FS7: Sony FX9 v Canon C500 mkII

SlashCam.de usually seemed to do pretty good thorough tests, including with charts. Their FX9 skintone footage didn't look as good as their C500 II footage (which doesn't match e.g. the REDUser test where the FX9 looked a bit better than the C500 II), and their test charts make the FX9 look quite poor, lower resolution, even blurry- this doesn't match any FX9 footage I've seen. What lens, was it in focus (chroma issues)? The charts look swapped based on the example FX9 & C500.

Again, the charts seem backwards....

Oh... Kay.. I’ll just let that simmer. :/) FWIW. The slashcam results were published several weeks apart, the c500ii before they even had a copy of the fx9 to review. So it bends credibility to suggest that somehow the samples were switched. They were quite critical of the detail from the fx9. These results square very well with the detail we see compared in the reduser tests. Again I’m not making any claims. Simply pointing to the evidence. I don’t think it really matters which camera is more detailed, only that if we’re making pronouncements, that we get it right. My assumption is that they probably display about the same, except in higher iso /under exposure, where the Sony is applying more noise reduction and thus smearing detail.
 
Oh... Kay.. I’ll just let that simmer. :/) FWIW. The slashcam results were published several weeks apart, the c500ii before they even had a copy of the fx9 to review. So it bends credibility to suggest that somehow the samples were switched. They were quite critical of the detail from the fx9. These results square very well with the detail we see compared in the reduser tests. Again I’m not making any claims. Simply pointing to the evidence. I don’t think it really matters which camera is more detailed, only that if we’re making pronouncements, that we get it right. My assumption is that they probably display about the same, except in higher iso /under exposure, where the Sony is applying more noise reduction and thus smearing detail.

Given the sensor resolutions are similar, I'd expect similar results if doing similar downsampling (there are many methods + OLPF will make a difference too). The SlashCam FX9 chart has green chroma issues- there's something wrong- lens, focus, etc. The Venice example I'd expect to look similar to the FX9 in the same way the C700 FF would be similar to the C500 II. The Venice chart doesn't have green chroma issues and looks acceptable (no lens or chroma issues). Note that even though these results don't look great for the FX9, I still posted it. And still stand by my earlier comment that the FX9 wins for IQ vs. the C500 II. Others here who've done detailed reviews have made a similar ranking (e.g. ranking the FX9 just behind the Alexa and Venice).

Again, haven't read anyone making 'too sharp' comments about the C500 II, and on 4K displays the FX9 does look more detailed from all the example I've seen. In the example from the AF tests, the FX9 has a lot more skin detail + more color variation. The C500 II's loss of detail/color may be part of Canon's processing of face/skin (which would not happen on a test chart) or related to the output / display mode. It's fine if they are doing that, though would be useful to be able to turn it off (same as other cameras with such features).

WRT color science, from: https://www.newsshooter.com/2020/01/02/acs-technical-committee-evaluation-of-the-sony-fx9/
The biggest surprise came from a test conducted by Pawel Achtel ACS. Pawel has a system where he can measure a cameras sensor response to certain colors. This is a very technical test, but it has the ability to map the color gamut that a sensor is producing. What was quite a shock, was that not only did the FX9 outperform the FS7 M2 by a considerable margin, but it also produced the best result of any digital cinema camera that has been tested. This includes most of the top tier cinema cameras that are on the market today.

Impressive if true, though when cost is no object Alexa Mini LF will likely be the top pro choice (where max IQ is the goal and AF isn't needed).
 
Note the WB is off on the FX9 shot (too much green):

View attachment 138362

FX9 +20 (magenta) in Photoshop ACR (only change):

View attachment 138363

To my eye the FX9 looks more detailed and color is pretty excellent compared to prior Sony offerings. The C500 II has lower contrast and a pastel look (lower satuation)- the FX9 could be adjusted to match if that's a preference. See the video full screen in 4K to examine detail (at smaller scales, higher contrast looks more detailed).

I am referring to AF, not WB or IQ.
 
That AF test/video was posted back on page 75 (#750) if you're interested in reading some reactions about it.
 
Doesn't the FX9 have the same or similar sensor to the Venice?

The FX9 does not have the same sensor as the Venice. This was straight from the slideshow at the Sony FX9 Training I attended and we asked the design engineer for the FX9. He did say that the sensor technology used in the FX9 is closer to the Venice than the FS7 MKII.
 
I watched and upvoted that post- agree the FX9 wins for AF. Most agree that's the case? IQ still has some doubters, even after the REDUser ramp tests (where the FX9 in a 709 profile did amazingly well against raw & log cameras). Note I don't currently own any Sony lenses (all Canon, mostly L)- comments based on examples and tests posted online so far. I prefer Canon stills division raw (amazing recovery possible) and color science for video (1DX II, EOS R). From what I've seen so far, the C500 II looks like a true 4K C300 II (owned and sold) and similar to the C700 FF (briefly tested). FX9 looks like an $11K Venice + AF and dVND.
 
