Arri Alexa and Mysterium-X...

[David Mullen ASC]

Ok David I get it, but the fact is that the native 5000K image I get, can be bended and molded far more than any other camera in the world.. Thats why I call it a multi whitebalance camera.

You CAN make it look like 3200K native. no?

Have you tested all the other cameras to know for sure that you can change the white balance much farther in post with the Red than with any other camera?

Because until the recent M-X sensor came along, that really wasn't that true, Red wasn't the best camera out there for 3200K shooting, there was too much noise in the blue channel except at the lower ASA ratings. NOW it's not a problem with the M-X sensor due to the much lower noise floor.

It's all about noise, color channel noise is what limits your choices in color-correcting.

A good test will be to shoot a daylight-balanced and a tungsten-balanced MacBeth chart on the Red and the Alexa (set for daylight and then for tungsten) and simply see if there is more noise one way or the other when correcting tungsten to daylight and daylight to tungsten. For example, if you accidentally (don't know how but this is hypothetical) shot the Alexa in 3200K balance for a daylight scene, would the corrected image have more noise problems compared to the Red version? Conversely, would the Alexa set to 3200K (baked in) have less noise for the tungsten chart than the Red (being 5000K native) corrected in post to 3200K?

All that really matters is the practical real-world results of these decisions.

All of these cameras are really "balanced" closer to daylight, in that they are less noisy in daylight balance. The Genesis/F35 as well, which also bakes in color temp, I've looked at charts on waveforms and seen the noise go down in daylight compared to 3200K. Yet practically speaking, the Genesis/F35 set to 3200K, though this was not "native" to the sensor, was less noisy at that balance than the old Red camera set in metadata to 3200K. Like I said, this has been solved with the newer, low-noise M-X sensor.

But my point is that what we are talking about is the ability to make a range of color-correction decisions without compromise and artifacts like noise. So the only thing that matters is a real-world test between the Red and Alexa in these situations. If the noise floor of the Alexa is low enough, it shouldn't matter too much if you want to take a 3200K recording and change it back to 5600K even though it would have been even better to shoot at 5600K in the first place.

 

[David M]

If you are asking if there is agreement that 3840 x 2160 is 4 X the pixel resolution than 1920 x 1080 pixels, well then yes I'd agree with Jarred's volume analogy and mathematics. I think the confusion is when you compare 2K to 4K as in film scanning and then we are referring to a scan as being twice the resolution not volume of pixels

Well, that was more or less what I said: A system with twice the resolution is going to generate four times the data/pixels, whatever.

In normal Industry parlance, resolution refers to the resolving power per unit measurement of the imaging medium, it is not necessarily an absolute value. The objection is that like many other technical terms with an established meaning, it is being re-defined for market purposes

For example, a particular negative stock might be able to resolve "60 lines per millimeter". That specifically means that with the right optics, it is possible to fit an image containing up to 30 white lines on a black background on each mm of the emulsion, with a useable level of contrast.
I doesn't matter if that film stock is then cut to 65mm, 35mm, 16mm or even Super-8 size, the available resolution in lines per millimeter is exactly the same.

The maximum number of lines possible in the output image is directly determined by how many mm wide (or) high the imaging area is.

If one imaging system can capture or display a pattern of 1,000 vertical lines across its width and another can only display 500, we say that the first camera has twice the resolution of the second.

To achieve that resolution, it needs four times as many pixels and four times as much data needs to be processed. It does NOT have four times the resolution. It only has twice the resolution.

 

[Nik Skjøth]

David. its not the sensor.. it's the file.... If my Canon Rebel can shoot raw and I can make multi whitebalance scenes, then surelly red can to.. Otherwise the R3D files are NOT raw.

 

[David Mullen ASC]

Then I dont know what RAW is apparently..

