Profile: SC 4.1 250ISO

[Miguel "Macgregor" De Olaso]

I hope this post helps people to better understand shooting RAW and get better looking images out of the RED ONE.

To start with the test let me set a high contrast scene to put the RED in trouble:


Camera settings:
rec709, 320ISO, 25fps, 1/50sec and as you can see, exposure has been set to the mid grey.
Images captured in a Cinetal monitor from the HDsdi 720p output.



And here you can see the reading with False color:




Now take a look at the readings from a lightmeter:




So this is a typical scenario when shooting outdoors with the RED ONE.
With the default 320ISO setting, the Red has a great amount of dynamic range in the shadows but it´s very limited in the highlights (something like +2 and -7 at 320ISO).

This is not good at all when shooting outdoors, which usually have lots of highlights and bright skies. And as soon as the image is clipped, we get that awefull video look. Check the 1st image again. It´s well exposed and still the brightest areas are completely blown out. It doesn´t look very good at all.

Now another problem for the old school DPs that want to use lightmeters when shooting digital, is that the readings from their lightmeter set at 320ISO do not match the false colour and readings when exposing with the RED ONE at 320ISO setting. Usually the difference is about 1 stop and 1/3 in 709 mode and 1 stop in RedSpace. This sets the real world sensitivity of the default RED ONE about 160ISO. Not good.

[Miguel "Macgregor" De Olaso]

To solve the DR problem, what we´ve been doing is raising the ISO, therefore "underexposing" from the default 320ISO (by closing the iris of the lense).
So for example, if we expose with False color with a 640ISO setting, we will be closing the iris by one stop, therefore, capturing 1 more stop in the highlights, but trading it for 1 stop in the shadows. However, since we´ve got a lot of dynamic range in the shadows, the trade off is good.
I´ve been recommending shooting 500ISO or 640ISO in most cases beacause all this reasons.
This helps getting more info in the highligths. However because of the way ISO settings work on camera, all that extra information in the highlights won´t be visible on the monitor. We need to swap to RAW view (which is not really raw anyway) to discover all that precious information in the highlights.
This is not good because:
- you need to be swapping to raw view constantly to check if there´s hidden info inthe highlights.
- the director/clients might complain about seeing skies blown out (and not, you can´t explain them all this shit)
- on editing your proxie files will still look blown out when they really have much more info in the highlights. Pain inthe ass for the client.
- on grading you´ll have to manually lower the exposure of all shots to recover those highlights. Pain in the ass for the colorist.

The R1 seems to capture about 11.3 stops of dynamic range. In real world we might be using 9-10. But they are not equally distributed (let´s say +4, N and -4). When we´re shooting scenes with less contrast than what the R1 can capture (let´s say 6 stops of DR), we will get the best image by exposing to the right without clipping. It will be easy since our camera is capable of much more than 6 stops. The ISO setting will make the image brigher or darker on our monitor, but as long as the RAW parade or RAW histogram shows no clipping we will be able to extract a really clean and nice image in post.

But what happens when we have a 15 stop contrast scene and we have a camera that only records 10-11 stops of DR? We need to choose: do we want to preserve highlights? are we more interested in the shadows? Usually you´ll want to have the same amount of DR above and under mid grey. But with the RED ONE this doesn´t happen with the 320ISO default setting.



So to solve all of the above, we´ve spent some time developing a camera profile that seems to fix everything. We´re not doing any magic. We´re just appling what we´ve learnt so far on this post.

Simply load the PROFILE SC 4.1 250ISO on the camera (with a SD card) and you´ll get:

- 1 stop more of DR in the highlights (compared to default 320ISO)
- No highlights clipping on LCD or HDsdi (what you see is what you will get, there´s not hidden info in the highlights, no need to swap to Raw mode, no need to lower exposure in post to get extra hidden info)
- Much better looking proxies (because the highlights dont clip in a linear mode now and because there´s no hidden info at the right of the histogram). So what you edit in FCP will be about the best you can get out of the RAW in terms of DR.
- Your ISO setting in camera should be very close to the ISO in a lightmeter (250ISO in camera, 250ISO in your lightmeter). So you can check lights and exposure with your lightmeter. (this means the mid grey/green colour of the false color matches the reading in a lightmeter)

[Miguel "Macgregor" De Olaso]

OK, now take a look at this:



This is the same situation, but exposed and captured with the SC 4.1 250ISO profile. See how much more info we are capturing? We see the ground, the sky, a bit of the clouds, and even the fences and aerials on the roof. The rollof from highlights to overexposed is smoother. Think in a better Redspace. Think in a Hypergamma from Sony.
But the mid grey is still measured as green and the overall brightness of the image remains the same.

Check the false color reading:





It´s important to notice that this profile has been written for a LIMITED RANGE mode. THis means, it uses the image range of TVs and field monitors (16-239 for 8bits). THat´s why it´s showing as much DR as possible on the field, there´s no white point clipping.
This is important to know during postproduction, beacause the proxies will be 709 legal as well as anything you export with the camera presets from redcine or redalert. But in case of grading for cinema, i would just delete all metadata and start grading from scratch (ie: in scratch) or remove the metadata, apply a redlog curve and grade from there.
But this is the point of shooting RAW, right?

