George A.
Well-known member
- Joined
- Apr 11, 2007
- Messages
- 1,965
- Reaction score
- 0
- Points
- 36
- Location
- New York City, NY
- Website
- www.eastrentals.com
Hello,
I'm doing lots of tests to get to know the camera. So far I am extremely pleased - I'm using the latest version of Build 16, and I experienced no codec errors and no bugs (except a one-time booting freeze). The images coming from RED speak for themselves. All in all, I'm super-happy with the camera.
The first - and probably most important step - is deciding on the best (most appropriate) exposure in every possible scenario. I mainly use the luma histogram for that, paying attention to the stop lights as well. I use ISO 320 in almost every situation.
I found that the ETTR method (exposing to the right) works best with RED - as opposed to exposing for the middle grey, as in the film world. Therefore, in a normal scene, I am always erring as much as I can towards to right side of the histogram, at the same time making sure absolutely nothing is clipped (except in difficult high-contrast scenes or the specular highlights - dealing with them is a different discussion). Pulling everything back in post gives amazing results, with very little noise and no clipped whites (if you were careful when ETTRed)
Now for the question... the expose to the right method will get you different results, based on the monitoring space you are in (REDSpace, RED 709 or RAW View). The histograms are, of course, different from each other. My tests show that once the REDSpace histogram starts to clip, you can open the lens half a stop (approximately) until the RED 709 histogram clips. Further on, you can open the aperture an additional half a stop until the RAW histogram clips. Since I suppose the RAW View gives you the accurate data of when the sensor actually clips, then there is a difference of about one full stop between clipping in REDSpace and the actual clipping in the RAW View and the sensor itself.
The first thought was that one should monitor in REDSpace (since it just looks closer to the real scene) but set the exposure in accordance with what the RAW View is showing. Again, using the ETTR, one wants to be as close as possible to the right edge of the histogram, without clipping - and REDSpace "makes" you close up your aperture roughly one full stop, defeating this very purpose. Monitoring in RAW takes you right where you want to be, since you accurately know where the exposure is "placed" on the sensor itself.
On second though, I can see the other side of the coin. Is it really best to be right at the edge of the sensor, even if you're not clipping it when shooting? Later on in post, one might easily clip the whites when playing with timing or when transcoding to an editable format, even without touching anything at all (due to different gammas and so on). Maybe it's best to set your exposure on REDSpace or RED 709 after all - still using ETTR - (just double-checking the RAW View for peace of mind), and use that one stop difference as a buffer - making 110% sure you don't clip anything, or using that buffer to gently roll-off whites in post etc
I'm only beginning to delve into the finer aspects of processing RED footage... but since everyone is saying you can dig even deeper into the shadows with Build 16, maybe the second option is indeed the best one. Of course, i'll be doing even more tests
I still don't have a definitive answer, so I'm throwing it out into the open. What viewing space are people using out there when setting exposure? Pros and cons with each one?
George
P.S. Also, my proxies look quite darker than what I'm seeing on the monitor when i shoot. Even when monitoring in REDSpace. Is that still an issue for everybody?
I'm doing lots of tests to get to know the camera. So far I am extremely pleased - I'm using the latest version of Build 16, and I experienced no codec errors and no bugs (except a one-time booting freeze). The images coming from RED speak for themselves. All in all, I'm super-happy with the camera.
The first - and probably most important step - is deciding on the best (most appropriate) exposure in every possible scenario. I mainly use the luma histogram for that, paying attention to the stop lights as well. I use ISO 320 in almost every situation.
I found that the ETTR method (exposing to the right) works best with RED - as opposed to exposing for the middle grey, as in the film world. Therefore, in a normal scene, I am always erring as much as I can towards to right side of the histogram, at the same time making sure absolutely nothing is clipped (except in difficult high-contrast scenes or the specular highlights - dealing with them is a different discussion). Pulling everything back in post gives amazing results, with very little noise and no clipped whites (if you were careful when ETTRed)
Now for the question... the expose to the right method will get you different results, based on the monitoring space you are in (REDSpace, RED 709 or RAW View). The histograms are, of course, different from each other. My tests show that once the REDSpace histogram starts to clip, you can open the lens half a stop (approximately) until the RED 709 histogram clips. Further on, you can open the aperture an additional half a stop until the RAW histogram clips. Since I suppose the RAW View gives you the accurate data of when the sensor actually clips, then there is a difference of about one full stop between clipping in REDSpace and the actual clipping in the RAW View and the sensor itself.
The first thought was that one should monitor in REDSpace (since it just looks closer to the real scene) but set the exposure in accordance with what the RAW View is showing. Again, using the ETTR, one wants to be as close as possible to the right edge of the histogram, without clipping - and REDSpace "makes" you close up your aperture roughly one full stop, defeating this very purpose. Monitoring in RAW takes you right where you want to be, since you accurately know where the exposure is "placed" on the sensor itself.
On second though, I can see the other side of the coin. Is it really best to be right at the edge of the sensor, even if you're not clipping it when shooting? Later on in post, one might easily clip the whites when playing with timing or when transcoding to an editable format, even without touching anything at all (due to different gammas and so on). Maybe it's best to set your exposure on REDSpace or RED 709 after all - still using ETTR - (just double-checking the RAW View for peace of mind), and use that one stop difference as a buffer - making 110% sure you don't clip anything, or using that buffer to gently roll-off whites in post etc
I'm only beginning to delve into the finer aspects of processing RED footage... but since everyone is saying you can dig even deeper into the shadows with Build 16, maybe the second option is indeed the best one. Of course, i'll be doing even more tests
I still don't have a definitive answer, so I'm throwing it out into the open. What viewing space are people using out there when setting exposure? Pros and cons with each one?
George
P.S. Also, my proxies look quite darker than what I'm seeing on the monitor when i shoot. Even when monitoring in REDSpace. Is that still an issue for everybody?