Barry Goyette
November 9th, 2015, 10:52 AM
Spent a good chunk of yesterday in my studio with a 15-16 stop test scene cycling though the myriad of gamma and color space options available on the new camera. I had two goals. One was to compare Clog on the original c300, which is what I'm used to, with Clog on this camera. The second goal was to get a handle on CLog2, to see what it's doing, and to see how much flexibility it gives, but most importantly to learn its quirks (and it has a few) so I don't make any huge mistakes on the first job or two with the camera.
Both tests are ongoing, and given my work schedule the next few weeks, I'm not sure whether I can do a detailed post or not. So, I thought given that I'm up early today, I'd take a few minutes to mention a some of the takeaways from what I've seen so far.
CLog -- New versus Old
This test needs a lot more work, but right off the bat I'm seeing lower noise overall with the new camera...but this comes with a caveat. With the old Clog we saw a relatively uniform noise profile across the tonal range of a given scene -- almost like it was added in (it was :-). The new camera exhibits about the same noise as the old camera below 25 ire and generally… shockingly clean results above that point. even at higher ISO’s. A slightly lower contrast scene shot "to the right" at ISO 20,000 saw almost no noise except in the deepest shadows that would normally get crushed in the grade. Phenomenal. Overall, there seemed to be more color range and saturation in the lower end of the spectrum on the new camera.
CLog 2 - All New
Canon is really stretching the limits of both it's sensor and the 4:2:2 flavor of it's codec (my cards arrived but no reader, so I'm currently limited to capturing pro-res on the Atomos), it's really important to get a couple of things right when exposing for clog2.
During the lead-up period to the C300 Mk II release, there were a number of early user videos, many of which, in my terribly arrogant opinion, had poor, or certainly challenged color. Sometimes this could be put off to the grade, but something told me there was something going on, on the camera side, that was causing some of these results. Many of the videos had a reddish tinge, but one video in particular had flummoxed me, shots of a yoga girl in Japan, as it appeared to have strange crossovers in much of the footage...greenish shadows with red mid-tones. At the time I put it off to inexperienced grading, but then on my first shots here in a "very" neutral set, I was seeing exactly same thing. Red overall with a sudden green shift at about 5 ire.
This freaked me out at first. But it turned out to be two things, both unsurprising, but still worth mentioning.
White Balance
It can be difficult to judge white balance when viewing Log Gammas, and Clog 2 is an order or two more difficult in this respect than the original Clog. (I wasn’t using a monitor LUT as I wanted to see what the native log tones were doing). The scene I set up was lit predominantly with a single Area48 LED with the daylight panel installed. These lights are among the most accurate LED on the market with a TLCI hovering between 95-98. With my C300, I've come to prefer a manual Kelvin setting of 5600 with these lights, and this is how I set the new camera at first. After that I cycled through the new AWB setting (it registered 5300K), the daylight preset, and a range of Kelvin levels. I also applied a manual WB which surprisingly, settled on a level of 4300 for these lights. From my experience with the C300, I've learned to be skeptical with led's and manual white balance and the fact that the AWB and manual WB produced startlingly different results combined with typically good c300 results at 5600K, I shot most of the test using the manually entered 5600k. Once I got the footage into my system though, I was in for a rude surprise: the daylight preset, AWB, and 5600 Kelvin settings were all quite red compared to the C300 and every other camera I own. The manual WB on the other hand was in fact quite accurate . I haven’t so much as taken this thing out to shoot a second of actual daylight, but I think I understand now why I've seen so much red in many of the videos posted so far.
ABB -- Auto Black Balance
With my C300 I mostly used this feature to clean up (or pre-empt) any hot pixels that occasionally occurred. With Clog2, it turns out that this is an essential procedure in terms of color balance, and one that requires perhaps more than a "do it and forget it" attitude. After my first round of shots produced the funny red with green shadows (the green shadow effect isn't subtle and is visible on the camera EVF and LCD even while in Clog2), I ran an ABB with my favored, lazy, lens cap technique (this is the big, coffee can lid lens cap that comes with the cinema lenses -- which has always worked fine on the C300) and watched on my atomos as it cycled through the test. When it finished I noticed now that the shadows were now red instead of green. Several other cycles of ABB using this technique produced similar results. At that point I decided to do what the camera says (use a body cap) and wah-lah...finally I'm seeing a neutral shadow. Certainly the argument goes that you should always ABB, I just think its worth noting that the effects of “not” black balancing when using CLog2 will be more noticeable than other gammas on other Cinema Eos cameras.... and hang on to your body cap.
