BMCC Dynamic Range Test -- INSANE!!!

[cpc]

No, I would slide the proper ND grad into my matte box and expose the whole scene as I intended. IMHO shooting raw should not be use as an excuse to get lazy at ones craft.

That's a bit cocky. There are hundreds of situations where a grad ND won't help because the darker shapes against the bright sky background aren't giving you a good line to align the filter by. Refusing to take advantage of technical achievements just because it seems "wrong" and so that you can use the old ways of handling problems and feel good about it is not very smart. Neither is it always cheap. The OP is absolutely correct that there are conditions when the only other way around the problem would be using a lot of light or huge reflectors. The viewer cares about the end result, not how you got there.
Huge DR is just another great tool. Use it when appropriate.

 

[Matt Gottshalk]

Your footage would most likely be way underexposed if you shot what appeared to be "correct" in-camera. Read more at the bmcuser forums. People have done a lot of tests, and they know how the camera "likes" to be exposed.

I've been shooting raw stills since cameras were able to do so, including on the Red since it was shipping. Just because you can somewhat bring back an image's tones into a somewhat normal exposure does not mean that the sensor "likes" to be overexposed.

That is the wrong way to shoot raw, period.

 

[morgan_moore]

ETTR is not wrong. Its right.

Because the further away you are from the ' dark floor' of the image the better your S/N ratio

ETTR should not be confused with clipping at the top end which is 'wrong'

If an ETTR image looks terrible when brought into the edit suite that just means the LUT on the suite is not correct

I find it weird that people talk about 'recovering' highlights - that just means the are operting on a light LUT and then pulling back which to me is arse about face..

S

 

[Tolin]

This whole discussion to me sounds like a bunch of grumpy men having woken up on the wrong side of the bed.
It's technology - We like technology - And nobody is saying we should over-expose everything just because we can pull it back in post
(And the thing about the sensor "liking" to be over-exposed I interpreted as it recovering highlight detail better than shadow-detail; if it's true or not I don't know, but let's not nitpick on the choice of words)

 

[David W. Jones]

Refusing to take advantage of technical achievements just because it seems "wrong" and so that you can use the old ways of handling problems and feel good about it is not very smart.

Your assumption is incorrect.

Like others here we started shooting raw video with the first Red One M. And we don't use old ways of handling problems to feel good about ourselves, we use tried and true methods because they work and are still relevant today.

 

[Kholi]

Raw coming from Blackmagic's 16-bit linear, stored in a 12-bit DNG file, and unpacks back to 16-bit linear in whatever program you use. Resolve for example.

Being that it's linear data, factual linear, the best thing you can do for your image is to create a "fat negative" which means getting to a place where your brightest brights are near clipping if you want to preserve the entire range of an image.

Exposing "Properly" is a short statement that would get most people in trouble when first using the camera. It's a very very different learning curve, and a lot of the comments here look curious to me after having used the camera for quite an extensive period of time.

Also, as Morgan says, overexposed is not the right term here. The DNGs are actually properly exposed for the camera, nothing is clipped, so everything is there. The image of the couple drinking wine, I've played with it days ago, every bit of sky detail is present.

 

[Kholi]

I've been shooting raw stills since cameras were able to do so, including on the Red since it was shipping. Just because you can somewhat bring back an image's tones into a somewhat normal exposure does not mean that the sensor "likes" to be overexposed.

That is the wrong way to shoot raw, period.

Noone has ever confirmed that Redcode is actually 16-bit Linear, and it is still compressed. The two formats are nothing alike. I shot plenty of RED, as well, and comparing them is wrong.

 

[Kholi]

No, I would slide the proper ND grad into my matte box and expose the whole scene as I intended. IMHO shooting raw should not be use as an excuse to get lazy at ones craft.

I agree about the grad.

Refusing to take advantage of technical achievements just because it seems "wrong" and so that you can use the old ways of handling problems and feel good about it is not very smart.

...


Huge DR is just another great tool. Use it when appropriate.

Exactly.

 

[morgan_moore]

Your assumption is incorrect.

Like others here we started shooting raw video with the first Red One M. And we don't use old ways of handling problems to feel good about ourselves, we use tried and true methods because they work and are still relevant today.

