DSMC Camera Firmware release 5.1.55 and 5.2.21 beta

[Nick Morrison]

Yep. The long and the short of it is that people were asking for something to help deal with "red specks" in shadows. What Graeme has ended up producing however is really interesting. It gets rid of the red specks, but also enhances the channel information via his adaptive debayer color science/algorithm/voodoo. D.E.B. has surprised me in that way. Red and Blue are cleaner and more detailed. Green is effected too, but not as drastically. What you end up with is slightly more detail overall within the image at any ISO. Pretty cool.

I'm testing all sorts of things. Green and Blue Screen, flesh both properly exposed and significantly over/under, dark things like tires, deep browns in wood, etc.... At the moment I'm working on anamorphic material and inspecting how it performs there. But so far I'm pretty impressed. Red's working hard to get this pushed out right and proper. Even at night, heh, just like me

Love how Graeme can make lemonade out of lemons. He's a true Hollywood Wizard.

 

[David Battistella]

Because instead of having to "recover", i can shoot it so it looks correct to begin with.

Just as a reminder. DEB does not eliminate noise, but it does eliminate the color (red dot noise) that people have been seeing. So, it's not like neat video. Noisy underexposed shots are still going to be noisy, that noise just takes on a monochromatic tone instead of being able to see the red dots when zoomed in to 1:1 or 1:4.

By and large if you rate at 2000 ISO the footage will seem grainier than if you rate lower, which we al already know. I think if you want that eliminated then you can send your dragon back for the old OLPF and knock off that stop and a half in the highlights, and deal with the big magenta blooming on low light sources and red dot patterning in the sun, etc.

I'm going to ride it out on the new OLPF because based on my testing I am building the exposure strategies to get the results I want and I do think the enhancements are good. Red might be working on an interchangeable solution, the software and firmware keep getting better. The new REDcine is making every thing look that much better.

Its the the thing about red. The stuff I shot in 2008 looks better because of software. Of course we want everything better now, and that will happen because it is a constant developmental process. That drives some people nuts, (I'm not saying you in particular) and some people embrace that. Some people see that as a positive, some people say it's RED making excuses and that they are inadequate.

I've always seen it as a positive thing and a great way for a camera company to operate.

Theres so much more in this new OLPF, like tungsten. Has anyone noticed how great this sensor and color science performs with 3200 sources. The color reproduction is off the charts with tungsten light.

My strategy is to get the best out of what's available to me today and the adapt and retest with the new builds, etc. even this latest new FW build has a bunch of subtle improvements. It's just an evolving process and I get that it is not for everyone.

Battistella

 

[David Battistella]

Yep. The long and the short of it is that people were asking for something to help deal with "red specks" in shadows. What Graeme has ended up producing however is really interesting. It gets rid of the red specks, but also enhances the channel information via his adaptive debayer color science/algorithm/voodoo. D.E.B. has surprised me in that way. Red and Blue are cleaner and more detailed. Green is effected too, but not as drastically. What you end up with is slightly more detail overall within the image at any ISO. Pretty cool.

I'm testing all sorts of things. Green and Blue Screen, flesh both properly exposed and significantly over/under, dark things like tires, deep browns in wood, etc.... At the moment I'm working on anamorphic material and inspecting how it performs there. But so far I'm pretty impressed. Red's working hard to get this pushed out right and proper. Even at night, heh, just like me

I can only echo everything you are saying here Phil. I have seen the slightest improvements even on stuff rated @250ISO in broad daylight. By tackling this red dot colour noise the overall image sharpness seems to have improved just the slightest bit as well, which helps with small detailed items in landscapes. Sometimes even just a few pixels makes the slightest difference when viewing stuff at full frame.

battistella

 

[Antoine Baumann]

Ok. I see that point of view. But I don't agree really that MX outperforms dragon at 800 properly exposed. I'm more confortable at 800 ISO on dragon than I ever was on MX.

Battistella

I'm not comparing MX vs Dragon, I'm comparing Dragon old OPLF old calibration with new OLPF with new calibration. And that's where you would freak out.

 

[David Battistella]

I'm not comparing MX vs Dragon, I'm comparing Dragon old OPLF old calibration with new OLPF with new calibration. And that's where you would freak out.

Can you post that? I have not seen any of those direct R3D's. I only have one camera here beside me.

Battistella

 

[Antoine Baumann]

well I have no direct comparison as I did not have both at the same time, but I have shot with the old olpf for more than 4 months before switwing to the new olpf, and the difference in the noise is obvious.

 

[Daniel Stilling]

... then you can send your dragon back for the old OLPF and knock off that stop and a half in the highlights, and deal with the big magenta blooming on low light sources and red dot patterning in the sun, etc. ...

Battistella

Please stop using this argument as a solution, when in the same breath you invalidate that solution with all the issues the old OLPF had. Stop patronizing people that feel the noise issue is a real problem.

 

[David Battistella]

well I have no direct comparison as I did not have both at the same time, but I have shot with the old olpf for more than 4 months before switwing to the new olpf, and the difference in the noise is obvious.

As I look at many of the R3D's posted from many different tests I have seen, I have to say that the "noise" pattern is exactly the same, the colour of the noise was new with the new OLPF but the pattern of the noise is exactly the same across all the many tests and frames posted by many different users at different times through the whole development phase.

The RED noise is there, but it's the colour that is different not the actual noise pattern. I've posted stuff from months ago that displays the same noise pattern without the colour noise. The forthcoming REDCINE X addresses the colour noise issue as a DEB switch. We don't know how it will be implemented.

For example. THIS IS IMPORTANT.

If in the new release of the SDK and the New version of RCX this was "always on" then essentially the fix is done, "in camera" because every time you, or your client open an R3D file they would never know that there was RED channel noise.

But people say, "fix it in camera". Well, the truth is that would be a bandaid because everything up to this point would have the noise and everything from here forward would not have the noise. I feel the way RED chose to go about it was smarter because, everything up to now and everything going forward benefits from the debater algorithm.

You see what I mean? When the new software is released the clients will never see it because it will not be visible on any footage, every shot until now and after now will benefit.

For now my suggestion is to use a 6 denoise and debater at full quality for final renders.

Battistella

 

[Paul J Steinberg]

Please stop using this argument as a solution, when in the same breath you invalidate that solution with all the issues the old OLPF had. Stop patronizing people that feel the noise issue is a real problem.

Amen.

Also please stop using "properly exposed" as a qualifier in these discussions. The suggestion is that people wouldn't be having the problem if they only knew how to expose the image. Dragon isn't causing light meters to malfunction or years of experience to fall out of cameramen's heads.



Your skill level with Epic MX:

Your skill level with Epic Dragon:

 

[Elsie N]

As I look at many of the R3D's posted from many different tests I have seen, I have to say that the "noise" pattern is exactly the same, the colour of the noise was new with the new OLPF but the pattern of the noise is exactly the same across all the many tests and frames posted by many different users at different times through the whole development phase.

The RED noise is there, but it's the colour that is different not the actual noise pattern. I've posted stuff from months ago that displays the same noise pattern without the colour noise. The forthcoming REDCINE X addresses the colour noise issue as a DEB switch. We don't know how it will be implemented.

For example. THIS IS IMPORTANT.

If in the new release of the SDK and the New version of RCX this was "always on" then essentially the fix is done, "in camera" because every time you, or your client open an R3D file they would never know that there was RED channel noise.

But people say, "fix it in camera". Well, the truth is that would be a bandaid because everything up to this point would have the noise and everything from here forward would not have the noise. I feel the way RED chose to go about it was smarter because, everything up to now and everything going forward benefits from the debater algorithm.

You see what I mean? When the new software is released the clients will never see it because it will not be visible on any footage, every shot until now and after now will benefit.

For now my suggestion is to use a 6 denoise and debater at full quality for final renders.

Battistella

David, you make some very fine points. Nice thinking.

 

[Christoph Miera]

But people say, "fix it in camera". Well, the truth is that would be a bandaid because everything up to this point would have the noise and everything from here forward would not have the noise. I feel the way RED chose to go about it was smarter because, everything up to now and everything going forward benefits from the debater algorithm.

