RED Monstro 8K VV & ARRI Alexa LF 4.5K Camera Test 2019

[Jarred Land]

Touché. Granted, I was referring to the resolution game -- a 2k camera substantially outselling a 4k camera 10 years later when 4k is actually viewable, is very telling (and hence why I mentioned the positive response to Gemini, even though it can "only" deliver 4k and there are 8k options available... used/old stock of Epic-Ws and Weapon Helium being cheaper).

Also, all Arri's cameras, from 2k Classics to 6.5k A65, look/react/expose/function the exact same, so they retain value. IPP2 is a step in that direction from a post perspective (and I *love* it for that), but each RED sensor exposes differently, captures different DR/Colour/artifacts/noise floors, and has different nuances that need to be learned to extract the most out of each. Production doesn't want to deal with that, though owner-ops don't mind because we own the tool and can test/finesse as much as we want.

(Side note: To be fair, Alexa was working/actively renting for 5x as long as the R1. It seems REDs are harder to get it out the door on a rental basis unless you're at the tip-of-the-spear sensor-wise, which ends up being more costly over the same timeframe).

Yup, there is no question that we are still learning. Arri is "easy" for the same reasons you listed. I like to think you can squeeze more out of a RED, but you need to invest in learning the tool. That's an asset but also a liability.

That's why when you look at the cinematographers that shoot RED, they are usually far more technical and they understand the technology behind the image a bit more than others. You don't need the Phil Hollands of the world on the ARRI side, because there just isn't that depth. That's why it is so polarizing.

Arri isn't a company that invents things... they just don't. They are a company that does a really, really great job of refining.

We are the opposite... We are pretty good at inventing things, and we kinda suck on the refinement . And that's what we are doing now.... Learning the refinement part.

 

[David J. Buchanan]

Yup, there is no question that we are still learning. Arri is "easy" for the same reasons you listed. I like to think you can squeeze more out of a RED, but you need to invest in learning the tool. That's an asset but also a liability.

That's why when you look at the cinematographers that shoot RED, they are usually far more technical and they understand the technology behind the image a bit more than others. You don't need the Phil Hollands of the world on the ARRI side, because there just isn't that depth. That's why it is so polarizing.

Arri isn't a company that invents things... they just don't. They are a company that does a really, really great job of refining.

We are the opposite... We are pretty good at inventing things, and we kinda suck on the refinement . And that's what we are doing now.... Learning the refinement part.

This is dope to hear. On the next one could we get the processing faster to 2ms instead of 5ms, and please ND filters built-in. I know we keep asking for it but... ya know. It would be dope!

 

[Mike P.]

Yup, there is no question that we are still are learning. Arri is "easy" for the same reasons you listed. I like to think you can squeeze more out of a RED, but you need to invest in learning the tool. That's an asset but also a liability.

That's why when you look at the cinematographers that shoot RED, they are usually far more technical and they understand the technology behind the image a bit more than others. You don't need the Phil Hollands of the world on the ARRI side, because there just isn't that depth. That's why it is so polarizing.

Arri isn't a company that invents things... they just don't. They are a company that does a really, really great job of refining.

We are the opposite... We are pretty good at inventing things, and we kinda suck on the refinement . And that's what we are doing now.... Learning the refinement part.

I hear ya. And I wouldn't be shooting red if that wasn't the case... I just think there's a balance that is being ignored, while boundaries that already exceed expectations are being pushed instead. But as you say, RED's learning the refinement (IPP2 and DSMC2 Unified being great examples of that).

...that said, it is kinda frustrating when anyone can pick up any flavour Alexa and crack off incredibly rad footage first attempt, and all they need to do is apply Arri's LogC-to-709 LUT. Or a better example, when Panny takes the same RED sensors we're using and says, 'yeah, we just made it substantially more film-like out of the gate... expose for ISO1600 and WYSIWYG.'

Speaking of which, when interchangeable OLPFs came out, I thought for sure you guys were going to have "Kodak OLPF with corresponding Log3G10-to-5219 LUT'" and "Fuji OLPF with Log3G10-to-Eterna LUT" etc, etc... Not 3rd party *almost* similar filmstock OLPF and LUT combos, but RED designed/developed/refined ones that extract the most possible. It seemed like a no-brainer and I think it would've dominated at rental houses (like as much as Ranger, maybe more). Just "so, you want Kodak flavour today or Fuji flavour?" No fuss, no muss. (and owner-ops/technical DoPs still could've tested/finessed/refined their own sauce with the regular OLPFs and their colourists).

