Welcome to our community

Be a part of something great, join today!

Log in
  • 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 :)

Banding in Red Footage

M

Macros

Member
Joined
Aug 6, 2008
Messages
11
Reaction score
0
Points
0
I have been working with some 4k red footage we shot recently and have found some issues with banding in my encodes.

It becomes more apparent when I export a scaled down sequence (1080). I have been using the Cineform codec and have gotten good results with the compression and almost lossless quality but found that even a dpx sequence has banding. After looking at some R3D's I noticed that a) very faintly I can see the banding (either that or I have been staring at this footage for so long that Im seeing banding where there is none)

and b) the footage is rather noisy which doesnt help with the banding.

Are we doing something wrong?

I render the clips straight using Cineform Remaster and also to dpx from Red Alert.

Color space: RedSpace
OutputLUT: Redspace
Chroma denoise : Max
OLPF Compensation High
ISO320
Kelvin 5600

I have also found that when grading with MB Looks, certain effects make it alot worse (Like the edge softness or vignette, exposure spot etc.)

Wanted to upload a Tiff but Cant on Reduser (Invalid File)
Going to need to host it myself somewhere it seems...
 
In all the RED footage I've seen, I've never seen banding in the actual data. I have seen it in the display of that data, be it something photoshop was doing wrong, or the monitor itself.

Graeme
 
Im think Im sure

Im think Im sure

Matthew Rogers said:
Are you sure the banding isn't do to looking at it on an 8bit LCD monitor?

Matthew
Click to expand...

Im using two Apple Cinema HD displays. They shouldn't have a problem, would they?
 
They're 8bit displays....

Graeme
 
Dear Graham,
thanks so much for your response. We're here at Mars Productions in Cape Town, and our boss, Ross says he's dealt with you in the past and is really happy you replied. Assuming the monitors are 8 bit and might be creating the banding in our viewing, what do you suggest we do (outside of buying a really expensive monitor) to make sure we aren't having a problem
Regards,
Chris
 
Best thing to do is spit out a 16bit tiff from RedAlert, take that to photoshop and look at it there 100% size, ie pixel for pixel.

Graeme
 
Macros said:
Dear Graham,
thanks so much for your response. We're here at Mars Productions in Cape Town, and our boss, Ross says he's dealt with you in the past and is really happy you replied. Assuming the monitors are 8 bit and might be creating the banding in our viewing, what do you suggest we do (outside of buying a really expensive monitor) to make sure we aren't having a problem
Regards,
Chris
Click to expand...

If you are seeing banding in the source, and not in the monitor, when you push the image the banding will get worse, if it doesn't it is just an artifact of the monitor. e.g. if you are seeing banding in a blue sky gradient, temporarily increase the contrast, and you banding should become pronouced with clear steps between the levels (but I doubt it will, there is no good reason to banding from R3Ds.) When you have proved it is not in the source, you can greatly suppress the monitor banding by adding 0.5% noise as an adjustment layer (so you won't render it in) -- dithering for you display.
 
Yes David, that's what I do - stress the 16bit tiff in Photoshop to see if it's in the data or in the display. Other thing to do is convert to 8bit in Photoshop as it generally does a good job of 16bit to 8bit, and then you're 8bit out to the display. That can sometimes show what's up too.

Graeme
 
Any ordinary computer LCD will show some banding, in particular if you have color-corrected it via a software LUT. Only hardware-calibrated monitors with a wider LUT in the display electronics are free of it.
 
Red produces banding 'prone' material more easily than ANY other camera system I have used.

Hoever it is NOT red's fault.

The combination of super clean images and really shallow Dof is trouble for 8 (and sometimes less) bit displays. Start pushing the footage around and it only gets worse.

We need great dithering routines on the back of finishing tools.

I spoke to one finishing tool maker and they saw no need (actually they though vanilla HDCAM was 10bit so I knew the argument was going to be difficult).

Really clever tweakable dithering PLEASE.

thanks

Michael
 
Macros said:
Here is a link to a tiff straight out of RedAlert.

area is question is the bottom right... the noise diffuses it a bit and its very subtle before color or grading. But then again this probably is my monitor as has been said.

https://rcpt.yousendit.com/627222267/3f14a41e4355631c0b9e118421244dfa
Click to expand...

Yes, I see it on 10bit monitor as well.
This banding is the main pain in the b..t when converting down to BluRay material.
Some guys know how to do it and some don’t.
No problem on any 16 bit material so far, unless I push the colors in some bad corner.

I did play with all standard codecs and color spaces that AF and PPr comes with and no luck. I wonder if working in native RED support in AE or PRr will produce clean 8 bit material.

Dithering helps a bit but the trick is I think, that only one combination (if any) of codec and color spaces will work.
Since there is many combination if you are not the expert in this colorimetry and you do not understand exactly what is going on, it is just try and error exercise.
 
So you see it too? this is straight from the R3D... Shot in 4k but output to tiff via redalert with no settings and max debayer etc. What should I do?
 
Before encoding to H.264 in 8 bit: just add a tiny little bit of noise! The images are too clean. We've seen this lots of times with CGI and DVDs.
 
Michael Lindsay said:
Red produces banding 'prone' material more easily than ANY other camera system I have used.

Hoever it is NOT red's fault.
Click to expand...

Dithering is an area where analog gain helps, and unfortunately, Red does not have analog gain. Since, adding noise, as suggested in the posts above, after ADC requires larger quantity of noise to hide artifacts, with the result that other parts of the image have more severe quality degradation. In fact, adding noise after ADC, perhaps may not even be called dithering as such, as dither was supposed to happen before ADC.

(For the technically inclined, ADC can result in the aliasing of probability density functions of the quanitized data. Hence, just like in Shanon sampling theorem, to avoid aliasing, in this case PDF aliasing, a low pass filter (not to be confused with a traditional low pass filter operating in frequency domain) has to be used. Adding noise before ADC, low passes the PDF characteristic function to avoid aliasing. However, if noise is not added before ADC, and aliasing of PDF happens, then a much stronger low pass will be needed to surgically remove the effects of aliasing. I.e., a noise with a wider standard deviation will be needed now.)
 
Joofa said:
Dithering is an area where analog gain helps, and unfortunately, Red does not have analog gain. Since, adding noise, as suggested in the posts above, after ADC requires larger quantity of noise to hide artifacts, with the result that other parts of the image have more severe quality degradation. In fact, adding noise after ADC, perhaps may not even be called dithering as such, as dither was supposed to happen before ADC.
Click to expand...

Joofa

Forgive my ignorance..

Imagine I have a 10bit files that when displayed on a 10bit monitor does not band. Truncating this file to 8bit causes banding. If i take this file and run it through a dithering plugin like the one offered free from http://krokodove.com/
and dither to 8bits. The edges where I would have seen banding have been spread (in a noise like pattern). Is this not dithering?

I would like to have the correct terminology so any help would be welcome?

thanks

Michael
 
You must log in or register to reply here.
Back
Top