be getting anything like that. Have you tried the pointing the camera at some other uniform colour in medium low light?

Goodness, what else is right with this camera??

Christian Horne wrote:That's not good, I think you should get in touch with BMD support, even in low light you shouldn't be getting anything like that. Have you tried the pointing the camera at some other uniform colour in medium low light?

I have tried pointing at other colours with a bit more light but still underexposed. It seems fine. Definitely grainy/noisey as I would expect at 1600 ASA but not getting anything 'grid like'. I usually really like the BM grain but it changed up for few shots (3 out of 20 shots had the issue) Probably need to get it checked. If someone knows a bit about 'grain structure' and these type of digital sensors would be great to hear your opinion..

If those shots are important to you, I think you could get rid of the grid with Neat Video... Canon 5D and 7D (at least) seem to have similar grid patterns that are visible in some shots (not as bad as this though) and Neat Video was able to clean them out.

sebasti wrote:If those shots are important to you, I think you could get rid of the grid with Neat Video... Canon 5D and 7D (at least) seem to have similar grid patterns that are visible in some shots (not as bad as this though) and Neat Video was able to clean them out.

Thanks for the info.

They are just test shots so aren't that important. It would be good to know whether this is standard behaviour on the camera. I can understand 5d h264 compression having a fair bit but ProRes HQ is seen as a higher end codec isn't it..?

Is this footage straight out of the camera?

I'm not sure that the grid comes from the sensor, actually i hope it's not the sensor but something software related, like downscaling gone bad, or some weird debayering glitch. If the footage was not directly from the camera but recompressed through Resolve or else this could as well be the reason.

Interestingly the file you put on dropbox is rated 1248x702 instead of 1280x720. This is rather odd, and may be the source of the grid like structures, because 13/20 downscaling (1080->702) may introduce more severe artifacts than the usual 2/3 downscaling (1080->720). But why is it 1248x702 in the first place? The odd numbers suggest that someone accidently typed 702 instead of 720 somewhere...

Anyway: does the problem prevail in 1080? If so, can you provide a sample 1080 clip out of the camera?

Mac Jaeger wrote:Is this footage straight out of the camera?

I'm not sure that the grid comes from the sensor, actually i hope it's not the sensor but something software related, like downscaling gone bad, or some weird debayering glitch. If the footage was not directly from the camera but recompressed through Resolve or else this could as well be the reason.

Interestingly the file you put on dropbox is rated 1248x702 instead of 1280x720. This is rather odd, and may be the source of the grid like structures, because 13/20 downscaling (1080->702) may introduce more severe artifacts than the usual 2/3 downscaling (1080->720). But why is it 1248x702 in the first place? The odd numbers suggest that someone accidently typed 702 instead of 720 somewhere...

Anyway: does the problem prevail in 1080? If so, can you provide a sample 1080 clip out of the camera?

I see your point. Because my internet upload is slow I exported out of Final Cut 7 from a 720p timeline. It's also LT - just so I could get an example up. I did see it straight out out of the camera though. It's not as severe (as 720p version but it's definitely there. I've started uploading the original camera file so will let you know when it's up. I don't think I messed up the downscale as it's one of the main Final Cut 720 selections and wasn't typing in the numbers. Out of interest the direct camera file says in Quicktime Movie Properties - Apple ProRes 422 (HQ), 1920 x 1080 (1888 x 1062), Millions 24-bit Integer (Little Endian), Stereo, 48.000 kHz.

I'm only technical to a certain point but what are the numbers in brackets?

1888x1062 is the visible area of 1920x1080

No need to worry

Do you have After Effects on your machine? Can you scale the image up 400x and save a photo file of just the effected area, might help by looking at the pixel level of what we are seeing?

Joel Frances wrote:Out of interest the direct camera file says in Quicktime Movie Properties - Apple ProRes 422 (HQ), 1920 x 1080 (1888 x 1062), Millions 24-bit Integer (Little Endian), Stereo, 48.000 kHz.

