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 :)

Dragon/Epic comparison

It is kinda silly.

I have images that show off what the Dragon can do 16 stops under the initial exposure, but because of the sillyness in the responses here, I do not feel like posting them.

What I have done, is to send tho R3Ds to a few knowledgeable people.

I have asked them to not post the images for the very same reason.

But if they want to comment, they are free to.
On noise, color, DR difference in the bottom, everything.

Bear in mind (if they chime in) that that is 16 stops below the initial image which did not break the Dragon in the top.

And an image exposed 16 stops bellow what is not the absolute top of a sensors range, will hardly look like a "nice iPhone picture from my Kitchen", but I do not want to spend more time explaining that right now.

And by that, I think I leave this discussion for a while.

Best.

G
 
Phil Holland said:
Dynamic Range with Dragon is significantly more than anything currently on the market.
Click to expand...

I'm shooting with Red since the beginning of 2008. I really like what they do. But be assured that Arnold und Richter, and Sony of course, know their stuff very well. Red has many advantages being the resolution, the price tag, and with Epic the size the biggest points. But DR and colour never was Red's speciality. With 6K pre-bayer resolution and an DR equally to that what Alexa delivers IS a major achievement and a big reason to upgrade to Dragon. But to say that Dragon's DR ist significantly more than anything on the market is a statement that has to be proved and in the current state pretty bold.

Honestly, marketing by hype that never will fulfil the high expectations later when the products ships is something Red is pretty good at. See the discussion about noise in this thread and in the unlucky other Dragon vs MX thread. It clearly show that people expect miracles from Red's Dragon sensor. Very unlikely to happen.

Regardless of noise et al. I find Gunlike's Log encoded frame grabs pretty encouraging. Perhaps it's me but since I swapped the RedOne to an Epic I finally have a good substitute for the Arri 235 which has been my favourite camera for some years. I still fiddle around with LUTs and grain to accomplish looks that help me to come near to that what I once had. Perhaps Dragon's colour science will finally reach a level where I will say, ok film is history and I have something new that is better and not only a makeshift that makes production faster and cheaper.

Hans
 
Gunleik Groven said:
It is kinda silly.

I have images that show off what the Dragon can do 16 stops under the initial exposure, but because of the sillyness in the responses here, I do not feel like posting them.

What I have done, is to send tho R3Ds to a few knowledgeable people.

I have asked them to not post the images for the very same reason.

But if they want to comment, they are free to.
On noise, color, DR difference in the bottom, everything.

Bear in mind (if they chime in) that that is 16 stops below the initial image which did not break the Dragon in the top.

And an image exposed 16 stops bellow what is not the absolute top of a sensors range, will hardly look like a "nice iPhone picture from my Kitchen", but I do not want to spend more time explaining that right now.

And by that, I think I leave this discussion for a while.

Best.

G
Click to expand...

If someone disagrees with you, or calls you out on something...it's silly? Common Gunleik, we appreciate your posts and your tests, but don't expect everything you share to be some golden measure, or every result you backup with the big names to be accepted without question. This is the internet. You're a knowledgeable guy, and you do a lot for these forums which is commendable and much appreciated. I for one will never take anything as golden until I try it for myself though, regardless of the name that's attached to the proof. That's not to say I disrespect or deny peoples findings on-line, but I and others aren't being silly. We are just searching, communicating and questioning....nothing personal.
 
Matthew Scott said:
Ground-breaking new stuff like better dynamic range, better color, a "smoother" image, higher resolution, etc, etc...BUT.....virtually the same noise as MX, pixel for pixel, true? And let's not argue that sub-sampling and color science fixes that, because pixel for pixel, it's very, very similar. That's not so bad. I think it's fine! So, you guys are certainly not silly incompetent fools, but maybe aren't bold enough to agree on something like this here on Reduser? For the sake of what? RED's image? The hype? That's what frustrates me. The upgrade to Dragon is clearly well worth the money. No Doubt. The images you get from the new sensor are superior to MX, no doubt. So what harm is there in talking about the strengths AND weaknesses, rather than cleverly typing around the weaknesses. When Les posted his noise example, it shed light and revealed the true characteristics of the sensor in regards to noise, at a 1:1 viewing. This is what what I'm interested in, along side your test which was has revealed a lot...although, what are you worried about with posting your .R3D's? People asking you stupid questions? Or creating forum "noise"? (pun intended). That noise is precisely what pushes this industry into new places...to keep a lid on it, restrict it, or shun it elsewhere doesn't help anyone...other than the top dogs who seem to have authority around these parts.
Click to expand...

Couldn't agree more Matt, on all points.

However, although MX blown up to match the frame size of Dragon does look identical noise-wise, the benefit of higher resolution images is that with the Dragon image shrunk down to the same size as MX - the noise structure shrinks too. So pound-for-pound it should actually be a measure less notable on Dragon than it is on MX when you're viewing comparable images on the same screen.

I think the big thing here is clearly the added highlight latitude, Dragon looks like it pulls up alongside current filmstocks and Alexa in holding on to details in the high-end. Which means you can expose a face 'hot' and still have enough room in the highlights to hold bright skies or windows (which has always been the biggest trouble with digital IMO).

Thanks for the tests Gunleik, they're informative.
 
Hi Matt.

no problems. Just a little surprised.

This is the 1st time I have worked on a camera that I can in a plausible way say and document that can capture 16+ stops.
THAT got me a bit excited. Sorry for that.

I do not deny what you see in the noise at that exposure and without scaling.

That changes when you go down. For now you will have to take my word for it.

All good

G
 
gunleik,

thanks for the r3d files. I know this is a minefield, but I will try to explain why I like your testing methods.


At 16 stops under, it is possible to push each sensor past it's usable limit, they aren't meant to perform their best there.

By by recovering each image and developing them in exactly the same way, it's plainly obvious that there is more usable image data being captured with dragon. To be clear, you'd never underexposed an image by 16 stops and then boost the ISO 12800 and expect a usable image, but at these levels you can see that the noise floor has been significantly lowered.

Even if the noise pattern or characteristic were/seem very similar, this kind of test shows that you have to go much further into under exposure to see it. So this makes dragon less noisy than MX because it's the point at which you begin to see and discern noise that is VASTLY different. It doesn't matter much if the noise characteristics are similar, (I think that is actually a good thing and it makes sense)

Neither image boosted to this extent would be usable, so you can't shoot 16stops under and get usable images, but this deep into the engine it's clear what is going on under the hood. Dragon is outperforming MX, it's not better or worse, it's outperforming. It's clearly an improvement and there is more usable image.

Like going from R1m sensor to R1mX, but on HGH and steroids. There clear more image data in the dragon. Which means resulting images from the dragon will be less "noisy". The way it was from the start and the original r1 until now.


I hope some others that got the r3d can chime in.

Battistella
 
You guys who don't see the difference between Dragon and Mysterium-X sensor should get a minepick and start digging the R3D files. R3D doesn't brake anymore!
 
Personnaly I coudn't care less about the noise level improvement, because their is eventually lighting kits for that.
I care more about the high light improvement, wich seams obvious by watching the many sample posted already on RU.
Thanks Gunleik for sharing the love. ;-)
 
Nope, we see a difference. All I have been saying, over and over, is that dragon was supposed to MUCH cleaner than MX noise wise, clean at 2000 I remember once reading. And now, we see, that at 1:1, it's not cleaner, it's pretty much the same. That's it! I'm not disappointed....I'm not angry, I'm not confused and I don't need convincing that Dragon will/does out perform MX.
 
saying it looks like "anything" at "ISO X", does not really make much sense without taking into account the amount of light that hit the sensor at the time.

Of the few I have sent the -16 stop R3Ds, some has attacked it at ISO 2000, others at 12800. Result is pretty similar on Dragon to Dragon and epic to epic comparisons of those, since it all boils down to the real underexposure of the sensor due to actual photons available to eat.

