Jump to content

Too much noise in Red One footage???


Mo Kamal

Recommended Posts

I've been seeing more and more feature films shot with the Red One, at the same time, all the Low light Red footage I've seen online were in my eyes much too noisy for feature film use. So I am wondering if Soderbergh, Jackson or any other mainstream directors have actually used the Red One in night time or low light situation. Anyone know of any such footage available online?

Link to comment
Share on other sites

I've been seeing more and more feature films shot with the Red One, at the same time, all the Low light Red footage I've seen online were in my eyes much too noisy for feature film use. So I am wondering if Soderbergh, Jackson or any other mainstream directors have actually used the Red One in night time or low light situation. Anyone know of any such footage available online?

 

Remember, the camera shoots to a raw format. The recorded image contains data all the way down into the sensor's noise floor. With most other cameras, your blacks are to some extent already crushed by the camera's on-board image processing system, so you don't see how much noise was actually there. With the Red One, you can choose to crush the blacks and kill some of the noise, or keep the shadow detail and live with the noise.

 

The Knowing trailer contains a bunch of night shots, presumably shot on the Red One like the rest of the movie, though of course it's impossible to know exactly how well lit they were. (They don't particularly look like available light.)

 

We have some decent-looking low-light footage (mostly from test shoots) shot with Red 404 + Superspeeds. If I remember I'll grab a few frames when I'm at the office this afternoon.

Link to comment
Share on other sites

  • Premium Member

"Too Much" noise or grain is rather subjective, like too much salt in your soup.

 

Once you are not talking about a grainless or noiseless image, then who's to say what it too much?

 

I've tested the RED ONE all the way to 1000 ASA ratings in available light at night, and the noise level wasn't "too bad" (subjective, again), but it was also complete dependent on how you color-corrected the image because most of the artifacts were in the blacks (so a lower-contrast look would exaggerate the problems and a hi-con look would mask some of them) and most of the noise was in the shadows too, not the highlights. And most of the "chunkier" noise was in the blue channel only, so how warm or cold you timed the image would bring that out, or mask it.

 

Trouble is that when you start talking about "low light" you have to define what you mean, because low light doesn't necessarily mean underexposure -- if you shot wide-open at f/1.4, that may get you a well-exposed image in low light.

 

What you really are asking is how much noise is there when underexposing the image, essentially rating it faster than the recommended 320 ASA, coupled with how much blue channel noise is there in 3200K rather than 5600K settings. 500T is definitely noisier than 320D, but by how much is a subjective call, and as I said, color-correction decisions will play into it. Because of color-correction issues, I've had some 500T shots look fairly clean and others look somewhat noisy in the blue channel.

 

Here are some frame grabs from "Stay Cool", the last RED feature I shot. The first two were shot at 500 ASA, output at about 3800K balance:

 

900%20Stay%20Cool%2084.jpg

 

900%20Stay%20Cool%20140.jpg

 

This one was at 400 ASA, 3800K:

 

900%20Stay%20Cool%20141.jpg

 

While there IS some noise, I don't think it could be labeled as "too noisy for feature use" -- the noise isn't any worse than the grain of shooting on 500T film stock.

 

Yes, I wouldn't say that the RED ONE is the best low-light digital cine camera on the market right now -- it's closer in behavior to its nearest cousin, the ARRI-D21, also a CMOS Bayer-filtered imager. I think the Sony F23, F35, and Genesis probably all behave better in low levels of tungsten light, noise-wise. But I certainly think you can shoot the RED ONE in those conditions if you know what you are doing.

Link to comment
Share on other sites

Granted a bigger FF35 sensor will offer another response as far as sensitivity goes, I am wondering which outcome a fast aperture might or may offer at low light.

 

"Too Much" noise or grain is rather subjective, like too much salt in your soup.

 

(...)

 

Trouble is that when you start talking about "low light" you have to define what you mean, because low light doesn't necessarily mean underexposure -- if you shot wide-open at f/1.4, that may get you a well-exposed image in low light.

;-)

Link to comment
Share on other sites

So, why *wouldn't* you use the least-noisy digital camera or film option available in a low-light situation?

 

Several primarily-35mm films have opted to shoot night scenes digitally in recent years because they produce a less-grainy/noisy/busy image.

 

Seems like renting the most-appropriate tool for the job would be ideal here.

 

I know you only have experience with the F-23's predecessor and the RED David, but could you hazard a guess as to the top low-light performing camera?

 

While I was thoroughly disappointed by "X-Files" use of the Genesis, it looked like it did quite a nice job in Apocalypto.

Link to comment
Share on other sites

The Knowing trailer contains a bunch of night shots, presumably shot on the Red One like the rest of the movie, though of course it's impossible to know exactly how well lit they were. (They don't particularly look like available light.)