I'm only technical to a certain point but what are the numbers in brackets?

If you are opening the file in QuickTime and it's scaled down, the dimensions in brackets are what size QuickTime is displaying the file.

Christian Horne wrote:Do you have After Effects on your machine? Can you scale the image up 400x and save a photo file of just the effected area, might help by looking at the pixel level of what we are seeing?

I've attached 200% and 400% in AE. I'll have to log out soon as it's getting late here but thanks for your input. I'll still try and get original camera file up soon though..

Looks like a Sensor problem to me, I would get in touch with BMD and send them all the files you have posted here. It doesn't appear in the deep blacks of the footage but anything just above black there is a problem, also the grid moves with your panning which is why I think it's fixed to a Sensor issue.

Would this be the downscale algorithm in Resolve? Sometimes Resolve makes strange patterns out of noisy footage when downscaled that is otherwise non-existence when viewed at the original resolution. I have seen similar things when noisy BMCC footages are downscaled in Resolve.

I think he said he could see the anomaly straight out of the cameras prores file? If not then it could be something else...

The question is: is the problem caused by the camera, or later by software? Is it visible in-camera as well, and if so, only during replay (=> firmware problem) or also during recording (=> sensor problem)? If in-camera you can't verify the problem, chances are that it's only caused by your pc/mac's software, eg. a broken codec, bad quicktime installation, etc.

It can't hurt to contact BMD customer service, but it would be a shame to send it in unneccesarily.

In any case that has to be fixed - the camera is no low-light-monster, but it definitely can do better than that!

Here's the original camera file - finally got it up. You'll have to click the download rather than play. At the end of the tilt up near top of skyscrapers I can see it.. Let me know your thoughts. Downscale to 720 also - if you have the time.
https://www.dropbox.com/s/8il0z6m10mvzc ... _C0003.mov

I'm out tonight as it's late. Thanks again for the help though.

I have downloaded your clip and have run MB denoiser over the footage which shows the problem more clearly, http://www.flickr.com/photos/christianhorne/9874487045/

Thank you for the upload!

I can now also confirm that the "pattern" is actually part of your video, so the camera is to blame. I'd first try to do a firmware upgrade, but if that doesn't help send it back in, and add your footage - preferably a 2x or 3x magnification - to the description of the problem.

Now we can just hope that this is not the next issue that needs "recalibration"... btw: is this camera pre- or post-calibration?

Mac Jaeger wrote:Thank you for the upload!

I can now also confirm that the "pattern" is actually part of your video, so the camera is to blame. I'd first try to do a firmware upgrade, but if that doesn't help send it back in, and add your footage - preferably a 2x or 3x magnification - to the description of the problem.

Now we can just hope that this is not the next issue that needs "recalibration"... btw: is this camera pre- or post-calibration?

I picked up the camera last Wednesday 18th Sep. The supplier had received the day before so I was thinking post-calibration. I haven't done a firmware upgrade yet though so will see if its a fix..

I don't think the firmware update will help, but would do it anyway, just in case. Then you can say you've done all you can when you return it...

Still i hope you camera is just faulty, and what you found isn't a side effect of the new calibration - keeping my fingers crossed.

If that's all the light you could get a 1.8 and 1600, you're getting that pattern because you're severely underexposed.

Sensor's rated 800 ISO, looks like you're still pretty far under. Without noise reduction in camera etc. that's what's going to happen.

I've seen it when trying to shoot 1600 with my Sigma 1.8 very far underexposed, but when I can shoot and expose at 800 the patterns are nowhere to be found.

Here are 800 ISO SLR 25/0.95's wide open.

Zero noise reduction or sharpening.

Kholi wrote:If that's all the light you could get a 1.8 and 1600, you're getting that pattern because you're severely underexposed.

