Need beginners help for low light shoot

[masterzor]

Lol, sorry to bother everyone. I Just bpught a DSMC2 Dragon-x, and i want to test how to shoot low light, but i am an absolute beginner. Youtube doesnt really nail things for a beginner in Cinematography with the Red. I heard the Dragon-x could be a wrong choice for low light. But i didnt know the threshold.

Can someone givecme tips on settings and advice on:

What are the settings to improve low light with no noise. Between aperture, ISO ?

I dont mind being pointed in the right direction to do a little homework. But i am half handicap with Toxic Encephalopathy , and i am avoiding being overwhelmed.

So far i ordered a low light OLPF and i got a Canon 24mm 1.4L ii, to aid me.

 

[Patrick Tresch]

You are going to be fine with the Dragon. I love the noise of this sensor.

- You need to be sure to be calibrated for the right sensor temp (try to calibrate at high temp. Low temp will give you only more sound noise from the fans. If you change FPS you will also have to have a sensor temp calibration preset for it.

- You need to lower your compression ratio. It's better to have 1:6 instead of 1:10 as noise is considered as detail and increasing compression will make the noise look muddy.

- Test your Iso settings so you know what is ok for you. Noise floor is a very subjective opinion even if there are some technical measurement. I would be ok to shoot your camera at 3200 iso. Test it at 6400 to see if it's still ok for you (with and without denoise).

Remember increasing your iso value will protect your highlights. The trade off is noise floor but debayering at full res will help to clean it further.

Have fun.

Patrick

 

[M Harvey]

All great tips from @Patrick Tresch.

The other broad subject to think about for low light cinematography is your overall lighting approach. ISO and lens aperture are about adjusting the overall illumination level in your scene-- to simplify, they affect both highlights and shadows the same. So you also want to think about where the light is coming from (whether you're only using what's already there, like a lamp or a window, or introducing your own artificial light) and where you need it to fall in order to get the image you want.

Watch some low light scenes you like and notice where the light falls. For most movies or TV shows or music videos, shot on the very same camera you're using or similar, they didn't crank up the ISO to 3200-- even in scenes where they weren't using big expensive movie lights. Remember that you don't necessarily need to see EVERYTHING-- maybe a figure can be somewhat dark in front, and you just need to position them relative to a window in the background so that they have an edge of light that separates them a bit from the dark background.

This is a subject that can take years to learn and invites creative solutions. So have a good time learning and try things out!

 

[Les Hillis]

You need to shoot some tests to see for yourself what degree of noise you find acceptable.

The easiest one I can think of is to set up or find a controlled lighting and subject situation, set a base or middle f-stop/ISO combination (eg. f/5.6-ISO800), adjust the light to get a normal exposure for that f/5.6-ISO800 combination, then just open up and close down the lens one stop at a time without changing the light or ISO. Then take those deliberately under and over exposed shots into Redcine-X or Resolve or whatever, and in the R3D RAW settings, change the ISO to normalize each of the under and overexposed shots to the same normalized exposure, as described below (using ISO800 as the middle ISO and f/5.6 as the middle f/stop) -

Set lens to f/1.4 --- 4 stops over ------- Adjust to ISO50+- in post to normalize exposure
Set lens to f/2 ----- 3 stops over ------- Adjust to ISO100+- in post to normalize exposure
Set lens to f/2.8 --- 2 stops over ------- Adjust to ISO200+- in post to normalize exposure
Set lens to f/4 ----- 1 stop over -------- Adjust to ISO400+- in post to normalize exposure
Set lens to f/5.6 --- Normal Exposure - ISO800 no adjustment in post needed to normalize exposure
Set lens to f/8 ----- 1 stop under ------ Adjust to ISO1600+- in post to normalize exposure
Set lens to f/11 ---- 2 stops under ----- Adjust to ISO3200+- in post to normalize exposure
Set lens to f/16 ---- 3 stops under ----- Adjust to ISO6400+- in post to normalize exposure
Set lens to f/22 ---- 4 stops under ----- Adjust to ISO12800+- in post to normalize exposure

Looking at the adjusted clips will give you a basic idea of how noisy the camera gets when underexposed as well as how much you can reduce the appearance of noise from normal levels when overexposing as well.

You can also use the clips to check out how noise reduction, lift/gamma/gain/contrast/saturation and other post adjustements can influence the appearance of noise.

There are better and more accurate ways of testing for noise as it relates to exposure, but what I've described above is something you should be able to shoot for yourself at any time and get some immediate results from if you're just starting out.

