Dragon Noise Confirmation

[Elsie N]

Same old discussion even with Dragon huh....

It does seem that way, doesn't it. I remember years ago, Jim Jannard said expose all the way to the right and then back off smidge and you've captured the majority of what's in the image. He was talking about the Epic. But the Dragon opens up the possibilities of capturing an image within a "range" of exposure that lets anyone willing to become proficient at that, to capture magic straight from the camera... without having to rely on doing that in software.

The good news is, those of us still ingrained in the old method of ETTR and tweaking in software (especially new improved software with ADD) can continue to function in the same way, but with Dragon, better.

I think it comes down to time. The ETTR way requires more time to get to where the Exposure Pros get to, but they have done much of their work in camera and can probably get it out the door much quicker.

As for noise... I've erased that from my mental dictionary. I think noise is something one adds with software as a creative choice rather than worry about as an image downer.

 

[Medavoym]

Agree with Mark all the way...

A quick additional note here.

WHY are people so obsessed with noise on Dragon/Alexa/Epic-X etc?

I come from shooting 35mm. These digital cameras are simply amazing, compared to film. They are super-clean. In fact, for me no noise (or no grain) means plastic-looking images: fake, artificial, texture-less.

And, in any case... go to the movies, or go to film festivals... tons of "noisy", "grainy" footage out there. It's night and the light levels are low? People accept grain and noise. They don't even notice it (even when it's heavy). They follow the story.

Seriously... divert the energy spent on how to make your footage slightly cleaner, and start thinking of your stories. Who cares if that night shot in the cab, almost no light, is a bit noisy?

And on Dragon? Shoot that on ISO500 (or pushed ISO500) 35mm and compare. We are so lucky to have Dragons with so little noise.

Go out and shoot! The noise on Dragon is a NON-ISSUE.

 

[Björn Benckert]

It's not a massive color shift, it's slight. Dragons color fidelity is pretty amazing through all ranges. It's a tough call. These sensors are all about 5000K, so they will look sharpest when lit with Daylight sources and slightly softer with tungsten sources. So when you are in a living room with available mixed light sources and you underexpose, you are going to maybe get hit with a bit more noise, which translates to slightly softer image.

Just try it yourself. Download the R3D and process the shot at 2000 ISO using curves and the scopes. Save the preset. Now process it gain at 400 using curves and the scope and "match" them. Theoretically they should be equal but any kind of gain you add to an image can affects it's color/noise performance. Now, if you paste between the looks, you might see ever so slight shifts in color depth and tone. Curves are one kind of gain, ISO is another kind of gain.

IN the end, it's the look you prefer. If you did it with well lit charts you might also see slight shifts. Its up to you to decide what is accurate and where you want the over/under stops of middle grey to lie.

It depends on the scene. EXT daylight might call for higher rating 800-2000ISO to control overexposure/clip. INT NIGHT might want 320ISO so you can clearly know how many stops under you are or how much light to add to keep the picture fairly clean.

Battistella

But the actual shift is minuscule and practically imperceptible.

Dragon is very accurate. I am nit picking.

Dragon is extremly linear. Pushing the image up and down the histogram change nothing but noise levels. As long as you do not go outside the goal posts.

 

[David Battistella]

Dragon is extremly linear. Pushing the image up and down the histogram change nothing but noise levels. As long as you do not go outside the goal posts.

thats what hat I am seeing in the examples that have been posted.

best word is elastic.

Still, i did say it was pixel peep in and nit picking, zoomed way in and exploring, nothing you'd notice on a typical frame.

The flexibility of this tool is what is blowing my mind.

battistella

 

[D Fuller]

David Fuller, I'm really looking for consistency across all shots and I think ETTR would give me that consistency across two cameras. It would also require less time spent in trying to ensure each camera is identical.

But as you suggest, total lack of post and color timing would be folly. So, I'm thinking by shooting a simple ETTR would give me a little latitude to get footage from two cameras and two points of view to look in sympathy with one another, but not identical? Hearing from both David and Mark that color fidelity remains throughout the range, ideally I would like to put a single curve on all the footage and keep the differences in time of day noticeable. And still leave me the option of making a part of a scene "pop" for dramatic purposes.