Sensor's rated 800 ISO, looks like you're still pretty far under. Without noise reduction in camera etc. that's what's going to happen.

I've seen it when trying to shoot 1600 with my Sigma 1.8 very far underexposed, but when I can shoot and expose at 800 the patterns are nowhere to be found.

Here are 800 ISO SLR 25/0.95's wide open.

Zero noise reduction or sharpening.

I see your point about it's native ASA. The shot above is just an extreme test. I did it to see how it is in low light while pushing ASA. What interests me is the 'grain structure'. On some other cameras - even the BMCC the grain is more randomly displaced and doesn't form any patterns in these really under-exposed areas. It has a film quality and it would be great to see the same in its baby brother. Most of what I shoot will be exposed correctly but sometimes there will be dark areas within the image and I'm interested in what it does.

I'm in contact with BMD support and they've been very helpful thus far.

I too got this pattern. I was shooting 1600ASA with a sigma 1.4. The grid seemed to appear when a red light sources was coming from the city street... I've noticed it subtly on some other footage, but never this pronounced.

Its hard to see in the JPG because it compresses it, but you can see it in the screenshot from Davinci. Its much more pronounced and pops up several times, sometimes only taking up half of the screen.

mine does the same pattern as well at 1600 ASA. Run it through Neat video and I can typically clear it up but I shouldn't have too.

I noticed the same kind of pattern, it actually looks like a maze. But I only saw it in one clip that was staring directly at a bright LED light on a piece of computer hardware. And it wasn't really apparent unless I was really peeping, like 200%...

A lot of people are talking about NeatVideo. And I just need to add that it's one of the most useful plugins ever made. I use it on almost everything, even if it's not all that noisy. The sharpening works really well too and if you're running your footage through something like film convert the film grain looks better if you're not stacking it on top of video grain.

It's really an indispensable piece of software.

I third the Neat VIdeo comment.

Especially with the noise pattern on the BMPCC, Neat does an AMAZING job at creating a clean, detailed image.

I've never seen neat video work as well as it does with this camera.

Still a bummer about the sensor grid pattern in footage. I'm going to have to experiment more at ASA200 to see how clean this is...

Same exact thing has happened to me as well. Looks like it may just be the camera. I do have to say though that it has only been somewhat pronounced when staring at a bright light source that is overexposed. Correctly expose the shot and it goes away, or at least becomes much less visible incrementally the further you stop it down.

Same here.

How it relates with battery?

I thouhgt its my problem with strong color grading. Well, its already noticeable at low level unsharp mask.

OK, so last week I'm posting in the hot pixels thread, and now here.

Did some more tests this evening with my PCC and I'm having this same issue. Grid-like pattern appearing across half of the image when shooting in city natural light situations at 800 ISO.
Anyone talk to support about this? What did they say?

Was able to upload a clip to show what I'm talking about. This was with only slight sharpening applied in Premiere...

rhyduk wrote:OK, so last week I'm posting in the hot pixels thread, and now here.

Did some more tests this evening with my PCC and I'm having this same issue. Grid-like pattern appearing across half of the image when shooting in city natural light situations at 800 ISO. Yes, it's a low-light situation, but once again, this is unacceptable. I've about had it with this camera.

Anyone talk to support about this? What did they say?

You're right - it does also happen at 800asa. I get it on flares from certain coloured streetlights. Doesn't seem to like magenta flares at all. I spoke to support a few weeks ago and they said it can be fixed with a software update so it's not a sensor issue.

