Going insane with noise

[CinematicFanatic]

Yeah I color graded it and I can get it looking decent. But I just feel like for a $7000 camera the picture should come out somewhat noiseless in broad daylight. I need to keep experimenting. What would you set the zebras to?

I have to somewhat agree with you.

I found my FS100 gave a cleaner, crisper image straight out of the camera with a PP that I found on Philip Blooms website.

The only reason I honestly own my FS700 is because of the slo mo and built in ND's. Or give me the FS100 any day of the weak.

 

[Neutral]

I have to somewhat agree with you.

I found my FS100 gave a cleaner, crisper image straight out of the camera with a PP that I found on Philip Blooms website.

The only reason I honestly own my FS700 is because of the slo mo and built in ND's. Or give me the FS100 any day of the weak.

That just can't be though. I've seen great footage shot on it. I've read a million reviews on it and I've never seen anyone critiquing the footage/noise/artifacts or anything. It's in fact the opposite. So many people praised the image quality and I graded fs700 footage before I got one which was the deciding factor, I loved it. But for some reason I can't seem to get what I even consider "decent" picture out of it.

 

[waefre]

You are expecting too much from the dynamic range. In your test shot you have a building in the background getting blasted with full sun but not blown out while your subject is in a dark alley in a black shirt. Clearly You are either stopped or NDd down to have an underexposed subject which is leading to the noise. Yes for the sake of your highlights you underexpose, but that doesn't mean you don't sacrifice something. Do another test and bring the exposure on the girls skin to around 65% using Zebra or false color or whatever you have at your disposal. Yes you will lose details as the building in the back blows out but the blacks and shadows in the ally will be much cleaner.

 

[Neutral]

You are expecting too much from the dynamic range. In your test shot you have a building in the background getting blasted with full sun but not blown out while your subject is in a dark alley in a black shirt. Clearly You are either stopped or NDd down to have an underexposed subject which is leading to the noise. Yes for the sake of your highlights you underexpose, but that doesn't mean you don't sacrifice something. Do another test and bring the exposure on the girls skin to around 65% using Zebra or false color or whatever you have at your disposal. Yes you will lose details as the building in the back blows out but the blacks and shadows in the ally will be much cleaner.

Thanks I'll definitely try this out!

 

[paulears]

I must admit to noticing at the trade shows that what comes out of the increasingly higher quality cameras, to the eye, is not that nice looking. What seems to happen is that the data is detailed enough to allow the end product to be stunning, but in the recorded form, it's a bit uninspiring. Zooming in the sample to pixel level is quite impressive compared with some of my own stuff at lower definition, but it just looks murky. I'd love to see the same settings on a colourful and vibrant scene for comparison.

 

[Colinelves]

A few things strike me as wrong here. Firstly you mentioned that you're exposing so the historgramme just hits the yellow line - that's potentially a problem. The yellow line doesn't represent the clip point of the camera, it represents the level of the zebras, if you're hitting that and stopping you're likely not using the full highlight range of the camera (unless zebras are set to 109%).

Generally I would say exposing by the histogram is a bit crazy anyway - who cares whether or not a bit of the image is at white clip and another is at black clip? The important thing is exposing your main subject properly. If you're using regular (non-slog2) profiles, you should be getting those white skin tones at a little under 70 IRE. Set Zebras to 70 and expose the skin highlights so that they're just getting the zebras. The internal codec really isn't too grading friendly so you really want to nail your main exposure and image in camera, not leave it to post as it will make you cry.

Finally I'd also point out that a big part of what you're seeing is the extended dynamic range of the camera compared to the DSLRs you're used to. Sure those cameras have clean blacks, but they get that by choking off the shadows (restricting the DR). They remap the exposure range of the camera to a limited range in the highlights. If you shoot cinestyle or Flaat profiles you'll see more range - at the cost of extra noise in the shadows (as you're pulling the extra range out of those noisy shadows. You would get clean blacks on the FS700 too if you overexposed by 2 stops then brought the exposure down in post (you'd also lose two stops of latitude that way).

I was shooting on the C300 (cinelog) and 7D this week, the 7D certainly had nice clean blacks, but it also blew the highlights more than the C300, which had noisy blacks (noisy enough to notice anyway) - that's the trade off with digital cameras these days: range vs noise. You can get clean footage, but you have to lose some of the range to get it.

To be honest, that's essentially what the gamma profiles of the FS700 are doing: Cine 2 (base ISO 320) will give you the cleanest shadows, but least DR, slog2 (base ISO 2000) gives you max DR but noisy footage. It's the same chip so there's nothing magical going on between the two settings, its simply a question of how the scene is interpreted further down the pipeline: Slog2 encourages you to stop down (because the 2000 ISO makes it look over exposed otherwise) to prevent highlight over exposure then gives you extra stops by stretching the shadows. Cine 2 encourages you to open up the stop: overexposing the highlights and bringing the shadows out of the noise floor and then it pulls the whole exposure level down. You lose those highlight stops but get cleaner shadows.

Try shooting the exact same scene, exact same exposure settings (f stop, shutter, ND, lowest ISO setting) but change the profiles over - then compare them in post, grading each to try and match (maybe do it twice, once exposed 'correctly' for slog2, once for cine 2). You may then get a bit of an idea of what they're doing.