Jump to content

Horror Lighting. Very low Key Inquiry


Recommended Posts

Team, How does one get something like this without the darkest part of the scene get artifacts? I am trying to copy this lighting for a short skit but the blacks are getting such horrible artifacts. I light and turn it down in post. Sorry for my english.

 

http://kissthemgoodbye.net/horrors/albums/theconjuring/The__Conjuring_2013_kissthemgoodbye_0406.jpg

Edited by Jae Solina
Link to comment
Share on other sites

  • Premium Member

What sort of artifacting? Noise?

 

If you underexpose an image and try to make it brighter in post, you're always going to be digging into the noise floor of the image - and there are very definite limits to how far you can go before the image starts to break down.

 

The key is to expose your image properly, expose it for how you want it to be.

 

On most of the modern digital cinema cameras, you can hold some usable, visible detail, down to about 5 stops under key. At 6 stops under you want to black, because it's seriously noisy down there.

 

For low-key scenes, if you want a face to be dark, but still clearly visible, I find 2-stops under is a good level to be at (you can crush it down further if need be, but you can read the expressions on your actors faces still). At three stops under, people are going to be straining to see detail, and at four stops under, there's virtually nothing left to see.

 

Light your low-key scenes with all of that in mind, and you shouldn't have any issues with noise, since you won't be pushing your image up from the depths.

Link to comment
Share on other sites

What sort of artifacting? Noise?

 

If you underexpose an image and try to make it brighter in post, you're always going to be digging into the noise floor of the image - and there are very definite limits to how far you can go before the image starts to break down.

 

The key is to expose your image properly, expose it for how you want it to be.

 

On most of the modern digital cinema cameras, you can hold some usable, visible detail, down to about 5 stops under key. At 6 stops under you want to black, because it's seriously noisy down there.

 

For low-key scenes, if you want a face to be dark, but still clearly visible, I find 2-stops under is a good level to be at (you can crush it down further if need be, but you can read the expressions on your actors faces still). At three stops under, people are going to be straining to see detail, and at four stops under, there's virtually nothing left to see.

 

Light your low-key scenes with all of that in mind, and you shouldn't have any issues with noise, since you won't be pushing your image up from the depths.

When referencing key stops do you consider Key at around 60-70IRE?

Link to comment
Share on other sites

  • Premium Member

When referencing key stops do you consider Key at around 60-70IRE?

I was referring to exposure readings from your light meter Jae. So if your key light is f/4 and you want to be 2 stops underexposed, you'd need to put your iris at f/8 (or f/4 with two stops of ND in front of it).

 

IRE isn't a good scale to learn these things by, as it varies from camera to camera depending on the particular gamma curve you're recording in.

Edited by Mark Kenfield
Link to comment
Share on other sites

I know for clean blacks rating the camera ISO one stop lower is going to yield a 6db improvement in the signal-to-noise ratio. So also if you expose at key and bring it down in post you will get improvement. Right?

 

Hence I'm wondering if DP's prefer to either 1.) actually record the way they want to image to look (ie. in-camera with say 2 stops underexposed from key light) which then means you're getting the camera rated signal-to-noise ratio, or 2.) shoot properly exposed for key and bring the image down in post which technically increases the signal-to-noise ratio and improves the noise/grain from the first option.

 

If the camera has an exposure index feature or loadable LUT's (where you would load a -1 or -2 stop LUT) then it's easy to work with seeing the image 2 stops under exposed whilst actually recording it at normal exposure, and later lowering it in post for that improved clean black.

 

Thoughts?

Link to comment
Share on other sites

  • Premium Member

I know for clean blacks rating the camera ISO one stop lower is going to yield a 6db improvement in the signal-to-noise ratio. So also if you expose at key and bring it down in post you will get improvement. Right?

 

Hence I'm wondering if DP's prefer to either 1.) actually record the way they want to image to look (ie. in-camera with say 2 stops underexposed from key light) which then means you're getting the camera rated signal-to-noise ratio, or 2.) shoot properly exposed for key and bring the image down in post which technically increases the signal-to-noise ratio and improves the noise/grain from the first option.

