FS7: Sony FX9 v Canon C500 mkII

[Guest]

Venice was the standout to my eye - I know it's a great camera, but I wasn't expecting that.
The Venice does a noticeably better job of keeping the colours pure under pressure though (under and overexposure), which is more useful.

I’ve done some side by side with the Venice, Alexa Mini, and the Gemini. I’ll do something more comprehensive at some point. Mini got noisy surprsingly early. Like 1600. And then was pretty unsuable by 3200. The LF at a demo on a 24” monitor exhibited similar results. However, the Venice and the Gemini were able to go to 4000 pretty confidently. That is, if one is the type to push an Alexa to iso2000-2500, then yeah 4000 or even 6400. In real world testing, the Venice and Gemini are not far behind in practical latitude. They still both lack that thing that Arri has. They both feel more video. However, the Arri looks less 3D. It has a sort of gauz that unifies the overall image. Looks more painterly. It was pretty easy to nudge the Venice or Gemini into that feel, but they definitely don’t do it naturally. And in challenging situations, i’d imagine they can’t hold a scene as organically. But i intend to test for that at some point.

Between the Venice and the Gemini, it would come down to crew size and workflow, for me. These two cameras are neck and neck for me. Venice would win in flat out iq and internal ND, but the Red wins in size weight and battery life. And to some lesser degree, workflow. The Alexa Mini and the Venice bulk up pretty quickly, and cameras like the Red or Fx9 or C500ii open up shooting to people that aren’t up to carrying around big metal cameras.

My one pause for caution, is seeing how uniform the colours on the Venice are, makes me wonder how much decision making the camera is taking away. If i see a sublte colour in the light, i like to bring that out sometimes. If Venice is just going to say “no”, you get skintone only, i may find myself gravitating back towards Arri or Red. Venice is uniform, but it wasn’t my favourite colour.

 

[Guest]

To bring this thread a little back on topic, seems that the Fx9 faired well in a “C500ii vs Fx9” sort of way:
One person commented:

The surprise for me was the FX9, $10k camera that has superior highlights than 500mkii, monstro, etc..!

http://www.reduser.net/forum/showth...hootout-2020&p=1889063&viewfull=1#post1889063




There is an ongoing thread for general new camera thoughts. I try to occasionally update and summarize the opinions in the thread on the ratings list at the top:
http://www.dvxuser.com/V6/showthread.php?366226-The-New-Cinema-Camera

 

[Grug]

My one pause for caution, is seeing how uniform the colours on the Venice are, makes me wonder how much decision making the camera is taking away. If i see a sublte colour in the light, i like to bring that out sometimes. If Venice is just going to say “no”, you get skintone only, i may find myself gravitating back towards Arri or Red. Venice is uniform, but it wasn’t my favourite colour.

I suspect that's an unfounded concern, having stable and consistent colour from top to bottom means that the subtle colour you like the look of, is going to hold that colour regardless of its exposure level. You'd get more of it, not less.

It doesn't look to me like the Venice is compressing the colour gamut down to a more limited palette, it just seems consistent across the full breadth of its exposure range.

 

[robotfist]

So what is the point of the over and under test? To see how the camera fares if you're an absolutely terrible shooter? For the over exposure test my guess is it's to illustrate highlight rolloff, detail in the highlights and how well the image holds color in the brightest part of an image? And the under test is to show noise pattern of the sensor and detail in the shadows? Just seems like to get the kind of results you see in those tests, you'd have to really do a bad job with your exposure, lol. So why do these tests matter if you know how to shoot and expose properly?

 

[morgan_moore]

So what is the point of the over and under test? To see how the camera fares if you're an absolutely terrible shooter? For the over exposure test my guess is it's to illustrate highlight rolloff, detail in the highlights and how well the image holds color in the brightest part of an image? And the under test is to show noise pattern of the sensor and detail in the shadows? Just seems like to get the kind of results you see in those tests, you'd have to really do a bad job with your exposure, lol. So why do these tests matter if you know how to shoot and expose properly?