 

[Les Hillis]

Another fairly easy way to test for noise is to change the shutter angle/speed in-camera per shot instead of the lens aperture. The advantage of doing the test that way is that it eliminates the optical and exposure variables that come with changing the exposure via the lens aperture.

You can use the exact lighting setup as previously (to get a normal exposure with ISO800/f.5.6), but because there is more leeway for changing the shutter angle towards under-exposure vs. over-exposure, compared to adjusting the exposure with the lens aperture, you need to use a wider aperture for your standard middle exposure, in this case going from an f/5.6 as your middle exposure, to a constant f/2 for all shots.

Set Camera to ISO800/Lens to f/2 then -

Change Shutter to 360°(1/24sec) Shutter Angle ----- 4 stops over ------ Adjust to ISO50+- in post to normalize exposure
Change Shutter to 180°(1/48sec) Shutter Angle ----- 3 stops over ------ Adjust to ISO100+- in post to normalize exposure
Change Shutter to 90°(1/96sec) Shutter Angle ------ 2 stops over ------ Adjust to ISO200+- in post to normalize exposure
Change Shutter to 45°(1/192sec) Shutter Angle ----- 1 stop over ------- Adjust to ISO400+- in post to normalize exposure
Change Shutter to 22.5°(1/384sec) Shutter Angle --- Normal exposure --- No adjustment in post needed to normalize exposure
Change Shutter to 11.3°(1/765sec) Shutter Angle --- 1 stop under ------ Adjust to ISO1600+- in post to normalize exposure
Change Shutter to 5.6°(1/1543sec) Shutter Angle --- 2 stops under ----- Adjust to ISO3200+- in post to normalize exposure
Change Shutter to 2.8°(1/3086sec) Shutter Angle --- 3 stops under ----- Adjust to ISO6400+- in post to normalize exposure
Change Shutter to 1.4°(1/6171sec) Shutter Angle --- 4 stops under ----- Adjust to ISO12800+- in post to normalize exposure

Of course you can just change the lighting too, measuring it with a light-meter or using the in-camera exposure tools to see what the noise is like, although that itself can introduce variables into the test, coming from the quality of light itself or it's effect on the subject and surroundings as the light intensity is increased and decreased.

As M Harvey mentioned above though, eventually it comes down to the subject-matter you're filming and how you choose to shoot it. There's a diference between shooting something that ends up looking like it was low-lit, and trying to shoot in an underlit way that leaves you just trying to salvage useable footage. It can also change shot to shot, with some subject-matter looking fine with crazy levels of post-boosting, while other subjects just fall apart even with minimal correction. The basic problem comes with trying to make something that was actually dark look like it wasn't. Better to light it how you want it to look, or shoot it for what it is and present it as that.

I will say though, imo, if you find your lenses f/1.4 image-quality is acceptable, and you don't mind the level of noise in Dragon footage shot at ISO1600, you've got a pretty big range of exposure to work with and any noise you do find will look pretty natural and organic .

 

[Les Hillis]

Here are some shots from a test done using the shutter-angle method mentioned above -

It was shot on a Scarlet Dragon that only does 6K at 12fps, so the noise is a bit more erractic than it would be at 24fps. The camera was fitted with the Skin-Tone Highlight OLPF too, so there's also a comparative loss in low-light performance from that. As I understand it the DSMC2 Dragon-X also has a slightly improved performance over the original Dragon Camera's, but I think it's a fair enough comparison to make to give an idea of what the Dragon sensor noise can look like.

The test isn't showing or suggesting any particular exposure settings, it just shows one middle exposure with a certain amount of noise, with two noisier more underexposed shots preceding it, then two cleaner more overexposed shots following it. The exact same frame sequence is then repeated, but with every shot boosted again in post by about another stop.

The colour chart on the left is lit one stop lighter than the middle chart, and the chart on the right is lit one stop darker than the middle chart, just to show some more exposure variance within the one frame.

The camera was set to 4:1 RedCode and was black-shaded/calibrated as necessary for every change in exposure and kept within the indicated temp and exposure range while shooting.

.............................................

Here's another quick test showing some Dragon noise in a low-light scenario. The noise is still visible when viewed at 4K but just gets blurred at the lower resolutions -

Has been interesting re-looking at this stuff.

 

[masterzor]

I appreciate the feedback. I will go out and choose an area and change those settings 1 at a time to see the performance. I bought the Dragon-x because the Gemini was $2000 more than my affordability at the time.

And i gambled that relatively speaking the Dragon-x couldnt be bad at low light, if i knew what i was doing. Now i just need to know what i am doing .

I bought it for the immediate goal of shooting a live band in a bar (low light). But knew i will shoot out doors and with controlled light, later.

