One correction, RedCine is indeed available for build 20. You are no longer limited to RedAlert
Posted by Geoff Bailey on 08/28 at 08:52 AM
An excellent and fair article. Thanks.
Posted by leeberger on 08/28 at 03:03 PM
Art, are you sure you’ve really captured the idea how to work with a CMOS camera and the raw footages? Let me be provocative. If one can’t get more dynamics out from Red One than 9 stops, one does not know how to use the tool.
So, let’s sort this our and start from the fundamentals. As Art says the ‘gamma is a straight line down into the murk of noise’. This is not a drawback but instead the key feature. Consequently the next line from Art: ‘still, setting a light meter between 160-250 seems to work for most people, depending on taste and comfort with noise’ is about driving a car backwards.
Why? Well, as said the gamme curve is a straight line, which in more precise terms is to say that the data recording the amount of light is directly proportional the light on the sensor. The practical advantage is, as long as one is below clipping it does not matter whether one overexposures in the traditional sense or not. For, one can always revert back by scaling in post. Notice, this is true ONLY for the raw image.
But, there’s also a definite advantage in shooting to the right. The point is, every additional stop of light increases the bit depth and the dynamics. Exposuring one stop below the maximum on the right implies one employes only a half of the available tones. Two stops below means only one quarter of the available tones are recorded. As the post systems have always more bits available than what the camera can store, the camera is the critical part of the chain ans one should always try to maximize the amount of tones recorded. Maximizing tones imply also the maximum distance from noise which is what is called dynamics.
The practical workflow is this. Set the exposure using the raw metering. This means that the image on the LCD/EVF is typically too bright. But, to get an idea how the image will appear in post, don’t touch the exposure but instead dial the ISO setting downwards. The ISO setting is then recorded to metadata and when the R3D raw files are opened in REDCINE/REDALERT, this ISO number appears as the default value. Obviously, if needed the value can be changed at will without any sideeffects, thanks to the linearity of the sensor.
It should be now clear that color grading raw files will likely take more time than processing something like RGB images. But, for a good reason: dynamics is maximized. Once the gamma curve is hardwared to the recorded file all these nice features are lost.
So, once more, the ISO setting is only for viewing. It’s silly to first set the ISO value and then make the exposure to depend on the ISO value that is there only for previewing. Better to forget neutral grays and all the traditional issues and focus on the highlights. Everything else can be handled in post. And when doing so nothing is lost but instead quite the opposite: everything is gained.
Posted by .(JavaScript must be enabled to view this email address) on 08/29 at 11:28 AM
I’m wondering/puzzled as to whether Larry is expecting a direct response from the person he refers to in the third-person. Or whether he invites readers to respond.
Posted by wsmith on 08/31 at 10:20 PM
Yes, it was indeed bit difficult to formulate the comment for on one hand would like be generate a cultivated discussion among the readers of this nice forum that benefits everybody. On the other hand this and the previous text by Art raises the question whether he criticizes Red One because he tries to fit the metrics of filmcameras to a modern digital camera having a CMOS sensor.
The obvious risk is that the comments blow out of hands and the discussion forcuses on people instead of techincal issues. Of course it would be nice if Art among other made comments. Especially so, as he show some strong words and opinions in his article so one may perhaps expect he is prepared to defense his view if needed. Still would like to stress out that no intention to hurt anybody’s feelings. It’s all about technical issues that deserve to be clarified.
Posted by .(JavaScript must be enabled to view this email address) on 09/01 at 11:09 AM
Art,
Great article having worked a lot with Red I usually find the sensitivity to be right around 175 in Raw. Thanks for cutting through the hype that is Red Digital and bringing some great answers. DPs, learn your cameras and work within those limitations to make great images! I agree 10000% with software bugs, workflows etc… and I would ad to that build quality for accessories as well..ugh!!
Matt Jeppsens rant from the point of a weary AC, comes from a shoot where we saw a lot of these problems crop up.
Posted by Kendal Miller on 09/01 at 01:31 PM
Larry,
You’re describing exposing to the right, which is a valid way to work with this camera, but not the only way. Under controlled lighting circumstances, using a light meter (and rating the camera at a certain ISO) is a more desirable way to work for many DPs. (And recent revelations about color shifts under different exposure conditions is a very valid reason to work this way.)
And don’t take Art to task for his ‘9 stops’ figure—in another article (about build 20—‘Build 20 Torture Tests’ at http://provideocoalition.com/index.php/aadams/story/red_build_20_torture_tests/) Art describes how he arrives at the 9 stop figure (9.5 under daylight, if I recall correctly). It’s quantitative, and it’s good.
When it comes to latitude, the only subjective part is where you think the image is ‘lost to noise.’ If you tolerate noise better than most, you may think the camera has as many as 11 stops (that’s the most you should claim, though—Red released response curves for the sensor that show 11 as the theoretical maximum).
Personally, I like super-clean images, and I don’t mind 9 stops of latitude. It’s more than I would get out of any other camera I would buy, by a decent margin. (And frankly, the only time I’ve been unhappy with my Red is when it’s been starved for light, and the image gets noisy.)
Cheers,
Ryan
Posted by .(JavaScript must be enabled to view this email address) on 09/22 at 01:11 PM