Assuming you are comptent you are going not need maybe +/- 2 stops. Why would a competent person get it wrong? Well IMO you would not get it wrong in a black tent with a 32in monitor. But most of us dont operate in those conditions so our monitors can mislead us,
And you exposure can change..
sun goes in during interview
background artist with white outfit hits the sun
Dolly/Steadicam shot moves from INT to EXT

And then there is post.. the classic is the 'beachview hotel', where you expose for the EXT and then want to hugely pull the INT up in post.

And also 'arty' post.. pulling down the the sky to make the mild grey day into a sea of almost black rolling clouds.

But if you dont have a post budget or habit then to some extent latitude is not needed?

 

[Guest]

I suspect that's an unfounded concern, having stable and consistent colour from top to bottom means that the subtle colour you like the look of, is going to hold that colour regardless of its exposure level. You'd get more of it, not less.

It doesn't look to me like the Venice is compressing the colour gamut down to a more limited palette, it just seems consistent across the full breadth of its exposure range.

You’re right. Still, i’ll look for it. Unlikely to be a big concern. I actually sold my DSMC1 in preference for the F55, but over time, found i’d occasionally miss the MX. Even though the F55 was better in many measurable ways. So, i find myself playing devil’s advocate with the Venice. But unless i was buying or worked in broadcast i’d probably still opt for the Arri at this time.

And i hear you about colour consistency. Was grading a project where everytime the exposure shifted, we had to Match the colours. Was a huge pain. Wasted many extra hours in post. That said, the footage was poorly exposed at times. So, this is only an issue if the DP messes up or isn’t experienced/disciplined. Anyway, pre testing the Venice, i can only really admit to being thenother side of the spectrum. It is a nice camera.

 

[Airwolf]

Downloaded the pro-res file and did some comparisons. Thanks to Evin for making this available.
These are my non-scientific findings so far going off the LF as the benchmark for color.

From a C500 II perspective, I think it matches closest with the LF and Monstro, very close with Monstro in my opinion. This stands out the most in the night scenes, but the blues on the C500 are desaturated much more than any other camera. In the night scene the garbage bin in the background does not stand out as much. I'd say on color quality the C500 is pretty good compared with a $60k monstro. What I do think is funny is that Monstro is a pretty noisy sensor, and so is the C500, and they match the closest. But while you can underexpose the Monstro a ton, you can't at all with the C500.

In daylight, I can't quite put my finger on what's wrong with the C500's skin tones. They seem more desaturated then those of the Arri. Less orange than the FX9 though. The Arri has just the right amount of red in her face to make her look alive. She looks pale with the C500, yellow with the Venice, and orange ont he Fx9. Pretty good on the Monstro. Some saturation boost with the C500 helps a lot to fix that issue.

FX9 color is nice, clean in low light, but doesn't match the LF that well. The color in S-Log is odd, a bit green I think.

I do question the shooting conditions though, not sure what time of day this was shot at. But i'm assuming it took a bit for them to re-rig each camera on their gimbal/steadycam and the position of the sun would have changed. The sky looks a bit orange on the FX9 Slog shot so not sure if it dipped in golden hour there.

So, with the C500, expose correctly or you're SOL. Apparently with the LF you can shoot drunk and blindfolded and still get great stuff.

 

[Guest]

So what is the point of the over and under test? To see how the camera fares if you're an absolutely terrible shooter? For the over exposure test my guess is it's to illustrate highlight rolloff, detail in the highlights and how well the image holds color in the brightest part of an image? And the under test is to show noise pattern of the sensor and detail in the shadows? Just seems like to get the kind of results you see in those tests, you'd have to really do a bad job with your exposure, lol. So why do these tests matter if you know how to shoot and expose properly?

