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

[Gunleik Groven]

Hm... I apparently had several NSFW things in me...
Even when people are fully dressed...

Keep them money coming Adrian, that is in many ways smarter and you get more sleep

And even if it was a jab at me, I am fine with that!

Axs to the issue at hand, RAW is fortunately not fully translateable o to a preview on set. No magic in that. Just a way to store data.
Initially (back in 2008) I was very intrigued about the idea of "finishing" on set.

As I have picked up more knowledge about signalchains and what to do when in what part of the chain, I must say I have changed my mind radically...

But I am also very much in the mind of:
Whatever works. is good.

And thus I think it is good with Toias "no bullshit, I finnish on my laptop on the plane" approach and Geoffs "insisting on ACES and controlled environments".

There is only a problem if the result does not meet your expectations, and THAT is where the signal-chain stuff gets interesting...

 

[Mark Toia]

I am also very much in the mind of:
Whatever works. is good.

My thinking exactly.

We do staggering amounts of internal work flow testing to achieve only one goal.... Find a way to have flawless results in the shortest time frame possible with minimal amount of gear.

Time is money, Content is king...

I'm sounding like a broken record again aren't I.


On a side note.
I don't like this little test, who ever did it. Yikes... ! Lets hope the RED team can pull that RED / noise back a touch on the new OLPF. 12800 IS0. ???? .... what ever.

[/URL][/IMG]

 

[Hrvoje Simic]

Are these noise tests with the motion mount too ?

 

[Stacey Spears]

Are these noise tests with the motion mount too ?

There is a thread where people are mentioning a possible bug in the motion mount. Might be worth re-running those tests by using the work around. i.e. change color temp to something else and then change back. I will try and repro this week once my camera returns from a job it is on. Then again, if the mount was used on both cameras, they would have a similar issue.

 

[Paul Harb]

Yes, it still looks damn good though.

IN ANY CASE...

Did we give up on the idea that Dragon has 16+ stops? It seems so, judging from the comments in this thread.

It looks like Dragon is still struggling to match Alexa - which is a 13-14 stops camera.

Can someone explain clearly what's going on with the 16+ stops claim?
If Dragon is a 16+ stops camera, then it would easily exceed Alexa in both highlights and shadows. No need to push and pull and re-arrange the available DR like crazy (and still not match - or, at best, barely equal - Alexa).

Where are the two extra stops of DR? I am honestly confused, not trying to pour gasoline over fire.

P.S. I love RED.

 

[M Most]

YES - we should be able to drive the camera with WIFI - in a CDL mode - which locks down all RAW parameters except white balance (K + tint) and ISO and outputs a LEGAL(SMPTE) OR FULL signal down one of TWO independently processed monitor paths. We should be able to swing the CDL values (lift, gamma, gain & saturation) with an iOS app in realtime and see the results on a 17" well-calibrated monitor on a C-stand near the camera for the DP. We should be able to load CDLs into the camera via WIFI and pull them off via WIFI. The CDL values should be stored in the RAW metadata header and the RMD.

Well, this is one of those admittedly rare days on which I agree with almost everything you just said. I would pass that "Legal OR Full" note on to Sony as well (they seem to give you either one or the other or a combination of the two, and don't tell you what it is, even if you ask, at least on the F55.....). I would also point out that CDLs are not Lift, Gain, Gamma, Saturation. They are Offset, Power, Slope, Saturation. That may not mean much to some here, but there is a significant difference in how they manipulate the image and the way you should use them. In many ways, the presence of the Offset parameter in a world dominated by log formats is a very good thing, but only if you understand what that control is doing (it's basically an Exposure control) and how to use it in conjunction with Power and Slope to achieve a very DI-like "exposure/contrast" grading approach. Software that exposes a LGG mode but outputs CDL values is basically manipulating Offset and Slope together, in opposite directions, to achieve what feels like a Lift control but isn't. That's OK for those that haven't worked in a true log based DI type environment, but it usually yields much more "video like" results. I would also point out that one of the things Codex does on the Arri side is allow you to embed settings in the RAW file headers (RAW settings, CDL's, basically anything the camera can keep track of and hand off to you in either an ALE or XML file), and while it is usually very useful, it is also a bit dangerous unless those processing the dailies understand how to either pass those values on or bypass them. In the long run, though, I do agree that it's a worthwhile feature.

I suspect that can't give us FULL out the HDSDI - or they would have already done that.

I'm curious as to why you think that might be the case, as opposed to the notion that they might have just decided a long time ago that SDI should always be scaled for proper display on a Rec709 type display....

 

[luigivaltulini]

On a side note.
I don't like this little test, who ever did it. Yikes... ! Lets hope the RED team can pull that RED / noise back a touch on the new OLPF. 12800 IS0. ???? .... what ever.

MArk, I've never called test.
I just showed this to 12800 to highlight the differences ...
Although during a particular shot at ISO 800 with the 2 dragon we noticed that the new was more nervous about blacks, so I tried to figure out why.
If I had the new OLPF probably I would not have even noticed.

 

[Mark Toia]

IN ANY CASE...

Did we give up on the idea that Dragon has 16+ stops? It seems so, judging from the comments in this thread.

It looks like Dragon is still struggling to match Alexa - which is a 13-14 stops camera.

Can someone explain clearly what's going on with the 16+ stops claim?
If Dragon is a 16+ stops camera, then it would easily exceed Alexa in both highlights and shadows. No need to push and pull and re-arrange the available DR like crazy (and still not match - or, at best, barely equal - Alexa).

Where are the two extra stops of DR? I am honestly confused, not trying to pour gasoline over fire.

P.S. I love RED.

? what , where, when... ? Have I missed yet another controversial test. ?

Honestly...

16 stops, 13 stops, I don't really care now... The results Im getting are way better than anything I've got from Alexa. And its smaller, cheaper, lighter, more res, blah blah...

These pixel comparo arguments are exhausting. is anyone out there even working. ?

Ok.. now back grading my TV AD, matching GO PRO with Alexa, with 5d, with EPIC... ooooh look at that !!!! Its all looks the the same now ))))

Everyone Chill. It's only a camera. Its what you capture is what makes you money.

 

[Mark Toia]

MArk, I've never called test.
I just showed this to 12800 to highlight the differences ...
Although during a particular shot at ISO 800 with the 2 dragon we noticed that the new was more nervous about blacks, so I tried to figure out why.
If I had the new OLPF probably I would not have even noticed.

It's all very interesting to be honest.

It's good for the RED to see this anyways, Keeps the sensor tuning lads on their toes. I'm hopping hey will crack out a solution quickly.

Thanks for sharing.

 

[Mark L. Pederson]

Well, this is one of those admittedly rare days on which I agree with almost everything you just said.

I would also point out that CDLs are not Lift, Gain, Gamma, Saturation. They are Offset, Power, Slope, Saturation.

Yes - you are correct - brain fart - sorry about that. I will correct my post.

I'm curious as to why you think that might be the case, as opposed to the notion that they might have just decided a long time ago that SDI should always be scaled for proper display on a Rec709 type display....

Well ... if that is the case ... they should change their minds. But much more relevant for 3D LUT workflows for features than episodic TV.

 

[M Most]

Well ... if that is the case ... they should change their minds. But much more relevant for 3D LUT workflows for features than episodic TV.

Not entirely irrelevant for TV, though, especially if "live grading" is involved. The problem on the F55 and some other cameras is that what is coming down the SDI pipe does not always match what is being written into the file. So to adjust "live grading" to be accurate, you need to force the SDI feed to match the file recording, which should usually mean a simple 1D scaling LUT. But that doesn't work if the manufacturer does things like send a 64-1023 signal out the SDI pipe (i.e., neither scaled nor full, but an odd combination of the two, probably to allow "normal" results on a Rec709 display, but a misguided way of doing it). And what gets recorded in the file changes with the different file formats, forcing you to waste valuable prep time trying to figure it out. One of the things Red actually avoids by recording only Raw is that type of scenario, but to some degree, the problem is really just passed down the line to the transcode stage, and once again, either gets scaled or not depending on the file format. The real solution I would want is some way of setting up the camera so that I know the SDI matches the recorded file, which, as you implied, should always be the case if you send full range down SDI and record Raw.

 

[Mark L. Pederson]

The real solution I would want is some way of setting up the camera so that I know the SDI matches the recorded file, which, as you implied, should always be the case if you send full range down SDI and record Raw.

Yeah ... like the SMPTE/FULL setting on the RED ONE camera Yes we had it. Now we don't.

 

[Gunleik Groven]

Early on, Deanan said it would eventually be possible with different LUTs to different outputs on the Epic. I guess that practically didn't work out, as we would probably have seen that by now. Somehow I still hope the necessary extra processing power for that is buried somewhere in the new Dragon Hardware, though.

There are a couple of those early Epic roadmap items, that kinda still pop up.

To send an SMPTE/FULL signal does not sound undoable to unleash, though.

 

[Adrian Jebef]

The real solution I would want is some way of setting up the camera so that I know the SDI matches the recorded file, which, as you implied, should always be the case if you send full range down SDI and record Raw.

This is all I want. I honestly could care less about the other stuff. Let it be black magic and voodoo. If I can get my picture on-set to match my picture in Post I'm a happy camper...

 

[Adrian Jebef]

Hm... I apparently had several NSFW things in me...
Even when people are fully dressed...

Keep them money coming Adrian, that is in many ways smarter and you get more sleep

And even if it was a jab at me, I am fine with that!

Ah, GG! I don't jab. At least not personally. The internet makes intent hard to decipher. Truthfully, I'm jealous of your work. If I had the chance I'd be doing the same thing. Right now in my career I've found myself a place in the larger industry. Shooting shorts is on my list but not this year. Maybe next. I'll bide my time until I blow your mind...

Hopefully, our paths will cross sometime in 2017 and I can finally buy you that beer I've owed you since 2007...

 

[Patrick Tresch]

Its what you capture is what makes you money.

Please define money.

Pat

 

[Mark Wuerthner]

I stand corrected. But, I also blame what must be a typo in this graphic (near the bottom of the page).

It states, "11 Stop Greyscale Guide" when it should probably read, "11 Step Greyscale Guide."

http://library.creativecow.net/adcock_gary/magazine_charting-your-camera/1

What do the brightness differences in the steps represent? I read a little further and I guess they're showing different degrees of reflectivity.

How useful is knowing the various reflectivity values?

From DSC Labs - http://www.dsclabs.com/Images2/PDFs/DSC Labs Basic Chart Use Matt.pdf :

"Neutral 11 Step Crossed Grayscale - For testing and setup of lighting, exposure, gamma and dynamic range."

How is DR measured if the steps don't represent stops?

In the interest of clearing up some misinformation in regards to the DSC charts I would like to explain and to note in a friendly way that Gunleik's answer to your question was incorrect :smile5I also think that Gunleik has a cool name). I have seen all sorts of tests out there on the web on all sorts of cameras and people insist on sticking the DSC Chrome Du Monde chart in the shot. The chart has its purpose but as a way for us pixel peeping guys to evaluate dynamic range this chart makes it somewhat difficult. I talked with DSC and they sent me a chart of the "photographic" values (f-stops) of the gray chips on their charts. I have taken those values and put them on the appropriate chips of the charts, and I have also included the DSC fstop chart and and a scale of f-stops made in quarter stops. As you can see the difference between chips is not 1 full stop but rather about .75 of a stop. This difference is not always constant either. So, going from chip #8 which is f2.8 to chip #11 f5.48 is about 2 stops, slightly less. Going from chip #8 to chip #5 is about 2 stops, slightly more. The chart itself has I believe about 6.5 stops of total range of reflectivity and from what I've read it is pretty much impossible to make a front lit chart that can have a greater dynamic range than this. DSC does make back lit charts that can cover a 20 stop range but that is a whole other thing( it would be nice if someone could aim the dragon sensor at a XYLA chart and put up the accompanying image and waveform image-then all this supposition and arguing about dynamic range would cease...sheesh). I wish people who shoot these tests would include the Gamma and Density chart as it consists of whole stop increments. Then they could shoot the over/under exposure tests in half stops and we could then accurately and easily see the camera's latitude to within a half stop.

 

[David Battistella]

The greyscale part of the chart is best used for selecting white balance in post. It also can show chroma st certain levels, with certain cameras.

Battistella.

 

[Gunleik Groven]

THX Mark!
I stand corrected!

I had a bird whisper the whole stop difference and took it as "granted"

 

[Brian Boyer]

In the interest of clearing up some misinformation in regards to the DSC charts I would like to explain and to note in a friendly way that Gunleik's answer to your question was incorrect :smile5I also think that Gunleik has a cool name). I have seen all sorts of tests out there on the web on all sorts of cameras and people insist on sticking the DSC Chrome Du Monde chart in the shot. The chart has its purpose but as a way for us pixel peeping guys to evaluate dynamic range this chart makes it somewhat difficult. I talked with DSC and they sent me a chart of the "photographic" values (f-stops) of the gray chips on their charts. I have taken those values and put them on the appropriate chips of the charts, and I have also included the DSC fstop chart and and a scale of f-stops made in quarter stops. As you can see the difference between chips is not 1 full stop but rather about .75 of a stop. This difference is not always constant either. So, going from chip #8 which is f2.8 to chip #11 f5.48 is about 2 stops, slightly less. Going from chip #8 to chip #5 is about 2 stops, slightly more. The chart itself has I believe about 6.5 stops of total range of reflectivity and from what I've read it is pretty much impossible to make a front lit chart that can have a greater dynamic range than this. DSC does make back lit charts that can cover a 20 stop range but that is a whole other thing( it would be nice if someone could aim the dragon sensor at a XYLA chart and put up the accompanying image and waveform image-then all this supposition and arguing about dynamic range would cease...sheesh). I wish people who shoot these tests would include the Gamma and Density chart as it consists of whole stop increments. Then they could shoot the over/under exposure tests in half stops and we could then accurately and easily see the camera's latitude to within a half stop.

Thanks Mark,

As I posted a little later - http://bit.ly/1eWFgxd - I realized I was confusing the Gamma & Density charts (which do have whole stop increments) with the DSC charts and erroneously concluded the gray steps were showing the same thing.

As for the Xyla chart, John Marchant posted this a little while ago but I guess some people have a hard time believing its efficacy:

I'll just leave this here... DSC labs make a multi-thousand dollar chart for a good reason.

Side note: I didn't share this at the time, as the camera was pre-release and undergoing significant change and improvement. I'll have my new OLPF setup in hand very shortly, and will re-test of course.