 

I watched that trailer (the 720P version) and some of the footage did look extremely noisy in the shadows to me. One shot that jumped out was one of a creepy-looking guy with blonde hair pointing his finger towards the lens. It jumped out because it was noisy enough that it noticeably did not match the shots that came before and after it. I think these kinds of problems are often exaggerated by trailers, because footage from every part of the movie is thrown together in a way that was never considered when the color correction was being done, so things don't necessarily match as well as they might in the context of the actual film, but nevertheless I did notice the high level of noise several times in that trailer. And sometimes it wasn't in the "dark" scenes you would expect, but in an otherwise bright scene there might be a gray object or wall that showed an excessive amount of noise. I also don't think frame grabs really illustrate the noise level very well, because noise (or film grain) is much more apparent when you seem it in motion.

Link to comment
Share on other sites

  • Premium Member
I also don't think frame grabs really illustrate the noise level very well, because noise (or film grain) is much more apparent when you seem it in motion.

Hmmm -- I've always thought the opposite. Freeze frames look grainier to me than the part of the shot that's in motion.

 

 

 

-- J.S.

Link to comment
Share on other sites

I also don't think frame grabs really illustrate the noise level very well, because noise (or film grain) is much more apparent when you seem it in motion.

 

As John Sprung mentioned, moving images should give an impression of smearing of noise and less perceptability. Additionally, I have an impression that Red does not do temporal filtering for noise removal before compression and that manifests itself in low light situations. Similarly, lack of analog gain on Red can also increase perception of noise in certain low light situations.

Edited by DJ Joofa
Link to comment
Share on other sites

Hmmm -- I've always thought the opposite. Freeze frames look grainier to me than the part of the shot that's in motion.

 

Well, I guess it's just my own perception then. I agree that this is the case most of the time -- looking at a still frame gives you the time to really focus on details like grain, but I think excessive grain or noise becomes more apparent when it's in motion, precisely because you don't normally notice grain when it's in motion.

 

I was doing a DI once and was really pushing one particular color in a scene, to the extent that it was adding excessive noise to areas of the scene that were that color. It all looked fine on the monitor, because we were just looking at one frame, but when we actually applied the effect and played the scene, the noise in those areas became immediately apparent and we had to go back and tone it down. Watch that trailer in high res and see if some of the noise moving around in the background of some of those scenes doesn't jump out at you.

Link to comment
Share on other sites

Thank you guys for the responses, and David Mullen: I am in the process of purchasing a Red One so Its really good to know that pro DPs are using the Red One at higher than 320 ASA with such good looking results.

 

We will be mainly using the camera on our own indy projects, so really the only thing that worried us about the Red is the noise levels.

Link to comment
Share on other sites

As John Sprung mentioned, moving images should give an impression of smearing of noise and less perceptability. Additionally, I have an impression that Red does not do temporal filtering for noise removal before compression and that manifests itself in low light situations.

If they did it would be a reason for me to not use this camera. Temporal real time noise filtering in the master source, no thanks.

Link to comment
Share on other sites

I watched the trailer and noticed the noise too. Although, when I watched the trailer awhile back and wasn't looking for anything in particular, just looking to see what the red looked like on a big budget production, I didn't really notice it. I don't think the scene with the guy pointing was in that one though, I'm pretty sure I would have noticed it. I am also wondering what the compression to H.264 did to enhance the visibility of the noise. I also wonder if it's possible that the shot with the blond guy was flubbed in production and this was the best they could do to fix it...sometimes murphy sneaks up on you when you are tired, stressed, rushed and using gear you are not entirely familiar with yet...

 

now that I think about it. The over all look seems like they really pushed the midtones up a lot, like in the classroom scenes where you can see the noise in the wall. maybe they were going for a really sunny room look on a not so sunny day or in a room that never really get's that way to begin with. Once again it becomes a trade-off between stylization and technical perfection...

Edited by monday sunnlinn
Link to comment
Share on other sites

If they did it would be a reason for me to not use this camera. Temporal real time noise filtering in the master source, no thanks.

 

Temporal filtering is a well-established practise in camera industry. Since the first access point you would have to your "master source" will be after decompression in this case, the corresponding original compression will be relatively poor if done without temporal noise removal. A few things are industry norms. For e.g., if a camera does not have OLPF you need to wonder what was the reason? May be there was a reason. Similarly, absence of temporal filtering needs strong justification.

 

BTW, there are some powerful nonlinear ways of dealing with temporal noise that only impact the noise part and leave the signal part more or less unchanged under many circumstances, and of course not all circumstances. Perhaps your notions of noise removal are colored by spatial noise removal where the typical linear methods employed effect both signal and noise.