The over/under isn’t all about recovering the shot, it is about seeing how the sensor behaves in the extremes. Does it shift colour? Where does it become too noisy for certain things? How far can you push it, and how does it look when it breaks? Can you over expose and still retain nice higlights- or should you expose normally and push in post? And on and on. I find recovering the very last thing i am thinking about in the over/unders.

Also, just because the C500ii shifts green, or the Fx9 has some anaomlies at -6, doesn’t mean you might ever see that depending on the shooting style. The artifacts on the Fx9 shadows may show up in a colour grading selection, but probably won’t effect the final image too much.

On the overs, it can show if color channel separation is happening, but most of these cameras seem to have that pretty well controlled.

All digital cameras clip pretty cleanly. There is no detail. So, the way they look recovered is as much an indication of the users pull in post as anything else. The ones that look the ugliest, were just pulled back the furthest after clipping. In these tests, the difference between a camera that clips at 5.5 stops and a camera that clips a 6stops is exaggerated, because the information tends to either be there, or not.

 

[FrameFarmMedia]

Apparently with the LF you can shoot drunk and blindfolded and still get great stuff.

This is exactly why every DP I know asks for an Arri on set. It just makes good images and when thats the job why risk it.

 

[Guest]

Downloaded the pro-res file and did some comparisons. Thanks to Evin for making this available.
These are my non-scientific findings so far going off the LF as the benchmark for color.

From a C500 II perspective, I think it matches closest with the LF and Monstro, very close with Monstro in my opinion. This stands out the most in the night scenes, but the blues on the C500 are desaturated much more than any other camera. In the night scene the garbage bin in the background does not stand out as much. I'd say on color quality the C500 is pretty good compared with a $60k monstro. What I do think is funny is that Monstro is a pretty noisy sensor, and so is the C500, and they match the closest. But while you can underexpose the Monstro a ton, you can't at all with the C500.

In daylight, I can't quite put my finger on what's wrong with the C500's skin tones. They seem more desaturated then those of the Arri. Less orange than the FX9 though. The Arri has just the right amount of red in her face to make her look alive. She looks pale with the C500, yellow with the Venice, and orange ont he Fx9. Pretty good on the Monstro. Some saturation boost with the C500 helps a lot to fix that issue.

FX9 color is nice, clean in low light, but doesn't match the LF that well. The color in S-Log is odd, a bit green I think.

I do question the shooting conditions though, not sure what time of day this was shot at. But i'm assuming it took a bit for them to re-rig each camera on their gimbal/steadycam and the position of the sun would have changed. The sky looks a bit orange on the FX9 Slog shot so not sure if it dipped in golden hour there.

So, with the C500, expose correctly or you're SOL. Apparently with the LF you can shoot drunk and blindfolded and still get great stuff.

I felt the Canon was appropriately desaturated given the lighting. But it isn’t going to jump out as a win here. Haha. The Sony’s have that golden honey or “brown” monotone mojo going on (sans slog clip), but it is so subtle, as to Grug’s point, i am just playing it up. Alexa found a way to be both appropriately muted and still look alive.

 

[robotfist]

The over/under isn’t all about recovering the shot, it is about seeing how the sensor behaves in the extremes. Does it shift colour? Where does it become too noisy for certain things? How far can you push it, and how does it look when it breaks?

Maybe it's just me, but I 'm not really worried about needing to push a sensor to its absolute extreme. Maybe if you're in Iraq, hiding in a bunker shooting war footage? But even then 6+ stops for a margin of error is huge. This footage has shots where the shooter is literally under a bridge, in a ravine, under tree cover, while shooting the sky in noon sun and you can still see everything just fine. It doesn't come to close to needing to push the exposure to the level seen in the over/under test...

 

[TheDingo]

From a C500 II perspective, I think it matches closest with the LF and Monstro, very close with Monstro in my opinion.