Also i am trying test the crap out of it properly , in order to make sure everything works fine , before Ebay 30 days is over.

 

[masterzor]

I had this notions that just because the Gemini was the low light King, doesnt mean the Dragon-x cant do low light to a close 3rd and coupled with the notion that in lots of light , or enough light, the 6K Dragon-X can be better than 5K Gemini.

 

[Les Hillis]

Sounds like you've got a good grip on the situation.

I think you'll be fine as long as you keep what you want to see clearly and without noise properly exposed (you can still darken it down in the grade to look low-lit) and are willing to let everything below that fall into darkness or get crushed to black.

It's a just a matter of deciding how noisy you want your 'normal' exposure to be (my personal preference is for no more than 2 stops max), and not falling into the trap of actually underexposing everything and trying to pull it all up in post.

You don't need a hell of a lot of actual light to get a clean exposure with the Dragon sensor shooting at ISO800 and using an f/4 or wider aperture. In fact, you could spend just as much time trying not to clip your highlights. The Traffic Lights on the Histogram and the Gioscope Tool are a good quick way to check in-camera the exposure levels of objects in frame as you're shooting.

 

[masterzor]

Thanks for that reminder, about the traffic lights. I will keep an eye on that.

 

[alpharion]

The article below is from Dominic Ramirez and Jason Bowdach , i found it useful, so shared:
DR:
I own a Scarlet and I understand your pain. I've learned to tame the beast through extensive testing and frustration.

Temp and fan control are more critical than iso setting
Set the fan to a target tempeture of 64°
Never use the auto or manual settings
To calibrate run the camera to get it to a stable temp.
Put the body cap on and put the camera in a darkened area
Record for 8-10 min
Format media
Then BLACKSHADE

You should be able to shoot up to 1600 with no issue. Daylight is always better since the sensor is native to about 5000k. Going lower introduces noise in the blue channel.

If while shooting your TE light is yellow or red, stop and Black shade or deal with the noise the hard way.

Nuke can clean up the noise from a bad calibration. I haven't tried Neat but it sounds like it might work. Resolve, and AE NR don't cut it in this case, lumpy.

JB
Avoiding noise is a lot more complicated than adjusting ISO to lowest possible. You may have gotten less noise using the native ISO of the camera, which I believe is 800asa (someone correct me who works w more RED footage if Im mistaken), even if it is higher compared to 320. There could be an entire book on this subject, but it also depends which RED you shot with (Scarlet, MX, One, Dragon etc), and what type\level of lighting you used, what frame rate, and several other factors. I'm going to guess the sensor is a little starved for lighting, causing some sensor noise. Not much you can do besides try and crush it if its in the shadows, and gently noise reduce if its else where. Best of luck!

 

[Les Hillis]

Not sure I totally agree with a lot of that.

Blackshading and keeping things in the proper temperature range is important, and I agree that trying to keep your exposures within ISO1600 is a good idea if you want to keep noise under control, but there are different valid reasons for using the Auto and Manual fan settings.

Keeping the T/E light in the Green doesn't always have to be the highest priority either, though imo it does become more important if you are shooting in low-light, as the side-effects of not staying in the Green are more noticeable in underexposed shots.

Here's an example comparing the visual effect of shooting within the Green/Yellow/Red T/E temperature ranges. This test was done starting out with the camera temperature too low for the calibration, then letting it warm up and become too hot, then I just cut out the Green/Yellow/Red parts and edited them together for a direct comparison -

In this particular example, the results of going out of calibration are there but not really obvious. In other cases it would be more noticeable, especially if going even further into the Red.

I'd add that the whole blackshading and staying in calibration range thing is just as much about maintaining colour consistency and accuracy than avoiding noise, although the two are related.

Something else to factor in when dealing with exposure, noise, calibration and fan control matters, is the duration of your shots. It's one thing to set things up perfectly in-camera for one scenario where you're just rolling for a few seconds at a time, but it can make a difference if you're rolling for minutes at a time. It can be a bit of a balancing act to keep everything wthin your chosen parameters, sometimes requiring changing settings even as you're shooting instead of just setting and forgetting and hoping for the best.

Hope it's helpful, but also just mentioning all this as a way of making sense of it for myself.

 

[astra2019]

Here are some shots from a test done using the shutter-angle method mentioned above -

It was shot on a Scarlet Dragon that only does 6K at 12fps, so the noise is a bit more erractic than it would be at 24fps. The camera was fitted with the Skin-Tone Highlight OLPF too, so there's also a comparative loss in low-light performance from that. As I understand it the DSMC2 Dragon-X also has a slightly improved performance over the original Dragon Camera's, but I think it's a fair enough comparison to make to give an idea of what the Dragon sensor noise can look like.