RCx has IMO a huge limitation on the fact that you cannot (AFAIK) do secondaries, unlike Resolve so I dont know how you get round that one, easily

As for a ND grad - it is IMO poor practice to use such on a RAW camera in all but the widest DR scenes

A grad will just induce a form of distortion of the image.. a digitally 'burned' sky offers a lot more options to the colorist especially in situations where the camera moves or the transition between 'scene' and sky is complex

grads are obsolete for 90% of shots IMO (when shooting RAW)

To be truly 'relevant' today one needs to know where a file can go in post, and more challengingly, communitcate and control that from set to final delivery

S

 

[S. Matthews]

Noone has ever confirmed that Redcode is actually 16-bit Linear, and it is still compressed. The two formats are nothing alike. I shot plenty of RED, as well, and comparing them is wrong.

http://www.reduser.net/forum/showth...eems-to-know&p=1067326&viewfull=1#post1067326

Graeme's responses in this thread and the others linked seem to be the closest you will get, if you trust him, which I do. Generally any post by Graeme on this subject is very enlightening. It seems(to me) even the most informed people still have difficulty fully understanding these concepts, and where old terminology and tech no longer apply, and when it does. This (in my opinion) causes mass confusion and pointless endless discussions about numbers that at this point mean very little compared to how the camera and post tools process that data no matter the numbers on the page. Just look at what some of the 8bit cameras are capable of these days compared to their equivalent counterparts from even only a few years ago.

Yes ETTR is important, but the issue is there is a point of no return. It's great to pad the shadows but I would never do it at the sacrifice of clipping the highlights, unless that was the artistic intent. Shadows in digital are much easier to pull back, blown highlights are gone forever. It's a fine balance like everything, and yes expose properly in camera, the ability to recover in post is great but not a crutch to rely on day in day out.

have a good one!

 

[David W. Jones]

Well I guess I'm the odd man out, as I have never noticed any distortion using a Grad filter. 99% of what we do is TV, so we monitor with a Rec 709 LUT and try to expose as close as possible to achieve proper results for that color space.

I think that many folks seem to forget that every sensor camera has a raw stage in the image capturing process. What each camera does with that raw information is how they differ.

 

[Inprogress]

I think there is a lot of confusion. I read up on the whole ETTR thing recently, funny, whilst reading up on Medium Format Film photography - looking forward to that! If you shoot raw, (and I assume DNG which is "raw" as well), you shoot "overexposed" but DO NOT clip the highlights. It comes down to the sensor construction and operation of capturing light information - the highlights have a much (MUCH) higher tonal range than the dark's (excuse the bad terminology). It is a different way of exposing correctly. It is also only useful if you can control that raw file info before it gets compressed. I assume the BMCC shoots and outputs untouched in-camera raw footage, which allows you to pull back the highlights. By shooting "overexposed" you are taking full advantage of the sensor tonal range, as oppose to treading the sensor just like film. I've played with raw pictures today and it is amazing to see this ETTR applied.

So, in essence the same discipline must still be applied to exposing correctly for a image sensor, but the level of exposure compared to film, will be referred to as overexposed. Given I haven't shot the BMCC, only played with various raw pics today where I exposed correctly as learned in film, and overexposed but not clipping highlights, and the results, the creative freedom to play with the image (from ETTR), both darks and lights, especially lights is amazing - and so little noise.

But, if no CC is applied, or a budget doesn't allow for it, expose like film. Oh, and I have no experience with LUT's either, I am merely looking at the technology and how it works and if I shoot raw and can CC (another article discusses why manufacturers still apply film exposure technology to digital sensor technology), then I will apply the ETTR principle cause at least, in my little experience with stills, it makes sense. But I stand to be corrected - I only learned about the figures, haven't had the pleasure of experience.

P.S. Someone want to upload to bits of footage from the BMCC in raw. One exposed with ETTR and the other like film?

 

[morgan_moore]

Well I guess I'm the odd man out, as I have never noticed any distortion using a Grad filter

So if you pan up during a shot you slide the filter down or what?

Im sure your technique is top notch but I dont see how a grad can have the same flexibility as a properly executed digital burn - it just cant.

Of course if you shoot well on location then there are savings to be made in post - and that is where the craft still applies.

But a LUT based on a 'right' image will still render a cleaner file, it's noted as '200' on the BMC or '320' on the Red

The shoot in the OP he (in effect) exposed for 200, imported at 800 and then pulled it to 200 - seems like a strange journey to get back to the same place!

S

 

[Shooter]

There is another angle to RAW and exposing it.

Firstly and to get it out of the way...I acknowledge that the OP was just presenting an example of Dynamic Range. And it is impressive without doubt.