I think this is just the way the sensor handles low light...I can't really be 100% sure but from the response I got from BMD support they'll probably say the sensor is not designed to handle such extreme low light conditions. I have seen the grid patterns, I have some hot pixels, and even a strangely shaped grey blotch that shows up on lighter shadows when shooting in ASA1600 and when the footage is way underexposed. I think all sensors on this camera will behave the same way...again I can't be 100% sure but I plan to keep my second camera until I get the replacement one from BH and compare.
I shot this last weekend and none of these problems seem to appear under normal conditions when the clips were slightly/naturally underexposed and pushed to normal levels in post



I'm not defeding BMD but I'm just saying don't expect night vision capabilties in this sensor and perhaps these tests while valid don't really mean there is a problem with the sensor.

ihabali wrote:I think this is just the way the sensor handles low light...I can't really be 100% sure but from the response I got from BMD support they'll probably say the sensor is not designed to handle such extreme low light conditions. I have seen the grid patterns, I have some hot pixels, and even a strangely shaped grey blotch that shows up on lighter shadows when shooting in ASA1600 and when the footage is way underexposed. I think all sensors on this camera will behave the same way...again I can't be 100% sure but I plan to keep my second camera until I get the replacement one from BH and compare.
I shot this last weekend and none of these problems seem to appear under normal conditions when the clips were slightly/naturally underexposed and pushed to normal levels in post

I'm not defeding BMD but I'm just saying don't expect night vision capabilties in this sensor and perhaps these tests while valid don't really mean there is a problem with the sensor.

I don't think it's a sensor issue either. Codek..maybe? I'm not convinced it's a low light issue though anymore.. With 'Greeches' example - the practical red/magenta light is flaring on the lens and creating the pattern. The overall exposure doesn't look too under-exposed. Flares from daylight coloured lights seem fine but I think lights with a colour temperature below tungsten are problematic. I live in Melbourne where most street lights are daylight temp but some look like they're rated between 2000k and 2500k and I think they're best avoided. I love this camera and if this is the only 'work around' I'll be pretty content but I do think it's a minor issue. A sunrise test would interest me as it would be a similar colour temp to these lights..

A sunrise test while getting lens flare

I tried to provoke this for 2 hours now, but no dice - looks like you need a certain combination of flare, lens and light.

But I saw this on a Sony F3 before, actually it was the sensor mirror image, that was mirroring in the rear lens element, but only in the flare portion. Took me forever to find this out back than.
Needless to say, that a mattebox instantly cured the problem.

I guess it's a similar thing here.

Hey guys. I just ordered my BMPCC and I have some questions.

1. Does any of you guys use PC for editing BMPCC videos and in what program?

Thank you guys.

Frank Glencairn wrote:I tried to provoke this for 2 hours now, but no dice - looks like you need a certain combination of flare, lens and light.

But I saw this on a Sony F3 before, actually it was the sensor mirror image, that was mirroring in the rear lens element, but only in the flare portion. Took me forever to find this out back than.
Needless to say, that a mattebox instantly cured the problem.

I guess it's a similar thing here.

Did you try at ASA1600 with an underexposed uniform background? If you do that and push it a bit in post you'll probably see the grid pattern if I'm not mistaken.

I've spend most of today doing various tests with the BMPCC.

I was able to create this pattern but only in some circumstances. So I tried a few other things and I think Frank has a point.

I'll try explain. I used to get issues with the FS-100. If I had light in the frame it would cause internal reflections where the light would reflect off the sensor onto the rear element and then back onto the sensor. So you start getting ghosting images are strange flares. I think this is being caused in a similar way.

So this is how my test ran. Camera set to 1600ISO, aperture around f/5.6. Dark room with charcoal curtains as my back drop. My light source was a mini maglite touch.

I put an old Nikon 55mm f/1.2 lens on the BMPCC with the Metabones SpeedBooster. Shine the light across the lens and create flares. Catch the light direct in frame. Shane the light on the curtains. When I review the footage I get noisy footage as expected but I'm not seeing the grid that others have posted.

So I remove the Speedbooster and use a normal adaptor. Same process. When I review the footage most of it is normal noise. But when the torch shines across the lens a pink flare is produced and the pixel grid appears.

So I then take the lens off and create light leaks with my hand. No grid.

