6sec / 141 frames at 24fps = 704 MB. I can imagine that after first trials a lot of people will switch to ProRes. But really awsome.

@kholi @ shian

Overexposed is an interesting term. Is it overexposed ? I would say no because nothing is clipped. You're saying it LOOKS overexposed because...that's how the preview looks when you open it in Resolve ?

I guess the exposure methodology of protecting highlights means many tend to ETTL but I have a different way of setting exposure.

To me, and I do this with Alexa and RED, I want to know what is clipping, so using Zebra or Other similar tools, I set it to 100%. ON the BMCC the Zebra is based on SENSOR clipping, not a 709 LUT of the video space.

So when shooting RAW, the 100% zebra absolutely tells me when the sensor is really clipping.

Once I know what's clipping in the scene I then "choose" my exposure based on what I'm happy to let clip. In a round about way I guess it's like ETTR, but I like to have as much "meat" of my exposure above the noise floor. it depends on the screen though and that's something you have to judge. The DR of the scene itself will affect my judgement of where to "position" the exposure.

So yes Sian, the numbers you're looking at on the waveform are about right. I actually set the exposure in this by looking at her white dress when she was standing up. I then used the zebra to see where it was clipping, and then gave myself some headroom by backing off another stop. The sparkles on her dress where still clipping but I decided that was OK and let those "go". There's no point in trying to hold that kind of highlight detail obviously at the cost of the skin tones.

I'd love to ask both of you...you're reacting against the high levels of skin tones..being up higher in the exposure scale in this shot. Is there a downside ? If nothing is clipped, isn't all the information there ? How does it change the way things behave ?

There are other low keys scenes where the exposure isn't set in the same way at all. In those scenes (night, low con) then you make a much more subjective judgement of exposure. Wide open and highest ISO and trust the camera will hold the DR in the highlights of point sources !!

jb

the sensor is very small, there is no flip lcd, the lens mount is the big mistake because it could have a smaller flange to allow more different brand lenses to be used including vintage 35mm, m43 lenses and sony nex lenses, why not? not just pl canon eos and nikon as it does. wide angle with this camera is hard to do due to the crop factor and big flange. no way to go. change the sensor size or change the flange, or both. if it could use m43 lenses the wide angle would be more easy to do. or need to use a sigma 8-16mm nikon mount. it would be better to buy the bolex raw. it allows lots and lots of lenses to be used.

Clearly it's not overexposed otherwise the detail would be completely gone. I'm just attempting to understand the DNG color space.

With R3D RAW we get something that looks like a 35mm film scan. Its in LOG (or we can put it in log) there's a fine white "mist" over the image, that those of us who have worked in film are used to seeing, we know exactly where everything will sit on the IRE scale in 32bit floating point.

These images don't behave that way. They come in looking rather "overexposed" this is easily corrected using the interpretation module. But even then there is something very odd going on in the skin tone range. I can't quantify it, in any other way but to say it's not behaving the way that it should. I should be able to place skin tone around 60-70IRE and have it look perfect. For some reason, it looks muted. Like someone crushed over-exposed highlights. I'm still playing with them, and I'm sure by the time I figure out what's going on, someone from BMD will come along and explain it. Probably a simple gamma setting or curve adjustment will compensate for whatever it is. But it is just damn peculiar.

I wasn't accusing anyone of bad cinematography. I just wanted to know what the image looked like to you vs what the camera is recording via DNG.

But for the fact that it is reeeeeeeally sluggish, I love working with the footage.

But like I said from the beginning what will make me wait to buy is more lens options, and the manifestation of the options that are "coming". For those who ordered, enjoy. It seems BMD done good. Here's hoping the damn thing ain't as fickle as the RED can be.

@shian

I get what you're saying, as well. Although RED doesn't immediately come in as flat, and these days we're using REdgamma3 etc to do things. ARe you guys using REDlogfilm or whatever?

Again, I also understand where John's coming from and what'll really drive things home is having hands on camera and running exposure tests to see where you want to start protecting flesh.

Either way, it still looks damned good.

I'm top lister over at Markertek (right under Brian), just hoping we get some info on when it's going to ship officially. With a handful of days left in August, doesn't seem likely? Maybe.

@shian I certainly wasn't taking your comments as criticism of the photography. I'm still learning how to get the most out of the camera and with the restrictions until now, it's been very difficult to get feedback from colourists and I really want to hear how the footage is sitting with those that I have a key creative relationship with.

Sounds like you're working in AE at the moment ? I'd love to hear if you have Resolve how it performs there in terms of system performance. Lots seem to think it handles DNG's better than Adobe (in terms of speed) ! I'm sure that will change soon.

The sensor is recording 16 bit linear, which is converted to 12 bit log, recorded, then expanded back out to 16 bit when you load the DNG's in Resolve. I assume it's the same in AE, but perhaps there is a difference. - Though you seem to be saying you can still recover all the range ?

Would you say that there's something wrong with the skin tones at that exposure level, or that it doesn't behave the way you expect, but you can get it looking right once you've made adjustments ?

jb

I prefer getting 60-70IRE in the first place if possible. Not saying ETTR is a bad thing though.

Maybe this is a good camera to advertising. advertising uses big crews, lights, needs the best image ever to show products, make people dream and droll, to full hd broadcast or to theaters. also a feature film with money to pay the crew, the raw storage, and so on... today low light means iso 12800, 1600 is not enough.

@johnbrawley yeah the free (1920x1080) version of Resolve doesn't want to play with the DNG files.

But if you look at the first couple of pics I posted in that first thread (not embedded just linked below the post) that when adjusting skin tone to where skin tone is supposed to go, it looks squashed. It doesn't make sense unless there's something going on in the interpretation phase that is placing luminance values into the spectrum incorrectly.

@bwhitz nicely done

@kholi we're purists we always work in LOG.

@shian

The CinemaDNG's work fine in my free Resolve Lite. I have the latest beta3 Mac version.

Made a grade, pretty natural looking

Awesome stuff for 3k!

@jhero What version of Davinci Resolve you using? I'm using Resolve 9b3 and I'm having using issues importing the clips on OSX 10.7.2 The clips come out green and no playback..

@TrackZillas I'm on mountain lion if it helps