Jump to content

SONY FX3 new camera to be announced


Trankilstef
 Share

Recommended Posts

Another issue with Sony footage (at least lots of FX6 stuff I've seen lately) is that it's quite prone to some nasty gamut clipping in S-Gamut3.cine at least. It looks even worse than BMD footage with HR turned on. Thankfully NR doesn't seem to be a huge problem on the FX6 at last.

Is there some S-Gamut3 footage showing bright saturated lights available as well?
 

https://filmplusgear.com/sony-fx6-test-files-for-download/

Link to comment
Share on other sites

15 hours ago, Llaasseerr said:

For me personally, any ETTR method is not relevant and I'm only interested in exposing for middle grey. Each to their own, though. Also, is there any attempt to bring these images into a common baseline space to compare them, via camera vendor supplied colour transforms? There is no mention in the translated article.

And yes, NR is a factor, that is why I would advocate for DR tests without it. Such as ProRes Raw or FX6 with NR "off", compared to an Alexa that does not have NR anyway, and stick to the Arri method for noise floor calculation with the Xyla/IMA tests for consistency.

In the end, I can do NR in post with Neat Video or a deep learning method, and get a better result than an in-camera method. I never shoot with NR enabled though.

Just showing how much latitude (over- and underexposure range) is available.
The S5 (S1H/S1) perform remarkably well in that regard. Also no NR nor any Gamut issues.

ProRes RAW on the S1H (NR-1)/S5 also shows how little NR S Series cameras use which renders PRR basically redundant on these cameras, same with the FX6 and 9 judging from the demo footage I've downloaded. On the A7s3 however PRR is necessary. 

Link to comment
Share on other sites

6 hours ago, Llaasseerr said:

Tbh from my experience working with a lot of handheld Alexa footage, the CineD estimate for the XT in I assume 2880 seems more credible (around 3ms). I've worked with a lot of open gate XT footage and it did not look like 13ms either. But I didn't measure it, so ?

The only time I've noticed artifacts is with strobe lights. With the 65, I've seen whip pans def have more of a diagonal though. But still only ~10ms in my estimation.

Alexa's rolling shutter values have been scientifically and comprehensively measured by 3DEqualizer so that it can be accurately compensated for VFX use.

They call it "Framewise Time Shift", the time it takes for the sensor to scan from top to bottom.

Link to comment
Share on other sites

5 hours ago, androidlad said:

Alexa's rolling shutter values have been scientifically and comprehensively measured by 3DEqualizer so that it can be accurately compensated for VFX use.

They call it "Framewise Time Shift", the time it takes for the sensor to scan from top to bottom.

Thanks, I'll take a look. I have total respect for 3DE. Germans being precise about stuff.

Link to comment
Share on other sites

7 minutes ago, Llaasseerr said:

This actually makes me hopeful about whatever Sony is intending with the FX3 since clearly it's not just the "a7sIII with S-Cinetone" as some have stated.

Actually that makes it even more sound like just an A7sIII in another housing since both may as well use exactly the same firmware now.

So you're still handicapped by horrible NR, oversharpening at -7 detail, no shutter angle, no 3D Lut support for output via HDMI, no scopes, no 48p, no true 24p, no DCI 4K etc. 

But yaaay Cinetone 🙄

Link to comment
Share on other sites

7 hours ago, deezid said:

Just showing how much latitude (over- and underexposure range) is available.
The S5 (S1H/S1) perform remarkably well in that regard. Also no NR nor any Gamut issues.

ProRes RAW on the S1H (NR-1)/S5 also shows how little NR S Series cameras use which renders PRR basically redundant on these cameras, same with the FX6 and 9 judging from the demo footage I've downloaded. On the A7s3 however PRR is necessary. 

PRR is not redundant because it does not subsample chroma information. It's also 12 bit log (EDIT: on the Sony cams) vs 10 bit and has less compression artifacts than an internally recorded h.265 type of codec that's really designed for final exhibition, not for capture.

Of course, this may not be important to your particular situation.

These baked colour latitude tests do not really show very much because they do not reflect the original sensor captured image. In addition, any DP that needs to correct footage that was incorrectly exposed by more than 3 stops should be fired.

Link to comment
Share on other sites

9 minutes ago, deezid said:

Actually that makes it even more sound like just an A7sIII in another housing since both may use the exact firmware now. So still horrible NR, oversharpening at -7 detail, no shutter angle, no 3D Lut support for output via HDMI, no scopes, no 48p, no true 24p, no DCI 4K etc. 

But yaaay Cinetone 🙄

We'll soon know what they are thinking. For interoperability with the FX line, at minimum true 24P and I suspect DCI will be required.

