Enter the Dragon - 20-Stops of Dynamic-Range

So you have to pay for the upgrade obviously, but will the new sensor increase the base price of a new Epic? That's the one thing I haven't heard any mention of yet.
 
"You must have skipped the part about: "Don't look at image quality here... this is a dirty debayer with no black calibration, no offset correction and no processing on a special 21 stop DR Chart from DSC.""


No. I didn't. Calibration and noise elimination are two entirely different things. After literal years of R&D, they chose to put this image out, and I'm responding to what's in it. Are we only allowed to notice positives, without negatives?

Perhaps it'll give a clean 15-16 stops. Or perhaps it'll give a clean, usable 12.x stops, as the image actually shows. The ball's in their court. This is an experiment at this stage, and nothing will be shipping for a long, long time.
 
"You must have skipped the part about: "Don't look at image quality here... this is a dirty debayer with no black calibration, no offset correction and no processing on a special 21 stop DR Chart from DSC.""


No. I didn't. Calibration and noise elimination are two entirely different things. After literal years of R&D, they chose to put this image out, and I'm responding to what's in it. Are we only allowed to notice positives, without negatives?

Perhaps it'll give a clean 15-16 stops. Or perhaps it'll give a clean, usable 12.x stops, as the image actually shows. The ball's in their court. This is an experiment at this stage, and nothing will be shipping for a long, long time.

apparently you only notice negatives. And calibration actually affects how the camera processes noise. While I was expecting to see more of an image than just a dirty dsc chart at this point, Red is well known to be ambitious on their dates and they admit this to be one of their many failings. What they're doing is obviously not easy, their R&D history is extremely short for what they are doing as well as compared to any other developers out there.

This is obviously not a sample image of best of what the sensor can do and to make a judgment as it's a final word are from those that are oblivious. This is a what potentially could be in the coming future, a sensor capable of much higher dynamic range.
 
"You must have skipped the part about: "Don't look at image quality here... this is a dirty debayer with no black calibration, no offset correction and no processing on a special 21 stop DR Chart from DSC.""


No. I didn't. Calibration and noise elimination are two entirely different things. After literal years of R&D, they chose to put this image out, and I'm responding to what's in it. Are we only allowed to notice positives, without negatives?

Perhaps it'll give a clean 15-16 stops. Or perhaps it'll give a clean, usable 12.x stops, as the image actually shows. The ball's in their court. This is an experiment at this stage, and nothing will be shipping for a long, long time.

They said the lens mount had a light leak... which was strange to me because, it's not that hard to get a lens mount light proofed... however, that would be like flashing film and if similar, would increase the lower end DR, because the darks get lifted.

Just throwing it out there.

Also, no matter what the DR, it will likely be higher than now, and that's ok with me. In fact, I'm sure I'll be adding contrast and whatnot, because unless shooting uncontrolled exteriors, I never had ANY issue with 12-14 stops on my interiors where I could fully control lighting. Just saying. .... again.

And yes. This is the first image off the sensor they have shared. Expect another 6 months at least.
 
They said the lens mount had a light leak... which was strange to me because, it's not that hard to get a lens mount light proofed... however, that would be like flashing film and if similar, would increase the lower end DR, because the darks get lifted.

Not too strange, just speculating here but as engineering is concerned easy access to internals often helps with development process, so if they're only testing firmware and hardware and not looking at the image quality at the moment light leaks might not be priority:-P Tho I wouldnt say that it increases the lower end DR, but rather lifts the blacks and wash out the lower end of the chart.
 
Yea I was going batty with all the n00b FB posts and Twitter posts last night about "usable 20-21 stops of DR ahh!" Calm down folks. It's awesome. The fact a sensor even sees that much without all the processing, black calibration, color science, etc. is amazing...but I hate it when folks don't read, don't think, and post threads with titles like this one.

Relax, mate. I did write; "How many of the stops can be effectively used is anyone's guess - from the picture though, I'd say 16, or 17 stops."
 
Not too strange, just speculating here but as engineering is concerned easy access to internals often helps with development process, so if they're only testing firmware and hardware and not looking at the image quality at the moment light leaks might not be priority:-P Tho I wouldnt say that it increases the lower end DR, but rather lifts the blacks and wash out the lower end of the chart.

There isn't a roll of gaff tape sitting around at RED HQ somewhere? :)
 
Relax, mate. I did write; "How many of the stops can be effectively used is anyone's guess - from the picture though, I'd say 16, or 17 stops."