ISO, Noise, and CLog2
This is a broad topic and one that’s been covered in a lot of places from Cinema5d to CML. I don’t intend to challenge any of what’s been printed thus far. I had a couple of specific goals in mind as I ran through the various gamma settings and ISOs. First up, like all Cineon type curves, CLog2 shows quite a bit of noise in the shadows compared to other gammas, even at base (800) iso. One could argue that the noise in Clog 2 is a bit more than other cameras by other manufacturers, and is more “electronic” than what we’re used to with canon’s 8 bit cameras, and I’d agree with that, adding that in minimizing noise reduction with this camera, Clog2’s shadows seem to harbor more detail than many of the other cameras with similar curves. Also, when properly black balanced, this noise is mostly luma noise, making it much easier to grade than some cameras I’ve seen.
Taking it higher.
One of the interesting things I noticed was that in high ISO situations, with the same exposure, sometimes there’s benefit to shooting at a Higher ISO than you need, as long as you’re not clipping anything. Canon is doing some pre-processing gain on the sensor and in my tests, I found that I could use ISO to “shoot to the right” and achieve a result that was a bit cleaner at a given exposure level (f-stop). I also did a shot at 51,200 and cleaned it up with Neat Video and it really looks stunning. Virtually no detail loss and plenty of dynamic range.
Below the belt - The Low End.
Base ISO on this camera is supposed to hit a happy medium of shadow noise and dynamic range. But some might find that the shadow noise at this setting bugs them, (and require a certain amount of hand-holding when delivering ungraded footage to a client which I occasionally do). The general rule here is to rate at 800 but to overexpose by a stop or more, and this works, but in high dynamic range scenes, one may not want to give up that stop or more of headroom to control noise in the shadows. I experimented with lowering the ISO down as far as 200 (the native ISO of the sensor is actually 100), and I was quite surprised by what happened. As you lower the ISO below 800, the camera starts lowering the maximum IRE level accordingly. Essentially whatever maximum exposure you’d use at ISO 800 is the same at ISO 500…400, 320, and 200…they all peak at the same Fstop. Whats interesting though as you lower the ISO below 800, is the overall dynamic range stays the same, the shadows compress a little and the noise characteristic in shadows changes dramatically (it gets really pretty). Essentially you get a look that is similar to the original Clog, but with a full 15 stops of range and very clean shadows, the caveat is that more of that range is below 18% grey than base iso (similar to Clog but with more highlight range). Interestingly, when I graded ISO 800 and 200 samples back to “normal”, meaning I was stretching the mid tones up on the 200 clip and pulling the shadows down on the 800 clip…the result was still a significantly cleaner shadow on the 200 clip. Again..we’re talking about the same exposure here. I have to say I really liked the look of the lower ISO’s on this camera and I think for any low key situation or anything with “lower than 15 stops DR), I’ll be choosing to shoot at lower ISO rather than the recommended “shoot 800 and meter at 400” method.
I’m thinking this approach might be a bit controversial, (and I haven’t tested this in the real world or with real skin tones) so I’m interested in hearing what other users have to say.
Much more to come on this topic.
Both tests are ongoing, and given my work schedule the next few weeks, I'm not sure whether I can do a detailed post or not. So, I thought given that I'm up early today, I'd take a few minutes to mention a some of the takeaways from what I've seen so far.
CLog -- New versus Old
This test needs a lot more work, but right off the bat I'm seeing lower noise overall with the new camera...but this comes with a caveat. With the old Clog we saw a relatively uniform noise profile across the tonal range of a given scene -- almost like it was added in (it was :-). The new camera exhibits about the same noise as the old camera below 25 ire and generally… shockingly clean results above that point. even at higher ISO’s. A slightly lower contrast scene shot "to the right" at ISO 20,000 saw almost no noise except in the deepest shadows that would normally get crushed in the grade. Phenomenal. Overall, there seemed to be more color range and saturation in the lower end of the spectrum on the new camera.
CLog 2 - All New
Canon is really stretching the limits of both it's sensor and the 4:2:2 flavor of it's codec (my cards arrived but no reader, so I'm currently limited to capturing pro-res on the Atomos), it's really important to get a couple of things right when exposing for clog2.
During the lead-up period to the C300 Mk II release, there were a number of early user videos, many of which, in my terribly arrogant opinion, had poor, or certainly challenged color. Sometimes this could be put off to the grade, but something told me there was something going on, on the camera side, that was causing some of these results. Many of the videos had a reddish tinge, but one video in particular had flummoxed me, shots of a yoga girl in Japan, as it appeared to have strange crossovers in much of the footage...greenish shadows with red mid-tones. At the time I put it off to inexperienced grading, but then on my first shots here in a "very" neutral set, I was seeing exactly same thing. Red overall with a sudden green shift at about 5 ire.
This freaked me out at first. But it turned out to be two things, both unsurprising, but still worth mentioning.