I need to do more testing by I think, as Frank mentioned, that it might be to dow with internal reflections between the sensor and the rear element of the lens. Looking at the other shots that have this issue it seems that it happens where there is light flare.

Screen shots here:
https://www.dropbox.com/s/bkt91flu1l3np ... test_1.zip

Hope this can be helpful to someone...

I've noticed this pixel grid in some of my footage, although very little. At first I thought it was only some kind of moire pattern created by the recorded pixel grid of the sensor superimposed onto the pixel grid of my computer display because if I viewed the footage at 100%, most of the time it disappeared. However, after viewing some new footage at 100%, I could still see some of the pattern, albeit very faint.

So, then, in Resolve, I tried shifting the image in the frame (in the SIZING panel: tilt;pan) and shifting it by HALF a pixel in a horizontal as well as vertical position made the grid pattern disappear (it works in FCPX and AE also).

So, I don't know exactly what the cause is (although I still suspect that it's a moire pattern) and if my fix will always work, but at the moment I am able to fix it quite easily.

It's not a reflection. I had this issue randomly. With same stings at 400iso and 800 iso. With light flare and no light in the shot. Once I see it on the screen, i just replace battery and it's gone. It's definitely debayering glitch.

Chiaroscuro wrote:So, then, in Resolve, I tried shifting the image in the frame (in the SIZING panel: tilt;pan) and shifting it by HALF a pixel in a horizontal as well as vertical position made the grid pattern disappear (it works in FCPX and AE also).

Nice find. That works a treat.

For what it's worth, I had a similar issue with my pocket cam. I sent my camera back to Blackmagic along with some footage for them to look at. After 10 days, Blackmagic's support team concluded that it "does not believe the grid effect is a hardware issue with the camera and does not believe a different camera will yield different results in the same circumstances. They are continuing to investigate this on a deeper level and are looking into possible debayer or render issues or other reasons behind the effect seen in the footage."

From that response it seems as if all the cameras have this reaction. Blackmagic knows about it and are looking into it. Hopefully this can be fixed in firmware? For now I'll have to try that trick in Resolve.

Chiaroscuro wrote:
So, then, in Resolve, I tried shifting the image in the frame (in the SIZING panel: tilt;pan) and shifting it by HALF a pixel in a horizontal as well as vertical position made the grid pattern disappear (it works in FCPX and AE also).

.

Now that's interesting. Just can't wrap my had around it.

Shifting half a pixel is the same as applying a slight blur, as each new pixel will be interpolated from four pixels. So while this may work, it reduces the resolution. I'd rather have a in-camera solution.

Mac Jaeger wrote:Shifting half a pixel is the same as applying a slight blur, as each new pixel will be interpolated from four pixels. So while this may work, it reduces the resolution. I'd rather have a in-camera solution.

Yes, I agree: I would also rather have an in-camera solution. I like things to work and work properly and not rely on some kind of "hack" to make things work. So again I'll wait for BM to come up with a proper solution and for the time being I'll use the half-pixel-shift-hack if I really have to.

Meanwhile you might want to check this out:

http://community.sony.com/t5/F5-F55/F55 ... d-p/201051

Hi everyone,

Just got a BMPCC as a B cam to my BMCC and I also noticed the grid pattern in flares. Are there any news on a possible fix?

Also (unrelated), is it normal that the battery door feels quite flimsy? Tapping it sounds like a cheap plastic toy, especially when there's no battery inside, but also when there is.

Yep, I am also wondering if there is any news about this. I would love to get rid of the grid pattern. Happens with every flare.

You don't even need flare to reproduce the grid pattern. Apparently all you need is a lens with a large aperture. Almost every shot on my Voigtlander 17.5mm f/0.95 with the aperture between 0.95 and 1.4 has the grid pattern (with or without flare, so I am led to believe it is low light related). At least I now know there is a "fix."

Has anyone seen this on RAW footage, or is this only happening on ProRes footage? I haven't seen it on any DNGs myself.