Red's new Mysterium-X sensor at ISO 2000

That's another thing to consider for 3D. Most people aren't aware that there is about .5 to a stop and half light loss when using a beam splitter rig for shooting stereoscopic cameras. Having used the low light capabilities of my 5D MKII a lot, I will say that having extreme light sensitivity in a digital cinema camera is a handy thing to have.

Dan
 


One of the cams came back on Friday, took it out on Saturday with the T2.1 Standard sPeeds and a Swing and Shfit lens.

Matty_G and I operating.

The experience is hard to put to words, honestly. I've tried explaining how it was to shoot with this beast to others but it took showing the raw footage for the vicious cursing to begin.
 
Jason Ramsey said:
I saw that over on REDuser Kholi. Looking good.

by the way, the footage from the Di'Caprio frame grabs has been posted for anyone who hasn't seen it yet.

700MB
http://red.cachefly.net/LEO.mov

Original thread here.
http://reduser.net/forum/showthread.php?t=40947
Click to expand...

Thanks, Jason. Completely new belief in RED, over here. We couldn't shut up about it ALL night. LOL, actually, Matty was cutting this footage while a Superbowl After party was going on, and Geoff_Reisner and I were popping back and forth geeking out.

Of course, the Di'Caprio footage is fifty times better! That match flare is quite insane.
 
glad to hear people outside of red have been shooting with the new mysterium x, its been hard to believe all the hype when the product hasnt even existed
 
The Hype Machine has been bowling over whole cities. People that I know whom rarely ever bother with technology were approaching me and asking about it.

However, this one lives up to the hype all the way. I reserved any comments until I had the footage in post and we all looked and stared at pixels for hours.

Real deal, man.
 
Is anyone actually impressed with that demo reel? The shots in daylight look great, but the shoots in the garage/workshop look very flat and washed out with no pop at all. I'm starting to suspect that the 5D still has less DR drop-off at higher iso than even the new MX sensor.

Has anyone seen this clip? http://vimeo.com/7058755

I know the Mysterium-x is sharp as hell, shoots RAW, and is a better overall "image". But I've yet to see anything come out of the Red camera with color depth or pop like that Egypt clip... except maybe in "Knowing". But defiantly nothing on the web.
 
Last edited:
bwhitz said:
I'm starting to suspect that the 5D still has less DR drop-off at higher iso than even the new MX sensor.
Click to expand...

Don't know about MX, but 5d II is indeed impressive (actual measurements in STILL mode):

iso200 11.2 stops
iso400 11.1 stops
iso800 10.9 stops
iso1600 10.5 stops
iso3200 9.5 stops
iso6400 8.4 stops
iso12800 7.3 stops

Its ISO800/1600 is as good as ISO200 in a good large pixel sensor, and after that it deteriorates one stop per doubling of iso as expected. If the large sensor didn't force you to stop down 2 to 3 stops and need that higher ISO for getting some practical DOF for video, it would have a great advantage.
 
...But defiantly nothing on the web...
Click to expand...

If final delivery is only going to be for 720p web, or whatnot, then a 7D is almost all the camera one would need...

...(actual measurements in STILL mode)...
Click to expand...

I'd be more interested in ISO comparisons in Video mode. I'd also be interested in comparisons of Still vs. Still mode, when there is a DSMC that has that capability, from the photographic side of things.

Just saying, stills vs. stills and motion vs. motion would be the more relevant comparisons, as one can't shoot anything approaching 24fps (or 100+) with those 18MP on a 7D.
 
bwhitz said:
Is anyone actually impressed with that demo reel? The shots in daylight look great, but the shoots in the garage/workshop look very flat and washed out with no pop at all. I'm starting to suspect that the 5D still has less DR drop-off at higher iso than even the new MX sensor.

Has anyone seen this clip? http://vimeo.com/7058755

I know the Mysterium-x is sharp as hell, shoots RAW, and is a better overall "image". But I've yet to see anything come out of the Red camera with color depth or pop like that Egypt clip... except maybe in "Knowing". But defiantly nothing on the web.
Click to expand...

Quite an odd statement... you're comparing a clip in Daylight yet you're concerned with the RED in the garage? Odd. It surely wouldn't be hard to oversaturate an image, though. I don't like the color in the Egypt footage, especially the skin. It looks hokey.

Trust me... the MKii has little on a RED M, and nothing on a RED M-X.
 
Kholi said:
Quite an odd statement... you're comparing a clip in Daylight yet you're concerned with the RED in the garage?
Click to expand...