Dont tell me that if i shoot redraw in tungsten light Im as much screwed as I would be if I shot still DSLR jpg's balanced to 5000k

I repeat.. the only relevant question here is.. do you have to decide what whitebalance to shoot with on the red prior to post or not? If YES why raw then? what advantage do I get from raw if everything is natively preset.

The sensor is balanced close to 5000K so in a 3200K situation, what gets recorded in RAW will look too orange.

Now if you're asking what's the difference between shooting RAW and converting to 3200K versus in JPEG and "baking in" 5000K in-camera and having to fix it to 3200K in post, you're basically talking about the problems of JPEG as a recording format, that's where you get the advantage of RAW, you are working with the entire information captured by the sensor (though compressed by RedCode in RED's case.) So you aren't "as screwed" with RAW compared to working with JPEG's where a lot of data would have been tossed and compressed heavily. But in terms of color, you have to do similar things to get the image to look color-balanced for 3200K in terms of the RGB levels.

Graeme can explain more clearly than I can about how color temp settings are created / applied from RAW to RGB. I'm not sure if there is first a basic conversion RAW to RGB and then RGB levels are changed, or in the transformation itself, some pixel / photosite values can be changed so that RGB comes out already with the new color temp setting.

Yes, in theory, it's all better left for post where your processing power is going to be much higher and advanced than what a camera has internally anyway, but I'm really only interested in the real-world results of these different methods, not theoretical differences. A difference that makes no difference is no difference, to quote Spock I think...

I'm not sure how I can say it any more simply, that the camera responds to a certain color temp better than others and that RAW records everything that the sensor responds to. So basically the Red RAW files are "naturally" close to daylight balance and have to be manipulated in some way to get other color temps to look correct, but you are in essence working with a daylight-balanced recording.

As for the Canon, as with the ARRI, it seems some camera's "RAW" mode bakes in white-balance and some do not, so I can't say what your Canon is doing one way or the other. Graeme seemed to imply that color temp and ASA are "baked into" the Canon's RAW files. Or maybe not.

 

[Martin Stevens]

RAW is like a Film Negative.
Baked in 8 bit JPEG is like a Release Print.

Well.............sort of.

I would not want to grade a Release Print.

 

[Rick Burnett]

David. its not the sensor.. it's the file.... If my Canon Rebel can shoot raw and I can make multi whitebalance scenes, then surelly red can to.. Otherwise the R3D files are NOT raw.

You are entirely missing the point of this discussion pertaining to white balance. The question pertains to the mathematical application of white balance to raw pixel data and whether it is better to leave the pixel data alone and just use meta data in camera and finalize the white balance adjustments in post, or, if applying the white balance to the raw data in camera and possibly having to adjust it later is going to produce any significant degradation in quality of the image. David is pointing out that regardless of the mathematics involved with the white balance, the noise of the sensor is a huge factor in what is possible REGARDLESS of when it is done.

NO ONE is arguing about JPEG related to Red Raw of Arri Raw. Calling something multi-white balance makes ABSOLUTELY no sense. With your description of applying different white balances to the same scene is VERY MUCH dependent on the sensor noise floor.

 

[Nik Skjøth]

ok ok... I see what you are doing here.. You are speaking about the technical definitions which make the image appear in a very teoretical manner, and alltho Im sure you are right you never really approached the relevant question which this whole discussion is about..

Do we agree that:

1. you do NOT preset a particular Kelvin value while shooting for anything else than reference? This Kelvin value has no impact what so ever on the post work?

2. the same preset can be used in both tungsten and daylight scenes?

3. you could split an image into two and whitebalance it differently, making it LOOK like it was shot with two cameras preset to each its own white balance?

 

[David M]

Reminds of that joke that the Americans spent years and hundreds of millions of dollars on developing a ballpoint-pen that will generate "gravity-effect" in space for it to work, and the Russians simply used a pencil.

That story is an urban myth.
The pen in question is the "Fisher Space Pen" which used a pressurized ink cartridge so it could write in any position. The pen was not developed by NASA, it was designed and made by a private company and later sold to both NASA and The Soviet Union's space program. NASA paid the same price as everybody else, not millions of dollars.