Link to comment
Share on other sites

9 minutes ago, Llaasseerr said:

PRR is not redundant because it does not subsample chroma information. It's also 12 bit log vs 10 bit and has less compression artifacts than an internally recorded h.265 type of codec that's really designed for final exhibition, not for capture.

Of course, this may not be important to your particular situation.

Unless you're shooting lots of greenscreen or mess up WB/Tint/Exposure all the time, at least on the S1H using its 4K ALL-I and also 6K modes isn't really that important. The external 6K PRR footage pretty much looks almost the same even when pixel peeping. The S1H set to -1 NR isn't doing any destructive temporal noise reduction, only some spatial chroma nr which to be fair may cause some loss of chroma information.

Keep in mind, you have to export as ProRes 422HQ at least to have any visual advantage over the S1H's H265 6K even, ProRes 422 already performs significantly worse with noise or lots of moving detail and texture.

Still waiting for proper Davinci Resolve support here so I don't have to use Assimilate Player Pro all the time.

 

Quote

These baked colour latitude tests do not really show very much because they do not reflect the original sensor captured image. In addition, any DP that needs to correct footage that was incorrectly exposed by more than 3 stops should be fired.

These ETTR tests are more helpful than any synthetic IMATEST results which can easily be tricked by NR since the ETTR method actually shows what can be buried out of shadows and how much hues shift throughout the exposure range and the A7sIII performs particularly bad in that regard with the FX6 slightly better due to active cooling and way superior internal processing.

Link to comment
Share on other sites

7 hours ago, deezid said:

Another issue with Sony footage (at least lots of FX6 stuff I've seen lately) is that it's quite prone to some nasty gamut clipping in S-Gamut3.cine at least. It looks even worse than BMD footage with HR turned on. Thankfully NR doesn't seem to be a huge problem on the FX6 at last.

Is there some S-Gamut3 footage showing bright saturated lights available as well?
 

https://filmplusgear.com/sony-fx6-test-files-for-download/

I can't speak for S-gamut3.cine, but just a note on S.gamut3. It is a very large gamut like Arri Wide Gamut and if not handled properly in post, it will clip. You need a big enough bucket and a smart gamut transform  to your viewing device that deals with the edges in an aesthetically nice way. Recent ACES developments in this area is one example.

If you just randomly grade Slog3 for Rec709 output, it will be more likely to clip.

Part of the issue is also that software like Premiere is really only designed for Rec709, so this needs to be done in Resolve.

This clipping issue is fallout because camera manufacturers put industry-level tools into consumer cameras but the workflow is not explained. We all benefit form this democratization though.

Link to comment
Share on other sites

Just now, Llaasseerr said:

I can't speak for S-gamut3.cine, but just a note on S.gamut3. It is a very large gamut like Arri Wide Gamut and if not handled properly in post, it will clip. You need a big enough bucket and a smart gamut transform  to your viewing device that deals with the edges in an aesthetically nice way. Recent ACES developments in this area is one example.

If you just randomly grade Slog3 for Rec709 output, it will be more likely to clip.

Part of the issue is also that software like Premiere is really only designed for Rec709, so this needs to be done in Resolve.

This clipping issue is fallout because camera manufacturers put industry-level tools into consumer cameras but the workflow is not explained. We all benefit form this democratization though.

The problem shown in these examples shows up in the source footage, which can be downloaded. That's some pretty nasty gamut clipping being even worse than BMD Film Gen 4/5 also using Bt.709 primaries. I can't reproduce anything like that on RED or Panasonic (full V-Gamut) cameras using Bt.2020 basically. I don't think that's the reason though?

Link to comment
Share on other sites

5 minutes ago, deezid said:

Unless you're shooting lots of greenscreen or mess up WB/Tint/Exposure all the time, at least on the S1H using its 4K ALL-I and also 6K modes isn't really that important. The external 6K PRR footage pretty much looks almost the same even when pixel peeping. The S1H set to -1 NR isn't doing any destructive temporal noise reduction, only some spatial chroma nr which to be fair may cause some loss of chroma information.

Keep in mind, you have to export as ProRes 422HQ at least to have any visual advantage over the S1H's H265 6K even, ProRes 422 already performs significantly worse with noise or lots of moving detail and texture.

Still waiting for proper Davinci Resolve support here so I don't have to use Assimilate Player Pro all the time.
 

Look, it works for you so that's fine. I can assure you for the work I do, I notice and value the difference and it goes way beyond the scenarios you are suggesting. 

As an aside, WB/Tint/Exposure is not the province of raw footage. It can be done in the same exact way with any footage if it's transformed to linear or a pure log like ACEScc.

It does not take long for 4K ALL-I to fall apart for what I do, so I value the affordable option of things like PRR for my own projects.

I would never export PRR as 422HQ. I would export as either EXR, DPX or ProRes 444.