How so? I was just pointing out that all the daylight shoots with the MX look great. But once the light went away... so did the dynamic range and color. The last shot (guy with the light bulb) looked pretty clean and noise free, but also significantly less DR 5-7 stops maybe vs the 12 or so the Volkswagen shot appeared to have. I'm sure it will still look great on big screen though.

Jason Ramsey said:
If final delivery is only going to be for 720p web, or whatnot, then a 7D is almost all the camera one would need...
Click to expand...

Exactly, 7D is a killer music video/"digital short" camera. Red seems more an more like a camera I would rent to rent if I knew for sure it was going to the big screen, but then again, if I knew for sure, it would probably be a studio deal with a budget for a Genesis or Arri.
 
Last edited:
Jason Ramsey said:
I'd be more interested in ISO comparisons in Video mode. I'd also be interested in comparisons of Still vs. Still mode, when there is a DSMC that has that capability, from the photographic side of things.

Just saying, stills vs. stills and motion vs. motion would be the more relevant comparisons, as one can't shoot anything approaching 24fps (or 100+) with those 18MP on a 7D.
Click to expand...

The video mode has the same dynamic range performance, it's just not improving it (no binning, just using a small part of the total pixels). Most 5d II shots do have low dynamic range because the shallow DOF of the full frame sensor forces higher ISO in the real world compared to Red One and smaller sensors. Most night shots with the 5DII use extremely high ISO settings so they end up equal or even inferior to a camcorder with smaller sensor/pixel size. Limitations aside, it's one of the best sensors I have ever seen. But it cannot compare to a professional camera due to aliasing and low quality interframe codec. It wasn't meant to. The 7d is more than one stop inferior. The format is also key to the dynamic range of a camera. This measurement was in uncompressed RAW still mode. Even a 12stop camera is not much use on a codec with a 5 or 6bit PSNR that is even worse in demanding scenes. It takes high data rate intraframe to capture the shadow detail.
 
Otis Grapsas said:
Most 5d II shots do have low dynamic range because the shallow DOF of the full frame sensor forces higher ISO in the real world compared to Red One and smaller sensors.
Click to expand...

Not to completely disagree with you, but Isn't the exact opposite happening? The full-frame sensor (being as large as it is) collects more light faster, which allows a lower real-world ISO, and therefore more dynamic range when compared to the higher ISO on a smaller chip camera (Red One) that you would need to achieve the same exposure. Even though it looks clean resolution wise, the new MX sensor still looks less efficient at maintaining DR at high ISO than the 5D sensor.
 
Ian-T said:
Or...you can be Rodney Charters who has half a dozen 5Dll's on set of "24'....a show that also has a budget.
Click to expand...
Yes, but WHAT DOES HE USE THEM FOR?

He uses them for the little security-camera insert shots on computer monitors. He uses them for background plates that get rear-projected behind their process cars. Same kinds of things he used to use the JVC HD100s for. Same kinds of things he used to use a SONY PD150 for!

Do you think Rodney Charters is shooting a multi-million-dollar television show on a $2700 DSLR? No. Or, more appropriately, HELL NO.

He uses them for appropriate things. Things where a $2700 still camera is appropriate. Maybe a background plate here or there. Or a security camera. But the show itself is shot on 3-perf Super35 film, not on a 5D.
http://digitalcontentproducer.com/di/depth/video_workflow/
Charters is a technology freak, and professionally, he has investigated just about every kind of high-definition and video acquisition technology known to man. And yet, he steadfastly insists that 24's image acquisition belongs in the traditional 35mm film universe
Click to expand...

Charters estimates his camera crew shoots about 4 million ft. of film each season
Click to expand...

Here's him showing how he uses the 5D, to make background plates for driving sequences. As in, it's the stuff they show outside the windows of the car. Not exactly A-camera stuff.
http://blog.planet5d.com/2010/02/how-to-make-plates-for-foxs-24-on-the-canon-eos-5d-mark-ii/

Here's an interview from him from just about four months ago, talking about what he shoots and how. He shoots film. Even though there's tremendous pressure on him to convert to video, he shoots film.

So how does he use the 5D? He uses it for the same job that they used to do with PD150's. When they started the show he used the standard-def, interlaced Sony PD150 to shoot background plates. A few years later he changed to a 24p HDV camera, the JVC HD100, and then recently he's been using the 5D.

So let's have an end of the "Rodney Charters Uses The 5D on 24" stuff, unless you're willing to put it in context -- which is, "Rodney doesn't use a 5D to shoot anything that gets seen on screen as an actual full-frame image, but he will use it to do the same job he used to consider a PD150 good enough for."
 
Thanks for that Barry. Seriously, 35mm is still king. A local shooter that uses the Red told me (this is his take on it so I don't know how true it rings) that 35mm film is about 4K resolution but when it comes to color space in post production it is about 12K.
 