It’s a term used when you want others to stop exploring hypotheticals that aren’t in line with your experience and knowledge.


(Originally: Fear, Uncertainty, Doubt. Which is referencing a disinformation tactic used to manipulate a population or a market. Perhaps on the internet it is just casually passing along bad information. Which would be “fear uncertainty and disinformation”. In a way, it could be said that Liam wants the speculations to stop, so he uses the term FUD to cause fear, uncertainty, or doubt in those that may have different views. This fear of public shame and doubt of one’s own opinions could decrease or stop the amount of observational or speculative comments made. Which is kind of how the FUD tactic works. So, he is using the term FUD to cause a form of FUD. Ah, but look at me, i’m naval gazing... )
 
Last edited:
You guys are obviously having fun posting in this thread, else why spend time here: is irony ferrous humor?

Earlier in this thread it was stated that people can't agree on color, then proceeded to take a color perception test and everyone got great or perfect scores. That was pretty funny, right?

Searching for truth is fun, like solving a puzzle. If the arguments aren't to your liking: Where-can-I-go-specifically-on-the-Internet-to-argue-with-people
 
It’s a term used when you want others to stop exploring hypotheticals that aren’t in line with your experience and knowledge.


(Originally: Fear, Uncertainty, Doubt. Which is referencing a disinformation tactic used to manipulate a population or a market. Perhaps on the internet it is just casually passing along bad information. Which would be “fear uncertainty and disinformation”. In a way, it could be said that Liam wants the speculations to stop, so he uses the term FUD to cause fear, uncertainty, or doubt in those that may have different views. This fear of public shame and doubt of one’s own opinions could decrease or stop the amount of observational or speculative comments made. Which is kind of how the FUD tactic works. So, he is using the term FUD to cause a form of FUD. Ah, but look at me, i’m naval gazing... )

I used the term FUD because of the increasing amount of comments in this thread that are misinformed or plain wrong. I'm sure people are well-intentioned, but trust me, it's sucking the joy out of owning one of these cameras. I'd rather hear from people using the FX9 or the C500 MKII, how are they enjoying their camera, what things are great, what things are not so great? These are great tools, let's share best practice...
 
I’ve been using my FX9 since super bowl week. I find it to be a great little camera with some room for improvement (which will come via firmware updates). In my experience, the AF is flawless when you have 1 person at a time. I have used it for stand ups and for sit down interviews, and it’s downright amazing. Now, when you have multiple people on screen, even just 2, it gets tricky. There is a something called face registration where you get an asterisk letting you know that face is always the priority on your frame. The catch is that if that person let’s say is facing to the side even slightly, the registration is lost and the af will hunt until the person looks forward again. Obviously this is not good, and I found when doing 2 person man on the street stuff you need to go full manual to avoid focus hunting.

This was a sit down with the 70-200 GM on AF. It never once missed focus for 30 minutes.

B8DC53BF-1C04-4E8A-B64B-173CECD1BEB8.jpg

Hers an example with the 24-70 also AF with a reporter who didn’t sit still. Once again never failed.
1D5A30DF-CF99-4569-8E46-657F8E2A04E1.jpg
 
That's how it should be (meaning it shouldn't fail)...that's how the Canon and Sony DSLRs/mirrorless' have worked for years. But it's good to see the technology solidify itself in the Sony cinema lineup after extended consumer/prosumer beta testing.

As far as multiple faces, if you use the touch screen and manually choose a face, does the AF still try to go back to the other face(s)?

I have a Canon M6 Mark II and if you touch the screen the AF turns into 'tracking'. So if you touch a face that tracking turns into manual face-tracking and stays on the face you choose. However if the 'tracking' isn't activated (by touching the screen), the regular face-tracking may look for multiple faces.
 
I used the term FUD because of the increasing amount of comments in this thread that are misinformed or plain wrong. I'm sure people are well-intentioned, but trust me, it's sucking the joy out of owning one of these cameras. I'd rather hear from people using the FX9 or the C500 MKII, how are they enjoying their camera, what things are great, what things are not so great? These are great tools, let's share best practice...

Right on, man. :-)
 
That's how it should be (meaning it shouldn't fail)...that's how the Canon and Sony DSLRs/mirrorless' have worked for years. But it's good to see the technology solidify itself in the Sony cinema lineup after extended consumer/prosumer beta testing.

As far as multiple faces, if you use the touch screen and manually choose a face, does the AF still try to go back to the other face(s)?