My 2 cents: There are two possibility's of "fix it in camera": First is a fix of the RAW Image in Camera. That comes with the issues of fixed and none fixed footage, as you describe. Second the recorded RAW Image is the same after all, but in Camera comes DEB for Display, SDI and so on.

So second is in my opinion the best way to go, as RAW is RAW and the processing can always be optimized. And why not call it Reddebayer 2, as there are also different Redgamma and Redcolors.

 

[Jan Reiff]

Phil helped me today and rendered one r3d of my footage with DEB. first thanks to Phil, 2nd: this is really an improvement, and if it´s available in all running programms (da vinci ...) this will help a lot to get the best out of Dragon. some people i talked to always forget: This is a 100% RAW camera. And you have to know the limits as you have to know it with every camera. Still, this is not a "i push to ISO2000 and all is bright and clean" camera. i think some should stop to repeat this ISO2000 statements a long time ago. Especially with DEB, Dragon is much superior to MX. Actually, there is noise WHEN you crop 200% and push and push and push, ... but the image with DEB stays sharper, more details, no chroma noise.
as i don´t have redcine right now, only this one frame, i am sure, that a well exposed and graded Raw at ISO 800 even up to 1000 is State of Art at the moment (with DEB), because color , tungsten-look and resolution are things that seem to be pushed more and more in the background since weeks, BUT they are so much more important (especially colors in tungsten) than an underexposed area (that should stay black for a natural looking image,) ... you will have noise right now when you force it, but you will have great images too, ... if you want them. i welcome the fix IN camera too, but if it´s a tool of the Post - it´s okay for me too. i remember days when we shoot film, and the things were running fine or bad in ... the film processing. so, nothing has changed so much. there may be cameras out there that gives you the perfect image (clean, best colors, best dynamic, best resolution) directly out of the box. i don´t know them. someone can tell me ?

 

[David Litchfield]

Arri must be finding all of this hilarious.

The fix needs to be in camera otherwise directors and producers are going to ask questions. The answer to which is hassle in their minds.

The fact we’re even talking about a fix (of a fix) for a widely late sensor that was supposed to be the digital equivalent of the second coming is ridiculous.

 

[David Battistella]

Please stop using this argument as a solution, when in the same breath you invalidate that solution with all the issues the old OLPF had. Stop patronizing people that feel the noise issue is a real problem.

I don't feel that was patronizing. It is a legitimate choice offered by RED. When they went this direction with the NEW OLPF there were tradeoffs, take an extra stop in the highlights, clean up the issues mentioned, etc. They are shipping this NEW OLPF. It's the direction they are going. I can not change that.

There are trade off's (Even RED is acknowledging that) to this kind of science (and that is what it is, a science, with variables). So, they, like apple might do, move in a direction and stick with it as the way forward.

Now some users say, "hey, I liked that old one". What does RED do, they say, "Hey, we are telling you what the tradeoffs are, but if you want it the old way, we will give it to you."

Me repeating what you might not want to hear is not patronizing. If I am restating what RED has already reiterating then you are basically saying that Jarred is patronizing by giving you options.

The dance between OLPF (which is far more complex than a piece of glass), how a sensor receives light, how that is compressed and decompressed and debayed is complex science. It's not like ordering up a Big Mac.

There are choices, RED made one and they offered you and every other user one. Just because I am reminding you that you have a choice, I'm being patronizing?

Scratching my head here.

Battistella

 

[Mark Wilkinson]

Arri must be finding all of this hilarious.

The fix needs to be in camera otherwise directors and producers are going to ask questions. The answer to which is hassle in their minds.

I also think the fix needs to be in the camera.

 

[Jan Reiff]

Arri must be finding all of this hilarious.

The fix needs to be in camera otherwise directors and producers are going to ask questions. The answer to which is hassle in their minds.