Refinement is arguably the best part; it's when things go from "great" to "perfect".

 

[David J. Buchanan]

Also what's crazy about this, I can't really see the difference (a few minutes in - watching on YouTube). At this point, it almost seems like preference.

 

[Brad Grimm]

Thanks for doing this Phil. Monstro is a monster. Please Jarred, just keep improving what we get off of this sensor. I've been upgrading with RED for a long time, but we need to slow down and steep for a bit while we perfect this amazing technology.

 

[Jarred Land]

Refinement is arguably the best part; it's when things go from "great" to "perfect".

Amen.

 

[Phil Holland]

Also what's crazy about this, I can't really see the difference (a few minutes in - watching on YouTube). At this point, it almost seems like preference.

I love YouTube, but YouTube is a lie I can't upload a master file there for you to review because all streaming sites re-encode whatever you give them to save space. So you upload a high quality master, in this case the ProRes 422 HQ 4K file I provided the download link to.

That reveals a whole lot more about the cameras as a medium.

Yup, there is no question that we are still learning. Arri is "easy" for the same reasons you listed. I like to think you can squeeze more out of a RED, but you need to invest in learning the tool. That's an asset but also a liability.

That's why when you look at the cinematographers that shoot RED, they are usually far more technical and they understand the technology behind the image a bit more than others. You don't need the Phil Hollands of the world on the ARRI side, because there just isn't that depth. That's why it is so polarizing.

Arri isn't a company that invents things... they just don't. They are a company that does a really, really great job of refining.

We are the opposite... We are pretty good at inventing things, and we kinda suck on the refinement . And that's what we are doing now.... Learning the refinement part.

Heh. There was a time when I was more on the Arri side of things, but from about Dragon on I realized I could do a lot more, both on creative and technical levels, within the RED ecosystem. Plus there was a bit more of a film-like workflow and mentality with the downsampling POV. I needed 4K for the films I was working on, which led me to RED initially and eventually plopping down coin and jumping in. But the second you guys fixed the IR stuff and took the color up a notch compared to Mysterium-X, well, that's what I needed from a filmmaker's perspective.

ARRI was always clear early on that ALEV III and the original Alexa, which was built on the shoulders of the D20 and D21, that it was for a 2K finish. Even now I suspect their 2020 4K S35 camera option will likely manifest in an around 6K larger format camera. That's actually where this gets much, much more interesting. That's not just because RED did that or currently Canon and Sony doing that, but more about the conversations happening right now about 4K finishing behind certain doors.

That all said, I could easily have made a "pretty footage" test, but I really wanted to show some practical and technical stuff people could look at and examine. Everybody will see what they want, but the core stuff is there and not something that can be really argued about as the newer pixels are a bit different than the well known older ones. A DP just shot me a great email regarding his findings and he's the first to bring up the lack of details and colorization on ARRI material, but there's some aspects that ARRI does well, some will favor that color. I'm mostly highlighting the differences and also bringing them together.

I think the Dynamic Range and Color Stability which in effect impacts the workable latitude between these systems is notably superior on Monstro, and I'll say this, against any system currently. Though Canon does stuff a bit differently with their processing. To that point for a 4K straight finish where you work on the ISO scale with ARRI is something you need to be really aware of. ISO 800 is sort of the top-o-the world. ISO 400 and 800 seem to be where people live at the moment with the LF for 4K finishing and this is even outside my previous tests. If you can handle ISO 1600 noise and or NR in the pipeline, there is that. Monstro-side, due to the oversampling and all that, ISO 3200 might be the equivalent top-o-the world from an equivalence perspective. I've pushed it further a couple times into the ISO 5000 realm. The real juice comes in the form of being able to work at lower than 8K resolutions with Monstro and still net a better image quality in those specific directions. Similar to the productions doing the 6K or 7K extractions.

