FS7: Red Raven and FS7

[Gene Sung]

Have you shot on the MX or Dragon sensor Reds?

I only say this because you stated you are looking forward to learning how to use one. If you haven't used one and learned the workflow inside out it is hard to appreciate how good the image quality can be if you work with it properly. Of course, it is equally easy to treat it like a regular camera and get it horrifically wrong - something a lot of people did in the past and then relayed their horror stories of working on Red to everyone else...

I definitely hope to be super wowed as I am buying the camera. You have way more experience shooting Red than I do as I've only worked with them a handful of times. However, I do quite a bit of post work and have worked with tons of Red footage on that end. In fact, doing some post work right now on some Red footage. I guess I've never been super wowed by Red footage in the same way as Alexa which is truly beautiful stuff.

But Red is still a good camera with nice footage and like I said, it's a beneficial camera to know well, so I'm excited to get mine.

 

[DLD]

... Ironically with all this Red talk, if you look at the Panavision TV credits list and what camera was used, Sony F55 is used at least 3x as much as Red on their list of major TV productions.

http://www.panavision.com/TVCredits...

The list is basically Alexa + everyone else and everyone else only accounts for a small portion of Alexa. It'd be interesting to see what happens when 4K becomes the standard.

 

[Harry Lime]

Kind of you to say, ad I wasn't trying to set myself up as any sort of authority on the matter but not only have I researched the shizz out of this stuff, I've owned both and put them through the mill on many occasions.

It's somewhat disingenuous of me to say the FS7 doesn't hold up because it's so good at what it is for, but the reasons why you wouldn't go pitch a Superbowl commercial shot on the FS7 are

1) Just a smidge too much noise when pushed for the top end, and it's not an organic sexy noise.
2) Harder (for me) to precisely zero in on a shot that has to be very exact. Shooting Slog, even with a LUT on a monitor, always feels like 'I THINK we got it, let's go back and check in the edit'.
3) The way it treats not just shadows, but anything in the first stop or two above the shadows can, compared to High end cameras, look like pixelated mush very very quickly. I have had this with perfectly exposed clothes my talent is wearing and whilst it wouldn't be an issue on youtube, is IS one for broadcast.

My most recent findings were based on shooting 240fps 2k to the 7Q, but it also holds true internal 4k 25fps recording too for me.

What format are you shooting the FS7? XDCAM? Proves 444? RAW?

 

[yoclay]

Have you shot on the MX or Dragon sensor Reds?

I only say this because you stated you are looking forward to learning how to use one. If you haven't used one and learned the workflow inside out it is hard to appreciate how good the image quality can be if you work with it properly. Of course, it is equally easy to treat it like a regular camera and get it horrifically wrong - something a lot of people did in the past and then relayed their horror stories of working on Red to everyone else...

The point I have been making is that in a world of online/reality/zero budget drama/corporate/music promos etc the FS7 will do just about everything you need it to do. And do it well enough for most applications. It's only if you have a job where you are really trying to push the image to that next tier of production it becomes apparent that the FS7 image starts to not hold up as well. I am dealing with this now after shooting some commercial studio stuff for a large Beer company because I thought I could get away with not hiring a Dragon.

My old Red was a precision instrument that I tried to shoehorn into doing a lot of jobs it was not best suited to - ie run and gun reality/corporate/lifestyle stuff and it was an unwieldy pain in the arse for a lot of those jobs. But when you had the kit/crew/location/art direction/time and wanted to make something look perfect for the higher/highest end, it always delivered. The new super lightweight Weapon/Raven etc put them back on the table for me, even if they end up being a lot more expensive than the initial list price suggests.

On a personal level I am at a point where most of my bills are paid by doing the - not lower, but REGULAR - type of work, but I am on the cusp of getting more of the higher end stuff and would really like to push to do more of that. I don't imagine I'll sell the FS7 but there is certainly scope for something better in the not too distant future on my horizon.

You have described very well the dangers of being an owner operator. I see this all the time. Trying to shoehorn your jobs into the one camera you own is really a prescription for disaster.

I have shot and graded MX as well as Dragon. MX has big noise problems in the shadows too. The Dragon, less so. I guess people could argue about how organic the noise is of one manufacturer compared to the other, but I think that is personal taste. I don't feel Raven will be on the plus side of this because the sensor is much smaller, but then no one really knows this do they? I will say this, there is very little footage out there that cannot benefit from noise reduction later in post. Including the Alexa. With Dragon, the real issue is with which OLPF you use. The Skin Tone OLPF really helps with guess what? Skin tones. But both the standard and the skin tone OLPF's definitely can have noise issues in the shadows. Which is why they have a low light OLPF. Getting it right with Red is a tricky business. I don't know if it is vastly superior to what can be achieved with an FS7 and Sony RAW, but I have definitely seen RAW from an F55 which absolutely on par with anything coming from Red.

