Barry_Green
Moderator
Today's testing was to see what the 7D can deliver as far as total dynamic range. To do so, I used my brand-spanking-new DSC Labs Billups VF/X 28 chart, and also my DSC Labs ToNi chart, in combination.
Billups VF/X chart:
ToNi chart:
Now, you'll see that both charts offer grayscales, so -- why use two? Mainly because I really wanted to design a scene that has extreme differences between light and dark. Massive, huge differences. So, I put the Billups chart over the ToNi, and lit the scene with some low-wattage fluorescents, and then I aimed two HMI lights at the Billups (and flagged the light off the ToNi). That way I had about 725 watts of HMI blasting the Billups, and 130 watts of fluorescent barely lighting up the ToNi. The ToNi has a cavity black on the left side, that's actually a hole in the chart with a cave made from black velvet (or something) so it's a complete "black hole" of light, giving us a perfect absolute-black reference.
I then adjusted the lights, spotting in or flooding out, and testing with my spotmeter, until I got the range I was looking for. In this test scene we have readings ranging from f/0.7 all the way up to f/38 (at ISO 200). That's a total range covering 11 and a half stops. That should be more than enough to exceed the range of the 7D, thus we should be able to determine the 7D's actual ability to render from dark to light using this setup.
(and yes, after going through this process, I also immediately ordered a Stouffer 41-step wedge to make this so much easier next time!)
But, anyway, back to the 7D... I configured the 7D as Stu Maschwitz recommends on his ProLost blog for flattening the 5D. Contrast -2, highlight tone priority on, and Neutral style. I put the camera on 1080/24p mode, used the stock lens, ISO 200.
I then shot the charts at multiple shutter speeds, which caused the predictable but annoying artifact of rolling-shutter bands scrolling throughout the footage on any shutter speed higher than 1/60th. Grrr. But, I worked around it, selecting frames where the bands weren't visible on the chips of the charts. I chose multiple exposures to see how far in the shadows we could see before crushing, and then by stopping down, how far into the highlights before clipping, and taking multiple shots into account, trying to figure out the actual f-stop differential between being able to render detail, and losing detail completely. The goal here was absolutely to push the highlights into blowing out, and the darks into crushing; if we can see blown-out highlights and crushed darks, then we know that what remains in between them is the usable dynamic range.
So this chart is not going to look as "pretty" as the marketing chart at the top of the post -- it's not supposed to. We want to see some blown-out parts, that tells us we've hit the max. We want to see some crushed parts, that tells us we've hit the minimum.
And note, this is not about "usable" dynamic range, this is about absolute maximum, top-to-bottom, the same as would have been found using a Stouffer. For comparison, the Red One was found to deliver 11.3 stops of total dynamic range from top to bottom; an EX1 does about 10.5 stops, an HPX300 does about 10 stops.
So how did the 7D do?
On this chart, it has blown out highlights on the top left of the Billups, and what appears to be crushed shadow detail on the darks in the ToNi. From visual observation, I'd say that it looks like we're seeing detail in all the light chips of the ToNi, crushed at the first dark chip; I put an "X" on the darkest chip I think is individually discernible. That means we're rendering down to the f/2.0 level. And we can discern detail in the whitest chips of the Billups VF/X over to just barely the 4th-brightest (with the "X"), which puts the f-stop reading at about f/22.5. That gives us a total visible range of about 7.5 stops.
But wait -- there's more! What happens if we use Premiere's Shadow/Highlight effect to crank up the shadows and drop down the highlights -- is there more detail there than meets the eye? There sure is! Check this out:
There's definitely a lot more shadow detail there than met the eye initially, and this is detail that can be yanked back in post. Looks like there's nothing in the highlights to be recovered (which, frankly, is to be expected; I was using Highlight Tone Priority and should have already been extracting the most from the highlights). But in the shadows, we can dig in quite a ways, all the way to the f/1.2 mark. That gives us another 1.5 stops of dynamic range, making the total range the camera can image, about 9 stops.
Is that 9 "usable" stops? That's up to you to decide; those deeper stops come at the expense of quite a bit of noise. But they're there.
On a Stouffer, the Red One resolves about 11.3 stops, but most users seem to put the "usable" range at about 9. On the 7D, I'd say it resolves 9 stops, with a "usable" range of about 7.5, maybe 8. Not too shabby, seeing as the 7D costs 1/10th the price.
How does it compare against a comparably-priced video camera? I ran the same test with the HMC40, and found the HMC40 (with DRS on, equiv. to HTP, in CINE-D) to deliver a range from about f/32 on down to f/1.7, or a total range of about 9.5 stops, and when digging into the shadows with Premiere's Shadow/Highlight feature enabled, you could technically discern all the way down to the f/1 block but that's really pushing it. It also brought out some cleaner separation of the top steps, adding maybe 1/2 stop at the top and bottom, bringing the total potential dynamic range to 10.5 stops. However, the shadows are so noisy and grainy that I wouldn't say it's got 10.5 "usable" stops. In terms of visibly usable stops, with DRS off, I'd call it 8.5 without DRS, 9.5 with DRS.