I agree on refinement and pushing towards the future in big ways. At the same time ARRI really needs to add to their focusing toolkit in camera. I've talked to Bill Bennett about this at length and even he agrees as it was literally his biggest issue on his first LF shoot with the dirt bikes. To that point, that's also an area where RED could develop a bit on the magnifcation-while-rolling side of things. But the other part of that is 4K monitoring and 4K 7-ish inch screens too as well as in larger production monitors on set.

 

[Mike P.]

From the youtube encoded 4k, viewing the log at ISO3200, Monstro definitely seemed perfectly usable, whereas the MLF definitely didn't seem usable (though when normalized it wasn't nearly as bad.)

Does Arri do internal sharpening, because LF looked noticeably sharper at 4k/184%?

The low-light/treasure chest test showed the patina of Alexa's colour, which I think is a.) very reminiscent of film and b.) more visually pleasing... Plus, it might help with lowlight, as the blue-channel is noisiest/green is strongest in CMOS sensors.

Another note on colour; yellow in skin tones never looks good to me unless it's done intentionally in the grade. Alexa appears to still have less yellow in the skin tones. I think a few of Phil's LUTs, and Hjorve's "Improving the Colour" LUTs, and TrueColour's REDtoAlexa LUT tackle yellow in skin tones directly, but I wouldn't mind seeing it in an official capacity (e.g. RED's own normalizing LUTs/SDK.)

 

[Brendan H. Banks]

Yup, there is no question that we are still learning. Arri is "easy" for the same reasons you listed. I like to think you can squeeze more out of a RED, but you need to invest in learning the tool. That's an asset but also a liability.

That's why when you look at the cinematographers that shoot RED, they are usually far more technical and they understand the technology behind the image a bit more than others. You don't need the Phil Hollands of the world on the ARRI side, because there just isn't that depth. That's why it is so polarizing.

Arri isn't a company that invents things... they just don't. They are a company that does a really, really great job of refining.

We are the opposite... We are pretty good at inventing things, and we kinda suck on the refinement . And that's what we are doing now.... Learning the refinement part.

Wow. Very cool to know your thoughts and it’s good to hear as a person who feels similarly. There’s a certain massaging involved, but I’ve found the juice to be more than worth the squeeze.

 

[Drew Thomas]

Yup, there is no question that we are still learning. Arri is "easy" for the same reasons you listed. I like to think you can squeeze more out of a RED, but you need to invest in learning the tool. That's an asset but also a liability.

That's why when you look at the cinematographers that shoot RED, they are usually far more technical and they understand the technology behind the image a bit more than others. You don't need the Phil Hollands of the world on the ARRI side, because there just isn't that depth. That's why it is so polarizing.

Arri isn't a company that invents things... they just don't. They are a company that does a really, really great job of refining.

We are the opposite... We are pretty good at inventing things, and we kinda suck on the refinement . And that's what we are doing now.... Learning the refinement part.

Thanks for taking us all on the journey and being so open with where y'all are at. It's honestly one of the best parts of owning a red.

 

[Roman Alaivi]

Hi, thanks for the comparison.

Do you have any Log C frames you can post? I've shot quite a bit of Log C material over the past few years and I suppose I'm just curious about the look of some of the shadows in this comparison, they feel a little crunchier / darker than I'm used to with really any Alexa material. I'd love to see just a bit of log material before your grade.

 

[Aaron Lochert]

Hi, thanks for the comparison.

Do you have any Log C frames you can post? I've shot quite a bit of Log C material over the past few years and I suppose I'm just curious about the look of some of the shadows in this comparison, they feel a little crunchier / darker than I'm used to with really any Alexa material. I'd love to see just a bit of log material before your grade.

Which Log C > Rec709 LUT are you used to using? I've notied the one included within Resolve has a lifted black level compared to the one you might download from Arri's own LUT generator.

 

[Phil Holland]

Hi, thanks for the comparison.

Do you have any Log C frames you can post? I've shot quite a bit of Log C material over the past few years and I suppose I'm just curious about the look of some of the shadows in this comparison, they feel a little crunchier / darker than I'm used to with really any Alexa material. I'd love to see just a bit of log material before your grade.

This is the LogC to Rec.709 LUT Arri provides via their LUT creator tool, same in the SDK if I am not mistaken. The Warehouse Scene is lifted up to ISO 3200 to see exactly what's going on in the shadows when comparing Log3G10 and LogC.

