noise - compare

nobbystylus

Well-known member
Just did a multi cam shoot, 3 cams - Sony Z1 shooting HDV (mostly wide open, but some shots stopped down to 4.0), HV30 shooting HDV (in cinemode, exposure down to about -4) and GH1 shooting full HD 1080 25p, 1/50, 400iso (with stock lens wide open at f4.0). Clearly the GH1 has the nicer visual image. But i'm so disturbed by the noise and banding in the GH1s image that is only just slightly underexposed that basically i'm having to cut around the GH1 stuff almost entirely. This is certainly a result of being a bit underexposed, but if i upped the iso ( to expose correctly) it got much much worse.. Also, the noise is so much less of an issue on the HV30 (in cinemode) that the GH1s buzzy wavey lined noisey blacks are almost an embarrassment. Yes i can denoise, yes the image is nicer, but its at 400 iso and it looks soooooo sooo messy. Buzzy buzzy noise all the way through plus that awful banding. The Z1 is clean, the HV30 is a bit noisey, but the GH1 has banding, low light noise and serious amounts of zig zaggy / wiggly noise AT 400 ISO!!!

This is using the ipowell hack also, so can't be a codec issue.

I'm wondering if i can really get a clean low light image with dark backgrounds with this camera? I've shot stuff at Iso 100 and its still had noise... maybe its a faulty unit?
 
Last edited:
I'm wondering if i can really get a clean low light image with dark backgrounds with this camera? I've shot stuff at Iso 100 and its still had noise... maybe its a faulty unit?

Any chance you can post some frame grabs? It's difficult to give feedback without seeing how bad the problem is..
 
frame grabs don't really do it as its buzzing noise which generally shows up in moving footage more.. i'll try and post some clips once i've digitized stuff in..
 
it's strange the noise seems to appear only in certain things. i've shot in super underexposed night outdoors and ZERO noise. then i've shot in medium underexposed daylit room at like 320 ISO and there's banding so atrocious that the footage is utterly unusable/worthless.
 
What about your movie mode and contrast?

More contrast -> less banding

Movie mode (from more banding to less)
Nostalgic -> Soft -> Standar -> Natural -> Dynamic - Vibrant
 
What about your movie mode and contrast?

More contrast -> less banding

Movie mode (from more banding to less)
Nostalgic -> Soft -> Standar -> Natural -> Dynamic - Vibrant

I don't believe this to be true at all, I really think it solely depends on the exposure/lighting of the shot. I've done side by sides between Nostalgic + Dynamic and got MORE banding with Dynamic....but again I think it depends on what the composition is. Is the shot of a white wall, or is it of trees+shrubery??? these factors completely play a roll- there is no magic formula.
The formula is ever changing and requires the DP or 'user' to make adjustments accordingly to reduce unwanted affects.
 
it's strange the noise seems to appear only in certain things. i've shot in super underexposed night outdoors and ZERO noise. then i've shot in medium underexposed daylit room at like 320 ISO and there's banding so atrocious that the footage is utterly unusable/worthless.

agreed
 
Which Lpowell hack? MJPEG or AVCHD? Sorry but your statements lack the detail we need to really help.

I've found that in certain conditions that some of the different hack settings will break up under certain conditions, sometimes in strange ways. We need your information to pinpoint possible specific failures in specific situations.

Please state all of your settings and give screenshots of the problem so we can help.
 
I'm using the ipowell AVCHD stability hack with 1080p 25p footage using the stock lens. The scene was a 'studio' theatre space with performers, generally well lit, by a DP with two 5k side lights and a bounced blonde as a fill/key. In cam it looks great, but its very very noisy when looked at on my computer. I was at iso400 for most of it, but did have to poke the iso up to around 800 on occasions. I'll have to run the whole lot through a denoiser to make it usable, probably meaning a 24 hour render. Its not just the noise, but the banding shows up.

The space was lit for the Z1 (as the wide master), but i thought my GH1 (even with the stock lens) would still be better in low light, but the buzzy moving noise, fixed pattern noise and banding was a real bummer....

