Llaasseerr

Just chiming in after doing a few tests with random DNGs I've found online. I decided to revisit the camera since it's having a bit of a moment on Youtube and it's had a number of firmware updates, but I don't own one. I'm converting the DNGs to ACES open exrs, generally in Resolve. The reason for this is that Resolve builds an IDT on the fly based on looking at the DNG metadata, and then it dumps a linear RGB image to disk that is similar to the raw file. It takes the native sensor gamut and dumps that into the larger bucket of the ACES container. From what I gather, for DNGs the camera is mostly ISO invariant, based on looking at the ISO diagram in the manual. In absolute terms, for an ISO 800 image the sensor seems to be clipping at around 3.0-5.0 in the RGB channels - it varies per channel if unclipped. In the small number of test images I have, for ISO 100 I'm adding +2.5-3 stops because the images are underexposed, and it holds up well. Not sure if they were intentionally underexposed, so I should really do my own tests. The clipping point isn't great, but considering the images are so noise free in the shadows it does suggest that the camera could be underexposed 3-5 stops to bring it more in line with other cameras. This also assumes you don't mind a bit of noise. So I would be interested to shoot at say ISO 800 with an ND and push the highlights as far as possible. For reference, Sony cine cams clip at about 11.7 when shooting S-log3 (~1.5-2 stops), or a tad higher in ProRes Raw for the alpha cine cams, and the Alexa clips at about 55-ish (~4 stops). Amazingly, the BM Pocket 6k Pro clips around the same as an Alexa when highlight recovery works. I did not believe it when other people mentioned it, until I tested it out for myself. Re: monitoring, there was a test I did a few years back where I inverted the "none" profile through the ACES Rec709 inverse IDT and the result is similar to the raw DNGs imported into ACES, but without the full highlight range. It was probably close enough for general monitoring if you want to check grey card exposure but are not too worried about checking highlight clipping, and also if you don't mind adding an external monitor. With the new false color feature, this might be less important if you just want to get the exposure in the right place without needing an accurate monitoring image. Overall, the low highlight clipping point and the mediocre rolling shutter are a turn-off, along with the difficulty of monitoring a raw image accurately. But there's still something really compelling about this camera.

I noticed the ugly falloff and stepiness around the blooming blue lights on the left after I posted this image. Without doing some pixel peeping in the chroma channels, I think this may be to do with chroma subsampling and lossy compression. Again, it would be interesting to shoot a scene like this in raw. You would also get the extra 2 bits in log for colour transitions, so that's 4096 vs 1024 discreet integer log steps per channel which is very efficient. It's not that I think raw is a panacea, but it does everything that something like a log encoded ProRes444 does and a little bit more. For me, the bench mark is always a film scan and then Alexa log C. And tbh I did not have any plans to shoot with any of these Sony cams in anything but raw, unless absolute convenience was required. I really wish more cameras shot internal raw or log ProRes444.

Re: gamut clipping, I finally took a look at one of the shots on this guy's page. Here's a shot bought into ACES which of course has a larger gamut than sgamut3.cine, then it's remapped to display on a P3D65 monitor. I would say that the issue you're seeing is that on the web page there is no correct gamut transform from a wide gamut to something like rec709. However there is still some funkiness from the digital numbers in the top right corner. I'm not sure if that's a compression artifact, or possibly to do with chroma subsampling, or it could be something that is still out of gamut and requires something like the "Neon Suppression LMT" in Resolve. Or maybe the sensor is clipping - it looks like the red channel has hit the dynamic range limit. I have a feeling it all might have been okay if shot raw though. EDIT: It seems that behind the scenes the OS color management or the web browser is transforming it so it should display correctly on a regular sRGB monitor. Hard to tell how this will look for you though!

Do you have the receipts?

IMO The S1H is a "theoretically nice" camera but I would rather get the a7sIII/FX3/FX6 because of the S1H's appalling rolling shutter.

Lol fuck Netflix. But yeah, it is a major pain that this camera has no TC in. Using the Atomos as a workaround is nowhere near as easy. I actually only just realised you can't jam sync with the AtomX Sync module, I thought it had a TC-in. It seems it can only be used as a master and then send a signal over Bluetooth. Way too many things could go wrong on a set. EDIT: obviously there is a lock-in to using the UltraSync Blue (owned by Atomos) with this Bluetooth approach, plus maybe a handful of other devices that allow TC over Bluetooth - the new Zoom F2-BT? I'm unclear on that.

Gotcha.

Possibly dumb question, but does this camera do true 24P with ProRes Raw? If you check the official Sony page, all the frame rates for the internal codecs list 23.98. But under the raw specs, it says 24P and also emphasises that you can crop to DCI from 4.2K. I don't think they would say that unless they could fully support the DCI spec with true 24P. https://www.sony.com/electronics/interchangeable-lens-cameras/ilme-fx3/specifications#features

Can't agree more. This is one to watch. A bit disappointing they didn't roll it out with true 24P though.