What becomes pretty obvious when you push hard, though is that neither noise or character is similar.
besides the simple fact that in the end there simply is not an image in the Epic frame, while the Dragon remains a "something".

If a cam has 16+ stops, as RED claims, and I agree, it does not mean that you can underexpose by 16 stops... ;)

What I am really surprised by, was I started the highest exposure without really clipping the Dragon, and still 16 stops under there is "something".

It "should" kinda have gone at least a stop earlier.
But it did not.

Also comparing effective noiselevels without taking scaling into account, is kinda counterintuitive, unless you are aware of some 6k cinemas somewhere.
I am not.
 
That is the perfect description.

You push it to 12800 and there is a "something", that is the part that is eye opening.

Battistella
 
Matthew Scott said:
And now, we see, that at 1:1, it's not cleaner, it's pretty much the same.
Click to expand...

That is not really true, because you are getting much more usable image at 2000iso with dragon than you would with MX. That is the difference. Too see the same amount of noise with the MX you might have to be at 5000ISO and you still would not have the same amount of image detail.

The dragon is better on both the low and high end. We all know this. That makes the noise profile much different in real world applications. It means if you lit a scene the same way, dragon would give you more signal to manipulate and probably at a LOWER, cleaner ISO.

I'm going to stop there for now.

Battisella
 
David Battistella said:
That is not really true, because you are getting much more usable image at 2000iso with dragon than you would with MX. That is the difference. Too see the same amount of noise with the MX you might have to be at 5000ISO and you still would not have the same amount of image detail.

The dragon is better on both the low and high end. We all know this. That makes the noise profile much different in real world applications. It means if you lit a scene the same way, dragon would give you more signal to manipulate and probably at a LOWER, cleaner ISO.

I'm going to stop there for now.

Battisella
Click to expand...

Okay, I like where you're going with this....but I'm looking at Les's post, and I don't see that.
 
Agree very much with this statement because I really think beautiful images, still will be created using all old tricks in the book including good lighting.

Thank you for the effort and congrats on receiving your Dragon Gunleik! :-)


Brice Ansel said:
Personnaly I coudn't care less about the noise level improvement, because their is eventually lighting kits for that.
I care more about the high light improvement, wich seams obvious by watching the many sample posted already on RU.
Thanks Gunleik for sharing the love. ;-)
Click to expand...
 
Hans Magnus Kvåle said:
Agree very much with this statement because I really think beautiful images, still will be created using all old tricks in the book including good lighting.

Thank you for the effort and congrats on receiving your Dragon Gunleik! :-)
Click to expand...

I completely agree that there is no replacement for great, artful lighting. However, I'm really sad that Dragon is not actually more light sensitive on the individual photosite level. As someone who works on productions with a budget range of $1,000,000 down to $1,000, I don't always have good lighting available. For that reason, I want a camera that will squeeze every bit of detail out of the available light. My trusty FS100U has been my low light camera of choice for years now, and I was hoping to pass that torch along to Epic Dragon. However, I can't do that if I'm forced to shoot at 6k to achieve better perceived light sensitivity. I just plain don't want to manage that much data. On top of that, my Red Pro Primes, made by RED wouldn't cover the full dragon sensor, made by RED. I'm feeling a bit frustrated.
I'm still going to upgrade to the Epic Dragon, because that's what my clients want. It's just disappointing to find out that 4k on the dragon won't be better in low light than 4k on the MX
 
Last edited:
dragon is a lot more sensitive in the bottom.
i just decided against posting the lower exposures, due to the misreadings of test material. but it is in the region of 2+ stops more in the bottom

g
 
Gunleik Groven said:
dragon is a lot more sensitive in the bottom.
i just decided against posting the lower exposures, due to the misreadings of test material. but it is in the region of 2+ stops more in the bottom

g
Click to expand...

So the Dragon IS more sensitive than the MX? Another thread posted what was claimed to be a pixel to pixel comparison of the two sensors and both images were equally grainy and unusable
 
You must log in or register to reply here.
Back
Top