Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Alexa sensor and the Dragon sensor

And the REDUSER world thanks you! I know I'll do my part in spreading this proper information around. :)
 
As I mentioned, it's actually a very important part of HDR that you understand how this works. Although HDR is offering a wider RGB gamut as part of its appeal, we actually get to see more bright saturated colours because of the max brightness improvement than through moving the primaries further out. Moving primaries further out does help a bit (even on an SDR display) but the real benefit comes when you remove the traditional maximum brightness restrictions.

Graeme
 
Can we talk about the differences in what it takes to OUTRESOLVE each sensor?

Christopher Probst has often said Dragon sensor gets outresolved at 100lp/mm. See below:

Christopher Probst said:
Dragon 6K tops out at 100 lp/mm resloving power and based on the photosite ("pixel") pitch of that sensor... And prehaps surprisign to some of you out there, the same holds true for Weapon 8K. That resolving power is just spread out over a larger imaging area.

I'll explain... The easiest and most basic way to think about it is: You need 2 photosites to capture a line-pair off a chart. One photosite to get a black part of a test pattern, one photosite to capture the white next to it. So 6K is 6144 photosites across (not taking into account Bayer pattern or anthing on that level). So to capture a line pair you need 2 photosites, that means 6144/2 = 3072 pairs. Again, resolving power is measured in line-pairs per milimeter... And at 6K, the Dragon sensor is 30.7mm across... So 3072 line-pairs across a 30.7mm sensors is 3072/30.7 = 100.065 lp/mm...

Do that math with 8K and it's exactly the same resolving power. 8192/2 = 4096. 4096/40.96mm = 100 lp/mm. So until the photosite density is increased with a new sensor, Red Dragon sensors are capable of resolving 100 line pairs any day of the week. Now having a lens like a Master Prime or a Summilux that out resolves the sensor is a good thing when you take Nyquist sampling therom into account...
Click to expand...

By his math, does that means Alexa Open Gate gets outresolved at 60.6lp/mm?

3414/2 = 1707 pairs

1707/28.17mm across = 60.59637912673056

Is this correct?

And if so, what are the on set implications of this? Would this encourage a DP to use sharper or less sharp glass on Alexa vs RED? Or is this 40lp/mm difference btw sensors negligible?
 
Nick, my example is about as simplified and generalized as it possibly can be. It's not meant to be referenced as dogma, but as way to imagine a person by drawing a stick figure... Be prepared for some mind numbing response from Graeme now explaining why this example misses the point!
 
Last edited:
Hahahaha thanks for the clarification! I'll wait patiently for Graeme's "boring" response :)
 
There's another somewhat recent thread where Graeme discusses a sorta naive approach to signal reproduction, check his post history. It's much more in-depth than you'd probably even be looking for.

Chris Probst's is simply based on the fact that at 200 photo sites per mm, you could at most resolve 100 line PAIRS. That's not taking into account the debayer, the OLPF, nyquist frequency limit, or any other effects that limit resolution, and it doesn't say anything about MTF which is important to defining what it means to be 'resolved'.
 
With lens resolution ideally you'd have maximum contrast up to just below whatever the sensor can handle, and nothing after that, thus getting max resolution on the resulting image without aliasing. Of course, lenses don't work like that and nothing optical has a sharp cut-off like that (something to do with no such thing as negative photons - "darkons").

If your lens our resolves your sensor, you can get aliasing. If the sensor out resolve the lens you'll not get sharp images. Neither is a good position to be in. That's why the OLPF's low-pass function puts an upper limit on resolution and thus you can use sharper and sharper glass, get great MTF for the bulk of the image detail, but still avoid the worst of aliasing.

Graeme
 
Christopher Probst said:
Nick, my example is about as simplified and generalized as it possibly can be. It's not meant to be referenced as dogma, but as way to imagine a person by drawing a stick figure... Be prepared for some mind numbing response from Graeme now explaining why this example misses the point!
Click to expand...

No worries at all! I never thought your approach was gospel, but still seems like a great starting point to understand Dragon. When that "math" is applied to Alexa (and how it's sensor resolves), I'd love to know what the difference is practically on set btw the two (100lp/mm vs 60.6lp/mm), if any...