 

If the camera has an exposure index feature or loadable LUT's (where you would load a -1 or -2 stop LUT) then it's easy to work with seeing the image 2 stops under exposed whilst actually recording it at normal exposure, and later lowering it in post for that improved clean black.

 

Thoughts?

 

 

After having multiple low-key projects absolutely mangled in the grade (levels boosted WAY beyond the look the director and I had established on-set), nowadays I much prefer to play it safe, and put everything so far down in the toe, that they can't bring it back without drawing blood.

Link to comment
Share on other sites

I know for clean blacks rating the camera ISO one stop lower is going to yield a 6db improvement in the signal-to-noise ratio. So also if you expose at key and bring it down in post you will get improvement. Right?

 

Hence I'm wondering if DP's prefer to either 1.) actually record the way they want to image to look (ie. in-camera with say 2 stops underexposed from key light) which then means you're getting the camera rated signal-to-noise ratio, or 2.) shoot properly exposed for key and bring the image down in post which technically increases the signal-to-noise ratio and improves the noise/grain from the first option.

 

If the camera has an exposure index feature or loadable LUT's (where you would load a -1 or -2 stop LUT) then it's easy to work with seeing the image 2 stops under exposed whilst actually recording it at normal exposure, and later lowering it in post for that improved clean black.

 

Thoughts?

 

If if i were to shoot raw could could i change the ISO from 800 to 400 to lose 1 stop and 800 to 200 2 stops instead of using luts with -1 and -2? Since iso is just meta data i can just expose at 800 but change to 400 or 200 to see how it looks?

Link to comment
Share on other sites

 

After having multiple low-key projects absolutely mangled in the grade (levels boosted WAY beyond the look the director and I had established on-set), nowadays I much prefer to play it safe, and put everything so far down in the toe, that they can't bring it back without drawing blood.

I just hate how youtube kills lowlight videos. I get a lot of artifacts and blockiness when i upload.

Link to comment
Share on other sites

  • Premium Member

I just hate how youtube kills lowlight videos. I get a lot of artifacts and blockiness when i upload.

That's odd, usually high-key scenes with wide shots and lots of high contrast detail end up being the most difficult to compress, and thus end up with macroblocking unless you set a very high compression bitrate. Low-key scenes with lots of shadows and shallow depth-of-field should be reproduced accurately with a smaller bitrate. How are you compressing your videos before uploading them to YouTube?

 

Also, would you be able to post some frames of your ungraded shot straight from the camera and also your final graded shot? It's kind of hard to give specific advice without seeing exactly what went wrong.

 

Typically noise is the result of underexposure, usually when using high ISOs. It can be tricky because these factors are very much camera dependent in the digital era. Some cameras like the Blackmagic 4K do not like to be underexposed at all, so it's best to use 400ISO and expose brighter than you would like the final shot to look. And then lower the overall exposure in the color grade.

 

Other cameras like the Arri Alexa and Canon C300 can be underexposed several stops without a huge increase in noise, so you can go for the look you want to achieve in camera more closely.

Link to comment
Share on other sites

If if i were to shoot raw could could i change the ISO from 800 to 400 to lose 1 stop and 800 to 200 2 stops instead of using luts with -1 and -2? Since iso is just meta data i can just expose at 800 but change to 400 or 200 to see how it looks?

 

 

Yes if you shoot RAW on RED or BMD cameras you can adjust metadata later in post from 800 to 400 which is -1 stop, or 800 to 200 which is -2 stops, and get what you need.

 

As a side note; if you shoot non-RAW you can use plugins that adjust exposure in proper stops, or 3D LUTs that adjust exposure, or of course you can just use offset wheel in Davinci Resolve.

 

When you increase exposure in post it will always decrease the signal-to-noise ratio which means you increase noise. So as others have mentioned it is can be better to overexpose slightly (even with a low-key scene) than to underexpose it. My fear of lowering exposure in-camera by 1-2 stops is you have to be 100% sure your exposure is right on because if you need to increase exposure you are in trouble.

