Not neutral color (6 replies and 6 comments)
I shot a short film recently that I am not happy with color.
My camera settings:
color temperature: 3200
Red heads tungsten. lit from window outside.
The Look I was going for was "Prisoners movie". I wanted warm but It turned out differently. the camera I shot was Ursa mini 4.6K Pictures below.
Can anyone help me please :).
Thanks for you time.
Not Roger but my 2 cents,
I would say your stills look heavy underexposed. So there is no much information on the image and at the end it looks muddy, not much information. Each camera and sensor reacts to light differently so you need to test and re-test until you understand how expose for it. But if you follow the common rules for a thick negative you should have an ok image with any camera.
I did study Prisoners long time ago and if I remember correctly if you put all of its graded shots on a waveform monitor you will see you rarely hit 50/60 ire, its shifted really dark. But that doesn't mean you shot it dark, you exposed it correctly and then you bring it down in post.
Problem with cameras are always those low IRE values on the waveform monitor, you have to expose the image enough to move away from the floor noise of the low values and then bring it down in post. There is no way to expose straight to camera that dark and make it look good.
Well there is one that I know of, you could use a tweaked LUT on your monitor that looks this dark but really what you are doing is exposing correctly or even overexposing and the LUT brings all down heavily.
At the end you are still exposing correctly even if you see it dark on your monitor and the LUT.
I think if you read about Rogers method you’ll find he shoots how he wants the image to look, there is no overexposing then bringing the image down in post. Whilst overexposing is necessary for some cameras such as an fs7, not so for the Alexa. I think the issue here with the above images is down to lighting and probably starting from a Blackmagic LUT which will have skin tones swayed towards pink.
If you create your own LUT you can be overexposing but the LUT is bringing everything down to correct level or the level you want. This is done all the time with an iPhone or an Alexa, it’s how digital imaging works. It goes with the same concept of lowering the iso on dark scenes and higher iso when there is lots of light. It’s about shifting your sensor dynamic range. You can find reports on that on ASC mag and internet.
For what I’ve read on Roger method he creates or has his own LUT and he exposes for it on set. If you then in post apply same LUT as your base grade then you aren’t seeing how the LoG footage looks anymore.
As I said all the cameras struggle from 0 to 30/40 IRE (waveform monitor) as it’s inherent that in digital imaging it’s where there are less bits of information. It’s how it works. Have you seen how easy an Alexa recovers highlights? Don’t attempt to do it on the shadows. Again because that’s how all sensor works, shadow equals to less bits that equals to less information.
If you are on set and have a standard LUT in your monitor and expose all your shots between 0-40 ire, the values used for example in a film like Prisoners where you rarely hit pass 60 IRE then if you expose a regular REC709 lut like that on set you will end up with a mess of footage. Because you are using a small part of bits of information of the sensor. You are wasting the great dynamic range of your camera to just use a small part.
What you want to do is to use that sensor full capacity to have a “thick negative”. Also important to know when you move around with lots of blacks in your footage you want to stay away of those 0-10 IRE because there is where the digital noise is.
Knowing that and knowing what your final look is you can shoot normal and apply a LuT in post or you can load a LUT that takes a well exposed image and brings it down to look dark on camera. So when you are on set you are watching something really close to the final look. Some people like Roger likes that way other people prefer to watch a regular REC709 lut and do the look in post.
There is no way around it. The sensor has to be well exposed choosing one method or another. And it doesn’t matter if Alexa, red or Sony.
The magenta cast it’s a typical issue with underexposed footage on some cameras. I haven’t shoot with a black magic but I guess it’s the same.
This isn’t how it works with Alexa. Arri states the Alexa distributes bits equally across the stops of latitude / dynamic range it captures. Each stop interval from -7 to +7 is given the same amount of bits. The Alexa captures at 16 bit internally and then delivers at 12bit. When the Alexa is set at 800 ASA it has an even distribution of it’s available latitude between the shadows and the highlights. As an Alexa owner myself I have not felt the need to use the ‘overexpose then bring down In post’ approach (something like an fs7 is a completely different story and with that camera it’s essential). If something looks like I want it to on the Alexa and happens to have an IRE around 40 I still choose to expose that way because I’m not looking to lift the image in post, I’m looking to create the image I want in camera. It’s a different approach I think, you can take a technical view and overexpose your image in order to be safe and have more room in post should you need it or you can expose your images as you want them to look with the camera set at 800ASA with Arri’s rec709 LUT and that may well mean sometimes being low on the IRE scale, however many DoP’s choose to work this way. Harris Savides raised eyebrows when he revealed he would under expose film by 2 stops then also pull the development of the neg by another 2 stops! A thin neg was exactly what he was after. But looking at Birth the technique produced some stunning imagery.
So you can move the low IRE values around on the grading, for example giving more light to shadows without appearing noise or artifacts equal as you can with highlights, where you can recover them up to 3 stops without penalty? Ehem...
I’ll try to explain myself better:
Imagine the Prisoners scene you have the frame up here. You are about to shoot it. You put your lights by eye. But as the look of the film calls for darkness you don’t want anything over 60 IRE, that was agreed with your DIt guy and everyone in preprod, that’s the look everyone is expecting. You use a REC709 LUT on your monitor (What you see its what you are going to get) so knowing the look you decided highlights go straight to 60. Because that’s what it needs to be. What happens then? Your shadows probably will go to pure black. Because in order to maintain those 60 on the highlights you have had to close the iris. Not allowing enough light on the shadows. So in order to bring the shadows back you have to fill in them with a bounce or whatever you choose to. That’s one option, but I bet it will look too filly, and unreal. Because now your sensor range is 10 IRE to 60 IRE you have to contain everything in between. You are basically shooting your Alexa with half the dynamic range is capable of.