Agreed on Resolve, but I am happy to use Scratch as ingest. I realise that's a little expensive but it's worth it to me and I'm used to much more expensive ingest and dailies tools in my industry (Colorfront, Baselight). So Scratch actually seems like a good deal, and they thought out the PRR to ACES workflow nicely.

 

Quote

These ETTR tests are more helpful than any synthetic IMATEST results which can easily be tricked by NR since the ETTR method actually shows what can be buried out of shadows and how much hues shift throughout the exposure range and the A7sIII performs particularly bad in that regard with the FX6 slightly better due to active cooling and way superior internal processing.

 

That's assuming that noise reduction is enabled when the Imatest is done. I did not mention Imatests though, so I don't get your point. It's a DR test, unlike this. I measure DR above middle grey myself, but I also like looking at the IMA tests by Gerald and CineD.

In the context of talking about the a7sIII/FX6/FX3 sensor's DR, I have already stated that for me the idea of ETTR as a workflow is not useful. I prefer to just expose for a grey card. But more power to you if it's useful to you.

I never shoot footage with noise reduction, so that is just never a factor for me when assessing dynamic range.

 

Link to comment
Share on other sites

18 minutes ago, deezid said:

The problem shown in these examples shows up in the source footage, which can be downloaded. That's some pretty nasty gamut clipping being even worse than BMD Film Gen 4/5 also using Bt.709 primaries. I can't reproduce anything like that on RED or Panasonic (full V-Gamut) cameras using Bt.2020 basically. I don't think that's the reason though?

Cool, I'll take a look if I get a chance. I'd also be interested to see more raw footage from the FX6. So far I've only downloaded the CVP clips.

Link to comment
Share on other sites

13 minutes ago, Llaasseerr said:

Look, it works for you so that's fine. I can assure you for the work I do, I notice and value the difference and it goes way beyond the scenarios you are suggesting. 

As an aside, WB/Tint/Exposure is not the province of raw footage. It can be done in the same exact way with any footage if it's transformed to linear or a pure log like ACEScc.

It does not take long for 4K ALL-I to fall apart for what I do, so I value the affordable option of things like PRR for my own projects.

I would never export PRR as 422HQ. I would export as either EXR, DPX or ProRes 444.

Agreed on Resolve, but I am happy to use Scratch as ingest. I realise that's a little expensive but it's worth it to me and I'm used to much more expensive ingest and dailies tools in my industry (Colorfront, Baselight). So Scratch actually seems like a good deal, and they thought out the PRR to ACES workflow nicely.

 

 

 

You can do the same using a Log based workflow as well, unless your colorspace is limited which I assume S-Gamut3.cine might be too tiny for anyways.

Tried working with ACES but really at least the IDT for Panasonic V35 is quite a mess and leads to heavy blue and tungsten clipping which cannot be fixed within ACEScc/cct. It has to be done before IDT. Will see if I can fix it coming in via linear EXR.

Quote

That's assuming that noise reduction is enabled when the Imatest is done. I did not mention Imatests though, so I don't get your point. It's a DR test, unlike this. I measure DR above middle grey myself, but I also like looking at the IMA tests by Gerald and CineD.

In the context of talking about the a7sIII/FX6/FX3 sensor's DR, I have already stated that for me the idea of ETTR as a workflow is not useful. I prefer to just expose for a grey card. But more power to you if it's useful to you.

I never shoot footage with noise reduction, so that is just never a factor for me when assessing dynamic range.

Exposing for 42IRE here as well, what this test still shows is that despite doing that, cameras like the S1H have cleaner shadows with more texture and more highlight information as well than others.

Noise reduction on some cameras such as the C70 or even Pocket 4K and 6K cannot even be bypassed. I'm sure without its heavy NR the C70 would perform very similarly to other cameras in its price range (around 11.5-12 stops SNR=0.5). CineD currently plans to redo tests using ProRes RAW or Canon RAW on supported cameras. 

Shooting without any NR is the goal indeed, but this kind of luxury still didn't trickle down to cheaper cameras unless you're shooting ProRes RAW or Canon Raw.

Link to comment
Share on other sites

14 minutes ago, Llaasseerr said:

Look, it works for you so that's fine. I can assure you for the work I do, I notice and value the difference and it goes way beyond the scenarios you are suggesting. 

 

What type of work do you do? Not trying to question your legitimacy or anything just curious. 

It is pretty crazy how far compressed codecs have come but yeah won't cut it for every job. I wish 12 bit codecs would become a standard in cameras tho. Panasonic has always impressed me with the image they put out for the price. 

Link to comment
Share on other sites

6 minutes ago, TomTheDP said:

What type of work do you do? Not trying to question your legitimacy or anything just curious. 

It is pretty crazy how far compressed codecs have come but yeah won't cut it for every job. I wish 12 bit codecs would become a standard in cameras tho.

