ISO of Red

[Shawn Nelson]

Recently I've been doing more tech tests with my friend DP Ryan Walters. One test was a filter shootout and one segment was shooting a Macbeth chart. We lit it to 5.6 using his Sekonic and were surprised by how dark the results were. So we did a second test, an exposure setup, where we re-lit the chart to 5.6 and again it looked the same, underexposed. We double checked with two meters and they were both saying 5.6 right at the chart.

So I then turned on false color exposure in Redspace mode and put an 18% gray target in frame and adjusted iris until it was showing green (the false color 'good' for 18%) and looked at the lens...T4! We then kicked it into RAW false color and did the same thing, one tick under T4!

So I brought out the meter again and adjusted the meter until it read T4 on the chart...ISO160.

Looking at Red, the Macbeth looks great at T4! Nothing is clipped and its a nice fat waveform, perfect exposure. Now I know we dont shoot charts in real life but the Macbeth is a great spread, full white down to black and tons of shades in between, a properly exposed Macbeth seems a great representative of what a good exposure is for a certain quantity of light. Not a catch-all rule but a good rule of thumb if you will.

So does this mean Red is really ISO160? I'm not saying that, i'm asking that. Not to knock Red at all, everyone knows I'm a big fan if you will, just trying to up my game on the technical proficiency of exposure and camera knowledge.

I like Jim's thoughts on that we shouldnt think of Red as having a certain native ISO but rather think of it as an acceptable range varying on the project (depending on our tolerance for grain). This is a great way of looking at it I think as in film DPs will know how much they can push or pull a certain stock.

At the same time we need a middle 'aim for here' ISO that we can set our meters to and try to hit. So if we have the light and the time then try to get there. If we're off by a stop or two then we know we can handle it.

If Epic's 800 is what Red is at at 320, then it seems on Epic we could set our meters to 320 and it'd match the meter. Is that right? So given what I saw, why doesnt the text on Red's 320 actually say 160? What am I missing? Or is this a difference in what Sekonic thinks of as an iso320 amount of light vs what Red thinks is an iso320 amount of light? Help me out if i'm doing something stupid here. I know this is sort of a 'first week you get the camera' sort of test, sometimes you just have to circle back and fully understand the basics.

 

[Zakaree Sandberg]

So does this mean Red is really ISO160?

I personally rate at 160-200 because of this...

Ill set the camera to 320 (as per spec)
but My meter as well as whoever gaffs for me is set at 160-200

 

[CJ Roy]

Shawn-

Yeah, it's a slow, light hungry camera. 160 is about right and has been the results of many individual's tests.

 

[Blair S. Paulsen]

Contrast ratio / best ISO setting equation

Contrast ratio / best ISO setting equation

The advantage of rating the camera at 320 is avoiding clipping (full well). The philosophy being that with proper handling of the RAW data you can extract a decent amount of clean image information in the shadows but once you blow out a pixel its essentially gone (yes, I know there are some highlight recovery options).

Keep in mind that while a Macbeth chart may be reasonably representative of what you might point a camera at, your lighting condition is well controlled and there are no specular highlights or hot rims. Rating for any scene that has moderate contrast at 160 should work out fine, ETTR certainly helps signal to noise ratios. High contrast day exteriors are another animal. YMMV.

Cheers - #19

 

[Tyler Black]

Shawn,

I did the very same test when I first got my camera. I thought I may have been doing something wrong, but the meter never lied. I've always rated my meter at 240, allowing for a sort of "middle ground" of error. But I would say realistically that the camera is closer to 160 in reality.

 

[Dylan Macleod CSC]

Yep. This makes sense. I think RED have always said that their rating (at 320) builds in some highlight protection. If your using a meter then you are better off at 160.

 

[Tim Whitcomb]

Just shot at ISO 200 myself today screwing around with the TOKINA 11-16 we just received. In traditonal photography. I always thought a low ISO meant less grain (noise)

 

[Shawn Nelson]

Just shot at ISO 200 myself today screwing around with the TOKINA 11-16 we just received. In traditonal photography. I always thought a low ISO meant less grain (noise)

meaning you set Red to ISO200 or thats what your meter was set to?