The test isn't showing or suggesting any particular exposure settings, it just shows one middle exposure with a certain amount of noise, with two noisier more underexposed shots preceding it, then two cleaner more overexposed shots following it. The exact same frame sequence is then repeated, but with every shot boosted again in post by about another stop.

The colour chart on the left is lit one stop lighter than the middle chart, and the chart on the right is lit one stop darker than the middle chart, just to show some more exposure variance within the one frame.

The camera was set to 4:1 RedCode and was black-shaded/calibrated as necessary for every change in exposure and kept within the indicated temp and exposure range while shooting.

.............................................

Here's another quick test showing some Dragon noise in a low-light scenario. The noise is still visible when viewed at 4K but just gets blurred at the lower resolutions -

Has been interesting re-looking at this stuff.

no visible noise in fps 12 yet in 24 will be much

 

[Les Hillis]

If you keep the light levels the same for 24fps as 12fps (using a 24fps project base rate), then yes you'll be shooting 1 stop underexposed and so you'll get more noise.

There shouldn't be any differences just based purely on your frame-rate though, if your exposure is adjusted and you're calibrated for the different capture frame-rates.

Noticing by the way that the difference between calibrating for 12fps and 24fps isn't that extreme with the Dragon sensor. Using the 24fps calibration while shooting at 12fps only turns the 'E' calibration warning indicator yellow, not red.

Just to check the difference, here's an example of the differences between shooting at 5K at 12fps and 24fps, with a 24fps Base project frame-rate and the light source adjusted to maintain a proper exposure -

This video also includes an example of the difference in exposure if using the light level required for 12fps when shooting at 24fps (shooting 1 stop under then pushing the exposure back up in post).

 

[astra2019]

If you keep the light levels the same for 24fps as 12fps (using a 24fps project base rate), then yes you'll be shooting 1 stop underexposed and so you'll get more noise.

There shouldn't be any differences just based purely on your frame-rate though, if your exposure is adjusted and you're calibrated for the different capture frame-rates.

Noticing by the way that the difference between calibrating for 12fps and 24fps isn't that extreme with the Dragon sensor. Using the 24fps calibration while shooting at 12fps only turns the 'E' calibration warning indicator yellow, not red.

Just to check the difference, here's an example of the differences between shooting at 5K at 12fps and 24fps, with a 24fps Base project frame-rate and the light source adjusted to maintain a proper exposure -

This video also includes an example of the difference in exposure if using the light level required for 12fps when shooting at 24fps (shooting 1 stop under then pushing the exposure back up in post).

hi i wonder , is filters 4x4 instead of 5,65 are suitable for my camera dragon epic x 35 mm ?

 

[Les Hillis]

They could both work with the Dragon S35 sensor.

It's more about the size of the front of the lens you want the filter to cover, and the way you want to mount the filter in front of the lens.

The rectangular filters are more of a standard for the traditionally more rectangular frames of the motion-picture formats, while the square filters are more oriented towards the multi-aspect-ratio, more lightweight and often hand-held stills camera shooting (and with the circular screw-in filters being even more of a traditional standard for that).

The larger rectangular filters and cine-style matte-boxes can allow you to angle the filters, move them in and out easily, stack them and have them placed further from the front of the lens (and not be attached to the lens itself), while the main advantage of the smaller filters and their holders is their being relatively cheaper in price and more light-weight.

There are also filter holders/adapters that allow you to use the smaller square filters with some cine-style matte-boxes, at the the risk of cropping the sides of the frame if the matte-box places the square filter too far from the front of the lens.

The smaller, more photography-oriented square filters and holders can also potentially be of more varying quality if there are more brands making them than the more specialized rectangular cine-camera-oriented filters.

There can also be some differences between the same kinds filters of the same brand if they use different manufacturing techniques for the different sizes (eg. surface coatings vs. bonded).

There are some other variables that could come into play depending on your particular end-use case, but I think that's basically it.

 

[astra2019]

Thank you !

 

[Fred Potts]

lots of testing. I did one scene which was shot just in the dark using a flashlight and that turned out good

 

[GBP_RED]

I had this notions that just because the Gemini was the low light King, doesnt mean the Dragon-x cant do low light to a close 3rd and coupled with the notion that in lots of light , or enough light, the 6K Dragon-X can be better than 5K Gemini.

Whats the verdict? How is the Dragon-x 6k with low-light did you find a method that works or are you still wanting the gemini? Thanks