A 12 bit debayered Log codec like BRaw is actually all you need. Only gripe with Braw I have is its strong processing which e.g. on the EVA-1 produces an image looking worse than its 10 bit internal counterpart. Blackmagic does a great job applying the right linearization and transformations though even for different colorspace outputs.

Link to comment
Share on other sites

1 hour ago, TomTheDP said:

What type of work do you do? Not trying to question your legitimacy or anything just curious. 

It is pretty crazy how far compressed codecs have come but yeah won't cut it for every job. I wish 12 bit codecs would become a standard in cameras tho. Panasonic has always impressed me with the image they put out for the price. 

I work in feature film VFX and post, so for better or worse the industry workflows and quality standards have bled into my own creative work with affordable cameras. There are expectations about plate integrity and the compressed images just make the work very difficult or impossible, and no-one has any patience for that.

It's really what is the tool for the job, right? f you are just editing and doing a quick grade before delivery for a fast turnaround then it's fine. For all the progress with internal compressed codecs, they are only now getting near ProRes 422 HQ which has been around forever. But for acquisition I'm coming from a place where as a starting point I want no subsampled chroma resolution.

If you think about it, with subsampled chroma even that fucks with the noise (grain) regardless of disabling a camera's nasty internal NR methods. It makes it more blocky. You need to do away with both chroma subsampling and NR to see if the sensor has a nice noise profile.

I tried shooting 10 bit log 4:2:0 ALL-I 400mb compressed for a personal creative project, and although the image looks nice straight out of camera it just fell apart really quickly in post. Lots of blocky artifacts when doing a power window or an aggressive film LUT. If you convert the colourspace to YCbCr and look at the chroma channels they are just appalling garbage. They look like a highly jpeg compressed screengrab from an 80's Atari game. It's for delivery not acquisition.

Link to comment
Share on other sites

2 hours ago, deezid said:

 

2 hours ago, deezid said:

You can do the same using a Log based workflow as well, unless your colorspace is limited which I assume S-Gamut3.cine might be too tiny for anyways.

Tried working with ACES but really at least the IDT for Panasonic V35 is quite a mess and leads to heavy blue and tungsten clipping which cannot be fixed within ACEScc/cct. It has to be done before IDT. Will see if I can fix it coming in via linear EXR.

Exposing for 42IRE here as well, what this test still shows is that despite doing that, cameras like the S1H have cleaner shadows with more texture and more highlight information as well than others.

Noise reduction on some cameras such as the C70 or even Pocket 4K and 6K cannot even be bypassed. I'm sure without its heavy NR the C70 would perform very similarly to other cameras in its price range (around 11.5-12 stops SNR=0.5). CineD currently plans to redo tests using ProRes RAW or Canon RAW on supported cameras. 

Shooting without any NR is the goal indeed, but this kind of luxury still didn't trickle down to cheaper cameras unless you're shooting ProRes RAW or Canon Raw.

 

To replicate the behavior of the controls on the raw tab, the log curve needs to be  a pure log encoding curve like ACEScc otherwise it will not visually match because of the lifted toe on most log curves that are based on Cineon (Arri Log-C, F-log, Slog3, Red's latest etc). But yes there's nothing magic about those raw adjustment controls.

Agreed, the S1H appears to have better shadow performance. Personally I'm more interested in highlight range and better rolling shutter, and in my tests that is where the Sony sensor wins.

The Sgamut3.cine gamut is actually pretty big. It's just a bit smaller than the full Sgamut3 which represents the most the sensor is capable of. I personally would just use Sgamut3 since I think .cine is just a mild concession to colorists who do not want to do a matrix transform and then they complain about green skin tones. You would not grade a film negative before developing it.

You can definitely ignore ACES and just work with the right colour management with your camera's log and gamut. But sometimes ACES is great because it may already have solutions in place for when people complain about "bad Sony colours" or gamut clipping.

My first camera was a D16 so I got used to working with a lot of noise and appreciating it. And yes, I would work with ProRes Raw with a Sony cam to avoid the issues you're talking about.

 

Link to comment
Share on other sites

So, the A7sIII is getting S-Cinetone: https://www.sonyalpharumors.com/rumor-sony-a1-cine-colour-profile-will-come-for-the-a7siii-via-future-firmware-update/

The FX3 is getting more confusing (or, maybe why the A7sIII exists is the confusing part). Unless the FX3 has timecode and ND filters, not sure why you'd get it this if you already have the A7sIII. 

However, from the appearance of the body, buttons, ergonomics, top handle and XLR integration, and flip-screen being able to tilt flush to body (hopefully), I'd get the FX3 over the A7sIII, even if it was $500-750 more. Looks much more rugged as well. 

 

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

 Share

×
×
  • Create New...