This thread is open for discussion, so if you disagree with some of the final conlussions you are welcome to give your sincere oppinion.

Note: i´ll upload the PROFILE SC 4.1 later today. BTW, SC 4.1 stands from Salvaculos Curve version 4.1. ;D

Copy&paste from 2nd page for easier reading:

Before loading the profile on your camera you must be awere of this information:

The mysterium sensor (or the software applications that read the Redcode RAW) seems to have a colour shift depending on the amount of light that hits the sensor. This means that you can set your white balance at whatever you want and from that point if you were looking at a grey chart, every stop that you go up you´ll get warmer tones and every stop that you go down you´ll get cooler colours.
You probably have experienced this while matching 2 shots under the same light situation but with different ISO exposure. If you have a grey card on the clapper and you choose to use the pick WB tool from redcine, redalert or scratch, you will probably get two different readings (ie: 5600 for the lower ISO and 5970 for the higher ISO).
The reason is that by changing the ISO setting we are mapping the mid gray of our final image into a different zone (Ansel Adams´zone system again guys).
So when you set your ISO higher, you have to compensate the WB shift by some kelvin degrees. And the other way around too.
I belive this is the reason why overexposed areas at standard 320ISO have that yellow cast.





So since our profile trades shadow information for highlights by using a calibrated tone curve, we are mapping our mid grey into a lower zone. Therefore we´ll be getting a cooler image at the same given Kelvin setting.

In real world it seems there´s a 200º kelvin shift for each x2 increase on the ISO setting. And this is why the Profile SC 4.1 uses a 6000ºK setting for daylight balanced light sources. But don´t worry, the image will look about the same as the standard 320ISO setting at 5600º. Balanced.

The WB shift along different zones should be something RED should take care on their next sensor cameras. It shouldn´t be too hard to fix with software (if this is a hardware issue). Note: I´ve found this issue with RAW files from the canon 1DsmkII too, so maybe this is something in common with raw files and since the nature of raw is not to process anything, we should address this with CC.


So, once you load the profile all you have to do is set your camera to 250ISO. This information is not saved in the profile so that´s why you have to manually enter the iso setting. But don´t expect the SC 4.1 image to look anywhere similar to the standard 250ISO (never use such a lower ISO with your R1, btw).
You may change your WB as you wish. In case you need to go lower than 5000ºK i would recommend using a 80C filter. Changing your WB lower than 4000º usually affects exposure a little bit (besides introducing a lot of noise in the blue channel).

Make sure you have selected VIEW: 709.

When opening the footage in Redcine or redalert, make sure you have selected 709 as your project gamma.

In case you need to increase the sensitivity of the camera in a situation of really low light, you´ll encounter the same problem of highlights clipping on the previews but not on the raw. So the profile won´t give you all the benefits that it was designed for. It will still look better than any standard high ISO setting but have in mind that for every x2 ISO increase you´ll be hidding 1 stop of highlights information. The RED seems to show the full highlights information only at 250ISO, and that´s why i developed the profile from that starting point.

Also we are working on a SC 4.1 500ISO, so you can load different curves/profiles to match different sensitivities. More information later on.


Ok, here goes the SC 4.1 250ISO profile: http://www.macgregor.autoecstasy.com..._41 250ISO.RLK REMEMBER: this profile has been developed to only work well with the camera set at 250ISO and a 709 gamma.



Thread goes on here: http://www.reduser.net/forum/showthr...244#post429244

[Rick Darge]

Very nice work Mac. Looking forward to testing this out. Is this something you recommend for permanent use in all lighting conditions or just when you need that extra info in 15+ dynamic range scenes?

[Charles Angus]

Just to confirm - the second images are the same shutter and aperture, but with your custom look applied?

c./

[Miguel "Macgregor" De Olaso]

Originally Posted by Charles Angus

Just to confirm - the second images are the same shutter and aperture, but with your custom look applied?

c./

Nope, same shutter but different apperture, since with default 320ISO you get a real world 125-160ISO, while with the SC 4.1 preset, you are getting a real 250ISO, so it´s more sensitive to light, therefore, you´ll be closing your lense about 1 stop.

[mikeburton]

Nice! We have been rating the RED at 250 ISO the entire feature finding it to be more accurate as this is a metered feature. Can't wait to upload your profile and check it out!

[Tim Fassnacht]

Great Post MacGregor...

How bout a similar one for Tungsten?

[Liam Hall]

Interesting test, I understand how changing the gamma curve would get you more detail in the H/Ls but wouldn't increasing the ISO just introduce more noise in the shadows?

I'll have to lie down and work this one out! Any chance of a histogram showing each set up?

[Thom Steinhoff]

Great work! Can't wait to try it out!