Both the US and Soviets used pencils, but they were problematic in zero gravity because the tips can break off and float into electrical equipment. They mainly used grease pencils for this reason.

 

[Rick Burnett]

ok ok... I see what you are doing here.. You are speaking about the technical definitions which make the image appear in a very teoretical manner, and alltho Im sure you are right you never really approached the relevant question which this whole discussion is about..

Do we agree that:

1. you do NOT preset a particular Kelvin value while shooting for anything else than reference? This Kelvin value has no impact what so ever on the post work?

2. the same preset can be used in both tungsten and daylight scenes?

3. you could split an image into two and whitebalance it differently, making it LOOK like it was shot with two cameras preset to each its own white balance?

(1) This is not entirely true. Having the metadata for the desired look is a reference to what was seen during the shooting. Depending on preferences in the workflow with those involved, how the meta data is used is totally up to the project. It can be used OR ignored.

(2) Not if you want to get an approximation of what it will look like in post. The values are changeable so you can get the scene to look more appropriate to what you desire in post. The difference is, you are not locked into that value. The RED values in camera DO NOT change in RAW based on the white balance you chose.

(3) Yes. Which you could do with the Arri as well, it's just you will apply a second set of changes to the RAW data at this point. The results are unknown at this point because how far you can push the image in any direction other than the native built in to the sensor color temp depends on how noisy the floor is.

The two arguments going on here at this point are mathematical vs practical. The true answer will revealed when people can actually test both cameras and see.

 

[David Mullen ASC]

In Red's RAW recording mode, color temp values are merely metadata to be applied later. What you are recording is what the sensor is reacting to. But it naturally has a "bias" or preference for lighting with more blue wavelengths in it. So in effect, the RAW recording will convert to RGB "most cleanly", let's say, for a daylight-balanced scene.

If you split-screened the same shot color-corrected in different ways, what you'd see is the noise levels in each color channel are different -- mainly that the blue channel was getting noisier when you pushed the image bluer, like when correcting for a shot lit with tungsten light. But how visible that extra noise in the blue channel was just depends on a lot of other factors, like how well-exposed the sensor was overall, how large the image was going to be seen, etc.

Looking online, I see a lot of confusion stemming from photography books suggesting that in RAW mode, a camera is color-temp agnostic and you can pick any color temp you want later in post. This is not strictly true, it's just that with a low-noise image, your RAW recording, starting out with everything the sensor responded to, is going to be able to be converted to different color temps without much problem, though the lower ones like 3200K are going to have more noise in the blue channel.

The earlier Red camera, sensor, build, and color science, all had some noise problems with converting to 3200K in post due to the noise floor of the sensor. It worked OK at 320 ASA and lower, but not much higher or more underexposed than that before the blue noise was getting too noticeable. Hence all the talk about putting blue filters on the camera in tungsten light to get the image closer to the daylight balance that the sensor was naturally happier at.

Later builds and later color science improvements reduced the visibility of the blue noise at 3200K and the new M-X sensor, new color science, etc. has all but eliminated the problem, though the image is still the least noisy in daylight balance.

But it wasn't until the first Reds that people started learning that digital sensors were less noisy in some color temps than others. Or I'd even say it goes back to the Viper, when people used to video RGB white-balanced cameras wondered why Viper's LOG recordings were so greenish. You started to learn about the "native" sensitivity and color bias of sensors.

JPEG in still cameras is not a lossless recording format so it really limits your color-correcting options.

 

[Nik Skjøth]

Ah well..I was never talking about what the Alexa can or cannot do.. If it has the same options then hurray.. So that would make it the second motion camera in the world to achieve this? Others that shoot raw, which are in the stores now?

I simply replied to the question of Davids, asking why the heck you would care about if you can change the white balance after the fact or not. And my reply was to that question alone and nothing else.