I use Resolve too, but I ingest raw to dpx, exr or prores444 so it's not a problem for me. There's a bit of a trend at the consumer level to edit in raw, but I only see it as an acquisition format.

If Sony did later add true 24P and shutter angle control to the FX3 there is zero reason why they could not do that with the a7sIII. But it may be enough to tip me to buy this sexy beast. Are you listening Sony? Holding off for now...

I had a workaround in mind with the a7sIII where if you recorded external raw to a Ninja V you could use the AtomX Sync module for TC in - but I haven't tried it.

Welp, it's a very sexy design with Sigma FP vibes, but I'm really surprised it does not have true 24P and shutter angle control at minimum. It does not seem like it's truly compatible right now as a B or C cam with another Sony Cinema cam. And of course I was hoping they would add TC in. I do think there is bandwidth for further features to be added due to the active cooling. So maybe that is Sony's play here, and we will see some divergence from the a7sIII in the future. But we'll have to see. At this point I would definitely still look at the FX6 over this camera though.

That was always my take, maybe I got that wrong on the FP but my sense was it was a 6000 pixels wide sensor that output a 3840 UHD raw image at the full sensor width. But I don't know how that's possible while maintaining the integrity of the photosites and the bayer pattern, because clearly that's not line skipping.

Fair points. The issue with the X-Trans layout, I would still think it would be possible. Raw video doesn't have to be a Bayer pattern any more than raw still images. But I'm not sure how possible line skipping is with the X-Trans layout so maybe it's a moot point. I've seen some cameras with Bayer pattern sensors seem to do downsampled raw, right? The Sigma FP seems to offer HD raw? But yeah it doesn't make total sense to me. I guess because the X-T3/4 sensor is actually higher than 4K and video is downsampled, it's not gonna be possible to just line skip it anyway to get the same film back size.

I've found that shooting HLG and underexposing by 2 stops with a -2 ND, then bringing it back up in post gives a nice grainy look with the X-T3. In this scenario, I used a HLG IDT in ACES into Resolve. Re: the compression, the best way for me to combat it would have been to shoot everything DCI 4K to an Atomos at ProRes 422HQ and then scale down to 2K DCI. But I shot some stuff internally at 400 ALL-I then converted it to ProRes422HQ, and the compression starts to be a bit of a problem. Also, scaling from 4K to 2K does not fully remove the chroma subsampling artifacts. I basically want a more simple workflow with better results though, for what I do. BUT: I would love it if the Ninja V recording trickled down from the GFX100 to the X-T3/4. Also, I like the fact you can record DCI 2K with the X-T3 and I would love if you could record ProRes Raw DCI 2K.

It's true, the Fuji colours are exceptional at times. I'd love it if the X-T3 was able to output ProRes Raw or something.

Right, it's also a lot cheaper though. I ended up getting it looking pretty organic. It's definitely capable of some nice footage, or maybe we got lucky. There's just pain dealing with compression in post. I ended up getting some nice results shooting HLG and using a custom ACES IDT, partly because of @androidlad's HLG tests. There seems to be more DR shooting HLG than F-log, although the middle grey point is in a weird different place to where you would expect. At this point though, I'm looking for a more robust "small" camera with a more predictable workflow...like the FX3 maybe.

It depends if it's an indie film orr not. For low budget indie films and for TV shot on Alexa, they are generally staying as ProRes. I've worked on small-ish films where the ProRes was converted to DPX, which is more popular on the lower budget studio projects, whereas EXR is more for the big VFX blockbusters, but that's starting to change now because of ACES gaining more mainstream acceptance. If the footage is all ingested as EXR, then the VFX people will work linear but the DI will work log. That is handled transparently by ACES/Resolve where you pick either ACEScc or ACEScct as your working space. If the project is Arri log, it's common for vfx to work Arri wide gamut linear, but send back to production as Arri log dpx to drop into a log dpx DI timeline. There are situations where the ingested footage is Arri Wide Gamut EXRs, which would be converted to logC with a LUT or a lossless mathematical transform before the DI begins. This is very similar to ACES.

X-T3 I have never personally needed to deal with red footage in Nuke. The footage goes through the ingest department at E-Film or wherever and we got EXR plates.

Exactly. The processing on BRaw is off-putting. I believe on the one hand it's to get around the Red patent, but also it's a way to help mitigate the fixed pattern noise that still persists on some more recent Ursas by tackling it before debayering. I heard there was chroma subsampling. I could be wrong on all of the above though.

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.

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.

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.

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. 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.