Hey guys, when exposing RAW I expose to the right to try and eliminate some noise and then bring it down in post. Does this strategy make sense for the 4K2HD? Even though its still S-LOG2 and 2000 ISO it seems at first glance that the ProRes is a lot less noisy, should I still expose to the right or will this lead to blown highlights?
Exposing 4K2HD ProRes
- Thread starter ilnow
- Start date
Clermond
Veteran
I second this experience. In my first test I exposed as I would have exposed RAW and the highlights in the white fur clipped already. I am sure I could have recovered them in RAW.
Mitch Gross
Veteran
4K2HD in SLOG2 is a very rich image but it is still video. The reduction from 4K to HD hides a lot of noise and other issues within the signal. Try not to clip your highlights. You may find that 2000ISO works better for you here.
cuervo
Veteran
I'm finding the false color indicator pretty invaluable in determining exposure. Still experimenting with the best settings for this latest firmware release, but, it seems the best exposure index is to set the green to somewhere between 35 and 44%, then expose for the key features to be green. That usually leaves some headroom in the hilites without too much noise in the blacks.
Thanks, Cuervo. I think that's pretty close to the standard setting for green (mine is still at default and I see green is set from 38-45%) In my first few test shots this worked quite well.
Is anyone using the false color with different settings? It'd be great to hear other peoples' experiences with exposure (especially if they are ETTR in a more advanced way by setting a slightly higher exposure using the false color feature).
Colinelves
Well-known member
I Haven't shot much with 4KtoHD, but what I did shoot I exposed to the right and it came down fine, no problems at all (although I wasn't shooting people). Although I made sure nothing was clipping by setting my zebras to 108% and making sure nothing activated them (109% is too high, it never comes on).
As Mitch says, the downconvert may hide some of the noise, making ISO 2000 more acceptable (although personally I think it is still a little noisy). If you do go down that route, you can expose using the LUT and zebras at 70% (for white skin tones anyway) or just making sure they look right and relying on the extra slog2 headroom to hold the highlights (flipping back to slog2 to check if you're worried).
As Mitch says, the downconvert may hide some of the noise, making ISO 2000 more acceptable (although personally I think it is still a little noisy). If you do go down that route, you can expose using the LUT and zebras at 70% (for white skin tones anyway) or just making sure they look right and relying on the extra slog2 headroom to hold the highlights (flipping back to slog2 to check if you're worried).
paulcurtis
Veteran
I've been looking at the linear files and i think it's important to remember that with a linear format the bulk of the data is at the lower end. In linear the lower shadow stops have less detail and the detail (or the number of levels recorded) increase as you go up in brightness until you get to the final stop which holds a massive amount of unneeded detail.
You can see this is you open one of the DNGs and look at the levels, all the data is on the left. That's how linear works and it certainly looks like these files act like that too.
This is why ETTR is often mentioned as the best approach but that's fine for Stills but in a sequence that can very quickly become problematic, you can't be exposing shot by shot that would be a nightmare to grade.
The best solution is depending on your scene decide how the scene should be shot to capture the range you want. Use the false colour (the green allows for a low to hi) to peg something like 18% grey. Outside you might want to skew your exposure to highlights and therefore put 18% grey at 32% which is the recommended value for SLog2. Inside though you may choose to sacrifice some headroom and exposure that 18% at 40 or 50%. Then in post you would bring down the overall exposure to compensate. This would result in more detail allocated to the shadow areas.
I think you should go and shoot tests for yourself to see how this works.
Remember though that 'bring it down' in post only really works for RAW and Slog based files, a REC709 file has a different curve baked in and unless you reverse that curve lowering the values won't be photographically correct.
cheers
Paul
You can see this is you open one of the DNGs and look at the levels, all the data is on the left. That's how linear works and it certainly looks like these files act like that too.
This is why ETTR is often mentioned as the best approach but that's fine for Stills but in a sequence that can very quickly become problematic, you can't be exposing shot by shot that would be a nightmare to grade.
The best solution is depending on your scene decide how the scene should be shot to capture the range you want. Use the false colour (the green allows for a low to hi) to peg something like 18% grey. Outside you might want to skew your exposure to highlights and therefore put 18% grey at 32% which is the recommended value for SLog2. Inside though you may choose to sacrifice some headroom and exposure that 18% at 40 or 50%. Then in post you would bring down the overall exposure to compensate. This would result in more detail allocated to the shadow areas.
I think you should go and shoot tests for yourself to see how this works.
Remember though that 'bring it down' in post only really works for RAW and Slog based files, a REC709 file has a different curve baked in and unless you reverse that curve lowering the values won't be photographically correct.
cheers
Paul
Last edited: