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

Christoffer Glans

Christoffer Glans

Well-known member
Joined
Jun 11, 2007
Messages
4,993
Reaction score
36
Points
48
Age
42
Location
Stockholm, Sweden
Wait wait wait! This is not another "which one is best?", "...but it has less resolution", "but the colors are better on..." -thread. I was watching the wonderfully shot Penny Dreadful and it felt like it was shot on Dragon, which is sometimes easy to spot and sometimes not, the same for Alexa. So I was surprised that it was shot on Alexa instead and started thinking about the ACTUAL differences between these sensors. Not in terms of math, science, debayering, resolution and all of those things, but the actual look and feel, the perception we have of the images that comes out of them both.

So, what differences are there really between them? Let's say that the delivery format for a TV-series or a film is in 2K so that differences in 3.2K, 4K, 5K, 6K is out of the question.

I have extensive experience with MX and Dragon sensor and I've worked a few times with Alexa, but I don't have hundreds of hours with it. I do however love both sensors. But if we have a set with enough lights to do creative lighting to the best of our ability, how would the thought process be between the cameras and sensors if there's a directors vision already set before shooting? Cause the process seems to be exactly the same with either camera so what is it that makes any difference between these sensors? How can we spot the difference between them in the final result? What exactly, to our creative eyes (not scientific) is the difference between these two sensors?
 
And before you answer that question, or maybe after you answer that question, consider this: LLO vs. STH. I have shot DRAGON with both OLPFs, and they give a base image as different as Kodakchrome and Velvia. There is something really special about STH. And there is something really simple about LLO. I opted for STD with my WEAPON (the theory being more special than LLO, much more simple than STH). But I would argue that just the difference between LLO and STH can be as confounding a difference as DRAGON vs. ALEXA. And I would argue that the biggest difference (for me) between these two OLPFs is not only the final image, but the path to getting to the final image (more lighting required for STH, more care in dealing with noise, more beauty coming out the other end when you do it right).

BTW, I have two unopened STH OLPFs that I purchases for my DRAGON cameras that I won't be able to use for my WEAPON cameras after they are upgraded. If anybody with an LLO-based DRAGON wants to play around with having an alternative digital filmstock, drop me a PM.
 
Yeah, that's true, but in the end, if the OLPFs were used how they supposed to be used, we will still end up with a "Dragon image", because in the end, after all the processing, grading and noise reduction, STH was used for occurrences with a lot of light and then the other way around for the LLO so the result will be a perfect Dragon image. But let's say two scenes, one shot exterior daytime in bright light and the other at dusk or dark, first one with the STH and the second with the LLO and then another camera that's an Alexa. Or just go for the STD and compare, cause it's a balance between the two.
 
I'm not sure we should exclude the resolution. To get the (almost) same S35 image the Alexa uses 2880 pixels vs 5000 for the Red. Even downsized to 2K I guess we can see a difference with the bigger Alexa pixels. The image will be a bit softer, with a different grain structure.
There is a hue difference too.
 
Last edited:
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.
 
Ugo Del said:
I'm not sure we should exclude the resolution. To get the (almost) same S35 image the Alexa uses 2880 pixels vs 5000 for the Red. Even downsized to 2K I guess we can see a difference with the bigger Alexa pixels. The image will be a bit softer, with different grain structure.
There is a hue difference too.
Click to expand...

In other threads, Graeme Nattress has claimed that bayer-pattern sensors achieve 70.7% of monochromatic sensors at a reasonable (MTF50?) level. Which means that 2880 pixels should have a horizontal resolution of 2036. That's close enough (99.4%) to the 2048 of DCI 2K to really not make a difference. From an information-theory point of view, scaling down 2880 or 4880 or 8880 pixels to 2048 is not going to give measurably different resolution at 2048 pixels.

At 4K DCI, it's a totally different story. In that case you need 5793 pixels to get a clean 4096 DCI at a "good" MTF level. With up to 6144 pixels, DRAGON far exceeds (> 6%) that number. If 4K matters to you, and you can resolve 100 lp/mm edge-to-edge, DRAGON will wipe the floor with Alexa.
 
Last edited:
Michael Tiemann said:
In other threads, Graeme Nattress has claimed that bayer-pattern sensors achieve 70.7% of monochromatic sensors at a reasonable (MTF50?) level. Which means that 2880 pixels should have a horizontal resolution of 2036. That's close enough (99.4%) to the 2048 of DCI 2K to really not make a difference. From an information-theory point of view, scaling down 2880 or 4880 or 8880 pixels to 2048 is not going to give measurably different resolution at 2048 pixels.

At 4K DCI, it's a totally different story. In that case you need 5793 pixels to get a clean 4096 DCI at a "good" MTF level. With up to 6144 pixels, DRAGON far exceeds (> 6%) that number. If 4K matters to you, and you can resolve 100 lp/mm edge-to-edge, DRAGON will wipe the floor with Alexa.
Click to expand...

Problem with this is that it is just scientific. When the end result hits the eyes of the audience, it will have gone through many things from the master copy to the screen showing the result. Everything from different codecs, streaming, screens etc.
This is exactly what I meant with this thread, to not just look at it scientifically, because if you do, yes, the Dragon, on paper seems to be far superior than the Alexa. But there is something "there" on the Alexa sensor that just works great everytime which means that we can't just look at this through math and the science behind it.

If you shoot the same scene, with the same field of view (sensor crop) and with the exact same lenses, keeping the cameras very close to eachother so that the image is almost the same and then do the same kind of grading for them, or match them up as close as they could possibly be... what will the differences be? This is something I've never seen. The thing is that I've seen so many comparisons of charts and such, but the testings are always filled with things that makes the comparison hard. Like the cameras not shooting the exact same thing at the same time, that they are too far from eachother to make the comparison truly based on the same image, that the image compared is a very flat log image that is very hard to spot differences on without applying a gamma curve.

If we had such images to compare, it could almost be a slider over the image to really make it easy to spot differences. It's the lacking of these kinds of tests that makes it hard to see the difference.
 
Christoffer Glans said:
[...]

If you shoot the same scene, with the same field of view (sensor crop) and with the exact same lenses, keeping the cameras very close to eachother so that the image is almost the same and then do the same kind of grading for them, or match them up as close as they could possibly be... what will the differences be? This is something I've never seen. The thing is that I've seen so many comparisons of charts and such, but the testings are always filled with things that makes the comparison hard. Like the cameras not shooting the exact same thing at the same time, that they are too far from eachother to make the comparison truly based on the same image, that the image compared is a very flat log image that is very hard to spot differences on without applying a gamma curve.

If we had such images to compare, it could almost be a slider over the image to really make it easy to spot differences. It's the lacking of these kinds of tests that makes it hard to see the difference.
Click to expand...

Björn Benckert does some amazing stuff with motion control. I imagine somebody with his kinds of toys could set up a cool studio in which two cameras could trace identical paths as they scan across low-key, high-key, saturated, and desaturated scenes.
 
Based on curves the Alexa has better highlight roll off vs dragon STH but to me dragon is better at seeing the blacks. At least that is my take away and I'd say reality mimics the curves. Arri seems to love overblown slightly more, red seems to love crushed more. Very similar dynamic range but Arri is just a tad more optimized towards the ther end vs red. LLO changes this slightly but I'd say Arri has half a stop or less of dynamic range these days, generally speaking. Ps happy to rent my cameras out towards testing, or dedicate some of my time towards tests?
 
I think one of the elements is laziness. I see many videos with not much work on them. And I personally think Arri images out of the box are more flattering. Just applying redgamma on red footages gives very contrasty images. There are tons of contrasty Red videos on the net with poor colors or almost monochromatic colors. Just work on the middles, play with alchemy, never use the main saturation in Redcine (ugly results), use the post saturation instead. Play with hue in resolve to make blue more cyan etc. And I swear you can get images that look like alexa.

I was impressed by some colors here by the way: https://www.youtube.com/watch?v=zejyzr5vW3A
 
Last edited:
Totally agree, most don't really have s handle on grading, and yes Alexa looks better than red's standard gammas, but it can be changed easily
 
A DP I respect told me once "I shouldn't say this as a cinematographer but anything you shoot on Alexa looks good..." And he went on. So then I bought two of them. Sure they cost a lot, can't upgrade them really, but clearly people like them. I haven't used mine much compared to my red but the slow motion 2k stuff looks honesty better than when I shoot slow motion on my red to the speeds at which I need to go to say 2khd so on top of nice roll offs I get the most amazing over cranks with my high speed. I also love how Amira is pickup shoot vs build it like Alexa mini or red. But it can get heavy so after a while I want an easy rig... The main thing is give and take with everything, flex is amazing too. All the cameras on the high end rock... I'd love to shoot either but various advantages fall on them either function wise or dynamic range wise
 
Christoffer Glans said:
Problem with this is that it is just scientific. When the end result hits the eyes of the audience, it will have gone through many things from the master copy to the screen showing the result. Everything from different codecs, streaming, screens etc.
This is exactly what I meant with this thread, to not just look at it scientifically, because if you do, yes, the Dragon, on paper seems to be far superior than the Alexa. But there is something "there" on the Alexa sensor that just works great everytime which means that we can't just look at this through math and the science behind it.

If you shoot the same scene, with the same field of view (sensor crop) and with the exact same lenses, keeping the cameras very close to eachother so that the image is almost the same and then do the same kind of grading for them, or match them up as close as they could possibly be... what will the differences be? This is something I've never seen. The thing is that I've seen so many comparisons of charts and such, but the testings are always filled with things that makes the comparison hard. Like the cameras not shooting the exact same thing at the same time, that they are too far from eachother to make the comparison truly based on the same image, that the image compared is a very flat log image that is very hard to spot differences on without applying a gamma curve.

If we had such images to compare, it could almost be a slider over the image to really make it easy to spot differences. It's the lacking of these kinds of tests that makes it hard to see the difference.
Click to expand...

Huh?! Just scientific?!
Well, science is known to be based in reality.:biggrin:

Did You know that You never actually see The Difference? You always see either the A(lexa) or the R(ed) but never The Difference. That's because You can't actually subtract A from R or vice versa, but You can, say, 3 from 5. It's 2, huhuh, look, see, that's difference...:biggrin:
And that's exactly where science comes into play - difference is purely mathematical, calculated. Not "physical". Not "visible".
It's all about how much Lumens, meters, Luxs, pixels, bits, stops, etc. Red records You more & costs You less than Aexa, that's all.

Besides that, there can't be two same images. Only one is the same...that one.
:sifone:
 
Somehow I remember that tweet last year? I doubt they can make it mini size anytime soon though... But that will
be an interesting development! I want to try to get one whenever they do release it
 
Interesting post Christoffer. There's certainly a perception issue out there that Alexa produces much nicer pictures, this has little to do with science.

I have to agree that because of the price difference there is a lot more Red cameras out there on lower end productions that don't or won't grade.

I was having this discussion the other day and realised that many of the big shows and movies shooting on Red are doing so because they're FX heavy and like to have the extra resolution. These FX movies get pretty heavily stylised so it's hard to compare the bulk of the work.

But looking at 'Gone Girl' & 'House of Cards' you get a better idea of what the dragon looks like with a lighter touch in the grade and it's hard to say one is better than the other.
 
You must log in or register to reply here.
Back
Top