Elsie, I'm not a fan of ETTR except sometimes for single shots.

Here's why: brightness should differ for different scenes, and for different shots within the scene. Some scenes simply have more light and others have more darkness. They should sit in the histogram differently. If they don't, they won't cut together well. That's because by always exposing to the right, you will have to constantly change the development side of the exposure/development relationship. If you you expose a dark scene to the right, you change where middle gray is, you will have to pull the exposure down in post, and that will put it onto a different part of the gamma curve. It will have less noise than the shadow areas of the brighter scene that you ETTR, and when you cut them together, they won't look right together without a lot of post work.

What I recommend is to consider the exposure of the scene as a whole. Don't worry so much about individual shots. That's not to say it isn't a good idea to lift an exposure out of the noise floor a bit (I'm not that dogmatic) but it does mean that I'd rather let dark scenes be dark, with all the characteristics of underexposure that are natural to them, so that they will feel like part of a continuum from dark to light over the whole piece. Think of it like film. You wouldn't expose a dawn scene on film to make it look like full daylight, you'd let it be a stop or two under, because that would look right for the darker scene. I'd apply the same thinking to Digital Cinema.

There is enough dynamic range, and noise is controlled well enough in these cameras that I'd advocate revising the ETTR idea to something like "Lift everything you care about out of the noise floor" but beyond that, i think ETTR leads to an inconsistent look.

 

[Elsie N]

If I were doing that, I'd also want to look very hard at exposure as you go through the day. You will want to modify the ETTR idea a bit. It would be useful to do a few day-long time lapse recordings and watch how the exposure changes throughout the day. Histogram notes, if you will. To get the best-possible images, you''ll certainly want to modify the exposure from what a single-setting time lapse would give you through the day, but being familiar with the differences in what the exposure "naturally" looks like at dawn and noon and in between will help to re-produce the arc of exposure during the day.

The idea of doing this without color grading is a little too Dogma 95 for me, but certainly doing a most of the work in-camera is a very good way to approach it.

I more or less passed over this in your first post...

Elsie, I'm not a fan of ETTR except sometimes for single shots.

Here's why: brightness should differ for different scenes, and for different shots within the scene. Some scenes simply have more light and others have more darkness. They should sit in the histogram differently. If they don't, they won't cut together well. That's because by always exposing to the right, you will have to constantly change the development side of the exposure/development relationship. If you you expose a dark scene to the right, you change where middle gray is, you will have to pull the exposure down in post, and that will put it onto a different part of the gamma curve. It will have less noise than the shadow areas of the brighter scene that you ETTR, and when you cut them together, they won't look right together without a lot of post work.

What I recommend is to consider the exposure of the scene as a whole. Don't worry so much about individual shots. That's not to say it isn't a good idea to lift an exposure out of the noise floor a bit (I'm not that dogmatic) but it does mean that I'd rather let dark scenes be dark, with all the characteristics of underexposure that are natural to them, so that they will feel like part of a continuum from dark to light over the whole piece. Think of it like film. You wouldn't expose a dawn scene on film to make it look like full daylight, you'd let it be a stop or two under, because that would look right for the darker scene. I'd apply the same thinking to Digital Cinema.

There is enough dynamic range, and noise is controlled well enough in these cameras that I'd advocate revising the ETTR idea to something like "Lift everything you care about out of the noise floor" but beyond that, i think ETTR leads to an inconsistent look.

but now I better understand what you were driving at, I think.

That is, you are saying I should learn how the histogram looks for a certain time of day and use that as my guide, while leaving ISO and color temp setting the same? (Which is not a problem because those two settings on the UI could be locked as far as I'm concerned, as long as they are moveable in post)

Am I on the same page?

 

[Evin Grant]

One of the biggest adjustments I'm making with the Dragon is just seeing underexposure, there is so much DR that an exposure that is 1 stop under just doesn't look dark on the monitor, you really have to pay attention to the histogram, I'm even falling back on the incident meter more since it's a point of reference outside the system. This is something I'll get used to eventually, in the mean time I'm turning down the brightness on the 5" touch so I'm not as tempted to just go with it. That said the organic granular quality of the noise at higher ISOs is so pleasing I'm only really worried about consistent looks not so much noise quantity.

 

[Robert Horwell]

One of the biggest adjustments I'm making with the Dragon is just seeing underexposure, there is so much DR that an exposure that is 1 stop under just doesn't look dark on the monitor, you really have to pay attention to the histogram, I'm even falling back on the incident meter more since it's a point of reference outside the system. This is something I'll get used to eventually, in the mean time I'm turning down the brightness on the 5" touch so I'm not as tempted to just go with it. That said the organic granular quality of the noise at higher ISOs is so pleasing I'm only really worried about consistent looks not so much noise quantity.

I'm finding exactly the same thing Evin. Looking at the meters on Dragon and then my monitor, my brain shorts slightly and i'm tempted to stop down, but the meters tell me i am introducing noise if i do, i can see how easy it could be to underexpose on Dragon. I'm hoping that soon the histogram will show me the raw exposure all the time.

 

[Mark L. Pederson]

One of the biggest adjustments I'm making with the Dragon is just seeing underexposure, there is so much DR that an exposure that is 1 stop under just doesn't look dark on the monitor, you really have to pay attention to the histogram, I'm even falling back on the incident meter more since it's a point of reference outside the system. This is something I'll get used to eventually, in the mean time I'm turning down the brightness on the 5" touch so I'm not as tempted to just go with it. That said the organic granular quality of the noise at higher ISOs is so pleasing I'm only really worried about consistent looks not so much noise quantity.

Or ... you flip to RAW view and use the EXPOSURE meter. I find this FAR more informative than the histogram on Dragon. It's VERY obvious to see what is at or below the noise floor this way.

And yes ... Dragon makes you want to pull out a light meter.

 

[John Marchant]

Dragon sent me back to the light meter for sure - funny really

 

[Mark L. Pederson]

Dragon sent me back to the light meter for sure - funny really

We finally really got to digital "FILM"

 

[Robert Horwell]

Dragon sent me back to the light meter for sure - funny really

Never stopped using my Pentax digital spot. Just didn't feel right without it.

 

[Robert Ruffo New]

Agree with Mark all the way...

A quick additional note here.

WHY are people so obsessed with noise on Dragon/Alexa/Epic-X etc?

I come from shooting 35mm. These digital cameras are simply amazing, compared to film. They are super-clean. In fact, for me no noise (or no grain) means plastic-looking images: fake, artificial, texture-less.

And, in any case... go to the movies, or go to film festivals... tons of "noisy", "grainy" footage out there. It's night and the light levels are low? People accept grain and noise. They don't even notice it (even when it's heavy). They follow the story.

Seriously... divert the energy spent on how to make your footage slightly cleaner, and start thinking of your stories. Who cares if that night shot in the cab, almost no light, is a bit noisy?

And on Dragon? Shoot that on ISO500 (or pushed ISO500) 35mm and compare. We are so lucky to have Dragons with so little noise.

Go out and shoot! The noise on Dragon is a NON-ISSUE.

It is the scriptwriters' job to write the story, and the DPs to worry about the finer points of image, albeit in conjunction with story. I agree with however that a bit of nice noise like that from MX and Dragon is not often a problem, but you can't expect people not to look over the details of how to get the most out of a camera system. Deep knowledge is part of what makes the difference between a truly great DP and a passable one.

Noise is an issue on 100% of all cameras if you don't know how to handle it before it becomes a problem.

 

[Robert Ruffo New]

Sadly ... yes.

In this WYSIWYG world ... the basics of photography and cinematography seem to be skipped by many.

And this is very sad. You can't play Jazz well before you learn your basic scales.

 

[Robert Ruffo New]

It does seem that way, doesn't it. I remember years ago, Jim Jannard said expose all the way to the right and then back off smidge and you've captured the majority of what's in the image. He was talking about the Epic. But the Dragon opens up the possibilities of capturing an image within a "range" of exposure that lets anyone willing to become proficient at that, to capture magic straight from the camera... without having to rely on doing that in software.

The good news is, those of us still ingrained in the old method of ETTR and tweaking in software (especially new improved software with ADD) can continue to function in the same way, but with Dragon, better.

I think it comes down to time. The ETTR way requires more time to get to where the Exposure Pros get to, but they have done much of their work in camera and can probably get it out the door much quicker.

As for noise... I've erased that from my mental dictionary. I think noise is something one adds with software as a creative choice rather than worry about as an image downer.

yes and know. For example painting your camera on set is a waste of expensive set time, and many looks are only possible in post - that's not "fixing it in post", it's shooting in a way that is conscious of the possibilities and limitations of post, rather than pretending for some odd reason that post does not exist.

 

[Kenny Mosher]

Thanks for all of the amazing and helpful feedback guys! You rock

 

[Mark L. Pederson]

Elsie, I'm not a fan of ETTR except sometimes for single shots.

There's many contexts where ETTR is not what you want to do.

A histogram doesn't tell you how much light is falling on your subject - or how much light is falling on any specific part of the image.

If the histogram is your #1 exposure method ... you are selling yourself WAY, WAY short .... handicapping yourself bigtime.

Yes, I know MANY people choose their aperture setting using ONLY the MONITOR and HISTOGRAM ... and well ... that pretty much explains why we see such a wide range of results when compared to that silly celluloid stuff.

 

[Mike P.]

What about false colour, Mark? heh.

 

[Elsie N]

There's many contexts where ETTR is not what you want to do.

A histogram doesn't tell you how much light is falling on your subject - or how much light is falling on any specific part of the image.

If the histogram is your #1 exposure method ... you are selling yourself WAY, WAY short .... handicapping yourself bigtime.

Yes, I know MANY people choose their aperture setting using ONLY the MONITOR and HISTOGRAM ... and well ... that pretty much explains why we see such a wide range of results when compared to that silly celluloid stuff.

Mark, you are saying that even if the exposure is in the range, it can't be made good (great?) in post?

 

[D Fuller]

I more or less passed over this in your first post...


but now I better understand what you were driving at, I think.

That is, you are saying I should learn how the histogram looks for a certain time of day and use that as my guide, while leaving ISO and color temp setting the same? (Which is not a problem because those two settings on the UI could be locked as far as I'm concerned, as long as they are moveable in post)

Am I on the same page?

Yes, Elsie, I think you are. The thing is, some scenes are just dark. If you look at much of Gunleik's recent work, you'll see shots where there's not much at all in the right side of the histogram, and I'm sure the left goalpost is filled to overflowing. But the images are lovely. The under exposure is a creative choice.

Let me see if I can give an example that will shed some light. (for simplicity, I'm ignoring HDRx.)

Imagine a scene at dusk with a full moon rising over a forested hillside on a clear night. Let's say that at the moment you shoot, the moon is maybe four stops over middle gray exposure, and the forest ranges from say, three to six stops under. Your histogram will have a strong spike to the far right, and a large mass to the far left, and not much in between. You wouldn't move that mass on the left over to the right, because the moon would clip, so a good exposure will have the moon pretty close to the right edge of the histogram, but not clipping, and the forest falling where it falls. Hopefully, it falls above the noise floor, but it is likely that it will descend into the noise floor at its darkest, and that's OK, because you can crush the toe of the gamma curve just a bit to squash the noise, and the scene will look lovely. The full moon puts the exposure at f:11.

If you have another shot in this scene that doesn't include the moon, what do you do? By the ETTR method, you'd open up maybe six stops to fill the histogram with the forest scene, but that would completely change the noise profile of the scene, and on MX, the color rendition as well (on Dragon, the color would be affected much less). The two shots would not match each other. What I would recommend would be to use nearly the same exposure as the moon shot, opening up a stop or so to take anything you really care about out of the noise floor, but no more than that. When developed, the two shots will have similar noise characteristics, and will look like parts of the same scene.