BMPCC: Compression nasties or something else?

Got a BMPCC this week and took it out the box this evening. Shot a quick test into the sun which lead me here after noticing the maze/grid pattern.

A single raw frame is here:
https://dl.dropboxusercontent.com/u/131 ... 000064.dng

Shot as:
Raw
Voigtlander 25mm f/0.95
200 ISO
Stopped right down. Should have had an ND on of course but I literally just nipped out to bag the sunset in order to see what was recoverable from over exposed highlights.

I am keen to shoot into the sun during magic hour. The 5d mk2 shooting raw handles this without incident so I'm interested in any advice that might put me on the path to trouble free shooting in this regard (if, indeed, it's possible to shoot that way with this camera).

That looks to me like a lens artifact... the sensor is reflecting off of the back of the lens, back onto itself.

I could be wrong though. However I've never seen that issue before on my camera (though I don't have any lenses that fast either, so it may be a geometry issue that I'm just not seeing).

The fact that it occurs at the exact vertical center of the frame makes me think that this is the case though.

Got a BMPCC this week and took it out the box this evening. Shot a quick test into the sun which lead me here after noticing the maze/grid pattern.

A single raw frame is here:
https://dl.dropboxusercontent.com/u/131 ... 000064.dng

Shot as:
Raw
Voigtlander 25mm f/0.95
200 ISO
Stopped right down. Should have had an ND on of course but I literally just nipped out to bag the sunset in order to see what was recoverable from over exposed highlights.

I am keen to shoot into the sun during magic hour. The 5d mk2 shooting raw handles this without incident so I'm interested in any advice that might put me on the path to trouble free shooting in this regard (if, indeed, it's possible to shoot that way with this camera).

The purple grid pattern you are seeing in the highlight flares was addressed in firmware 1.8. Download the latest version here: http://www.blackmagicdesign.com/support ... al-cameras

The purple grid pattern you are seeing in the highlight flares was addressed in firmware 1.8.

Was it? I've heard contradicting statements about it. I haven't encountered it again in a while, but my last projects didn't involve critical lighting situations...

The purple grid pattern you are seeing in the highlight flares was addressed in firmware 1.8.

Was it? I've heard contradicting statements about it. I haven't encountered it again in a while, but my last projects didn't involve critical lighting situations...

From the 1.8 Read Me: "Improved instances where a grid-like pattern may occur in some flare highlights"
Under very rare circumstances you can get it so show up but it's nothing like it was.

This was shot almost entirely on the pocket with lots of lens flare and the grid pattern never showed up:
http://www.adamroberts.net/films/on-our-front-doorstep/

The purple grid pattern you are seeing in the highlight flares was addressed in firmware 1.8.

Was it? I've heard contradicting statements about it. I haven't encountered it again in a while, but my last projects didn't involve critical lighting situations...

From the 1.8 Read Me: "Improved instances where a grid-like pattern may occur in some flare highlights"
Under very rare circumstances you can get it so show up but it's nothing like it was.

This was shot almost entirely on the pocket with lots of lens flare and the grid pattern never showed up:
http://www.adamroberts.net/films/on-our-front-doorstep/

The problem is still there no change and my camera is 1.8 and I was internal getting some ugly flare patterns

Nice video Adam. I can see your camera is not having problems in this regard owing to the 1.8 firmware.

Like Mark though, my camera actually is using version 1.8. So it's a bit of a problem. I'd held off on buying the pocket camera, having thought the image issues had all been addressed.

I like to shoot into the light sometimes like you have done in your video, Adam. It's a very nice look.

Now I'm wondering whether attempting to do so with my camera will just leave me with ruined footage. Which would render the camera unusable as far as I'm concerned. I have seen elsewhere someone suggest an offset pixel method of overlaying frames to counter this but in my opinion that is a last resort as opposed to an acceptable pattern of workflow.

That looks to me like a lens artifact... the sensor is reflecting off of the back of the lens, back onto itself.

