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 :)

All the Epic/Dragon Vs. Alexa/Sony threads are belong to us...

If you are looking for answers please look to the camera original files. Personally I find this CML test useful for judging each sensor's reaction to extreme over and under exposure. Though there are greyscale charts included in the test I prefer dynamic range tests that shoot high-con scenes instead of charts. More real-world than technical. But, again, that is just me.

I've toyed with both the ALEXA and DRAGON raw files in the latest release of DaVinci Resolve. Here are a few screenshots and thoughts.



MAX OVEREXPOSURE

Cameras were developed at stated ISO to retain as much highlight information while producing the best overall placement of tones. Lift and Gain were left alone and only Gamma was brought down completely. This, in essence, pulls all of the retained DR from the highs into the mids and can give a picture of what sort of overexposed detail can actually be recovered.


RQCCvcO.jpg




dOrj1dR.jpg








MAX UNDEREXPOSURE

Cameras were developed at stated ISO to produce the clearest image possible. Extreme noise is a given. This is not meant to look pretty. It is meant to see how much actual detail and color "may" be salvaged in cases of extreme underexposure.



Bh6Fu7M.jpg






uLnegIK.jpg





Link to hosted images: http://imgur.com/dOrj1dR,uLnegIK,egBud5F,aakanWu,RQCCvcO,Bh6Fu7M,MIaOtca,NeKo4YH#0
 
Last edited:
Here is a set that I have purposefully attempted to grade back into a useable image. These images are "acceptable" to me. They may not be "acceptable" to anyone else.



MAX RECOVERABLE OVEREXPOSURE

The cameras were developed at stated ISO to retain as much overall dynamic range as possible and then graded (primary only L/G/G) down into an "acceptable" image. Special attention was paid to skin tone. If I felt the next step of overexposure began to clip or loose detail in the skin I reverted back to the previous frame.


+4 Stops

MIaOtca.jpg





+5.5 Stops

egBud5F.jpg








MAX RECOVERABLE UNDEREXPOSURE

The cameras were developed at stated ISO and then graded up (primary L/G/G only) to regain an "acceptable" image. This is all about the constant battle between underexposed recoverable detail vs. added noise. The level of added noise in these images is "acceptable" to me in an extreme case of underexposure. It may not be "acceptable" to anyone else.


-2.5 Stops

NeKo4YH.jpg





-3.5 Stops

aakanWu.jpg





Link to hosted images: http://imgur.com/dOrj1dR,uLnegIK,egBud5F,aakanWu,RQCCvcO,Bh6Fu7M,MIaOtca,NeKo4YH#0
 
Last edited:
A lot of actually understanding what you are looking at goes back to correctly understanding how digital CMOS cameras work in regard to scene-specific exposure. This has everything to do with dynamic range, latitude, ISO, sensor sensitivity, and intentional scene exposure / Post development.


The EXPOSING + NATIVE ISO thread is a good place to catch up. My suggestion is to start at page 17:

http://www.reduser.net/forum/showthread.php?112540-Exposing-Native-ISO/page17
 
Adrian Jebef said:
MAX RECOVERABLE OVEREXPOSURE

egBud5F.jpg
Click to expand...

Adrian, this image is clipping in the end two white bars. There is no separation like there is in the Dragon shot.
 
Matt Ryan said:
Adrian, this image is clipping in the end two white bars. There is no separation like there is in the Dragon shot.
Click to expand...


That ALEXA frame is also overexposed about a stop and a half more than the "comparable" DRAGON frame.

Again, these "recoverable" shots were graded for skin tone not for the greyscale or color charts. The top ends of greyscale charts are useless IMO if we are looking for a pretty picture. It is a compromise between how something important in the frame (like your actor's face) can be salvaged properly while other things are not (like top-end highlights). These grades were how far of an overexposed frame I could start with and still retain full detail in the face. These grades are acceptable to me. They may not be acceptable to anyone else. Honestly, it is extremely subjective.

BTW: I've labeled every frame I've used so feel free to load them up in Resolve or RCX and see if you can get any "better" or "more interesting" results.
 
Matt Ryan said:
Adrian, this image is clipping in the end two white bars. There is no separation like there is in the Dragon shot.
Click to expand...