I have a Canon M6 Mark II and if you touch the screen the AF turns into 'tracking'. So if you touch a face that tracking turns into manual face-tracking and stays on the face you choose. However if the 'tracking' isn't activated (by touching the screen), the regular face-tracking may look for multiple faces.

I don't think the FX9 has touch screen tracking like the canon's do. I read and interview with some Canon engineers awhile back and they talked about their approach being weighted towards allowing the operator to choose what was being focused on, versus the camera doing the deciding. (this of course could be an easy way out as their face tracking isn't as sophisticated or sensitive as the FX9's). I've found the range of tools on the C500II works well for me. Sometimes it doesn't recognize a face, and so you switch to area and touch, which works well...other times using face it seems to, without "registering", want to stick with the face it's been tracking, and go back to it after it loses it. I don't know if this is in their actual algorithm, but it seems to work for me. I photographed a staged meeting last week, long lens shooting into a table surrounded by talkers. Multiple Heads bobbing in and out, hands and wine bottles in the way, and I was really happy with how the camera handled things automatically, or let me select who was in focus. From what I've seen of the FX9, it's also a tremendous tool in this aspect.
 
I don't think the FX9 has touch screen tracking like the canon's do. I read and interview with some Canon engineers awhile back and they talked about their approach being weighted towards allowing the operator to choose what was being focused on, versus the camera doing the deciding. (this of course could be an easy way out as their face tracking isn't as sophisticated or sensitive as the FX9's). I've found the range of tools on the C500II works well for me. Sometimes it doesn't recognize a face, and so you switch to area and touch, which works well...other times using face it seems to, without "registering", want to stick with the face it's been tracking, and go back to it after it loses it. I don't know if this is in their actual algorithm, but it seems to work for me. I photographed a staged meeting last week, long lens shooting into a table surrounded by talkers. Multiple Heads bobbing in and out, hands and wine bottles in the way, and I was really happy with how the camera handled things automatically, or let me select who was in focus. From what I've seen of the FX9, it's also a tremendous tool in this aspect.

I know this is a FX9/C500 thread, but it's important to note and understand - even for anyone just passing by - that Canon's cinema cameras' AF is inferior to their DSLRs/mirrorless cameras.

The 1DX Mark II...2016...they did it right the first time and the only improvements that were necessary since then are more vertical and horizontal screen coverage and control of speed tracking, but for some reason when the AF made its way into the cinema cameras it became more troublesome at times even for basic operation.

1:22 shows the face-tracking/switching.

 
I know this is a FX9/C500 thread, but it's important to note and understand - even for anyone just passing by - that Canon's cinema cameras' AF is inferior to their DSLRs/mirrorless cameras.

The 1DX Mark II...2016...they did it right the first time and the only improvements that were necessary since then are more vertical and horizontal screen coverage and control of speed tracking, but for some reason when the AF made its way into the cinema cameras it became more troublesome at times even for basic operation.

1:22 shows the face-tracking/switching.


The cinema cameras also allow you to switch between faces. The cinema cameras have quite a few operations that are buried in the menus or can only be accessed by assigning a button. For example, buttons can be set to start tracking or to turn the auto AF on or off. I personally almost never use continuous autofocus and just use the manual focus guides. If I do use the continuous AF, I use it as face only so that if the camera loses a face, it doesn't focus on something in the background. It just stops tracking. Also, it is best to adjust the tracking speed and reaction time so that the camera doesn't start hunting accidently.

The C500 II has added the ability to have two tracking boxes onscreen that you can switch between. This allows you to more easily do a rack focus.
 
Indeed, but they don't work as well...as the concerns you're mentioning above simply don't exist on the stills cameras (or rarely).

All of the new, fun technology gets beta-tested in cheaper mirrorless cameras for the "millions" before moving onto the bigger, more expensive products. So anything the cinema cameras have, the mirrorless' had in their own way too (like being able to toggle tracking/on/off AF with a button - a physical one or one on the touchscreen).
 
To answer above about switching faces on the fx9. It gets very wonky and using the stick isn’t as intuitive as touch screen. Sometimes you’ll want to switch to a face, but the box won’t light up on that face because it’s not fully frontal. It freaks out once a face turns profile. In the scenario that there is more than one face on screen I recommend manual focus. I’m sure AF things will be tweaked in future firmware updates.
 
There's a lot going on in that camera vs. something like a simple 8-bit 100mbps 24p a6400 (flawless AF, btw), so the FX9 might just need some more work on its software.

I also imagine most people will be filming one person interviews vs. multiple persons interviews. For most, I think multiple people usually come in some form of b-roll and you'll just hope you have a higher hit percentage when you can't manually focus (or choose not to).
 
Back
Top