I agree. Middle grey is a point value surely. It doesn't occupy a block of an entire 1 stop of range. If a particular aperture gives you exactly middle grey and clicking 6 stops under gives you pure black (or noise) and clicking 4 stops up blows the sensor out, that is 10 stops of range, not 11. You only need to count up from -6: -5 is 1, -4 is 2 etc.
My own Sony A7s test - 4K resolution, dynamic range, noise and GH4 comparison
- Thread starter macgregor
- Start date
I agree. Middle grey is a point value surely. It doesn't occupy a block of an entire 1 stop of range. If a particular aperture gives you exactly middle grey and clicking 6 stops under gives you pure black (or noise) and clicking 4 stops up blows the sensor out, that is 10 stops of range, not 11. You only need to count up from -6: -5 is 1, -4 is 2 etc.
Illya Friedman
Well-known member
Hot Rod Cameras keeps the Simple Cine version of these in stock. Far higher quality than ebay, and the price iof most adapters are generally under $60 USD.
cpc
Well-known member
Look at it like a backlit grey scale DR test if it´s easier for you to understand it:
You count all the stops where there's gradiation. One of them corresponds to mid gray and of course you need to count that in as well.
Yeah, I figured that.
I still think that you should count intervals (of light doubling) though, and not points, because we are measuring a tonality range which is a continuous distribution. Seems more natural to me. Some older books actually use step isntead of stop exactly for the purpose of avoiding misunderstandings. Stop is overloaded with meanings and people seem to associate it with aperture stops/clicks even when it is supposed to be a measure of tonal/light range.
paulcurtis
Veteran
I'd pretty much concur about dynamic range results, 'useable' dynamic range.
I'd only add is that this difference between technical range and actual range is the same for most cameras out there, not just the A7s. The FS700 has lofty claims that just aren't there along with lots of others. When you are able to dig about in the RAW from cameras you can really see compromises.
As for stills mode i think people have got the wrong idea about what the A7s is. The bigger pixels can't really offer that much more than an A7 down sampled, the surface area on a CMOS is pretty much the same.
What the A7s offers is the ability to run at higher ISOs with comparatively little noise. So the difference in stills between 3200 and 12000 isn't as great as it is on other cameras. But the range is pretty much the same. So i have no problem running 12k ISO whereas i'd never go beyond 1600 before.
Also the choice of 3200 might not just be for Slog. In stills the noise ramps up between 100 and 1600, then goes down at 3200. So internally there's some sort of jiggery pokery going on!
The highlight aliasing *is* a problem, although it might not be there on 4K output. The reason is that these type of artefacts are usually the downsampling algorithm, not the debayer itself.
cheers
Paul
I'd only add is that this difference between technical range and actual range is the same for most cameras out there, not just the A7s. The FS700 has lofty claims that just aren't there along with lots of others. When you are able to dig about in the RAW from cameras you can really see compromises.
As for stills mode i think people have got the wrong idea about what the A7s is. The bigger pixels can't really offer that much more than an A7 down sampled, the surface area on a CMOS is pretty much the same.
What the A7s offers is the ability to run at higher ISOs with comparatively little noise. So the difference in stills between 3200 and 12000 isn't as great as it is on other cameras. But the range is pretty much the same. So i have no problem running 12k ISO whereas i'd never go beyond 1600 before.
Also the choice of 3200 might not just be for Slog. In stills the noise ramps up between 100 and 1600, then goes down at 3200. So internally there's some sort of jiggery pokery going on!
The highlight aliasing *is* a problem, although it might not be there on 4K output. The reason is that these type of artefacts are usually the downsampling algorithm, not the debayer itself.
cheers
Paul
Yeah, the FS700 has those too (or had, haven't used newer firmwares). The F55 had those highlight aliases but they fixed them later with firmware.
paulcurtis
Veteran
The 4K RAW out doesn't have them, it's the internally scaled footage that's the worst offender. You can mimic the effects in various software packages where you are able to change the scaling algorithm.
cheers
Paul
eddavid
Carbonite Member
I want to thank Miguel for doing these tests and I wish he would do tests for all the newest cameras when he has time - can we also give you a tip jar or something?