 

[Rob Ruffo]

Rob, I am curious why you make up 7/8ths of every comment on this thread lately? You often sound like you pull stuff out of your ass only
to take 4 or 5 more posts to then explain "what you meant" and its nothing at all like what you first wrote.

Dont take it personally, but can you please ignore the impulse to respond to every single post... several times over?

I cant speak for others, but I thought it was tolerable if you WERE a teenager... well apparently you arent a teen? OK/.

with all dude respect. can you at least let other people have a say and quit the ARRI trashing and misquoting, it makes us all look bad.

@ Graeme, Michael said they are still tweaking Alexaraw and that info isnt public yet. Hence my persistence stance lately to please please close this thread for a month or two...

@ Jim - please keep those sexy specs coming!

I did zero Arri trashing. I asked questions, hoping for answers. You twisted what I said into something ugly and I felt the need to clear my good name.

What I find makes people look bad is 40 posts of "cool awesome can't wait" on any subject. To me this is mindless. Debate is interesting. Debate helps you learn by seeing both sides. But I think you and I come from very different cultural backgrounds, and I can see this from your wording choices, not just the above post. (As I'm sure you will jump on me for this, I did not mean your culture is bad, just incompatibly different from mine.)

Ah well, time to get back to my real life I suppose, and my face-to-face friends who actually love my tendency to play devil's advocate. You've succeeded in chasing me away from Reduser. Good for you. I was procrastinating too much on these forums, even though I learned a lot. I wish you all luck with your shoots, regardless of which camera you chose to do them on.

 

[David Mullen ASC]

About the general "save it for post" versus "get it right in-camera" arguments, I know that as a practical person, you have to know which approach is better on a shot by shot basis, and obviously mistakes happen which is why you want flexibility in post to make corrections. But that's a different thing entirely than always shooting so that artistic choices can be delayed to post. That also sometimes happens (I had a set where the director couldn't decide what painting to hang on a wall, so they hung a greenscreen in frame and put the art in later in post -- but that was an expensive bit of indecision and it rarely quite looks as real as the real thing.)

I just finished color-correcting that TV pilot I shot on the Red One and we had a number of flashbacks where we decided to save the "look" for post because we didn't have time for testing in prep. So we must have wasted a couple of hours in an expensive D.I. suite trying out various degrees of processing to those images. Ultimately we mainly opted for a semi ProMist look with a ND vignette. But it also struck me that I had considered just doing that in-camera and if I had, we would have shaved a couple hours off of our D.I. costs. And it probably would have looked better. If this show goes to series, I'm going to suggest shooting the flashbacks with some filters and at least talking part of the work spent manipulating them away from the D.I. stage (we would still do the vignetting and whatnot.) In fact, I've had a couple of projects where we saved diffusion for post, only to have me wish in the end I had just done it in-camera due to the time it took to apply it to every shot and the endless monkeying around just to recreate a natural filter effect.

When the arguments here are about practical concerns, problem-solving, etc. I can relate, I just start to get a little annoyed when it drifts into the concept of saving artistic choices for post rather than making bold choices while shooting. I only believe in that when I feel it benefits the project in some way, I don't like it when it is merely a delaying tactic because no one wants to make a decision until later.

For example, monitor burn-ins. There have been so many times when I wished I could just burn-in a monitor image in post rather than deal with playback, but it was out of the budget. But today, it's so commonplace that now I have the reverse problem, no one wants to deal with on-set playback and they want to burn in the TV image in post. But that really limits you creatively when you want to get fancy with blurred people moving in front of the TV (and their out of focus hair) or racking back & forth, or having reflections of the person watching, etc. all while the glow of the TV set is interacting with reflective furniture, etc. All stuff that can be dealt with in post but rarely is due to post time & budget, you always get told in the last minute to not let out of focus hair cross the greenscreened TV, not shoot the screen out of focus, etc. There are times when you can creatively design a shot around a working TV set in the frame and putting in the image in post won't be as interesting. Other times, it's simpler to save it for post.