Unless things changed, Red should always be set at only 320, since you're not actually changing anything, the camera is what it is as "320". My point being that Red's idea of 320 is 160 on a meter

 

[Dan Hudgins]

White card reading

White card reading

Try a white card reading, use the back of the Kodak Gray card that is 90% white.

Adjust the iris so that the white card is at clip level, the flip the card over and use the gray side, you should need to open the lens 2 1/2 stops to get the gray side to clip.

Likewise, if the camera reads "green" for the 18% side, and you flip it over so you see the white side, you should need to stop down 2 1/2 stops so that the white side reads green (as a 18% gray would).

You cannot expose real subjects with that much light using a sensor camera because the highlights from oil on the skin and such will be way over 90% white, so you need to stop down at least one more stop to put 90% white at 50% signal on the waveform monitor.

Try adjusting the 90% white card to be 50% on the waveform linear sensor data reading (no gamma correction, just sensor ADC output) .
Then check to see what ISO that lighting comes out to with the T/ stop used.

90% white at 50% signal would be a good MAXIMUM exposure ISO to use for general subjects under controlled lighting. For high contrast shots with specular reflections you might want to go to 25% signal for 90% white card (i.e. two stops headroom).

You need to do these tests with the de-Bayer at daylight (sensor) balance since under 3200K without 80 series filters the RED sensor will be way out of balance with the BLUE sensor so there will be a blur to the waveform for the sum of the sensor data (or you will see the red and blue waveforms not reading the same levels, if RAW display is true).

 

[Ryan E. Walters]

meaning you set Red to ISO200 or thats what your meter was set to?

Unless things changed, Red should always be set at only 320, since you're not actually changing anything, the camera is what it is as "320". My point being that Red's idea of 320 is 160 on a meter

I'm still processing the charts we shot, I'll post them online when I'm done.

You are correct, IMO, that the Red should be set at ISO 320 (250 is also ok in my experience), but when metering using an ISO of 160 - 250 yields a better result- depending on your tolerance for noise. The great thing about the Red is that is can read deep into the shadows- but how deep is a matter of taste. So if you want to avoid noise and keep the image clean then metering at 160 - 250 and keeping the camera at 320 is the safer way to go.

 

[Shawn Nelson]

Try a white card reading, use the back of the Kodak Gray card that is 90% white.

Adjust the iris so that the white card is at clip level, the flip the card over and use the gray side, you should need to open the lens 2 1/2 stops to get the gray side to clip.

Likewise, if the camera reads "green" for the 18% side, and you flip it over so you see the white side, you should need to stop down 2 1/2 stops so that the white side reads green (as a 18% gray would).

You cannot expose real subjects with that much light using a sensor camera because the highlights from oil on the skin and such will be way over 90% white, so you need to stop down at least one more stop to put 90% white at 50% signal on the waveform monitor.

Try adjusting the 90% white card to be 50% on the waveform linear sensor data reading (no gamma correction, just sensor ADC output) .
Then check to see what ISO that lighting comes out to with the T/ stop used.

90% white at 50% signal would be a good MAXIMUM exposure ISO to use for general subjects under controlled lighting. For high contrast shots with specular reflections you might want to go to 25% signal for 90% white card (i.e. two stops headroom).

You need to do these tests with the de-Bayer at daylight (sensor) balance since under 3200K without 80 series filters the RED sensor will be way out of balance with the BLUE sensor so there will be a blur to the waveform for the sum of the sensor data (or you will see the red and blue waveforms not reading the same levels, if RAW display is true).

You bring up a good point about what skin tone under these conditions would be, I dont know. Thats a good follow up test.

We were under daylight, using some small HMIs to light the chart.

 

[Ryan E. Walters]

We do need to do a real world test- as I find those helpful. However, on the Macbeth chart we used one of the patches is for skin tone- so we do have that reference in there.

 

[Tai Wah Lim]

I'm still processing the charts we shot, I'll post them online when I'm done.

You are correct, IMO, that the Red should be set at ISO 320 (250 is also ok in my experience), but when metering using an ISO of 160 - 250 yields a better result- depending on your tolerance for noise. The great thing about the Red is that is can read deep into the shadows- but how deep is a matter of taste. So if you want to avoid noise and keep the image clean then metering at 160 - 250 and keeping the camera at 320 is the safer way to go.