I was using Smooth mode with all settings set to 0, apart from contrast which was -1.

The HV30 stuff looks great ;-)
 
I don't believe this to be true at all

Well, you're right. There is no magic formula; it is just MY OWN formula. I use it in low light situations, and it works quite well for me. I'm not trying to stablish a sacred holy set of rules for avoiding banding; it is just my plain and simple aproximation to the banding problem.

Let's say that my formula is Newtonian. Yes, it's not true, an Einsteinian formula would be better, but mine works quite well (for me) in some scenarios.
 
I personally think that saying any footage with banding must be underexposed is completely misguided, and it's repeated a lot on this forum. On the other side, saying that increasing contrast will reduce banding is true... but only because you're crushing the entire range that banding occurs in to black. Frankly, that's very undesirable if you want a decent dynamic range out of your image.

Banding is an inherent trait of the camera. I also don't think it's accurate when people say they "get banding" in some situations and other people say they "don't get banding." The camera has banding and FPN, always, period. It's there, and how much of it is particularly visible and obvious is the only thing that changes. Increasing ISO will amplify it, but it is there at base ISO.

Banding is most obvious in the 5%-25% range of the histogram (in other words, in shadows). Below this range the camera tends to crush towards black, so it's less visible. Above this range and the signal is strong enough to completely overcome the noise. If you underexpose, yes, much of your image is going to be in this range, so you'll get a lot of obvious banding.

However, unless you're shooting completely lifeless images with less than a couple stops of dynamic range in them, some part of your image will likely ALWAYS contain some information in this range (in other words, you probably have shadows or darker areas in your picture SOMEWHERE, regardless of how properly exposed you are). Exposing to-the-right will likely push most of your image past 25% and make it harder to see banding throughout the majority of your image, but it'll still be there in the shadows.

Sometimes though, ETTR will actually push your image INTO this noisy range, where it was previously crushed to black. This is often the case with night scenes where big parts of the frame have no lighting on them. So increasing your exposure in these cases will actually make banding much worse... although it's still probably prefable to do that while shooting, then crush those areas back down during post.

Unfortunately the recurring theme is that there is no way to completely avoid banding short of shooting a scene with completely static lighting. But if you can avoid large patches of 5-25% luminosity, it will be far less noticeable (it's hard to pick up on it in small areas of shadows... but it is there!)

Adjusting exposure will adjust where exactly that 5-25% range sits in relation to your scene, but it is wrong to say that properly-exposed scenes will not have banding. Almost every scene contains some amount of area with this luminosity and almost every scene therefore contains some amount of visible banding. I wish people would stop using the "you're underexposed" blanket statement. It can help IN CERTAIN cases, but it is an oversimplification and not getting at the real issue. It will also hurt in certain cases.
 
Last edited:
The zig-zag noise sounds strange... I've never seen that before, only the vertical bands. There might be something wrong with this particular camera.
 
I personally think that saying any footage with banding must be underexposed is completely misguided, and it's repeated a lot on this forum. On the other side, saying that increasing contrast will reduce banding is true... but only because you're crushing the entire range that banding occurs in to black. Frankly, that's very undesirable if you want a decent dynamic range out of your image.

Banding is an inherent trait of the camera. I also don't think it's accurate when people say they "get banding" in some situations and other people say they "don't get banding." The camera has banding and FPN, always, period. It's there, and how much of it is particularly visible and obvious is the only thing that changes. Increasing ISO will amplify it, but it is there at base ISO.

Banding is most obvious in the 5%-25% range of the histogram (in other words, in shadows). Below this range the camera tends to crush towards black, so it's less visible. Above this range and the signal is strong enough to completely overcome the noise. If you underexpose, yes, much of your image is going to be in this range, so you'll get a lot of obvious banding.

However, unless you're shooting completely lifeless images with less than a couple stops of dynamic range in them, some part of your image will likely ALWAYS contain some information in this range (in other words, you probably have shadows or darker areas in your picture SOMEWHERE, regardless of how properly exposed you are). Exposing to-the-right will likely push most of your image past 25% and make it harder to see banding throughout the majority of your image, but it'll still be there in the shadows.