Would that mean that sharper glass is preferable on Alexa vs RED? Because of Alexa's smaller resolving power? Or is softer glass preferable on Alexa, because Alexa needs less resolution to resolve pixels?

Or are the differences negligible?

Just curious if this fundamental difference - at a pixel level - has any impact on choices made on set, in particular optically.

With DR btw both cameras getting closer and closer, curious if this philisophical difference btw the platforms is significant or not!
 
Last edited:
I doubt it makes that much odds. Most good glass is going to out-reolve both cameras, and thus the low-pass element of the OLPF forms the "envelope" for the system. In measured tests on good Leica glass, we got a much higher MTF50 (a good proxy measure for perceived sharpness) on the Epic MX than Alexa.

Graeme
 
Going back to the end-on-end cube example, how does that pertain to the Alexa/ALEV? Like I understand why brighter colours lose saturation in RGB, but that doesn't explain how Alexa still kills it in the colour/saturation of its highlights. Also, are you suggesting that on HDR screens Alexa footage could end up looking weird/unpredictable (overly dark)?

I have noticed that recovered portions of an Alexa/ALEV image hold their saturation/colour a lot more "accurately" (though, "pleasingly" might be a better word) than even Dragon. Skin tones that are ~5-6 stops over can still be pulled down and actually look okay, which usually isn't the case with other digital cinema cameras. I know it's kind of moot (if you botched exposure that bad you're gonna be boned no matter what), but it certainly helps with recovering things like clouds/skies (parts of the image that naturally ride the high-line) and making them look colourful/detailed.
 
Last edited:
I think something that should be mentioned is that the Low Pass Filter is something we need with current technology due to the nature of a pixel's shape. I'm seeing some exciting and undesirable things on a few of the smaller than 5 micron pixel sized sensors from a few other companies. Aliasing, false colors, moire, etc... Which occasionally leads to other weird things.

Having a high resolving lens on a system with a Low Pass "shield" is still a good thing as it relates to how the image is drawn and in effect how much you can get out of it. That barrier is the only thing protecting us from a lot of post work.

There's different ways of "developing" the footage to get more out of it. Especially if working with the compressed raw format.
 
What I'm saying is that I see zero evidence Alexa image processing is doing anything special at all with saturation in highlights and what is seen is a natural consequence of the RGB cube.

Being able to pull back bright exposure is purely a consequence of where the DR of the sensor is placed relative to that exposure, and that is because we're talking about essentially linear systems. Of course, once colours clip, response ceases to be linear....

Graeme
 
Since DSMC2 bodies now support 3d luts, I wonder of it'd be possible to create a lut that will mimick the Alexa out of camera look and shoot ProRes using it. I guess KipperTie pearl OLPF should help with this too.
 
I'm guessing it's mostly possible if you wanted to do it, also I'd go with KipperTie Carbon 1 instead of pearl.

The big thing would be in mixed lighting, depending on where the peak sensitivities of the color channels are etc I think you'd still find that a 3D lut couldn't 100% transform one into another, unless the dragon was all around more precise in mixed lighting. I think.
 
Graeme Nattress said:
What I'm saying is that I see zero evidence Alexa image processing is doing anything special at all with saturation in highlights and what is seen is a natural consequence of the RGB cube.

Being able to pull back bright exposure is purely a consequence of where the DR of the sensor is placed relative to that exposure, and that is because we're talking about essentially linear systems. Of course, once colours clip, response ceases to be linear....

Graeme
Click to expand...

There's been talk in another thread about the saturation being radically different on the Dragon, but I've tried to see those differences and I'm not sure I do see them. But I'm experimenting with the LUM vs SAT settings in Resolve and I'm wondering what the "said differences" are? Is it that the saturation is dropping in highlights on the Alexa? Or the mid-tones? I mean if someone like that look, wouldn't the LUM vs SAT settings take care of that and match up good to the Alexa if that is the intended and desired look or starting point? Or, based on the pre-gamma/post-gamma sat, would it be better to dial back the RAW saturation and push up the mid-tones sat in LUM vs SAT as a post-gamma saturation change?