Your second option would be to use a LUT that compress all your camera sensor dynamic range possible between those 10-60 IRE, you can use the LUT on a monitor on set or in post. That doesn’t really matter at all.
With that special LUT on your monitor you put your highlights on 60 IRE but what it is really happening behind the LUT? Your sensor is really recording those highlights as 90+ IRE, hence, you are overexposing your sensor because the LUT is programmed to force you to do so. Then you go to the shadows. What happens there? Remember the LUT is remapping all IRE values? Let’s decide your 05 IRE shadow values are programmed to go to 15. You are watching on your monitor shadows at 05 IRE but again the camera is recording them at a safe area of 15. Your LUT it’s making you to overexpose For the look you want. You are shooting in one way on the monitor but the LUT remaps every value where it has to be. To the full dynamic range of the sensor.
Now you go to the suite to grade. Apply same monitor LUT on the Log footage and presto. Done. In top of that the colorist have a thick negative to match shots or correct things here and there.
It’s all about dynamic range and where to put each IRE value inside. Knowing the limits of the sensor and avoid their problematics. Maybe we are talking the same thing but I see it as overexposing trough a LUT and you don’t. But to me it’s to overexpose because the LUT it is forcing you to do so when it tells you your 15 IRE it’s really 05 IRE.
I think you are viewing the approach to lighting a scene and the use of a cameras dynamic range from a technical perspective that for me would be unnecessarily complicated. In your Prisoners example I would just nd windows if I want to bring the highlights down in a window In the background, or alternatively bring the highlights down in the grade if necessary. I think you’ll find that if you ask Roger how he exposed Prisoners it will be like every other project - as close to how he wishes it to appear in a cinema as possible. You’ll also likely find that his LUT is not underexposing, it will be making use of and mapping the full dynamic range of the camera like any good LUT. Roger uses one LUT and that’s it as he’s mentioned before. The look of Prisoners would have been achieved through lighting and shooting exteriors in overcast conditions, not via a grade or a LUT that underexposes.
As far as I know, BMD's Ursa tends to be pink in warm colors, and maybe making a standard LUT when shooting wool film will solve this problem.
Just my 2 cents. When shooting with the Ursa mini 4.6k - always perform a black shading calibration on the camera. Make sure your firmware is up to date - let the camera run for a couple of minutes (to up to operating temperature) then run the calibration (in the setup menu I believe) with the lens cap ON.
This is an update that has really helped in getting rid of a general color cast to the image
This needs to be done pretty much every day before the camera rolls and again if you have moved the camera to a location where the temperature is significantly different (indoor to outside for example.)
And I also agree that you may have upped your exposure in the images above
With cameras such as the 4.6k if you don't feed it enough light it gets a VNP, which is somewhat distasteful.
The Pro versions and G2 have rectified it to an extent but the images do get a weird pink/magenta tint, a visible pattern and muddy. Although like someone said Roger likes to shoot as it is in camera and not to alter the image too much in post. The Ursa is a budget camera and the 4.6k was there first run at the model as well. So bringing it down in post will probably be an idea.
Calibrate the sensor everyday, make sure the Dynamic Range is on film and make sure you're using nice filters as that tint is throwing me off even more than a regular 4.6k tint and in a couple of images the focus is a bit soft in contrast to the sharpness of the prisoner.
Also one large difference is the lighting as you can see from the images. Like I said before with the Ursa its good to feed it more and it then has less of a VNP. It really gets bad at above 800 and bellow 200, and for the prisoner like shot lower ISO is better as you hold more information in the shadows.
Thanks everyone for your replay and taking time sharing your knowledge.
I did a re shoot with applied some of your technique but I did a little research about camera color science. Arri cameras actually produce warm orange rather Red or magenta. that is the way the camera were manufactured So the Magenta color is something to do with the Camera not much the light or on Post Production process. here is a video talks in details.
here is my re shoot. I used an IR Cut filter.
I would take that video with a pinch of salt. I’d recommend watching DoP Steve Yedlins display prep demo and follow up on colour science and the resolution demo. Both videos bust a lot of myths and are very in depth. In the resolution demo he shoots imax film, 35mm film, Alexa, red, Sony demonstrating that they can appear almost perceptually identical to each other after going through his colour pipeline. Some fascinating observations on resolution also.
I have filmed with the Ursa Mini Pro for 3 years now and find it a strange beast. It's more of a system that is tied to Resolve. Despite many stellar reviews I find the image always needs a lot of work to get it looking balanced and nice. One thing I've found to be really useful is to try and set the camera to linear and then grade it by using colour space transform to Arri Log C - I've noticed some significant improvements in the ability to hold all of the DR - I then add some luminance mapping and then find the image is in a good place.
The colour tints in your stills are very common with the camera. I find they can be fixed to some degree in post - it's remarkable how much data you can pull out of BRAW. I find you can negate the pinkishness and any major casts with some balancing in post.
It's really only after 3 years have I felt kind of capable of understanding the camera - it's definitely not a case of what you see is what you get - that's certainly been my experience with a camera such as the AMIRA. With the URSA it's a case of understanding that you're going to have to play with the metadata of the image and craft it to your liking.
This is also the camera's weak point. I've been lucky enough to grade Alexa Mini on a few projects and the footage has often blown me away. Out of the camera i find it looks pretty breathtaking. This is rarely the case with the URSA but on the positive side after the required labour - you can get a pretty nice image out of it.