White Balance
It can be difficult to judge white balance when viewing Log Gammas, and Clog 2 is an order or two more difficult in this respect than the original Clog. (I wasn’t using a monitor LUT as I wanted to see what the native log tones were doing). The scene I set up was lit predominantly with a single Area48 LED with the daylight panel installed. These lights are among the most accurate LED on the market with a TLCI hovering between 95-98. With my C300, I've come to prefer a manual Kelvin setting of 5600 with these lights, and this is how I set the new camera at first. After that I cycled through the new AWB setting (it registered 5300K), the daylight preset, and a range of Kelvin levels. I also applied a manual WB which surprisingly, settled on a level of 4300 for these lights. From my experience with the C300, I've learned to be skeptical with led's and manual white balance and the fact that the AWB and manual WB produced startlingly different results combined with typically good c300 results at 5600K, I shot most of the test using the manually entered 5600k. Once I got the footage into my system though, I was in for a rude surprise: the daylight preset, AWB, and 5600 Kelvin settings were all quite red compared to the C300 and every other camera I own. The manual WB on the other hand was in fact quite accurate . I haven’t so much as taken this thing out to shoot a second of actual daylight, but I think I understand now why I've seen so much red in many of the videos posted so far.
ABB -- Auto Black Balance
With my C300 I mostly used this feature to clean up (or pre-empt) any hot pixels that occasionally occurred. With Clog2, it turns out that this is an essential procedure in terms of color balance, and one that requires perhaps more than a "do it and forget it" attitude. After my first round of shots produced the funny red with green shadows (the green shadow effect isn't subtle and is visible on the camera EVF and LCD even while in Clog2), I ran an ABB with my favored, lazy, lens cap technique (this is the big, coffee can lid lens cap that comes with the cinema lenses -- which has always worked fine on the C300) and watched on my atomos as it cycled through the test. When it finished I noticed now that the shadows were now red instead of green. Several other cycles of ABB using this technique produced similar results. At that point I decided to do what the camera says (use a body cap) and wah-lah...finally I'm seeing a neutral shadow. Certainly the argument goes that you should always ABB, I just think its worth noting that the effects of “not” black balancing when using CLog2 will be more noticeable than other gammas on other Cinema Eos cameras.... and hang on to your body cap.
ISO, Noise, and CLog2
This is a broad topic and one that’s been covered in a lot of places from Cinema5d to CML. I don’t intend to challenge any of what’s been printed thus far. I had a couple of specific goals in mind as I ran through the various gamma settings and ISOs. First up, like all Cineon type curves, CLog2 shows quite a bit of noise in the shadows compared to other gammas, even at base (800) iso. One could argue that the noise in Clog 2 is a bit more than other cameras by other manufacturers, and is more “electronic” than what we’re used to with canon’s 8 bit cameras, and I’d agree with that, adding that in minimizing noise reduction with this camera, Clog2’s shadows seem to harbor more detail than many of the other cameras with similar curves. Also, when properly black balanced, this noise is mostly luma noise, making it much easier to grade than some cameras I’ve seen.
Taking it higher.
One of the interesting things I noticed was that in high ISO situations, with the same exposure, sometimes there’s benefit to shooting at a Higher ISO than you need, as long as you’re not clipping anything. Canon is doing some pre-processing gain on the sensor and in my tests, I found that I could use ISO to “shoot to the right” and achieve a result that was a bit cleaner at a given exposure level (f-stop). I also did a shot at 51,200 and cleaned it up with Neat Video and it really looks stunning. Virtually no detail loss and plenty of dynamic range.
Below the belt - The Low End.
Base ISO on this camera is supposed to hit a happy medium of shadow noise and dynamic range. But some might find that the shadow noise at this setting bugs them, (and require a certain amount of hand-holding when delivering ungraded footage to a client which I occasionally do). The general rule here is to rate at 800 but to overexpose by a stop or more, and this works, but in high dynamic range scenes, one may not want to give up that stop or more of headroom to control noise in the shadows. I experimented with lowering the ISO down as far as 200 (the native ISO of the sensor is actually 100), and I was quite surprised by what happened. As you lower the ISO below 800, the camera starts lowering the maximum IRE level accordingly. Essentially whatever maximum exposure you’d use at ISO 800 is the same at ISO 500…400, 320, and 200…they all peak at the same Fstop. Whats interesting though as you lower the ISO below 800, is the overall dynamic range stays the same, the shadows compress a little and the noise characteristic in shadows changes dramatically (it gets really pretty). Essentially you get a look that is similar to the original Clog, but with a full 15 stops of range and very clean shadows, the caveat is that more of that range is below 18% grey than base iso (similar to Clog but with more highlight range). Interestingly, when I graded ISO 800 and 200 samples back to “normal”, meaning I was stretching the mid tones up on the 200 clip and pulling the shadows down on the 800 clip…the result was still a significantly cleaner shadow on the 200 clip. Again..we’re talking about the same exposure here. I have to say I really liked the look of the lower ISO’s on this camera and I think for any low key situation or anything with “lower than 15 stops DR), I’ll be choosing to shoot at lower ISO rather than the recommended “shoot 800 and meter at 400” method.
I’m thinking this approach might be a bit controversial, (and I haven’t tested this in the real world or with real skin tones) so I’m interested in hearing what other users have to say.
Much more to come on this topic.