Not seeing it at all. Going by the 6 headshots, the C500 renders her skin-tone much paler than the LF does ( it looks like it's missing some magenta ), and her brown hair is rendered cooler than the LF does. Where the Fx9 S-log renders her hair like the LF does, but her skin is on the yellow side of things.

 

[Airwolf]

I pushed around the images a bit and tried to get them to match the Arri. I'm not a great colorist by any means but I think I got them both pretty close.

 

[Airwolf]

Not seeing it at all. Going by the 6 headshots, the C500 renders her skin-tone much paler than the LF does ( it looks like it's missing some magenta ), and her brown hair is rendered cooler than the LF does. Where the Fx9 S-log renders her hair like the LF does, but her skin is on the yellow side of things.

Yeah the headshots were not the best comparison, but when you look at the whole frame there are other elements such as the color of the fence and the house. Additionally in the low light scenes, the shadows of the LF and Monstro are brown-ish, as is the C500. Whereas on the FX9 the shadows are more blu-ish which matches more with the Gemini I think.

 

[TheDingo]

I pushed around the images a bit and tried to get them to match the Arri. I'm not a great colorist by any means but I think I got them both pretty close.

Skin-tones look much better, but it's interesting that each camera renders her blouse a different shade of yellow. ( I'm wondering if any of this is color failure by the sensor, where the sensor can't capture every color 100 percent correctly )

 

[Guest]

Maybe it's just me, but I 'm not really worried about needing to push a sensor to its absolute extreme. Maybe if you're in Iraq, hiding in a bunker shooting war footage? But even then 6+ stops for a margin of error is huge. This footage has shots where the shooter is literally under a bridge, in a ravine, under tree cover, while shooting the sky in noon sun and you can still see everything just fine. It doesn't come to close to needing to push the exposure to the level seen in the over/under test...

Well, i can’t say you’re wrong. You just might be misunderstanding me is all.

 

[Airwolf]

Skin-tones look much better, but it's interesting that each camera renders her blouse a different shade of yellow. ( I'm wondering if any of this is color failure by the sensor, where the sensor can't capture every color 100 percent correctly )

I struggled with that. When I got one thing right that dress would be more orange than I wanted so I had to dial it back to sort of meet the difference. Out of the box, the FX9 matched the LF dress almost perfectly, whereas the C500 was much more orange. The C500 also did something weird in that the ground on the LF frame has more of an orangeish tint to it, but the C500 is more blue. I wasn't able to correct for that fully without altering her skin or the dress more.

 

[Guest]

@airwolf, the C500ii looks really good in person. Sometimes white balance skews a camera comparison.

And as robotfist mentioned in passing, that guy was under a bridge probably alone, and just shooting. A heavier camera, or one that is more battery hungry wouldn’t make that kind of shooting as easy. Considering the good reputation of the C series cameras, and just tacking on that this is their best one yet, what’s not to like?

 

[Liam Hall]

All week I've been editing footage I shot on the C500 MKII while on the road in Latin America for three weeks in January. I shot indoors, outdoors, mixed light, controlled light, high dynamic range, low-light - pretty much every type of lighting condition. I can report footage looks stellar. You can really push shots and there's bags of headroom on exposure. One standout feature is how well FF shots match with S35 and even S16. This was something I noticed looked really poor on Alistair Chapman's FX9 footage where the differing noise levels were really obvious. If you can't make the C500 MKII look great, you're an idiot.

 

[Grug]

I struggled with that. When I got one thing right that dress would be more orange than I wanted so I had to dial it back to sort of meet the difference. Out of the box, the FX9 matched the LF dress almost perfectly, whereas the C500 was much more orange. The C500 also did something weird in that the ground on the LF frame has more of an orangeish tint to it, but the C500 is more blue. I wasn't able to correct for that fully without altering her skin or the dress more.

I wouldn’t read too much into the colours of the outdoor test. They shot into sunset (you can see a streetlamp has turned on in one or two of the clips), so anything could have been happening to the overall colour temp and tint of the ambient skylight.