I see what you're saying there Matt but the face looks really good on the back wall as well, like there is wider low to mid range detail. Looks a bit flatter overall to the eye which lends itself to a portrait kind of look. Here is a case where maybe the highlights go where they go but the overall scene is "pleasant". Hair color, hair detail could be called more filmic. Both images are technically good, but I think many people would still lean toward the softer contrast.
 
Adrian,

Can you just, for reference add to the original post how many stops over you recovered from in the two samples you recovered from? At first I was think you were recovering from the same totally clipped files in the first example. I'll double check but I think those are the ones from 2 stops over?

battistella

Adrian Jebef said:
That ALEXA frame is also overexposed about a stop and a half more than the "comparable" DRAGON frame.

Again, these "recoverable" shots were graded for skin tone not for the greyscale or color charts. The top ends of greyscale charts are useless IMO if we are looking for a pretty picture. It is a compromise between how something important in the frame (like your actor's face) can be salvaged properly while other things are not (like top-end highlights). These grades were how far of an overexposed frame I could start with and still retain full detail in the face. These grades are acceptable to me. They may not be acceptable to anyone else.

BTW: I've labeled every frame I've used so feel free to load them up in Resolve or RCX and see if you can get any "better" or "more interesting" results.
Click to expand...
 
David Battistella said:
Adrian, an, can you just, for reference add to the original post how many stops over you recovered from in the two samples you recovered from? At first I was think you were recovering from the same totally clipped files in the first example. I'll double check but I think those are the ones from 2 stops over?battistella
Click to expand...
I dunno! I should have done the math but it's late and I've got an early call tomorrow morning. Suffice it to say that if you use the reel and clip numbers from the frame grabs you should be able to guesstimate what over/under stop interval I used for that second set of grades. Initially, I'm seeing the ALEXA able to pull down highlights and dig deeper into the shadows than DRAGON V1. But again I'm more concerned with skin tone than absolute extremes. I'll try to add that tomorrow while I'm at work...
 
David Battistella said:
I see what you're saying there Matt but the face looks really good on the back wall as well, like there is wider low to mid range detail. Looks a bit flatter overall to the eye which lends itself to a portrait kind of look. Here is a case where maybe the highlights go where they go but the overall scene is "pleasant". Hair color, hair detail could be called more filmic. Both images are technically good, but I think many people would still lean toward the softer contrast.
Click to expand...

I agree that the Alexa looks subjectively "better", but if were talking pure exposure the top end is gone. Skin still looks good though.
 
Adrian Jebef said:
I dunno! I should have done the math but it's late and I've got an early call tomorrow morning. Suffice it to say that if you use the reel and clip numbers from the frame grabs you should be able to guesstimate what over/under stop interval I used for that second set of grades. Initially, I'm seeing the ALEXA able to pull down highlights and dig deeper into the shadows than DRAGON V1. But again I'm more concerned with skin tone than absolute extremes. I'll try to add that tomorrow while I'm at work...
Click to expand...

Andrian,

I went and checked it out for you.

The Alexa Clip in your recoverable example is 5.5 stops over base for the test.
The Dragon Clip in your recoverable example is 4 stops over base for this test.

You processed Alexa and Dragon at 160ISO

Battistella
 
Medavoym said:
I honestly see a lot of dancing around the issue at hand.

Questions are answered with more questions - it's a bit annoying.

Can I ask a straight question? I need a straight answer. Either from Phil, Geoff, Gunleik etc - knowledgeable people who intimately know both cameras. So...

Based on these tests... and your personal experience with the two cameras... WHAT CAMERA HAS MORE DYNAMIC RANGE? EPIC DRAGON or ALEXA?

Clear answer please. No dancing around. We're not talking subjective art here... we are talking hard facts.

DRAGON or ALEXA?

Thanks.
Click to expand...

only on DR Alexa would be better but is it everything all we wanted?
We want triple times cheaper camera beat Alexa on every aspect.Honestly i dont giva a f.ck , if Alexa has better DR than Dragon.So be it
Dragon beats Alexa on everything except DR for now..What about that?
RED kept saying its all resolution they are trying achieve and they did.Look at outside everybody every company is following 4K revolution now.

im %100 SURE RED can produce 2.8K sensor beats Alexas DR by 3-4 stop but who would buy that camera? What will existing customers will think about that? i would say definetely hell no! sounds like going back to Stone ages for me.
 