I concur with the usual dynamic range found as I shot highlights in my real world testing - https://vimeo.com/102244108 where I would expose over middle grey and below middle grey it goes very dark. But I'm not here to promote my stupid test film.
The big thing though as a result of Miguel's test is that there is more dynamic range over middle grey which helps pose the usefulness of this camera dealing with intense highlight scenes.
And of course its low light abilities make it a really smart camera to use for certain stolen shots and at night wen you are stealing shots - that is quite useful.
I think its quite a useful tool for its price.
What of course also this test shows is how great the GH4 is - which I am happy about. I haven't been so excited for a Panasonic Camera since the DVX100. I had a HVX200 and then briefly had the Af100 which I sold after finding it a very noisy camera and I still owe a tremendous amount of gratitude to that little dvx100 - what a miracle little camera that was. At this point I still feel its a toss up between the Blackmagic Pocket Camera and the GH4 as super useful small cameras. I am not going to throw away my A7S - I have been using it a lot as a "breaking bad" go-pro style camera and a spy "stealing shot camera" and I'm really happy with Kholi's PicPro profile - maybe will work to make the greens less electric but I still have faith in this little Sony camera. The A8S - I hope they are looking at us. Also the Gh5 - can they look at us too and make it full frame or APS-C - or maybe that Canon 5d Mark IV will finally get it right or the next Blackmagic - I guess the point is and it always is - you just have to use the camera you have now - find its weaknesses and avoid them. Find its strengths and use that. And make whatever you have as good as you can. And if you have the money then read our tests and decide which one you want and you hate it rent another one or sell it. Man its tough some days to land on the right camera for the right job.
I concur with the usual dynamic range found as I shot highlights in my real world testing - https://vimeo.com/102244108 where I would expose over middle grey and below middle grey it goes very dark. But I'm not here to promote my stupid test film.
The big thing though as a result of Miguel's test is that there is more dynamic range over middle grey which helps pose the usefulness of this camera dealing with intense highlight scenes.
And of course its low light abilities make it a really smart camera to use for certain stolen shots and at night wen you are stealing shots - that is quite useful.
I think its quite a useful tool for its price.
What of course also this test shows is how great the GH4 is - which I am happy about. I haven't been so excited for a Panasonic Camera since the DVX100. I had a HVX200 and then briefly had the Af100 which I sold after finding it a very noisy camera and I still owe a tremendous amount of gratitude to that little dvx100 - what a miracle little camera that was. At this point I still feel its a toss up between the Blackmagic Pocket Camera and the GH4 as super useful small cameras. I am not going to throw away my A7S - I have been using it a lot as a "breaking bad" go-pro style camera and a spy "stealing shot camera" and I'm really happy with Kholi's PicPro profile - maybe will work to make the greens less electric but I still have faith in this little Sony camera. The A8S - I hope they are looking at us. Also the Gh5 - can they look at us too and make it full frame or APS-C - or maybe that Canon 5d Mark IV will finally get it right or the next Blackmagic - I guess the point is and it always is - you just have to use the camera you have now - find its weaknesses and avoid them. Find its strengths and use that. And make whatever you have as good as you can. And if you have the money then read our tests and decide which one you want and you hate it rent another one or sell it. Man its tough some days to land on the right camera for the right job.
BlackLegSanji
Member
Close at ISO200 yes, but dial the ISO up a few stops and the gap will be dramatic.