Tracey Lee said:
Thanks for that Barry. Seriously, 35mm is still king. A local shooter that uses the Red told me (this is his take on it so I don't know how true it rings) that 35mm film is about 4K resolution but when it comes to color space in post production it is about 12K.
Click to expand...
According to Kodak, it takes a 4K scan to capture all the detail that's present in a frame of film, so they're saying each 4:3 frame of film has about 12 million pixels in it.

As for color space, film is inherently full-color, so it'd need a 4K 4:4:4 scan to properly capture it. I don't know how this other shooter was coming at his numbers, but ... well, I guess if you wanted to compare film's "color sampling" to the Red One's 4K frame, yes it's about 4x as much (being inherently 4:4:4, versus more like 4:2:0 for the Red) and the Red's "real" resolution is just over 3K, so ... I guess the math could be juggled such that you'd be saying that 4x as much color x 3k means that film's color sampling is equivalent to a Red One frame if the Red One was shooting 12K? I don't know, I'm just trying to find a way that his statement would make sense?

And it's not about the color sampling, really, that Charters uses film. It's more about the dynamic range, the extreme latitude that film offers, plus he likes the aesthetic of the grain, thinking that it adds more of a gritty reality and texture to the show. It could be argued that a Viper or some other high-end digital imaging system would actually capture technically "better" imagery than film in those cases, but he prefers the aesthetic. He likes the grain and the grit.
 
bwhitz said:
Not to completely disagree with you, but Isn't the exact opposite happening? The full-frame sensor (being as large as it is) collects more light faster, which allows a lower real-world ISO, and therefore more dynamic range when compared to the higher ISO on a smaller chip camera (Red One) that you would need to achieve the same exposure. Even though it looks clean resolution wise, the new MX sensor still looks less efficient at maintaining DR at high ISO than the 5D sensor.
Click to expand...

If we assume similar technology, the light gathering and as a result the dynamic range, depend on pixel size. The pixel size for professional cameras is typically the same, roughly 5um. 1/3" cameras are 960x540 or 960x1080i (same effective pixel size), 2/3" is 1920x1080 and super35 cine is >4k. So they can all have roughly equal performance in dynamic range and sensitivity. And they actually do. Most good cameras are a basic ISO320 for best noise performance.

If all cameras had the same number of pixels (1920x1080 perhaps), then the 2/3" sensor would be 2 stops better than the 1/3" one and super35mm would be 4 stops better. But the extra area is typically used for more pixels, not for larger pixels. So this doesn't happen.

In low light situations, you need to gather as much light as possible but also get a DOF that allows practical cinematography. So, if a shot is f2.8 on Red One it will need to be f4 on the full frame to match the DOF. But f4 only gathers half light compared to f2.8. So the full frame sensor would better have good dynamic range and noise performance at an ISO of 2x compared to Red One. If not, the Red One will be 1 stop cleaner.

So, if the shots above are ISO2000 on Red One using f2.8 lenses, to match the DOF (which is not much), you will need ISO4000 using f4.0 on the full frame camera. You could use f1.4 for everything but is f1.4 on a full frame camera deep enough for focusing without major headaches? 50mm f1.4 can be used in still photography, but in full frame video it can be very limiting, even frustrating.

Cameras with lots of pixels that oversample the 1080p format, have an advantage over cameras of equal pixel quality. The output is scaled down for output. 4K to 1080p reduces noise a little bit and it also reduces grain size theoretically. Red one can take advantage of this downsampling. But The dSLRs do not do that, most pixels remain unused in video modes, and only active pixels gather light (1/4 or 1/6 of the total pixels). So, there is nothing to downsample. It's already the final output size. So, it's as good as a much smaller sensor (the active area is about 2/3") and the DOF can only work against you.
 
bwhitz said:
How so? I was just pointing out that all the daylight shoots with the MX look great. But once the light went away... so did the dynamic range and color. The last shot (guy with the light bulb) looked pretty clean and noise free, but also significantly less DR 5-7 stops maybe vs the 12 or so the Volkswagen shot appeared to have. I'm sure it will still look great on big screen though.
Click to expand...


Please, simply download the 2000 ISO Leo DiCaprio clip... no DSLR can hang with this. I had one of my 7Ds when I shot the Spriit Hoods footage and even in the same daytime scenario the 7D couldn't hang, it was pointless to even have it out.

Sorry, even for web delivery the 7D is a vastly inferior image. Sure, it's cheap, that's great, but it has nothing on a real cinema tool.
 
You must log in or register to reply here.
Back
Top