Now. There is something I would like to point out. I have long stated that I feel the F3 has in my subjective opinion one of the most organic images. As far as I am concerned it is also better in low light than most of the RED's. Why is this? I leave you all to consider the following pixel pitches:


Sony PMW-F3- 12 Microns
Phantom Flex 2K - 10 Microns
Arri Alexa EVIII - 8.25 Microns
Phantom Flex 4k - 6.75 Microns
Canon C300 - 6.4 Microns
Canon 5D Mark III - 6.25 Microns
Sony F55 - 5.8 Microns
Red Dragon - 5.2 Microns
Red Raven - 5.2 Microns
Red MX - 5.4 Microns

It doesn't tell the whole story but want to guess which cameras probably look best (and most "organic") to me?

 

[starcentral]

Dennis how does the 16bit Sony RAW compare to any of the RED RAW you have used?

As far as playing around in post and moving the image around, it is as you would expect with RAW from any camera. There is a lot of latitude to push tones, exposure and color around.

Of course there is always the question if you need that or not. To borrow Ben's phrase, "in a world of online/reality/zero budget drama/corporate/music promos etc", the RAW workflow is probably going to be overkill no matter what camera.

I think everyone agrees that 10-bit XAVC is quite excellent for grading and most of the time this codec is all you'll need for the majority of shoots. Like any codec, even if it's only an 8-bit codec (which many of us if not all of us have worked with in the past) you can still get great images from it and especially if you take more care with exposure and controlling scene contrast ratios, etc..

Do I think Sony RAW looks dramatically different than XAVC or SR Codec? With the same LUT applied they look quite identical and the difference is really only going to be seen in the grading. So comparing it to RED RAW I couldn't say you're going to get closer to it or better of a look. Each camera still has it's own color science and so that's where I feel a lot of the camera's look is coming from.

As it has been said before s-gamut3.cine, ACES workflow, or use of film emulation LUTs can help you get away from any Sony traditional look - that is if you believe there is one to begin with.

http://www.panavision.com/TVCredits

I don't know about all of them, but Blacklist and Marco Polo are shot in 4K RAW on the F55. There has to be a good reason why big productions chose RAW over XAVC, probably for the latitude they can get from it.

 

[Grug]

I'm watching Bloodline on Netflix at the moment (shot on the F55 and Panavision Primos), and it looks fantastic, not a hint of 'Sony' look about it. I get the impression that colourists are finally getting the hang of Sony footage - which is great news for all of us, as hopefully people will warm further to the Sonys as they see more and more examples of the results that can be had from them.

To answer Jack's question of how Redcode and Sony Raw compare. Having spend a fair bit of time working with both, I'd break it down like this:

- Redcode (at 5k or 6k) offers significantly more resolution
- Sony Raw at 3:1 compression, generally offers fewer compression artifacts on a per pixel basis (however, once the Redcode is downsampled to 4k to match the Sonys, the overall quality seems very close, at 3:1 compression Redcode offers a significantly more detailed image - but no one ever shoots at such a low compression ratio with Red)
- The noise floor on the F55 and F5 is significantly lower than the Dragon (with skintone OLPF) or MX sensors, the Sony's have a clear advantage in low light and at higher sensitivities
- Highlight latitude seems incredibly close between F5(5) raw and Dragon, MX is a long way behind both
- Redcode's variable compression ratios are awesome, and make shooting raw a much easier proposition (talking producers into buying the harddrives to shoot 3:1 4k Sony Raw seems significantly harder).
- Sony really needs to introduce the 6:1 compressed Sony Raw-Lite from the F65 into the R5 recorder for the F5(5). It would see a much higher take-up of shooting raw with the Sony cameras.

On the whole, comparing what seems to be the two most common types of raw shot with the two brands (5k 6:1 Redcode, and 4k 3:1 Sony Raw) I'd call them incredibly close overall. Either option will let you get a terrific result in finishing (assuming it's been well shot).

 

[Harry Lime]

Dennis how does the 16bit Sony RAW compare to any of the RED RAW you have used?

Keep in mind that the output from RED cameras is wavelet compressed (3:!, 5:1 etc.) Essentially it's Cineform or at least that is what I have been told. So, it's a sophisticated compressed codec in a RAW container.