Last edited:
In bright outdoor shots or well-lit studio shots, the GH4 has the advantage in that it can shoot 4K internally (if recording externally 10-bit vs. A7S's 8-bit). We shot our latest short with the A7S on location (many times with a variable ND filter and lowest ISO (3200 for Slog2)) however after testing it against the GH4 in the studio for green screen shots (A7S CINE or Slog2 gammas), the GH4 looks much better (Natural profile and ISO 200). The GH4 also affords us the option to reframe the 4K in post (1080p delivery). When not reframing, 4K to 1080p provides full 444 color (>8-bit effect hasn't been noticeable). GH4 green screen with A7S background video plates works well too. 60p A7S appears to have more usable resolution than the GH4 in 60p VFR mode. The GH4 and A7S make a nice team (the Canon 5D3 is a better stills camera (especially in terms of skintones and organic noise), so when the best stills are needed yet another item to bring on shoots). Someday one camera will be able to do all this and more.
Macgregor, thank you for this investigation!
Also a proud owner of the A7S I have few remarks/questions:
- When "usable" DR of S-Log2 with ISO 3200 is similar to Cine4 with ISO 200 lowering of the S-Log2 ISO should give more DR. If Sony would supply ISO 200 for S-Log2 with a FW update - is a DR of 14 EV a realistic hope? The DxO results suggest this to me.
- Anybody tried to enlarge the DR of the CineX profils via manual setting of Knee and/or Black Gamma?
- Why Color Mode Pro is used instead of Cinema for Cine2 during this investigation?
Thank you for comments!
Or may be SEBASTIAN WöBER is right on this page
http://www.cinema5d.com/dynamic-ran...ira-canon-c300-5d-mark-iii-1dc-panasonic-gh4/
when he stated 14,1 EV for the A7S with S-Log2 using very high quality equipment:
"We use the DSC labs XYLA-21, a high quality LED-backlit transmissive chart that displays 21 stops of dynamic range."
I think he used FHD for his measurements.
http://www.cinema5d.com/dynamic-ran...ira-canon-c300-5d-mark-iii-1dc-panasonic-gh4/
when he stated 14,1 EV for the A7S with S-Log2 using very high quality equipment:
"We use the DSC labs XYLA-21, a high quality LED-backlit transmissive chart that displays 21 stops of dynamic range."
I think he used FHD for his measurements.
Last edited:
Interesting... I'd say that the apparently minor difference in measured dynamic range (if correct) really isn't the whole story here. Having just spent a full week filming with both these cameras my personal "real-world" experience is the A7S is a really fantastic little camera that (despite a couple of annoying quirks) records a very attractive and malleable image, even under challenging lighting conditions. And the GH4 is also a camera... ;-)
We were shooting multi-cam recipe videos using two A7S (with e-mount 24-70mm Zeiss f4 and 70-200mm f4 lenses capturing 1080p internally at 800ISO using Cine4) and an unmanned GH4 (with 12-35mm f2.8 lens capturing UHD internally at about 400ISO using CineD) on a Redrock One Man Crew. The location was an old victorian factory building with very large windows behind the presenter, creating some very high dynamic range scenes. I didn't record s-log because I was worried about exposing it correctly, especially as the light levels were constantly changing due to the partly cloudy weather conditions outside.
DYNAMIC RANGE
Both cameras look good when exposed well, although the image from the Sony definitely has a more "expensive" vibe to it. As the lighting conditions changed there is a marked difference in how easy it was to deal with what was recorded from each camera. Even without the extra range of s-log, the shots from the Sony were relatively easy to recover under and over-exposed areas, pulling skin-tones back to where they should.
The shots from the GH4 were much harder to work with. Highlights and shadows roll off much more abruptly and shadows areas and skin-tones are flat and lacking in detail if they're pushed too far. The noise in the image becomes obviously pretty quickly too.
If you look at the shots above (uncorrected, other than bringing super-whites into line!) you can see that the image on the GH4 is much more contrasty and has less detail both in what's outside the windows and in the shadows on the vegetables in the display.