Exposing..
Exposure with the f-stop is not the art of cinematography. It is part of it but the true exposure is done with lighting.

The f-stop is a simple global (shapeless) exposure of the entire image ( under / over notwithstanding) ...but adding, subtracting and controlling light is "the art".

No amount of post will give it "shape and contour" even though the tonal levels can be manipulated in a "flattened: manner.

Sure ... shooting an exterior ( eg landscapes, street scenes, without actual subjects like an actor or a hero car or product) is easy in RAW...until you have to include actual subjects.


Another something else about shooting RAW.

I got bitten by it with an Alexa M shoot in studio last week . Decision (not mine) was to shoot Alexa M with anamorphic lenses. For the soft look . Fine and that suited the subject ( angels in clouds etc...).

But ...I had to shoot all the food and product shots with same camera (type) and lens set.

As you know..anamorphic lenses dont focus very close..so I was forced into using diopters. In one case I had to stack diopters ( +2 and +1).

This seriously messed up the "black level" and because its RAW... I had no way to "fix it". There is no ped adjustment in RAW!

But ...because this was a kind of lens and optics abberation...I could not control it with lighting!

Fix it in Post only solution but never the same as in the shoot.

 

[Kholi]

The primary thing with the camera's using Zerbas (at 100 percent) to judge what's clipping and what's not. At 100, it's showing what the sensor sees as clipping, so if you're not seeing zebras then you're not clipping.

 

[Shooter]

The primary thing with the camera's using Zerbas (at 100 percent) to judge what's clipping and what's not. At 100, it's showing what the sensor sees as clipping, so if you're not seeing zebras then you're not clipping.

Thats not entirely relevant in the world of RAW.

THe principle is correct and eventually one gets back to that though.

 

[morgan_moore]

Shooter

Your reel is humbling in the quality of the lighting and general standard of work, much of the colour excellent.

You are entirely correct that post wont fix lens/light work - for example a flare (or lack of) will only happen properly when caused by the interplay of the glass and the light sources, raw cannot create a shadow or remove one

..but

I really dont understand your recent posts about raw

The BMC zebras are the only tool that works (as is the raw histo on Red) and keeping exposure away from the noise floor creates a cleaner file - I dont really see how you can argue with those concepts

What we are basically seeing here is not an over exposed shot, just a LUT which did not match the way the OP footage was shot and a rathere bizarre workflow of importing the footage too light and then darkening it again!

Raw looks like this..

File + LUT + 'Adjustments Applied Manually in Post' = Look of image

You will see that if the LUT is wrong (too bright) that the 'Adjustments Applied Manually in Post' will be severe

The LUT technology on the BMC is minimal/severely lacking on the current firmware so the need for 'Adjustments Applied Manually in Post' to be severe does not indicate that the footage was shot wrong - it just indicates the lack of LUTs available on set and in post.

S

 

[Shooter]

The BMC zebras are the only tool that works (as is the raw histo on Red) and keeping exposure away from the noise floor creates a cleaner file - I dont really see how you can argue with those concepts

I dont argue against those "fundamentals". I live by them.

Thing is ...the D.R of RAW changes the game somewhat ...but not the principals of photography.

The D.P must control the levels within the D.R of any camera.

Control is the operative word and it is the art of lighting. ie the D.P lights for the levels ( Levels = shadows, gammas, highlights). Color is a secondary consideration.


Eventually what ever DR of what ever camera is sheeted home to fit into the scope and range of the display device. eg Broadcast TV = the IRE Scale. Cinema Digtal Projection / Hardware has its own "specs/limits". The web = who feckin knows!

Any camera can produce imagesn( D.R / levels) outside the specs of any display device/ system.

 

[morgan_moore]

Exactly. Just a raw cam has a bunch of LUTs choosable after the fact. So if you had the wrong look on set that does not mean the exposure was wrong and selecting the right LUT is not 'fixing it in post'

 

[Shooter]

Exactly. Just a raw cam has a bunch of LUTs choosable after the fact. So if you had the wrong look on set that does not mean the exposure was wrong and selecting the right LUT is not 'fixing it in post'

Sure...but exposure to me is not the global exposure of the image as per a simple f-stop setting that affects the entire range of levels across the image.

More important is light (shape/contour/ feel / contrast) and control of the levels ( black thru mids thru to highlights). I use lighting to expose and control these so I view on set with a LUT ( and WFM) that shows me what I am doing in all these regards

No way would I point and shoot in RAW thinking that a colorist will post light a shot for me.