Edited by DJ Joofa
Link to comment
Share on other sites

  • Premium Member
Temporal filtering is a well-established practise in camera industry. Since the first access point you would have to your "master source" will be after decompression in this case, the corresponding original compression will be relatively poor if done without temporal noise removal.

I sure know what an OLPF does, but I'm not sure what the absence of temporal noise filtering would look like. Would it be possible to find some with/without examples we could look at?

 

 

Thanks --

 

 

 

 

-- J.S.

Link to comment
Share on other sites

I sure know what an OLPF does, but I'm not sure what the absence of temporal noise filtering would look like. Would it be possible to find some with/without examples we could look at?

 

Hi John, I really wish I could provide examples here. However, since they are all work, I can't post them to a public forum, or make them available publicly. But, we have battled temporal noise before compression extensively and our consensus is that it is really helpful if employed before compressing the signal. The type of compression also plays a part here, and certain schemes, for e.g., those that use motion prediction, really take advantage of good temporal filtering.

 

Turning on and off the temporal filtering resulted in noticeable visual difference in the cases we have dealt with at least. As I said before, if a camera manufacturer does not use it, then, that is okay as long as they provide good justification for why not.

Link to comment
Share on other sites

How do these cameras even know in real time what the correct motion from frame to frame is?

 

Hi Michel,

 

If you are talking about motion prediction schemes in realtime, then, previous and sometimes "future" frames are kept in memory for processing.

 

BTW, on the issue of realtime, just like many people on this and other forums have incorrect perception of the utility of compression, a myth has been propagated by certain quarters regarding what can be accomplished in realtime, for e.g., if deMosaicing (deBayering), color processing, etc., can be accomplished in hardware processing in realtime. From our own experiences we have successfully dealt with these issues.

Link to comment
Share on other sites

  • Premium Member
on the issue of realtime, just like many people on this and other forums have incorrect perception of the utility of compression, a myth has been propagated by certain quarters regarding what can be accomplished in realtime, for e.g., if deMosaicing (deBayering), color processing, etc., can be accomplished in hardware processing in realtime. From our own experiences we have successfully dealt with these issues.

 

So what are you saying -- that the myth is that real-time debayering and compression is just as good as what can be done in post, or that the myth is that it can't be as good?

Link to comment
Share on other sites

So what are you saying -- that the myth is that real-time debayering and compression is just as good as what can be done in post, or that the myth is that it can't be as good?

 

Hi David,

 

The myth is that certain incorrect notions about the capability of real-time implementation of certain types of processing have been introduced, sometimes because if a typical manufacture could not accomplish things the best way then it can't be done that way. While, it is true that many operations are difficult to do in realtime hardware vs. software implementation. However, for the two things you mentioned, viz., compression and deBayering, compression is already being done successfully in real-time by many cameras, including the Red One. As for deBayering, very good quality hardware accelerated deBayering can be done in real-time.

Link to comment
Share on other sites

Hi Michel,

If you are talking about motion prediction schemes in realtime, then, previous and sometimes "future" frames are kept in memory for processing.

That's not what I meant. If cameras do genuine temporal filtering to get rid of random noise they have to filter along motion trajectories or they are comparing apples and oranges and filtering artifacts are unavoidable. To do that in non real time is difficult enough. I don't see how this is supposed to work in real time and be effective and free of obvious artifacts and indeed genuine temporal filtering. ??

Link to comment
Share on other sites

That's not what I meant. If cameras do genuine temporal filtering to get rid of random noise they have to filter along motion trajectories or they are comparing apples and oranges and filtering artifacts are unavoidable.

 

Hi Michel,

 

Things are a little different in temporal domain. Using some nonlinear approaches, the signal can be made fixed-points (or root signals) of filtering operations so that mostly noise gets suppressed.

 

To do that in non real time is difficult enough. I don't see how this is supposed to work in real time and be effective and free of obvious artifacts and indeed genuine temporal filtering. ??

 

You are right that sometimes temporal filtering will smear/blurr image information. However, careful adaptive filtering ensures that enough information is still left in an image even after some slight blurring so you can still recover trajectories, etc.

Link to comment
Share on other sites

  • Premium Member
However, for the two things you mentioned, viz., compression and deBayering, compression is already being done successfully in real-time by many cameras, including the Red One. As for deBayering, very good quality hardware accelerated deBayering can be done in real-time.

Yes, bottom line it's a matter of how much time and money can be put into optimizing a highly specialized solution. Rob Hummel said that he had good real time de-Bayering at Dalsa, because he had programmers who knew how to do it.

 

 

 

-- J.S.

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...