ROLLING SHUTTER
Rolling shutter is not great on either camera. On the A7S it's almost unusably bad in full-frame 24p / 25p / 30p for anything other than fairly static wide shots. APS-C mode on the A7S and 4K mode on the GH4 seem fairly similar – perhaps a little worse on the GH4 than the A7S. Mostly workable, but occasionally a little ugly.
HIGHLIGHT ALIASING
I sold my FS700 because I HATED the ugly, ugly highlight aliasing. Judging by other tests I've seen, it's not the only Sony camera to suffer (including the F55). Highlight aliasing doesn't occur as frequently as the FS700, and generally seems less pronounced.
EVERYTHING ELSE
• Ugh! The 24-70mm Zeiss lens on the A7S is a pain! Manual focus-by-wire is very (VERY) difficult to control. 70-200 is easier to work with.
• Lumix 12-35mm lens is very sharp, but dreamy shallow DOF is not easily possible on the GH4 for interview or presenter sized shots.
• Internal 4K recording on the GH4 is convenient, but you really needed an external monitor to focus during recording (no punch-in), so I'm happy enough to be recording with the Shogun (when it arrives) on the A7S.
• Don't particularly like the viewfinders or built-in monitors on either camera.
• No overheating experienced with either camera.
CONCLUSION
I'm very happy with the A7S. For all the benefits this camera brings (compact size, choice of lenses, and USABLE dynamic range) I've decided to live with possibility of occasional highlight aliasing, and the pain-in-the-bum rolling shutter. (It might seem a bit rash – but I'm not missing my C300).
I don't love the GH4 quite as much. But I'm keeping it for the moment... I think.
We were shooting multi-cam recipe videos using two A7S (with e-mount 24-70mm Zeiss f4 and 70-200mm f4 lenses capturing 1080p internally at 800ISO using Cine4) and an unmanned GH4 (with 12-35mm f2.8 lens capturing UHD internally at about 400ISO using CineD) on a Redrock One Man Crew. The location was an old victorian factory building with very large windows behind the presenter, creating some very high dynamic range scenes. I didn't record s-log because I was worried about exposing it correctly, especially as the light levels were constantly changing due to the partly cloudy weather conditions outside.
DYNAMIC RANGE
Both cameras look good when exposed well, although the image from the Sony definitely has a more "expensive" vibe to it. As the lighting conditions changed there is a marked difference in how easy it was to deal with what was recorded from each camera. Even without the extra range of s-log, the shots from the Sony were relatively easy to recover under and over-exposed areas, pulling skin-tones back to where they should.
The shots from the GH4 were much harder to work with. Highlights and shadows roll off much more abruptly and shadows areas and skin-tones are flat and lacking in detail if they're pushed too far. The noise in the image becomes obviously pretty quickly too.
If you look at the shots above (uncorrected, other than bringing super-whites into line!) you can see that the image on the GH4 is much more contrasty and has less detail both in what's outside the windows and in the shadows on the vegetables in the display.
ROLLING SHUTTER
Rolling shutter is not great on either camera. On the A7S it's almost unusably bad in full-frame 24p / 25p / 30p for anything other than fairly static wide shots. APS-C mode on the A7S and 4K mode on the GH4 seem fairly similar – perhaps a little worse on the GH4 than the A7S. Mostly workable, but occasionally a little ugly.
HIGHLIGHT ALIASING
I sold my FS700 because I HATED the ugly, ugly highlight aliasing. Judging by other tests I've seen, it's not the only Sony camera to suffer (including the F55). Highlight aliasing doesn't occur as frequently as the FS700, and generally seems less pronounced.
EVERYTHING ELSE
• Ugh! The 24-70mm Zeiss lens on the A7S is a pain! Manual focus-by-wire is very (VERY) difficult to control. 70-200 is easier to work with.
• Lumix 12-35mm lens is very sharp, but dreamy shallow DOF is not easily possible on the GH4 for interview or presenter sized shots.