The fact we’re even talking about a fix (of a fix) for a widely late sensor that was supposed to be the digital equivalent of the second coming is ridiculous.

iam director too, uncountable commercials, tv stuff ... i would never see a ISO2000 pushed redlog output at my monitor ... as a DP, i would never give the director this on a monitor. what is the difference to do it in post ? you can do 1000 things wrong with RAW in the post, not only with RAW- with ALL files, prores, .... i see everyday in germany massive clipped Alexa TVMovies, horrible grade and looks. noone is blaimg the camera for this. if i push Alexa footage up to ISO2000 and watch it on a 4K Screen, ... ähem (actually i love the image of Alexa, but it has limits too, it has, let´s be realistic)... i don´t get really into the problem to switch on DEB or name it as you like in a post-workflow when it´s available. it´s the same like the DRX slider with MX footage. if you know it, its fine. if not, you have this "issue" in the highlights - you must know the tools and what is possible. i can sit in the corner and cry the whole day, or i find my workflow. and i do a lot of great stuff with the Dragon, not released yet, but all love it (clients, producer ...)

 

[David Litchfield]

iam director too, uncountable commercials, tv stuff ... i would never see a ISO2000 pushed redlog output at my monitor ... as a DP, i would never give the director this on a monitor. what is the difference to do it in post ? you can do 1000 things wrong with RAW in the post, not only with RAW- with ALL files, prores, .... i see everyday in germany massive clipped Alexa TVMovies, horrible grade and looks. noone is blaimg the camera for this. if i push Alexa footage up to ISO2000 and watch it on a 4K Screen, ... ähem (actually i love the image of Alexa, but it has limits too, it has, let´s be realistic)... i don´t get really into the problem to switch on DEB or name it as you like in a post-workflow when it´s available. it´s the same like the DRX slider with MX footage. if you know it, its fine. if not, you have this "issue" in the highlights - you must know the tools and what is possible. i can sit in the corner and cry the whole day, or i find my workflow. and i do a lot of great stuff with the Dragon, not released yet, but all love it (clients, producer ...)

And like the DRX slider it might not get used. It's one thing the DP knowing his tools, but he/she then hands the footage over and can't always babysit it.

If it works for you then great. But I'm not the only voice asking for it to be in camera.

 

[Jan Reiff]

what was with filmstock ? you gave it to a producer - and he gave it in the hands of professional operators to develop. i think we must see, that a Red Camera is not a Handycam ... maybe in 5 years. not now.

 

[Jan Reiff]

of course it would be nice to have it IN camera. but i think we should learn especially with RED RAW, that there are some points in the post (not only DEB) that decide about the image - and to give RAW someone who just press the export button will bring not the best results in all image aspects...
i know so many different operators that bring you complete different results out of the same RAW file, so, of course this is a kind of science and talent, and those who knows the camera and its RAW very good will have better images. same again with filmstock since generations. i saw dramatic different qualities at the telecine with the same 35mm negative footage, depending on the operator. skintones, ... dynamic, ... noise, ...

 

[Bohdan Klos]

Once again this thread will turn into a lot of people requesting things be tweaked or problems fixed or a certain workflow considered. And then there will be a few members telling everyone else how the problem should be ignored or is insignificant and it is actually people voicing concerns that are wrong in their methods or their thinking.


It is YOU considered at fault if you dont agree with Reds decisions or product features, or you feel a problem should be a high priority. Somehow you are deemed less than professional, perhaps you are a prosumer who stumbled across Red Cameras in error? A rich benefactor bought you a camera? Are you an owner and if not why are you even here? Do you realise the camera is "not a camcorder"? No-one works like you except you. You don't know how to shoot...etc etc. I have heard all these and many more.


Then, down the road, Red will fix the EXACT same problems being discussed. EXCELENT! Well done RED! So those problems turned out to be legitimate? Who`d have guessed that? Not the few people telling everyone "there is no problem to be fixed". Those guys will move on to the next problem and again begin the process of discouraging conversation on that new issue. Rinse and repeat as they say.


All the time the discouragement to conversation is unsurprisingly keeping the conversation going.


Is it possible to stick to addressing problems rather than trying to justify why problems should be allowed to exist?