The obvious advantage are considerably smaller files, but in my experience RED footage lacks the depth and richness of something like ArriRAW, Sony RAW etc. due to the compression. It will band much quicker and personally I can see the wavelet compression that also shows up as artifacts when you're shooting green screens etc.

If you can get hold of some ungraded footage you should do some testing of your own to see which one works for you.

 

[Ben Scott]

What format are you shooting the FS7? XDCAM? Proves 444? RAW?

xvac-I internal or raw to pro res on the 7q. Gave up on Sony's awful interpretation of raw via the 7q a while back.

 

[Ben Scott]

Keep in mind that the output from RED cameras is wavelet compressed (3:!, 5:1 etc.) Essentially it's Cineform or at least that is what I have been told. So, it's a sophisticated compressed codec in a RAW container.

The obvious advantage are considerably smaller files, but in my experience RED footage lacks the depth and richness of something like ArriRAW, Sony RAW etc. due to the compression. It will band much quicker and personally I can see the wavelet compression that also shows up as artifacts when you're shooting green screens etc.

If you can get hold of some ungraded footage you should do some testing of your own to see which one works for you.

it absolutely 1000% will not band quicker than FS7 raw. I have never had a Red shot ruined by banding. Have it all the time with the FS7. I can't speak for the F55 or F65 but you would expect them to be better

And Just to to put the compression info in perspective, Prometheus was shot on MX sensor at 5k at 5:1 compression. Rust and Bone had a more indie vibe and beautifully organic aesthetic and was shot at 6:1 I believe. I shot plenty for broadcast documentary at 8:1. And Red Dragon sensor doing 4K 120fps at 13:1 (Raven's max) is cleaner than the FS7 doing 150fps in 1080p.

 

[Ben Scott]

I'm watching Bloodline on Netflix at the moment (shot on the F55 and Panavision Primos), and it looks fantastic, not a hint of 'Sony' look about it. I get the impression that colourists are finally getting the hang of Sony footage - which is great news for all of us, as hopefully people will warm further to the Sonys as they see more and more examples of the results that can be had from them.

To answer Jack's question of how Redcode and Sony Raw compare. Having spend a fair bit of time working with both, I'd break it down like this:

- Redcode (at 5k or 6k) offers significantly more resolution
- Sony Raw at 3:1 compression, generally offers fewer compression artifacts on a per pixel basis (however, once the Redcode is downsampled to 4k to match the Sonys, the overall quality seems very close, at 3:1 compression Redcode offers a significantly more detailed image - but no one ever shoots at such a low compression ratio with Red)
- The noise floor on the F55 and F5 is significantly lower than the Dragon (with skintone OLPF) or MX sensors, the Sony's have a clear advantage in low light and at higher sensitivities
- Highlight latitude seems incredibly close between F5(5) raw and Dragon, MX is a long way behind both
- Redcode's variable compression ratios are awesome, and make shooting raw a much easier proposition (talking producers into buying the harddrives to shoot 3:1 4k Sony Raw seems significantly harder).
- Sony really needs to introduce the 6:1 compressed Sony Raw-Lite from the F65 into the R5 recorder for the F5(5). It would see a much higher take-up of shooting raw with the Sony cameras.

On the whole, comparing what seems to be the two most common types of raw shot with the two brands (5k 6:1 Redcode, and 4k 3:1 Sony Raw) I'd call them incredibly close overall. Either option will let you get a terrific result in finishing (assuming it's been well shot).

Very interesting. I wonder if anyone has done a side by side of R5 Sony raw vs 7q raw DNG.

 

[Nirv]

xvac-I internal or raw to pro res on the 7q. Gave up on Sony's awful interpretation of raw via the 7q a while back.

As someone considering RAW via the 7Q could you please explain what Sony's awful implementation is? Please be as comprehensively damning as you feel you need to be I can take it.

I'm not too chuffed with XAVC and I'm looking into Prores internal, Prores external and RAW external (not with Sony's recorder) so any steering in the right direction gratefully received...

Cheers

 

[Ben Scott]

I have shot and graded MX as well as Dragon. MX has big noise problems in the shadows too. The Dragon, less so. I guess people could argue about how organic the noise is of one manufacturer compared to the other, but I think that is personal taste.

I always felt it was something I could dial out or expose for with MX, and the shadow noise has never forced me to bin a shot like I really wanted to on some of the first 7Q FS7 raw stuff I shot

Look at the first 5secs of this - https://vimeo.com/127490168

None of that shadow banding is Vimeo compression. It's all in the raw. Terrible. And by any reasonable reckoning it wasn't under-exposed. But if you shoot FS7 raw (or raw to pro res) you need to bring the shadows up so high to avoid this it can drastically affect the rest of your image's exposure and for me, it ends up being a false economy.