Yes, I personally would typically grade the material with a lower contrast curve, but that in effect increases the shadow noise. This test really isn't about how I would grade the material however. And even more honestly, similar to the Skin Tone lighting setup, it's best to add a bit more fill light like that anyways.

 

[Tom Gleeson]

Phil,

Bummer your Google Drive has maxed out and it is denying access to the ProRes.

 

[Phil Holland]

Phil,

Bummer your Google Drive has maxed out and it is denying access to the ProRes.

Can you screenshot that? It's still working here, but if that does happen I'll need to figure out a way to share a 60GB file with a few thousand more people apparently.

 

[David J. Buchanan]

Your right about the Prores seeing more in the image. YouTube really compresses the hell out of things. Anyway, that Low Light shot is staggeringly different, almost strangely so. I still think the two cameras are neck and neck in some things, but yeah the Monstro is destroying the ALEXA in noise floor and detail. Gotta give it up to ALEXA for that old ass sensor for keeping up tho.

 

[C. Burkhart]

I think my point of view towards the latest generation of RED sensors (which really are better than Dragon/Epic) vs the Alexa LF is that if information in the shadows or resolution is what counts (especially for crunchy commercial stuff) then RED has the upper hand, however if highlight roll-off or skin tones are what matters to me then I still prefer the Alexa.

 

[Alex.Mitchell]

I apologize for going a little off topic in a camera test thread, but this has been on my mind and I was wondering if anyone else can relate...

Something I really hope camera manufacturers work on is cooling. I respect that DSMC cameras are doing a lot in a small space, and the bigger RED bodies like Ranger and DXL probably run silent all day long, but I grit my teeth every time we get in to dialogue heavy setups because I know there's a good chance the fans are going to spin up at a bad moment. From a sound department perspective, nothing breaks my heart more than having to tell my director that the dialogue was contaminated.

Yes, we're using Adaptive Quiet Record. Yes, we've tried dialling in manual settings that will at least give a consistent sound, if not a quiet one. Yes, I encourage things like tail-slating for long takes so we can squeeze every second from these bodies. Yes, I always bring it up in pre-production. Inevitably, it always becomes an issue anyway. I implore camera manufacturers everywhere; please, throw us a bone. We're fighting so many battles to capture the performance and we need a little more help on this front.

Sorry if this came off a little melodramatic--I just got off a feature with a Monstro body and it was a punishing shoot for this very reason.

 

[andrewhake]

Really interesting stuff Phil.

And I have to say, having watched both on youtube and the ProRes version. If you really want to compare you really need to watch the ProRes if you have the connection to download it. The differences become MUCH more apparent without youtube's compression. Might be worth uploading the 422 HQ files to Vimeo directly instead of youtube for these type of tests.

Both cameras look great, but I don't think there is anything the AlexaLF can do that the Monstro can't easily replicate. Or even the DSMC2 Gemini or Dragon. But the same can't be said the other way around in terms of image, form factor and various other things. If you want a softer look for creative reasons, that is very easy to achieve with the Monstro, if you want the overall color profile that comes out of the Alexa, that is also very easy to achieve. I think most of the appeal of the Alexas come from people just liking the green shadow tint and higher contrast you get out of camera, and somehow thinking that is only achievable with those cameras.

 

[Bérenger Brillante]

That conforts me that I won't buy any Arri camera before they release a new sensor.

No surprises, I prefer the overall look of the Monstro but again, I have to agree, the 4 girls at 5600° studio, on skin tone Alexa looked better (but skin tone only, not various shades of skin, on that and everything else Monstro looked better).

In difficult situations, the Monstro is ahead every time, whereas the Alexa starts to show weird color shifting on skintone cheeks, (I guess where the color science doesn't detect skin tone) and in standard situations, both are excellent.

You should add to the video a side by side magnification of the X-rite color checker left side.
There is a big difference in the two cameras on base skin tone depending on the temperature source. This is the only significant part to me, as I think we already all knew Montro's a monster.
At 5600°, the skin tones and colors are different but no one can say one is better than the other but at 3200° the Monstro is a bit more orange-green on the skin, where the Alexa keeps the same tone.

Add that all the benefits of the DSMC², I just don't understand Red haters.