Welcome to our community

Be a part of something great, join today!

  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Noisy Blue Channel

JR Monteiro

Well-known member
Joined
Feb 24, 2012
Messages
87
Reaction score
0
Points
0
I just did a green screen shoot and noticed some blotchy noise in the picture in the skin tones and even in the background. I thought it might be my monitor but when I brought it back to the office and looked at a few frame grabs in Photoshop here is what I saw. Is this normal or do I have a problem with my camera? I was shooting with tungsten light as opposed to daylight balanced bulbs but would it make that much of a difference? The stills are marked as to what ISO and exposure settings I was using. The most prominent noise was seen on the woman shot at 640ISO and underexposed by one stop which would stand to reason if it were low light noise. But the noise indicator bar on my Red LCD monitor was clean. There are two frame grabs for each shot an RGB and a Blue Channel only. The Red and Green channels were clean. I remember seeing this on my old Red One prior to the MX chip. Is this the best I can expect from my new Epic under these lighting conditions or should I be contacting the service department?


http://www.redsolutions.tv/Red/1a.jpg
http://www.redsolutions.tv/Red/1b.jpg
http://www.redsolutions.tv/Red/2a.jpg
http://www.redsolutions.tv/Red/2b.jpg
http://www.redsolutions.tv/Red/3a.jpg
http://www.redsolutions.tv/Red/3b.jpg


 
Last edited:
Hard to tell much from this post...

Links are still bad...

What would be of more practical help, would be a single r3d snapshot from each exposure...
 
Well, this is about as I would expect it with the meta whiteballancing.

Is it causing you problems?

In that case, I'd maybe have tried to set the WB not lower than 4400 kelvin, and do the rest in a grading app.

But it's hard to really say anything meaningfull about this without having some r3d's to look into...
 
I repaired the links and they seem to work now. These are R3D Snapshots converted to jpegs
 
'couple of issues here.

3k, means less scaling of the noise. Strange choice for a keying shot.

Attached is a screendump of the waveform/histogram of the shot of the woman at 3g3/RC3/320/5000 which is closer to sensor native.

As you can see, there is very little exposure in the blue channel, and thus it has to be gained, and that generates noise.

Not really a noisy channel issue as much a noisy exposure issue...
 

Attachments

  • Skjermbilde 2012-05-09 kl. 00.53.15.jpg
    Skjermbilde 2012-05-09 kl. 00.53.15.jpg
    52.1 KB · Views: 0
Looking at the R3d, when you reset everything to default, it looks good, except the white balance. I typically rate the camera at ISO 320 when shooting Tungsten, so going even more yellow with the lights i'd rate it lower to reduce noise. Otherwise, use daylight balanced light at ISO800. That all produces good noise levels. What's happening here, is that you are not providing the sensor with enough blue light.

That's just my opinion. I'm sure others will have more definite answers.
 
That is what I was thinking because the footage rated at lower ISO looked cleaner. I just wasn't sure about the exposure limits of the Epic. I won't make that mistake again. Thanks.
 
Thanks Gunlelk that's very clear. I should have been looking at that display during my shoot.
 
Thanks that's very obvious. I should have been looking at that display during my shoot.

If you want to controll what WB the sensor "sees", 5000 is closer to it. "View RAW" could possibly be of some help (even though that isn't really RAW in any way... :) )
 
Currently there isn't any RAW exposure meters by channel, is it? That'd really clear a lot up for a lot of these noise threads, methinks...
 
The compression ratio is 8:1 - we had similar problems with a green screen shoot with the Epic set to 7:1. A previous green screen shoot at 6:1 was much cleaner - so you might want to review the compression setting. If you have access to NukeX then the deNoise filter will significantly reduce the noise in the blue channel and help the keying.

Z
 
If you want to controll what WB the sensor "sees", 5000 is closer to it. "View RAW" could possibly be of some help (even though that isn't really RAW in any way... :) )

Gunleik, what is the closest thing to monitoring RAW in camera? In post, if I set it to RedLog (Not RedLogFilm which has some weird roll-off in the blacks and whites), then I get what seems to be a very accurate histogram reading - this would be perfect if I could enable it in camera, but I can't. With anything else (RG2, RD3, RLF), the histogram lies....try pointing your camera up at the sky and watch the histogram NOT clip (sure the traffic lights will light up and colour channels will go through the roof, but the histogram stays well clear of clipping to the right hand side)...is this just my camera?

Also, using the "RAW" view doesn't help with this either.
 
Actually you cannot really... Linear gamma would give you a hint at what and how the camera sees, but it is not of much value for most people on-set.
The rolloff you describe in RLF, is how it is supposed to work...

If I need to controll an exposure on-set, I'd have set it at 5000k/320ISO/RLF. That gives you quite a good indication as to how the camera sees the colors AND good info on what is in the high and low end of the exposure. The image will look lowsat and lowcontrast, but it is packed with visual info of what is in the exposure.
 
Actually you cannot really... Linear gamma would give you a hint at what and how the camera sees, but it is not of much value for most people on-set.
The rolloff you describe in RLF, is how it is supposed to work...

If I need to controll an exposure on-set, I'd have set it at 5000k/320ISO/RLF. That gives you quite a good indication as to how the camera sees the colors AND good info on what is in the high and low end of the exposure. The image will look lowsat and lowcontrast, but it is packed with visual info of what is in the exposure.

Thank you :)
 
a002c0540501xo001000031.jpg


a002c03605015n001000003.jpg


Isn't it less noisy that way ?
 
Sorry a bit late here, but the Saturation is cranked all the way up in R3D snapshots, so that is not helping matters.
 
Back
Top