 

[starcentral]

Hi Ben, out of curiosity what was your method to determine exposure? grey card, zebras,? Did you rate the camera lower as to overexpose by any amount of stops?

Did you happen to record anything to internal cards for a comparison? I'd be curious to study the same shots to XAVC and see what they looked like, I'd bet there would be no banding so I'd have to question what it was in the 12-bit RAW to Odyssey chain that caused this but I might have a good idea….

With the way linear RAW works and assuming Sony's 12-bit RAW is a pure linear gamma curve there would actually not be enough data bits left at the lower end of the sensor's overall dynamic range of 14 stops.

This table from my FS7 ultimate exposure guide helps to illustrate what I'm talking about. As everyone knows with 12-bit you have 4096 total levels of grey. The highest stop takes 2048 of these levels (so half what is available overall) and then the 2nd highest stop takes 1024 levels, the third stop takes 512 levels, and so on and so on. By the time you get to the bottom you do not have any bits left to even cover the entire dynamic range of the sensor (14 stops). You'll actually notice the table only covers the first 12 stops and doesn't even get to the 13th and 14th stops as there would be no data available for them.

Now obviously Sony would not short change the coverage of data bits on a linear gamma so likely implements some rounding or bending of the linear gamma to help get some data bits covering the lower end of the tonal range. But there is no actual documentation by Sony on this, nor have they ever discussed it to my knowledge. Frankly we don't know how Sony is spreading those bits around.

16-bit RAW from the F5/F55 covers the dynamic range somewhat better because you're starting at 65,000 levels at the highest stop, but even with 16-bits you get down to 1 data bit coverage for the bottom stop position. Again some bending of the curve must exist.

Your best bet might be to try some tests and shoot ETTR where white points are just at a clipping point, then bring the shadows down in post. This should yield a cleaner and/or less noisier image.

 

[Ben Scott]

Dennis that chart is absolutely what I have found in real world terms and two other tech people from these forums have confirmed it to me privately that this is precisely what is happening and the cause of everyone's raw woes.


I've used a bunch of things for exposure depending on the setup/gig, from the Histo on the FS7 to the Histo on the 7Q to the Zebra 2 set to 61% on the FS7 and taking a reading off a 95% card to metering with a Sekonic 478.

Target exposure doesn't seem to be an issue, unless you have material in the bottom 3 stops but the scene has a wide DR and you need to preserve highlights.

An example (which I can't post yet) was a studio white infinity curve slomo shoot I did a couple of weeks ago. Background just shy of clipping and skintones on the models exposed lovely and as high as I dare do, but STILL if someone wore a dark top it would turn to mush.


(ahem) Incidentally it seems if I leave the focal reducer on my FS7 the meter readings are accurate for Cine-EI's EI2000 down as ISO2000 on the meter.

 

[starcentral]

Hm, you want to start a discussion on how Sony has rated their camera's ISO wrong? I'm down for that, already written about it numerous times and provided calibration constants from different light meters so their actual mapped grey values can be calculated rather than guessed… but I must go now, time for a meeting. Perhaps another day.

cheers,

 

[Ben Scott]

Ha, nope I've seen you write on the subject before. Just noticed that everything seemed to align when I left it on

 

[morgan_moore]

With this distribution one would want to be exposing for absolute clip point and then decide how many usable stops you have below (probably 6)

I have no idea if the exposure tools on any of the kit allow for this.

forget whites at 61, etc. IMO

Ben have you done a bracket test.

set a scene for about 5.6 nd then shoot it from 1.4 to 11 and then try and match the files in post..?

S

 

[Ben Scott]

Nope. On a day to day basis I don't have these sorts of problems, and I won't be using an FS7 on jobs of this ilk in the future. As it is, the best way to expose in this sort of environment is to do what you just said have it set to be at the absolute clip point and then bring it down 1 notch, trying to rescue the darkest (although I should stress NOT DARK) areas of the frame and then bring it all back down in post, but this is what brings you into the very imprecise 'I think we got it, let's check back home' way of shooting that simply isn't acceptable on higher end productions.

 

[morgan_moore]

Nope you have not tested or nope you have not got the exposure tools?

Lut calc?

S

 

[Ben Scott]

Nope I haven't done a bracketed exposure test. I don't really feel I need to if I'm honest. Like I say, 90% of the time everything is hunky dory, and in future I'm not going to try to fit a square peg in a round hole and will just rent a better camera for these jobs.