 

[Nik Skjøth]

I just finished color-correcting that TV pilot I shot on the Red One and we had a number of flashbacks where we decided to save the "look" for post because we didn't have time for testing in prep. So we must have wasted a couple of hours in an expensive D.I. suite trying out various degrees of processing to those images. Ultimately we mainly opted for a semi ProMist look with a ND vignette. But it also struck me that I had considered just doing that in-camera and if I had, we would have shaved a couple hours off of our D.I. costs. And it probably would have looked better.

Now this is where Im confused.. I thought that you COULD NOT do that in camera on the red One.. Can you shoot not raw or what? Did you cosider recording to a external device prebaking the image or howt? Yeah I know you can save every preset as metadata, but how does it make it easier to make that decision on set rather than i post.. You have the same options to tweak each setting, and I would argue that with the footage infront of you in the studio with the right edit hardware, would make it much much easier to set the image right, than fiddling with the menus on the camera on the set.. So arguing that on set preperations is a timesaviour puzzles me alot.

Im one of those people who hate to make image look descisions on the set, I hate the characteristics of say a camera like Sony Ex1, where the image you shoot is basically what you get... Maybe its because Im both the DOP and the editor, and I always wish that I had more latitude to change the look of the footage. Its rarelly about whitebalacing problems, and much much more about not having enough DR for example. eitherway.. I would choose to have as flat raw footage as possible so that I can decide what I want it to look like, in the studio, with the proper monitor.

 

[Rick Burnett]

Now this is where Im confused.. I thought that you COULD NOT do that in camera on the red One.. Can you shoot not raw or what?

The settings can be set in camera, but they are not applied to the data. In post, if you want to use the settings that were shot, you can use those as you convert from R3D to your final deliverable format. You just don't touch them.

 

[Rob Ruffo]

About the general "save it for post" versus "get it right in-camera" arguments, I know that as a practical person, you have to know which approach is better on a shot by shot basis, and obviously mistakes happen which is why you want flexibility in post to make corrections. But that's a different thing entirely than always shooting so that artistic choices can be delayed to post. That also sometimes happens (I had a set where the director couldn't decide what painting to hang on a wall, so they hung a greenscreen in frame and put the art in later in post -- but that was an expensive bit of indecision and it rarely quite looks as real as the real thing.)

I just finished color-correcting that TV pilot I shot on the Red One and we had a number of flashbacks where we decided to save the "look" for post because we didn't have time for testing in prep. So we must have wasted a couple of hours in an expensive D.I. suite trying out various degrees of processing to those images. Ultimately we mainly opted for a semi ProMist look with a ND vignette. But it also struck me that I had considered just doing that in-camera and if I had, we would have shaved a couple hours off of our D.I. costs. And it probably would have looked better. If this show goes to series, I'm going to suggest shooting the flashbacks with some filters and at least talking part of the work spent manipulating them away from the D.I. stage (we would still do the vignetting and whatnot.) In fact, I've had a couple of projects where we saved diffusion for post, only to have me wish in the end I had just done it in-camera due to the time it took to apply it to every shot and the endless monkeying around just to recreate a natural filter effect.

When the arguments here are about practical concerns, problem-solving, etc. I can relate, I just start to get a little annoyed when it drifts into the concept of saving artistic choices for post rather than making bold choices while shooting. I only believe in that when I feel it benefits the project in some way, I don't like it when it is merely a delaying tactic because no one wants to make a decision until later.

For example, monitor burn-ins. There have been so many times when I wished I could just burn-in a monitor image in post rather than deal with playback, but it was out of the budget. But today, it's so commonplace that now I have the reverse problem, no one wants to deal with on-set playback and they want to burn in the TV image in post. But that really limits you creatively when you want to get fancy with blurred people moving in front of the TV (and their out of focus hair) or racking back & forth, or having reflections of the person watching, etc. all while the glow of the TV set is interacting with reflective furniture, etc. All stuff that can be dealt with in post but rarely is due to post time & budget, you always get told in the last minute to not let out of focus hair cross the greenscreened TV, not shoot the screen out of focus, etc. There are times when you can creatively design a shot around a working TV set in the frame and putting in the image in post won't be as interesting. Other times, it's simpler to save it for post.

As my very last post to this forum (other than to the reviews section - I like writing reviews) I would like to thank David for his extreme generosity here. There are few people I have learned as much from about cinematography as him. We should all be very thankful for his time given free of charge.

And- I agree with you. "Leave it for post" can be a wimpy way of deferring balsey decisions to a later time, or to others. I also agree that filtering (diffusion, not so much color) never looks the same as a post effect, unless you are willing to SPEND, and even then, that you can get far more interesting effects in-camera. People seem to understand this for streak filters, but not for such wonderful tools as Classic Soft, Glimmerglass and others. In many ways a real set is (usually) the cheapest thing to fix and build, always(mostly) looks better, and no matter how fast compus get, a good artist still has to intervene and I do hope these will never get cheap as that would not be fair to them.

Anyway, thanks David. I mean that.

 

[Nik Skjøth]

The settings can be set in camera, but they are not applied to the data. In post, if you want to use the settings that were shot, you can use those as you convert from R3D to your final deliverable format. You just don't touch them.

Hey. Do you even read the whole post before replying? I know about the metadata, and the question remains. why does David need a D.I. to tweak THE SAME options in post, when he can do it himself on the camera.. DOES NOT COMPUTE.

 

[Jannard]

One point that needs to be made here... with RAW, you don't have to "leave it for post". You can set your WB (metadata) and never touch it through the chain.

BUT. The old days of having two kinds of light are over. And color science is based on proper white balance. For example... LEDs (now a legitimate option for lighting) has a heavy green tint. A daylight or tungsten setting won't cut it. Same issue with fluorescent lights... and mixed lighting. You need to have a way to set WB exactly for these and other "odd" color temps. We do it with WB options in post. There are other ways to do it I'm sure. We feel this is the best way.

Jim

 

[Nik Skjøth]

Thanks Jim.. Just the answer I was looking for... Straight and simple... and confirmed what I said all along. Im not crazy afterall...

 

[Rick Burnett]

"Leave it for post" can be a wimpy way of deferring balsey decisions to a later time, or to others.

EVERY person I have worked with video wise has always had the practice of trying to get it right in camera. I see people talk about "leave it for post" some here, but in my practical experience, everyone has tried to get it right in camera. I do this as well.

HOWEVER!

Mistakes happen. Having a data format that retains as much original data as possible and tries to eliminate as much destructive decisions as possible is there as a safety measure. Quantifying that risk is the hard part. Experience with cameras shows people where those limitations are in the formats they work in.

Not pertaining to you, but in general, RAW doesn't mean ANYTHING if the sensor is crap. RAW data out of my iPhone camera is pointless the quality is so bad. Until ANYONE here can sit down with both the Epic and the Alexa, EVERY response here is speculative pertaining to what you should expect.

It makes me laugh because of the time I bought a D90 camera when it first came out. I saw ALL these amazing videos on vimeo showing how amazing the video mode was.....till I got it. Once in my hands, I was absolutely shocked at how poorly the video was implemented. I have NEVER disliked a video mode in a camera more than the D90 in MY opinion. The point of this being, here I had made all these decisions about a product from samples other people posted, yet, they DID NOT even come close to telling the whole story. I didn't know what line skipping was then either, but boy do I know what that is now!

While I am VERY curious about how both Arri and Red implemented different aspects of the electronics, and also have my own opinions about different methods used, as I am an engineer, I absolutely agree with a lot of what David has posted in terms of focusing on the real questions.

Since NONE of these cameras are shipping yet, I think we all have too much time on our hands waiting to get our hands on ANY of these.