I wasn't talking about you. I said people on FB and Twitter...the rush of comments people made without even reading the thread thoroughly. They just saw the image and went batty lol.
 
Seems to me they posted it just in time a month or so before the new F's start shipping. :cool:

They promised months ago that a Dragon announcement would happen before the end of the year. I believe the original plan was to have a little more show and tell by now but Dragon has proved to be a lot more and hence the delay. In any case, I'll be trading in my Scarlet, but I doubt it will be anytime soon.
 
They promised months ago that a Dragon announcement would happen before the end of the year. I believe the original plan was to have a little more show and tell by now but Dragon has proved to be a lot more and hence the delay. In any case, I'll be trading in my Scarlet, but I doubt it will be anytime soon.

yup, at NAB earlier 2012 they projected that they would be rolling out dragon upgrades in December to early Epic owners:-P
 
Last edited:
It's kind of weird - but I would feel intimidated having 15+ stops of dynamic range, and sensitivity greater than the human eye. It's like being in a room with no walls - a bit of agoraphobia.

I need limits to be creative, damnit!
 
Yes I imagine it will be another year before the option is actually an available option - and who's to say it won't "change again"?

All that aside, who the hell really needs 20 stops of dynamic range and more importantly will 16-bit linear raw really be enough for that much dynamic range? A lot will be lost unless there is a good ADC on-board capable of even processing 20 stops off the sensor.

Like I said, "another year away" for sure.
Considering that image actually exists, both questions have already been answered: Yes, they apparently do have the electronics to do this, and yes, to debayer the image, the codec must have allowed for that much data to be stored in the first place.
 
This site sometimes - yesterday morning European time I posted a message on the AF100 forum on the Red dragon I believe I was the first to post on this site on this topic, the post was polite and linked to the red user site and the pic plus the stated 21 stops etc, only to have my post removed some hours later.
Yet other posts on this subject exist virtually identical in content (Sony F 5 and F55 forum).

I have no loyalty to Red I don't own one yet I can already see some who have an axe to grind.
If Red can't deliver on their claims then their posting will damage the companies reputation. I don't believe they would be so foolish, Red were under enormous pressure when Sony launched the new F 5 and F55 with many Red owners asking "where are Red in the game", now we know. I just sense a real anti Red feeling / vibe here.
 
Last edited:
It's kind of weird - but I would feel intimidated having 15+ stops of dynamic range, and sensitivity greater than the human eye. It's like being in a room with no walls - a bit of agoraphobia.

I need limits to be creative, damnit!

can't say I agree, I've had to scrap most of a shot involving a transition from an extremely bright exterior to a regularly lit interior, because the 13.5 stops of the F3 weren't quite enough to salvage the shot.

sure, if I'd had the budget I'd have brought in HMIs to balance the interior, but when that isn't an option more dynamic range can save the day - so it's absolutely a feature I think many could put to use.
 
From what I've read over the Reduser forum :

- No more update program for Scarlet owners but a trade-in program with Epic Dragon

- Need for more ultra fast ASIC's (so need for a new cooling scheme IMHO)

- More than the previously 6000$ to do the upgrade

-> RED should not upgrade EPIC's but design a new camera body with new optimized powering and cooling scheme.

If they keep the current thermal dissipation design of EPIC and Scarlet they will inevitably run with overheating problems unless they make the (stupid) front fan runnnig faster...
 
RED should not upgrade EPIC's but design a new camera body with new optimized powering and cooling scheme.

I think this is excatly what will happen, with another upgrade scheme for both Epic and Scarlet users.

I love the idea of modularity, but unless you get it perfect from the start, it can be a headache... and the current "brain" system is by no means perfect.
 
Its all development and fraken-rigged right now, what makes you think the new boards are already printed and ready to go?
It doesn't matter whether the technology is available in a ready-to-use format, but that it's there at all, stuffing it all into a small box is merely a detail that Mr. Moore is very good at solving. Assuming they used 16-bit ADCs, that would still cap the DR to 16 stops, so how are they reading 20 at once?
 
My suspicion is RED will have to move from the linear RAW format to a LOG based one. It is the only way to capture 20-bit precision in a smaller bit-depth format. Similar to ARRI and how their 10-bit LOG signal carries about as much information as a 12-bit linear signal does, and 12-bit LOG carries as much information as a 14-bit linear signal does.

Will Dragon bid farewell to 16-bit linear?

LOL that's the first time I've seen somebody link to themselves as if it's a source.

Pimp that blog.
 
Back
Top