Billups VF/X chart:
ToNi chart:
Now, you'll see that both charts offer grayscales, so -- why use two? Mainly because I really wanted to design a scene that has extreme differences between light and dark. Massive, huge differences. So, I put the Billups chart over the ToNi, and lit the scene with some low-wattage fluorescents, and then I aimed two HMI lights at the Billups (and flagged the light off the ToNi). That way I had about 725 watts of HMI blasting the Billups, and 130 watts of fluorescent barely lighting up the ToNi. The ToNi has a cavity black on the left side, that's actually a hole in the chart with a cave made from black velvet (or something) so it's a complete "black hole" of light, giving us a perfect absolute-black reference.
I then adjusted the lights, spotting in or flooding out, and testing with my spotmeter, until I got the range I was looking for. In this test scene we have readings ranging from f/0.7 all the way up to f/38 (at ISO 200). That's a total range covering 11 and a half stops. That should be more than enough to exceed the range of the 7D, thus we should be able to determine the 7D's actual ability to render from dark to light using this setup.
(and yes, after going through this process, I also immediately ordered a Stouffer 41-step wedge to make this so much easier next time!)
But, anyway, back to the 7D... I configured the 7D as Stu Maschwitz recommends on his ProLost blog for flattening the 5D. Contrast -2, highlight tone priority on, and Neutral style. I put the camera on 1080/24p mode, used the stock lens, ISO 200.
I then shot the charts at multiple shutter speeds, which caused the predictable but annoying artifact of rolling-shutter bands scrolling throughout the footage on any shutter speed higher than 1/60th. Grrr. But, I worked around it, selecting frames where the bands weren't visible on the chips of the charts. I chose multiple exposures to see how far in the shadows we could see before crushing, and then by stopping down, how far into the highlights before clipping, and taking multiple shots into account, trying to figure out the actual f-stop differential between being able to render detail, and losing detail completely. The goal here was absolutely to push the highlights into blowing out, and the darks into crushing; if we can see blown-out highlights and crushed darks, then we know that what remains in between them is the usable dynamic range.
So this chart is not going to look as "pretty" as the marketing chart at the top of the post -- it's not supposed to. We want to see some blown-out parts, that tells us we've hit the max. We want to see some crushed parts, that tells us we've hit the minimum.
And note, this is not about "usable" dynamic range, this is about absolute maximum, top-to-bottom, the same as would have been found using a Stouffer. For comparison, the Red One was found to deliver 11.3 stops of total dynamic range from top to bottom; an EX1 does about 10.5 stops, an HPX300 does about 10 stops.
So how did the 7D do?
On this chart, it has blown out highlights on the top left of the Billups, and what appears to be crushed shadow detail on the darks in the ToNi. From visual observation, I'd say that it looks like we're seeing detail in all the light chips of the ToNi, crushed at the first dark chip; I put an "X" on the darkest chip I think is individually discernible. That means we're rendering down to the f/2.0 level. And we can discern detail in the whitest chips of the Billups VF/X over to just barely the 4th-brightest (with the "X"), which puts the f-stop reading at about f/22.5. That gives us a total visible range of about 7.5 stops.
But wait -- there's more! What happens if we use Premiere's Shadow/Highlight effect to crank up the shadows and drop down the highlights -- is there more detail there than meets the eye? There sure is! Check this out:
There's definitely a lot more shadow detail there than met the eye initially, and this is detail that can be yanked back in post. Looks like there's nothing in the highlights to be recovered (which, frankly, is to be expected; I was using Highlight Tone Priority and should have already been extracting the most from the highlights). But in the shadows, we can dig in quite a ways, all the way to the f/1.2 mark. That gives us another 1.5 stops of dynamic range, making the total range the camera can image, about 9 stops.
Is that 9 "usable" stops? That's up to you to decide; those deeper stops come at the expense of quite a bit of noise. But they're there.
On a Stouffer, the Red One resolves about 11.3 stops, but most users seem to put the "usable" range at about 9. On the 7D, I'd say it resolves 9 stops, with a "usable" range of about 7.5, maybe 8. Not too shabby, seeing as the 7D costs 1/10th the price.
How does it compare against a comparably-priced video camera? I ran the same test with the HMC40, and found the HMC40 (with DRS on, equiv. to HTP, in CINE-D) to deliver a range from about f/32 on down to f/1.7, or a total range of about 9.5 stops, and when digging into the shadows with Premiere's Shadow/Highlight feature enabled, you could technically discern all the way down to the f/1 block but that's really pushing it. It also brought out some cleaner separation of the top steps, adding maybe 1/2 stop at the top and bottom, bringing the total potential dynamic range to 10.5 stops. However, the shadows are so noisy and grainy that I wouldn't say it's got 10.5 "usable" stops. In terms of visibly usable stops, with DRS off, I'd call it 8.5 without DRS, 9.5 with DRS.
Last edited:
