MX What ISO People Are Using Outside?

[Michael Rossi]

For those that have been using the new MX sensor what ISO seems to be the best for EXT sunlit locations? I've read that 500 seems to work well.

 

[Shawn Booth]

test, test, test, test, test

See what works for you, the story's look, the location and lighting conditions.

 

[Blair S. Paulsen]

It must be annoying that there doesn't seem to be a straight answer to what seems to be such a simple question. Get over it. Think signal to noise ratio rather than ISO.

IMHO the crux of the biscuit is to imagine what you would ideally like the scene to look like after grading, your "perfect vision". Then make sure nothing you really want detail, chroma and "creamy smoothness" in ends up in the noise floor. Enjoy all the highlight protection you have left yourself.

On a more practical note, I like to view the live histogram in RedColor viewing space and spin the iris around to see what the histogram does. It is my belief that by spreading out the histogram to capture a wide range of densities you will give the colorist the most discrete code values to manipulate. It also gives you a common sense reference point versus what your naked eye is seeing.

FWIW I like using 800 as it tends to lead me away from clipping and makes it easier to see color/detail on the monitor in high ambient light environments.

Cheers - #19

 

[Rich Schaefer]

I use between 400 and 800. I usually start at 400. It helps with less ND. But if I have a lot of hi-lights I will go to 800 to protect them, because I would be more underexposing the RAW data when I set my exposure based on 800iso in RedColor viewing & metering.

I just finished up a series of "Burts Bees" Spots and that seemed to work pretty well.

Prost,
Rich

 

[Michael Rossi]

Thanks, what iso does the raw image record at is it still 320?

 

[David Battistella]

Thanks, what iso does the raw image record at is it still 320?

yes. But you can go to 800 depending on how much noise you can handle. At 500 outdoor/daylight you can protect the highlights a little better.

David

 

[Rainer Fritz]

As what we saw so far on the big screen, it looks like that at 500 ISO is no noise visible, at somewhere between 640 and 800ISO is the first noise visible, but full acceptable, also if you want to go on a print... DR increased as well, one more stop on the upper end would be nice

best,
rainer

 

[Tom Mitchell]

Firstly let me say i agree with you findings that your setting you mid tones (ISO) to give you latitude where you need it, for hi contrast situations 500-800 asa seems to be the golden zone. (I would still shoot 500 even on a overcast day to account for light poping through the clouds)

The pickle come to when you go to post. on a DPX work flow, what ISO to export at, if you export at 800 (as was shot) you still loosing your highlights in the DPX but you placing your mids where you need them.....OR you export at 320 ISO The image will apear to be dark but you will have information in the hilights and mid-tones at least though i'm not sure how far you can push DPX before it starts to gain codex noise.

So the new question I have is in a DPX work flow what should be exported?

 

[Tim Sutherland]

I'm working on a series that shoots everything at 800iso. Interiors we often use an ND3 or 6, and exteriors we have IRND 1.8 and 1.2 hot mirrors that we wear almost all the time.

800 gives you a nice rolloff into blown out areas, not a sharp clipping point. If you need to regain that 1 1/3 stop back, then usually I use the FLUT and Shadow controls to bring the detail back into range before exporting to a codec. I would do this whether going to dpx or anything else.

One thing to note about FLUT and Shadow, the values you use indicate how many stops you're changing the image. So just make sure that whatever shadow/FLUT changes you make don't take the ISO above your acceptable ISO or below 320 ISO.

I really like the 800iso shooting for exteriors because it feels more like film, where you look at the monitor and you know you can bring back some of the highlight detail. I know using ND almost all the time is annoying, but in my experience the result is worth it. I've been able to pull back detail many stops over key, and I've been surprised many times how much I can actually pull out of the image. If you get the light controlled pretty well for 800iso, you can really fine tune it in the raw controls.

Tim

 

[Tom Mitchell]

Thanks Tim for your responce.

indeed I know I you can tweak the controls to pull everything in range.
But on fast turn around projects, or low budgets, red footage is often just exported with what ever meta data is on the r3d. in theses shoot scenarios all you have is some one downloading the rushes and not acting as a DIT like yourself or me.

taking out the process of being able to tweak individual shots. setting to the new colour since dose give you a good roll of, i agree but your still could be clipping on export to DPX at 800 but might still have all the information in the raw.

what i was saying is could you export at 320 asa, and bring it all up or would that be pushing the DPX to far, what is the happy medium to get the best in this situation?

 

[Tim Sutherland]

In that case maybe you'd be better off at 500 or 400 for the simplest solution. Then you'd get 2/3 or 1/3 stop highlight rolloff for nicer looking highlights and have the raw bring up the shadows a bit for you.

The other thing I'd look at is making a setting that has a base of 500 ISO and using a tad of FLUT (down) and SHADOW in camera - you have to be very careful doing that, but if you know what you're doing, you should be fine - and then maybe adjust the curve a little bit if you need to add contrast because of a shadow adjustment. That way you're not clipping much if anything extra, and you're providing a good middle ground so that you've brought up the shadows a bit in the raw and then you can do the rest from the DPX.

Basically, I wouldn't just export 320 and do the rest from the DPX, that's not taking very much advantage of the MX noise floor. You might get a slightly cleaner picture that way, but if that worked very well, people would have been doing that all along with mysterium to get similar results.

If 500 gives you an acceptable noise floor, then I'd rate at that, FLUT down .3, and maybe bring up the midtones in the curve a bit if you want to see better exposure in camera. If you're not worried about that, I'd let the grade from the DPX fix the midtones.

Hope that makes sense.
Tim

 

[David Mullen ASC]

I had a bit of this issue myself on my M-X pilot... we found that if we transcoded at a lower ASA value to RedGamma (let's say I shot at 500 ASA outside so we transcoded at 400 or 320 ASA) we could perhaps pull another third of a stop of detail in the highlights. Not much, assuming I exposed correctly, but there was a bit more that could be squeezed out of the footage.

However, this meant getting timed dailies, because the advantage of transcoding at the ASA value I used was that the footage came out at the correct brightness and would not need any timing for dailies. So if we had a bunch of stuff converted to look dark, that's fine if you are immediately going to do the final color-correction, but in our case, final color-correction wouldn't happen until a month later after production and editing. So the dailies process was faster and more streamlined if you converted the footage at the ASA rating chosen on the camera.

In the end, we decided to stick to transcoding following the metadata with the notion that in the final color-correction we could work with the original 4K HD R3D files if necessary to pull more information out. Otherwise, we would be working from 10-bit DPX 1080P files that we were creating daily at the post house.

 

[Tom Mitchell]

Thanks Tim.

I'm on the understanding that noise floor is different to the noise created when pushing a codex. with the mysterium, when you where pushing and pulling in the DPX you where seeing captured sensor noise first before you where seeing the codex breaking down.

assuming you capture a noise free image in put it in to dpx how many stops could you push it before it starts to break down..or look horrible?

As more and more grading suites are ingesting RAW it should hopefully be a thing of the past....DPX used to be the gold standard now it seems it's not good enough any more

 

[Tim Sutherland]

Not that DPX isn't good enough, just that it's always best to do what you can before baking in a codec.

We're on hiatus this week but I'll try some pushing of the image when I get a chance after we start back up and I can get to my cart on Monday. I'm not sure if I'll be able to post full images, but maybe some small pixel-for-pixel crops that don't show any production footage.

David - I've found that you always get the most highlight information at 320 (or the equivalent # of stops down with FLUT from your shooting metadata). I've worked out somewhat of a special sauce that brings the most dynamic range out of the image, but unfortunately it seems to vary a bit in different situations so there's not a specific combination of settings to get it right.

Our show is running everything through at metadata settings as well, unless there is a specific shot or sequence that I will give the lab special grading instructions for.

Tim

 

[Tom Mitchell]

That would be great Tim I look forward to it.

What David was talking about was one of the dilemmas i was thinking about when asking the original question, which as a DIT, is fairly fundamental. do you set your meta data for daily's(or MXF/QT) or set it for best output for DPX....again in a raw work-flow this would not matter.

 

[Tim Sutherland]

One more thing I will add is that very little of our footage would really need much more done to it than putting it through at 800 iso. Only if we're shooting at the sun or in one case where glare was shining off of sidewalk was 800 just by itself not perfect. The DP is lighting for 800iso and not planning on taking back some of that top 1 1/3 stop we had in the raw, so only special cases needed different processing. Remember that the new ISO settings are part of the FLUT color science, doing a smarter gain adjustment than just adding x number of stops at all ranges.

That's essentially what David was saying, that 800 works in most cases. In our show they may re-transcode a few shots for the final color work, but thats it.

But to comment on your question of what to do if there's no raw workflow, I treat it more like an HD camera, and get all the metadata settings in camera to go through the whole workflow, that way the DP (and more importantly, producers and client) see a correct image on set, and you're fine in most situations. The DP just has to be aware of the workflow and the implications it has on how he lights, not planning to bring much back in post because it's extra work in multiple places in the workflow.

Tim

 

[Tom Mitchell]

Great advice Tim

So in a sense your using 800asa it has no hard ceiling, because the FLUTis creating a nice roll off in the the whites/hilights......so your exposing it like film again where your no longer concerned about the hard cut off.

in that case,its worth then doing a dynamic range test in the DPX at 800ISO export.
(or perhaps a few different ISO to see which give the best range from roll off to noise)

I think your right in that in most cases your going to want to export as shot. But on bright high contrasty day, it's nicer to keep that blue sky and puffy clouds, and with flut it looks over even though it rolls off well.

Tom

 

[Nick Legend]

Does the MX sensor have a better dynamic range in the the daylight then the M sensor? Can the human eye tell the difference?

Thanks

N

 

[Dylan Macleod CSC]

Does the MX sensor have a better dynamic range in the the daylight then the M sensor? Can the human eye tell the difference?

Thanks

N

The MX sensor has at least 2 stops more DR than M.

So yes - the human eye will have more to look at!

The testing I have done shows WAAAY more to be had in the shadows. I wouldn't hesitate to rate it at 1000 in daylight to give yourself even more highlight protection.

I still think we should be referring to it as EI as opposed to ISO or ASA. The way the metering and "ISO" works is so "unlike" the ISO we know from film.

 

[Harry Clark]

Tim, excellent advice.
In a more controlled environment (studio/ greenscreen/ white cyc) I've been going with 400, since we have pretty good control over the highlight exposure, and I feel like that gives me more "meat" in the bottom end.
I also agree with Blair that getting the most out of the histogram is a good practice.
Obviously, judgement and experience is paramount, as is testing.
A HUGE improvement in response from the MX though, no matter how you slice it...
Cheers,
Harry