Ryan, I remembered you had a Sekonic profile to match Red 320 with the meter. Is that still valid? Lim

 

[Ryan E. Walters]

Yes the sekonic profile is still valid. The profile is created using the sekonic chart set according to their specs, so it correctly show the curve as well as the range of the sensor according to the chart. What is different is where you choose to rate the camera at when using your meter. So if you find that when you meter at 320 and set the camera to 320 your shots are underexposed to your liking, then you should adjust your meter accordingly- say to 160, 200, 250, etc. But keep the camera at 320. The profile / meter will still read correctly, as it has been setup and calibrated for the entire range of the Red.

 

[Eric Haase]

Recently I've been doing more tech tests with my friend DP Ryan Walters. One test was a filter shootout and one segment was shooting a Macbeth chart. We lit it to 5.6 using his Sekonic and were surprised by how dark the results were. So we did a second test, an exposure setup, where we re-lit the chart to 5.6 and again it looked the same, underexposed. We double checked with two meters and they were both saying 5.6 right at the chart.

So I then turned on false color exposure in Redspace mode and put an 18% gray target in frame and adjusted iris until it was showing green (the false color 'good' for 18%) and looked at the lens...T4! We then kicked it into RAW false color and did the same thing, one tick under T4!

So I brought out the meter again and adjusted the meter until it read T4 on the chart...ISO160.

Looking at Red, the Macbeth looks great at T4! Nothing is clipped and its a nice fat waveform, perfect exposure. Now I know we dont shoot charts in real life but the Macbeth is a great spread, full white down to black and tons of shades in between, a properly exposed Macbeth seems a great representative of what a good exposure is for a certain quantity of light. Not a catch-all rule but a good rule of thumb if you will.

So does this mean Red is really ISO160? I'm not saying that, i'm asking that. Not to knock Red at all, everyone knows I'm a big fan if you will, just trying to up my game on the technical proficiency of exposure and camera knowledge.

I like Jim's thoughts on that we shouldnt think of Red as having a certain native ISO but rather think of it as an acceptable range varying on the project (depending on our tolerance for grain). This is a great way of looking at it I think as in film DPs will know how much they can push or pull a certain stock.

At the same time we need a middle 'aim for here' ISO that we can set our meters to and try to hit. So if we have the light and the time then try to get there. If we're off by a stop or two then we know we can handle it.

If Epic's 800 is what Red is at at 320, then it seems on Epic we could set our meters to 320 and it'd match the meter. Is that right? So given what I saw, why doesnt the text on Red's 320 actually say 160? What am I missing? Or is this a difference in what Sekonic thinks of as an iso320 amount of light vs what Red thinks is an iso320 amount of light? Help me out if i'm doing something stupid here. I know this is sort of a 'first week you get the camera' sort of test, sometimes you just have to circle back and fully understand the basics.

I concur with your findings and have performed the same tests in the past. Set your meter at 160 (or 120) and you will have 18% grey show as green false color in RAW or REC709 (build 20) if you expose for what the meter says. If you set your meter to 320 (effectively "rating" the camera 320) you will have underexposed images. You can get away with this but will have less latitude in the shadows since you are in reality underexposing. I've only half understood why people claim the camera is 320 since if you expose based on 320 metered light you have a disproportionate amount of latitude above key to below key.

This is especially egregious in night scenes meant to be photographed with the majority of tones below "normal exposure." If you rate the camera at 500, you are really underexposing the image and bringing it up in post processing and sacrificing a lot of shadow latitude. It gets noisy fast and there are not a lot of tonal variations with that much underexposure. For me, the slow sensor is the number one issue with the RED ONE.

 

[Hans von Sonntag]

We've done this test just about 18 months ago. Yes, Red's "default" sensitivity can be interpreted, when using a lightmeter, at 160-200 ASA. I always found that one has to distinguish between Red's metadata ASA and lightmeter ASA, both beeing just about 1 Stop apart.

In film land Kodak recommends their 500 ASA stock to be exposed at 500 ASA. Most cinematographers expose this stock at 320 ASA if shadows are supposed to be as noise/grainfree as possible. The same with the RedOne but in the diametrical opposite. Here the manufacturer recommends 320 ASA but most cinematographers expose the redOne at 200 - 250 ASA for the same reason.

Exposing one stop under (320 ASA) exihibits not too much noise by creating the biggest dynamic range. Red thinks that for many this is the sweet spot. Or to put it in other words: the safe bet. Plus 320 ASA sounds better than 200 ASA.

I once exposed the RedOne at real world 500 ASA (light meter) which is 1 1/2 stop under, according to your test (which I concur). The footage was used for a cinema commercial. It was slightly on the noisy side but printed well. In a night interior/exterior situation with blooming highlights and crushed blacks such an exposure works well.

In a high key environment the texture would be too rough, IMHO. Here 200 ASA (lightmeter) are just fine.

ASA values should not be carved in stone. But it's good to know from where you start (160-200 ASA).

Hans

FYI: Check the histogram in RedAlert when exposing footage with clipped highlights at 320 ASA. If you want to safe all possible data from the RAW to RGB you have to dial in 250 ASA.

 

[David Mullen ASC]

I just set up a RED ONE for an classroom exercise and lit a face using my light meter set to 320 ASA and the camera set to 320 ASA - but when I tested the False Color guide, it basically suggested that the face was still about a 1/2-stop underexposed (and it wasn't a person with a tan or dark complexion).

Not sure what all that means, because it was like the camera itself was telling me to light and expose at 200-250 ASA when the camera was set to 320 ASA...

 

[Eric Haase]

I just set up a RED ONE for an classroom exercise and lit a face using my light meter set to 320 ASA and the camera set to 320 ASA - but when I tested the False Color guide, it basically suggested that the face was still about a 1/2-stop underexposed (and it wasn't a person with a tan or dark complexion).

Not sure what all that means, because it was like the camera itself was telling me to light and expose at 200-250 ASA when the camera was set to 320 ASA...

It means the camera is not 320 and if you follow RED's recommendation for rating it 320 you will be underexposing. You can always use a LUT like REDSPACE to lift it back up or dial in a custom curve but some areas of the image will be noisier because you've underexposed the image to begin with. If you rate 160 or 200 you will have an easier time reconciling your meter and false color and have a more even distribution of tonal range above and below key.

 

[Graeme Nattress]

Correct exposure is utterly scene dependent. A macbeth chart has a range of 5 stops from black to white. If you want a scene that has no more thane than a macbeth exposed as best as possible, ie as exposed as possible without clipping, you'll end up with a lower ISO than 320. If your scene has more DR and you're exposing similarly to the right, ISO320 will be appropriate. And again, as the DR increases, you may find ISO500 appropriate.

ISO320 was chosen because it's a reasonable starting point for a reasonably averge scene. That doesn't mean it's correct for every scene.

Graeme

 

[Eric Haase]

Correct exposure is utterly scene dependent. A macbeth chart has a range of 5 stops from black to white. If you want a scene that has no more thane than a macbeth exposed as best as possible, ie as exposed as possible without clipping, you'll end up with a lower ISO than 320. If your scene has more DR and you're exposing similarly to the right, ISO320 will be appropriate. And again, as the DR increases, you may find ISO500 appropriate.

ISO320 was chosen because it's a reasonable starting point for a reasonably averge scene. That doesn't mean it's correct for every scene.

Graeme

For me, the ISO definition is most helpful in discussing how much light is required to achieve a "normal" exposure on the sensor, whether it be 18% grey or caucasian skin tone. From that point, with testing, one can understand how much latitude there is a above and below normal exposure. You can then observe the scene and make decisions on where to place the exposure based on the tonal range of the scene and what you want it to look like. Changing how you rate the camera does not effect it's actual sensitivity of course. It just tricks you into placing the exposure in a different place, right?

If I rated RED ONE at 500 it would mean that what my light meter said is "normal" exposure would in actuality be on the sensor about 20 IRE (RAW). This means that what I think is going to be normally exposed tones according to my meter are actually very underexposed and will be lifted by an ISO metadata adjustment or a gamma curve in REDALERT or similar post processing in order to achieve the normal exposure tone.