 

I don't think Blackmagic has any built in tools to allow you do preview 400 ISO while recording 800 ISO, so you can use an external monitor with a -1 stop exposure LUT loaded on it, or for a poor man's method you can put a 1 stop ND filter on your lens and set exposure and then before you are about to record take off the ND filter. This is a good trick until you forget to take the ND filter off before pressing record :)

Link to comment
Share on other sites

  • Premium Member

Well you certainly went LOW key! All very nicely put together, and very well shot by your missus Jae (send her my compliments). However I fear you guys actually went a little too far in the end. With the regular lighting in my room on, I basically couldn't make anything out in hallways of the house. I had to turn off all of the lights before I could actually discern most of the imagery.

Keeping average viewing conditions in mind is important for these things. Because even if you create something that looks great (as you have here), if people can't see it just because it's daytime, or there's a light on in the room - they're simply going to think you stuffed up.

 

I find False Colour monitoring the single best exposure tool there is for real low-key work. Once you've worked out (in Resolve or whatever grading software you use), where the bottom falls out your camera's image on the waveform (i.e. where it needs to be crushed to black), then you can calibrate separate colours to tell you what parts of the image will read black in the end, and which parts will show small amounts of shadow detail. It allows you the greatest precision you can get in terms of black levels.

Link to comment
Share on other sites

Well you certainly went LOW key! All very nicely put together, and very well shot by your missus Jae (send her my compliments). However I fear you guys actually went a little too far in the end. With the regular lighting in my room on, I basically couldn't make anything out in hallways of the house. I had to turn off all of the lights before I could actually discern most of the imagery.

 

Keeping average viewing conditions in mind is important for these things. Because even if you create something that looks great (as you have here), if people can't see it just because it's daytime, or there's a light on in the room - they're simply going to think you stuffed up.

 

I find False Colour monitoring the single best exposure tool there is for real low-key work. Once you've worked out (in Resolve or whatever grading software you use), where the bottom falls out your camera's image on the waveform (i.e. where it needs to be crushed to black), then you can calibrate separate colours to tell you what parts of the image will read black in the end, and which parts will show small amounts of shadow detail. It allows you the greatest precision you can get in terms of black levels.

Mark! Thank you very much for all the information. I love false color but unfortuately, it doesnt work correctly with my smallhd 502 monitor when lut is enabled. That's exactly what i thought too I had to turn off all the lights to watch it properly.

Link to comment
Share on other sites

  • Premium Member

if you mean those wall gradients by artefacts I don't think you can do much other than use a compression method for the end product that hides those as much as possible.

you can also tweak lighting, like use edge lights etc. for the door frames, hallway etc. to draw the space and try to avoid the dark subtle wall gradients by using lots of negative fill and lighting the subject much brighter so that you can "print down" in grading.

 

video cameras only have certain pretty much fixed amount of stops between noise floor and clipping point and the ISO rating you use just determines how those stops distribute compared to middle gray (for example, EI 800 for certain camera has 4 stops of usable shadow detail before noise floor and 4 stops of highlights before clipping point. if you rate it EI400 you will have 5 stops of shadow detail and 3 stops of highlight detail. that's how video cameras (digital cameras, raw cameras, however one wants to call them) generally behave and that also includes the Blackmagic cameras, RED cameras and Alexa.

 

I generally light by eye, verify with the monitor if needed and then fine tune the exposure according to waveforms or histogram if needed to make sure that all the necessary details are captured "technically correctly" from extreme shadows and highlights. ISO rating stays the same unless absolutely needed to be changed because of technical reasons. tweaking ISO with video cameras works just the opposite way than you usually want: high ISO has less shadow details and low ISO has less highlight detail. If you need to raise the signal to get the compression (for example S-Log) to work more efficiently on the important shadow details it may be useful to gain the signal in camera though

Link to comment
Share on other sites

Jae,

 

I'm shooting on the Blackmagic Pocket for almost 3 years now, what I found when shooting raw is that exposing for ISO 400 gives me the best results in most circumstances, Exposing for ISO 800 makes sense in bright / high key circumstances to protect highlights but only then.

For very low key / low light scenes, exposing for ISO 200 usually gave me the best results as for detail/noise/colors.

Edited by Michel Rabe
Link to comment
Share on other sites

I just hate how youtube kills lowlight videos. I get a lot of artifacts and blockiness when i upload.

try adding a good amount of grain. It should look too much in your render but Youtube/Vimeo washes away a lot of it it.

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...