• Internal 4K recording on the GH4 is convenient, but you really needed an external monitor to focus during recording (no punch-in), so I'm happy enough to be recording with the Shogun (when it arrives) on the A7S.
• Don't particularly like the viewfinders or built-in monitors on either camera.
• No overheating experienced with either camera.
CONCLUSION
I'm very happy with the A7S. For all the benefits this camera brings (compact size, choice of lenses, and USABLE dynamic range) I've decided to live with possibility of occasional highlight aliasing, and the pain-in-the-bum rolling shutter. (It might seem a bit rash – but I'm not missing my C300).
I don't love the GH4 quite as much. But I'm keeping it for the moment... I think.
Last edited:
Birkkromann
Member
In regards to the Cinema5D test - I highly doubt that a professional chart in a lab would be 2 stops off. Furthermore their test seemed to exhibit quite a large difference between the A7s and the GH4 where this shows practically none?
Am I wrong to assume that yours is a tiny bit more subjective hence the "graduation" and "usable range" talk?
Am I wrong to assume that yours is a tiny bit more subjective hence the "graduation" and "usable range" talk?
Last edited:
Samuel H
Major Contributor
A few comments regarding DR measurements:
* Do you count middle gray as one stop? Some people do, some don't. I don't, because I follow the formulas in the wikipedia article, but I know lots of pros do, and some don't, so... just another reason to not compare numbers from different sources. All I care is whether A has X stops more than B.
* Usable DR: it really depends on how you measure it, so you need to devise a test that fits whatever you're going to shoot. I have done three preliminary tests, with three different results:
- total DR (if it shows up in the waveform, it counts; this would be the closest test to what cinema5d have done): slog2 and cine4 have around the same DR... and they are not much different from what I have in my 5n. Did you read that? Same DR as a 5N.
- usable DR, caring a lot small detail (I found a building made of black bricks, joined with dark-grey mortar; standing 20m away from it, with a 50mm lens, the shadows only count if I can clearly see individual bricks): slog2 has one stop better DR than cine4... and my 5N has 3 stops less DR than slog2. Did you read that? A 3-stop improvement!!
- usable DR, caring just about medium detail (same black building, same lens, but now standing just 5m away, so the bricks are much bigger in the screen; again, shadow detail only counts if I can clearly see individual bricks): slog2 has one stop better DR than cine4... which has the same DR as my 5N. So, just a 1-stop improvement this time, and only if I stick to slog2.
I had no tripod or light meter at hand when I ran these tests. I will prepare something better and upload it in the coming weeks.
* Do you count middle gray as one stop? Some people do, some don't. I don't, because I follow the formulas in the wikipedia article, but I know lots of pros do, and some don't, so... just another reason to not compare numbers from different sources. All I care is whether A has X stops more than B.
* Usable DR: it really depends on how you measure it, so you need to devise a test that fits whatever you're going to shoot. I have done three preliminary tests, with three different results:
- total DR (if it shows up in the waveform, it counts; this would be the closest test to what cinema5d have done): slog2 and cine4 have around the same DR... and they are not much different from what I have in my 5n. Did you read that? Same DR as a 5N.
- usable DR, caring a lot small detail (I found a building made of black bricks, joined with dark-grey mortar; standing 20m away from it, with a 50mm lens, the shadows only count if I can clearly see individual bricks): slog2 has one stop better DR than cine4... and my 5N has 3 stops less DR than slog2. Did you read that? A 3-stop improvement!!
- usable DR, caring just about medium detail (same black building, same lens, but now standing just 5m away, so the bricks are much bigger in the screen; again, shadow detail only counts if I can clearly see individual bricks): slog2 has one stop better DR than cine4... which has the same DR as my 5N. So, just a 1-stop improvement this time, and only if I stick to slog2.
I had no tripod or light meter at hand when I ran these tests. I will prepare something better and upload it in the coming weeks.