I think the only way to really and scientifically see the differences between these two sensors would be to do like a 3D cam setup so that the cameras shoot a subject at the same time with the exact same point of view, setting the Dragon to 5K in order to get s35mm FOV, using the same and really good lenses like the master primes set to T5.6 and then shoot something with them both, scale them down and up to 4K and have a wipe-slider while playing back. In that way we could really look at different parts of the image and see the actual differences. I mean, if there aren't much difference, it will probably be a good thing to show off that there's not much of a difference. And if someone do want the Dragon to look like Alexa, then that footage would be perfect to match up or create a LUT of some kind to adress saturation differences or whatever. In any case it would make it easier for productions that shoot both.
 
Christoffer Glans said:
There's been talk in another thread about the saturation being radically different on the Dragon, but I've tried to see those differences and I'm not sure I do see them. But I'm experimenting with the LUM vs SAT settings in Resolve and I'm wondering what the "said differences" are? Is it that the saturation is dropping in highlights on the Alexa? Or the mid-tones? I mean if someone like that look, wouldn't the LUM vs SAT settings take care of that and match up good to the Alexa if that is the intended and desired look or starting point? Or, based on the pre-gamma/post-gamma sat, would it be better to dial back the RAW saturation and push up the mid-tones sat in LUM vs SAT as a post-gamma saturation change?

I think the only way to really and scientifically see the differences between these two sensors would be to do like a 3D cam setup so that the cameras shoot a subject at the same time with the exact same point of view, setting the Dragon to 5K in order to get s35mm FOV, using the same and really good lenses like the master primes set to T5.6 and then shoot something with them both, scale them down and up to 4K and have a wipe-slider while playing back. In that way we could really look at different parts of the image and see the actual differences. I mean, if there aren't much difference, it will probably be a good thing to show off that there's not much of a difference. And if someone do want the Dragon to look like Alexa, then that footage would be perfect to match up or create a LUT of some kind to adress saturation differences or whatever. In any case it would make it easier for productions that shoot both.
Click to expand...

I tried that but it's not that easy. It has different hues, different saturations, different color lum. The reds on the alexa for example are darker with a different hue. The yellows are less saturated on the Alexa. Every color is different. Saturation levels too. Curves too. You can try to adjust them individually, modify the curve in the middle tones to fit the Alexa etc. But after a while you just degrade your image with weird stuffs especially when modifying individual colors and saturation.
 
Ugo Del said:
I tried that but it's not that easy. It has different hues, different saturations, different color lum. The reds on the alexa for example are darker with a different hue. The yellows are less saturated on the Alexa. Every color is different. Saturation levels too. Curves too. You can try to adjust them individually, modify the curve in the middle tones to fit the Alexa etc. But after a while you just degrade your image with weird stuffs especially when modifying individual colors and saturation.
Click to expand...

If so, then maybe not exactly the same, but doing small smoothed off adjustments with all those tools would maybe come closer, at least as a starting point node? I mean, I don't think the Dragon colors are bad, on the opposite, so maybe respecting those colors and dialing them half way to Alexa style colors might be a nice mid way?
 
Steve Sherrick said:
Two films I worked on recently intercut Alexa and Dragon a lot. Dragon often on the Movi. One of those movies is competing at Sundance and I'll be going to check it out next week. I guess seeing it on the big screen will tell me how well they match. Are there differences, sure. The question becomes how well can they be matched to eachother so that the general audience can't see any difference. If anyone will be at Sundance next week, the film is called The Birth of a Nation.
Click to expand...
great movie by the way. beautifully shot. and a great use of both cameras.
 
Demetri Zuev said:
Since DSMC2 bodies now support 3d luts, I wonder of it'd be possible to create a lut that will mimick the Alexa out of camera look and shoot ProRes using it. I guess KipperTie pearl OLPF should help with this too.
Click to expand...

An Ultracon 2 filter does the trick - it's the dirty little secret of the Alexa highlight rolloff
 
You must log in or register to reply here.
Back
Top