Sometimes though, ETTR will actually push your image INTO this noisy range, where it was previously crushed to black. This is often the case with night scenes where big parts of the frame have no lighting on them. So increasing your exposure in these cases will actually make banding much worse... although it's still probably prefable to do that while shooting, then crush those areas back down during post.

Unfortunately the recurring theme is that there is no way to completely avoid banding short of shooting a scene with completely static lighting. But if you can avoid large patches of 5-25% luminosity, it will be far less noticeable (it's hard to pick up on it in small areas of shadows... but it is there!)

Adjusting exposure will adjust where exactly that 5-25% range sits in relation to your scene, but it is wrong to say that properly-exposed scenes will not have banding. Almost every scene contains some amount of area with this luminosity and almost every scene therefore contains some amount of visible banding. I wish people would stop using the "you're underexposed" blanket statement. It can help IN CERTAIN cases, but it is an oversimplification and not getting at the real issue. It will also hurt in certain cases.

I don't know how qualified you are to make statements re the camera - but what you just said there is the best fitting explanation re banding I have seen yet (and I have read a LOT of threads on the issue.)

What you are saying fits exactly with my experiences - anyone care to disagree?
 
I think Dasbin means the gradient lines inherent to the compression algorithm, not color banding or FPN?

FPN is always present and looks like faint dark checkerboard pattern across whole frame.

Color banding looks like purple or yellow streaks that somewhat follow the same pattern as FPN but is only present in -2.5 or less underexposed video. Sometimes it's all over, sometimes it's just in certain areas of the frame. The more contrast in the picture, the less banding problem you have generally. This is why sometimes it happens and sometimes it doesn't.

Gradient lines occur on surfaces that are smooth and uniform in color/texture and generally follow the radius of light-to-dark transitions. This seems to happen regardless of the above problems and regardless of bitrates.
 
I personally think that saying any footage with banding must be underexposed is completely misguided, and it's repeated a lot on this forum. On the other side, saying that increasing contrast will reduce banding is true... but only because you're crushing the entire range that banding occurs in to black. Frankly, that's very undesirable if you want a decent dynamic range out of your image.

That's why I told him to contrast as much as possible, and not just to increase it to the max (+2). Yes, using nostalgic and lowest contrast settings, would be nice. But in low light conditions, this means visible banding. With vibrant movie mode and contrast +2, banding is much less visible. The point is to find the best setting to achieve as few banding possible with the lowest contrast possible. Nature movie mode with contrast set to 0 gives quite good results for me. But in some cases I need to boost contrast and I use dynamic or even vibrant with contrast set to +2. What I will not use never again in low light is nostalgic or soft movie modes with -2 contrast. That's crazy.

Increasing the contrast in post may help, sure. But remember that the picture has been compressed before post, so you have more "decent dynamic range", but a picture full of compressed noise.

Anyway, I'm 100% agree with you explanation of the banding issue.
 
Last edited:
I don't know how qualified you are to make statements re the camera - but what you just said there is the best fitting explanation re banding I have seen yet (and I have read a LOT of threads on the issue.)

What you are saying fits exactly with my experiences - anyone care to disagree?
That is a very interesting explanation dasbin. I have never personally seen banding in video footage (probably because most of it I shoot outside in broad daylight), however I have once accidentally shot some stills in difficult light in JPG mode with Intelligent Exposure ON and have seen the dreaded banding. RAW files from the same session did not have the same problem. It would be interesting to set up some tests to see if certain image processing routines really emphasize banding and FPN and which settings control those routines.
 
Increasing the contrast in post may help, sure. But remember that the picture has been compressed before post, so you have more "decent dynamic range", but a picture full of compressed noise.

That's an excellent point. I'm primarily a stills shooter who typically does crazy things to get max dynamic range and least noise out of the sensor (such as shooting in a forced UniWB), so the fact that we are not talking about RAW footage sometimes escapes my mind in these discussions. If you're absolutely positive that you can live with the crushed shadows, doing it in-camera is probably going to retain the most detail. So I'm back with you on this.
 
Last edited:
Back
Top