I could be wrong though. However I've never seen that issue before on my camera (though I don't have any lenses that fast either, so it may be a geometry issue that I'm just not seeing).

The fact that it occurs at the exact vertical center of the frame makes me think that this is the case though.

Interesting thought.
Like the Speedboosters, the fast Voigtlander lenses really add some low light muscle and shallow DoF capability to the m4/3 format. It would be a bummer if, for some reason, they aren't suited to the Pocket cam. And of course, I'm already invested in such lenses.

I have seen elsewhere someone suggest an offset pixel method of overlaying frames to counter this but in my opinion that is a last resort as opposed to an acceptable pattern of workflow.

This "countermeasure" works, but it only smoothes the grid artifacts, it doesn't remove the magenta spill in those highlights. I guess I'll have to wait what the next Concert shots reveal...

Nice video Adam.

Thanks.

I can see your camera is not having problems in this regard owing to the 1.8 firmware.

Like Mark though, my camera actually is using version 1.8. So it's a bit of a problem. I'd held off on buying the pocket camera, having thought the image issues had all been addressed.

I like to shoot into the light sometimes like you have done in your video, Adam. It's a very nice look.

Now I'm wondering whether attempting to do so with my camera will just leave me with ruined footage. Which would render the camera unusable as far as I'm concerned. I have seen elsewhere someone suggest an offset pixel method of overlaying frames to counter this but in my opinion that is a last resort as opposed to an acceptable pattern of workflow.

If you are getting the grid on footage you can "fix" it in post by shifting the clip 0.5px left and 0.5px down in your NLE or gradeing suit.

EDIT: Just seen your post about that.

Are all BMPCC's prone to that pattern, in the right conditions? And if so what are those conditions ?

I've never seen that pattern on my camera. I almost always shoot at 1600ASA, @ about f/2 - f/4 with a speedbooster. Raising the brightness in post from very underexposed shots has never given me a grid pattern. Grain, but not a grid.

Are all BMPCC's prone to that pattern, in the right conditions? And if so what are those conditions ?

I've never seen that pattern on my camera. I almost always shoot at 1600ASA, @ about f/2 - f/4 with a speedbooster. Raising the brightness in post from very underexposed shots has never given me a grid pattern. Grain, but not a grid.

The grid pattern appears when there are bright lens flares in the frame. It's not caused by raising underexposed footage so I doubt you would see it in your situation.

An easy way to test it to point the camera at a bright light and move side to side so the light source creates lens flares.

Large apertures are more prone to cause purple grids. I've only ever encountered it when using the sigma 30 f/1.4 or the sigma 18-35 f/1.8 wide open, and rarely in direct clipping areas, mostly in flares when bright lightsources were just outside the image.

To the best of my knowledge this problem appears on all pocket cameras, and since the new debayering also on BMCCs, so it might be a byproduct of the debayering algorithm.

Large apertures are more prone to cause purple grids. I've only ever encountered it when using the sigma 30 f/1.4 or the sigma 18-35 f/1.8 wide open, and rarely in direct clipping areas, mostly in flares when bright lightsources were just outside the image.

To the best of my knowledge this problem appears on all pocket cameras, and since the new debayering also on BMCCs, so it might be a byproduct of the debayering algorithm.

The BMCC did it before the latest firmware update (and still does it post new firmware unfortunately).

The BMCC did it before the latest firmware update (and still does it post new firmware unfortunately).

The purple grid? Didn't know that. Yet i don't own a 2k5 camera, so i'm following these threads less closely...

After upgradind the firmware to 1.8.2 I did the purple fringe test and I got the following: the issue was clearly visible on the Pocket´s LCD screen. Then I´ve did some basic CC on Vegas Pro, still the nasty issue visible in the shot (ProRes HQ). Finally, I´ve exported the file and the issue was much improved!? My guess is the BMD smoothed it out trough the previous firmware (1.8). Its not solved but mitigated.