Welcome to our community

Be a part of something great, join today!

  • 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

Not exactly true. You can adjust curves in-camera - or load a look with any RAW parameter adjustments from RCX into the camera. When you alter RAW parameters in camera - that information is saved in the header of the actual R3D recorded.

Mark, you cannot alter any RAW parameters in-camera (ok, you can change the iris etc....). RAW means raw sensor data. If you want to change the RAW you have either to change the light sources that hit the sensor, the exposure time, or the frame rate. Anything else will be manipulated later in post. What you can do with R3Ds is to create additional information how this raw should be treated later in post by the SDK (WB, gamma, colour science, curves and whatnot).

Ok. I guess it's a question of what one understands by the term RAW. For you I think it means RAW plus transcoding information. For me RAW means the raw sensor data.

----

All these endless possibilities and the lack of understanding how RAW eventually works makes working with the Epic for many unnecessary difficult and leads more than often to non satisfying results. Ok, one can say read the manual....

I said it elsewhere and be pleased to repeat it again: I wished Red and took the plunge and created with the introduction of Dragon a tool that transcodes the R3D to the best possible Log RGB file without anything to worry about, regardless how funky the camera settings are. This would avoid a lot of grief.

Hans
 
Mark, you cannot alter any RAW parameters in-camera (ok, you can change the iris etc....). RAW means raw sensor data. If you want to change the RAW you have either to change the light sources that hit the sensor, the exposure time, or the frame rate. Anything else will be manipulated later in post. What you can do with R3Ds is to create additional information how this raw should be treated later in post by the SDK (WB, gamma, colour science, curves and whatnot).

Ok. I guess it's a question of what one understands by the term RAW. For you I think it means RAW plus transcoding information. For me RAW means the raw sensor data.

----

Hans

Okay Hans - I have always used the term "RAW parameters" for the controls that you can adjust that are part of the R3D SDK and are displayed as "Source Settings" in 3rd party apps - like WB, curves, saturation, etc. etc. etc. - which effect how the RAW image is translated into an RGB image. And YES - you can alter these controls in camera and they are non-destructive - and these settings will travel in the meta-data of the R3D.

Perhaps I should stop using the "RAW parameters" term and use "Image Processing Controls" if this is causing confusion.
 
There is basically two "schools" for using the RAW stuff.

One is a meta-data tweaking WYSIWYG video-cam approach.

THAT is arguably very simple and works very well, with in-cam metadata and WB tweaking etc.

Downside is a bit more noise from the getgo and a few issues if you WB too far IMHO, both in the red and the blue channel, which are often better treated in RGB apps, like Resolve.

But it gives quick turnaround and you do not have to understand a lot about signalchains, image-generation etc etc etc to get very good results. And if you shoot good, you get good.

The other is the technically "better", but more knowledgehungry film-cam approach.
Basically leaving everything "flat" and work on the optical signalpath to get "as far as you can" and start off in log or linear from there for the grade.

That gives less noise and a bit more of dynamics and is arguably technically "best", but can be a very limiting approach for many.

I am not religious about which is "best", as the second approach exclude a lot of users on rather minimal nuances, for many of the things actually shot on these cameras.

The fun thing about Dragon...

Is that the in-camera WB kinda has a lot of fewer tradeoffs than on any other digital cam I have seen.

Result-wise the two approches are closing in. And then, of course; To the degree that this statement holds true, the in-cam/auto WB becomes a timesaver.


Hans... Actually RC-X comes with a default preset called something like "Film10", which more or less does exactly what you are asking for...


@ Mark:
The "what is really a RAW image and what are "RAW" control" is actually a bit interesting. I do see the potential of confusion bringing that into this comparison though, and thus I try to not bite too much into that part of the discussion...


Best

G
 
I think maybe we should take the WB discussion into another domain than this thread.

I will shoot a few examples next week.

The WB on the Dragon has less negative effect on the signal than it has on the Epic.
But I think this might get a bit elaborate... :)

I do not agree to the ref signal analogy at all, BTW... :)
I can understand it, but it no longer holds true once you touch the metadata.

(yup. I know that statement can be silly... hahaha. Just that... ah... let us do the wb thingy somewhere else!!!)

Anyhow...

Too elaborate, I guess.

I think both "what is raw" and different philosophies of signalchain workflows is very interesting, but probably both themes kinda confuse this thread... :)
 
Here's Gunleik's ISO 400 base exposure pushed up to ISO 6400 and White Balanced off the rear wall, which gets us 3725K with a slight tint using both REDgamma3 and REDlogFilm.

ProRes4444, 1080p, 69MB.

http://www.artbyphil.com/temp/reduser/gunDragonHighISOGamma.zip

Pay special attention to dark fabric in shadow.

This is indeed impressive.
 
Okay Hans - I have always used the term "RAW parameters" for the controls that you can adjust that are part of the R3D SDK and are displayed as "Source Settings" in 3rd party apps - like WB, curves, saturation, etc. etc. etc. - which effect how the RAW image is translated into an RGB image. And YES - you can alter these controls in camera and they are non-destructive - and these settings will travel in the meta-data of the R3D.

Perhaps I should stop using the "RAW parameters" term and use "Image Processing Controls" if this is causing confusion.

i would call them "grade starting point suggestions" That's really all they are, along with "final look monitor preview aids"
 
i would call them "grade starting point suggestions" That's really all they are, along with "final look monitor preview aids"

We are getting "deep and possibly confusing here"...

hahaha

Still, I agree

LOL
 
BTW, Eirik Tyrihjell and I are two of those who have actually seen Dragon exposed low and pushed to 3200 ISO under tungsten on a 4k large theatre screen on DCP screenings of "Sweet".

I guess he can jump in and say what he thought of how the noise performed...

It is also a very useable way to use the Dragon, if that is what you want.
 
Hi Gunleik, thanks for posting the test. Have you had a chance to test out IRNDs for Dragon? I'm particularly curious about the claim that normal (non-IR) NDs are sufficient even at six or seven stops, hard to imagine that being the case but only a side by side test will help resolve this.
 
I did this this summer. Had no need for IR filter at ND 1.8

John Marchant has published some pretty extensive side-by-side tests of this. Search up his threads.

IR is more or less a non-issue as far as I can see now.

http://www.reduser.net/forum/showthread.php?105573-Sweet&p=1259888&viewfull=1#post1259888

There are discussions about the different images as I post them in that link...

But I hope we finally are more or less done with the IR filters. They never did skin much favor....

G
 
The WB on the Dragon has less negative effect on the signal than it has on the Epic.
But I think this might get a bit elaborate... :)

There is a good reason why Dragon is better in this respect.

WB filtering always cut light, that is, it removes color information. But in case of RAW images this is justified when the color channels are significantly out of balance. That is, if one channel dictates the exposure, then it is better to cut the brightest channel downwards and lift the other ones upwards (by opening the exposure) to gain enough information on all channels. (If there's not enough light to open the iris or to increase the exposure time, then removing information from the brightest channel is neither the brightest thing to do.)

Since Dragon has more dynamics and less noise than Epic, the relative difference between the brightest channels and the lowest ones becomes less critical; The channels that become exposed the least have still enough information.

Film is quite different in this respect. When shooting to film, bright tones become compressed, and there is no way decompress the information that is packed. Furthermore, thanks to linearity, raw files have more headspace and are easier to adjust. Accordingly, when shooting with film one has to get closer to the desired outcome.
 
arri alexa…kodak film…dragon…this is great for cinema…keeping things competitive and evolving…use them all and benefit from it
 
Thanks for the link, Gunleik. Great to hear you agree with the assessment that we can get away with normal NDs. I did see the technical chart that John posted but without a 'control' reference of an MX chip I didn't know what to make of it. I'd still like to see what black synthetic fabrics look like under daylight with a normal ND 1.8. If we can get away with that then the issue is solved for me. I guess now that it's winter in Norway not much chance you'll be shooting that test, but thanks for all that you've shared, and enjoy the lovely dragon.
 
I def have done that under sunlight with snow on the MX, and will run a test on Dragon.

But I thought John did publish a comparison... Or maybe I just have seen it...

John, if you are around.... :)

It is pretty convincing to see the A/B... :)
 
Just to be on the safe side I think I'll keep my Schneider 750 Tru-cut IR Hot Mirror in my filter case. Based on notes from early testers I only expect to need it in extreme situations - fire, lava, summer in the desert with cast in synthetic black wardrobe, etc.

FWIW, it seems like RED is on the right track - an IR filter aggressive enough to kill all IR might be a handicap in some low light capture instances. I call it the 99% solution and consider it intelligent engineering. YMMV.

Cheers - #19
 
Hello Gunleik and Phil. Please help me up understanding this 100% about WB. Basically if you Auto WB in Camera, and then use that MetaData in post, you might get more noise. But if you discard that metadata and use the Raw shot by reseting it all in REDCINE-X and start from scratch, then you are better off to avoid noise?

I always AUTO WB just to have a similar look on the monitor of what I see a reference. But then in Post I like starting color grading from the RAW since I do not tend to keep a natural WB look but rather I bend more cinematic looks.

Thanks and sorry for a perhaps dumb question, but I got a little confused now after reading all these posts. LOL

Thanks

Alain
 
Hello Gunleik and Phil. Please help me up understanding this 100% about WB. Basically if you Auto WB in Camera, and then use that MetaData in post, you might get more noise. But if you discard that metadata and use the Raw shot by reseting it all in REDCINE-X and start from scratch, then you are better off to avoid noise?

I always AUTO WB just to have a similar look on the monitor of what I see a reference. But then in Post I like starting color grading from the RAW since I do not tend to keep a natural WB look but rather I bend more cinematic looks.

Thanks and sorry for a perhaps dumb question, but I got a little confused now after reading all these posts. LOL

Thanks

Alain

I think that you should most likely get the WB right in metadata from start.

And then Hans and I can make our "silly geeky WB" thread somewhere else! :)
 
Back
Top