Adrian Jebef said:
I dunno! I should have done the math but it's late and I've got an early call tomorrow morning. Suffice it to say that if you use the reel and clip numbers from the frame grabs you should be able to guesstimate what over/under stop interval I used for that second set of grades. Initially, I'm seeing the ALEXA able to pull down highlights and dig deeper into the shadows than DRAGON V1. But again I'm more concerned with skin tone than absolute extremes. I'll try to add that tomorrow while I'm at work...
Click to expand...

Hm... You start to sound like me... :)
 
Medavoym said:
I honestly see a lot of dancing around the issue at hand.

Questions are answered with more questions - it's a bit annoying.

Can I ask a straight question? I need a straight answer. Either from Phil, Geoff, Gunleik etc - knowledgeable people who intimately know both cameras. So...

Based on these tests... and your personal experience with the two cameras... WHAT CAMERA HAS MORE DYNAMIC RANGE? EPIC DRAGON or ALEXA?

Clear answer please. No dancing around. We're not talking subjective art here... we are talking hard facts.

DRAGON or ALEXA?

Thanks.
Click to expand...

I guess it is better that I than Geoff take that one over here... :) (Of course anyone are free to comment! : )

As of "now", I see more DR in the Alexa.
Currently, from what I see, the alexa records both higher and lower luma values. No "but" to that.


Seeing Adrians developments, I also see what I have seen since I first got the proto (which had less DR than the one I currently have): It was the juiciness of the image that had me over.
The image within range is to me more pleasing than anything else I have seen. And the practical range is wider or at least on par with the Alexa. (If you look at Adrians "max over/under recoverable exposures and compare the skin and color in those, I think they are informative).
Subjectively, I think RED has taken the "nicer image out of the box" throne Alexa to many used to have.

It is fun that we do not discuss any of the other cameras tested here... :)
WHere I see things, they take as many RED jobs as Alexa does, and Alexa gets the gigs a lot of the time out of on-set convenience.
But hey. :)

How the Alexa would measure if it was measured the way Phil and John have, I have no idea.
But their tests are good. So is the Dragon. As it is.

There are a few more things on the Dragons behalf, though :)
Like resolution (this test was for UHD delivery) and general image-flexibility in post.

I know that is not applicable for "all" situations, but HDRx is available as it is.
Not an "excuse" just a fact.



If you take the "base" exposure of both and grade them rough, I guess that will come through...
 
Last edited:
Tests

Tests

Hi,

Alexa may have the DR better than Dragon but after shooting with Dragon we find it it be more juicy,dense and 'grade' friendly....

and if I have to pay three times the money to get that extra Dynamic Range then it does not make business sense for me at least...and then to store that RAW footage
on tons of drives......so guys do not forget the power of REDCODE..

And it will become stronger with new DRAGON SCIENCE.......

and I really thanks Gunleik and Phil for doing what he is doing for us on this forum.

(Gunleik please change your avataar pic...earlier one was badaas...in this one you are looking tired)

Further just wanted to know which DP in the world will overexpose or underexpose so much:zombie_smiley::zombie_smiley:


Harcharan
 
For practical DR there are a lot of things that come in to play, which IMHO all points towards the Dragon.

Color/saturation throughout range, cleanlyness, resolution, and (subjectively) image character.
 
Agreed

Agreed

Gunleik Groven said:
Harachan, these are of course test-exposures to "learn" things about the cameras.

This, however is one of a few practical derivations of tests like that...
Click to expand...

I totally agree with you....

Harcharan
 
The elements of potential confusion in this particular situation are the facts that prior multiple comparisons between both Red sensors showed a significant difference in DR while this one does not.
 
There is a significant difference between MX and Dragon, if that is what you refer to.

A stop in the top, and a couple in the bottom, and significantly (to me at least) a much wider "good" picture in the DR, a beautiful highlight rolloff and infinitely improved files for post.

THAT is not what this test is about.

Actually the difference to Alexa in the top is very much as expected after seeing Dragon and MX together.
The surprise to me is in the bottom of the scale, but I think that is actually doable.
That is IF Alexas Luma-range is the view we want to have as the point of reference for Dragon.

I happen to think that that is not the only relevant way to use and think of it... :)
 
Last edited:
You must log in or register to reply here.
Back
Top