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 :)

Shooting in log?

Stu, what ISO did you rate the camera at for your tests?

Also, where did you do the conversion from rec709 to cineon log? The numbers you're showing seem more indicative of the conversion than exposure.

FYI, we recommend against shooting RED like video but instead shoot it like raw and expose not for IRE levels but for shadow detail vs highlight handling vs tolerance for noise. Video out is more like a 'video tap' at the moment than something to be used with scopes.

Tone mapping and how highlights are mapped to overbrights in cineon is then related to qualitative exposure choices. I believe you're saying the same thing but the examples seem a bit confusing at first glance because they show a video workflow.
 
Corcuera said:
But at this point I can only say one thing Exposing with REDONE from a 709 is ridiculous...
Click to expand...

It all depends on your point of view. If you're end goal is rec709, then exposing for what the output for rec709 will look like without any modifications is preferable to one crowd of people (not necessarily the film crowd).

When you start talking about non-rec709 tone mapping for exposure, the requests are all over the board from linear to redlog to pdlog to custom gammas to custom curves for "secret sauce" post pipelines. rec709 is what we started with but as we add features in the next few builds, selectable monitor tone mapping will appear.
 
Obviusly I am talking for a Film (unless the film is an advertisement and i got it wrong) , as he states in the article that the test starts for a film. So i wouldn´t make my decitions in reaction to what i see in a monitor with an REC709 incoming. That i believe is prety much what he comes up with even as you posted is a little bit confusing the exposition that he makes specialy for me a foreigner that doesn´t have a perfect understanding of the english. :D
Jon Corcuera
 
Stu,

I read your post. It was very informative, interesting, and a very carefully thought-out experiment.

I may not have gotten the full gist of it, but correct me if I am wrong, that you are arguing that we should not go from Rec 709 Log to film (via a lut), and might have used a straight linear mapping to film (via linear->log) as what the Panavision panalog is doing.

I concur with you as I have argued in the favor of a similar approach on this forum before, the only difference is that instead of Panavision model of linear capture -> log, I was more in the favor of log capture (via non-linear CMOS) -> linear (similar to film scanner digitization which is uniformly linear) -- just as what happens on Film negative. This approach might introduce a little more noise because of addition of a few more components before ADC, but I think such models are worth exploring.

You are welcome to have a look at that thread at the following link:

http://www.reduser.net/forum/showthread.php?t=8493

Thanks again for all your efforts and great images.
 
Deanan said:
Stu, what ISO did you rate the camera at for your tests?
Click to expand...

I have no idea, but it doesn't matter, since we exposed via waveform as much as meter. Paul's here so maybe he'll give some insight. I left everything at the default for the conversion in RED Alert, but I dropped EV down until I wasn't getting any more detail from the highlights.

Deanan said:
Also, where did you do the conversion from rec709 to cineon log? The numbers you're showing seem more indicative of the conversion than exposure.
Click to expand...

That was done in After Effects.

Deanan said:
FYI, we recommend against shooting RED like video but instead shoot it like raw and expose not for IRE levels but for shadow detail vs highlight handling vs tolerance for noise. Video out is more like a 'video tap' at the moment than something to be used with scopes.
Click to expand...

You can say that, but the only exposure assist RED One has is a (very clever) false-color mode that encourages one to place 18% gray at 45 IRE and skin tones somewhere north of 70%. If you branch out from there you're on your own. But yes, we have discussed here before the importance of branching out.

Along with custom gamma curves for shooting I'd love to be able to make custom false color modes.

-Stu
 
Joofa said:
I may not have gotten the full gist of it, but correct me if I am wrong, that you are arguing that we should not go from Rec 709 Log to film (via a lut), and might have used a straight linear mapping to film (via linear->log) as what the Panavision panalog is doing.
Click to expand...

Whatever color space you render to in RED Alert, you are going from linear to that space. I render to log in RED Alert because the resultant images drop right into my grading pipeline without much fuss.

The problem I'm identifying here is that exposing for Rec709 results in log images with either no headroom or dramatic overexposure.

I'm using Genesis as an example of a camera with a video preview system that encourages one to make film-friendly exposure decisions on the set.

-Stu
 
What are the possibilities of a variable exposure level false colour mode? If you could set up how much underexposure/shoulder room you wanted to leave then you could use this mode much more effectively, by the sounds of things.

Looking forward to part 2, Stu - that's going to help define good exposure practice quite a bit, I think...
 
ProLost said:
Along with custom gamma curves for shooting I'd love to be able to make custom false color modes.

-Stu
Click to expand...

Me too!

And this should be 100% possible as False Color is down stream from color settings -we just need to be able to load color settings/look file into the camera - which I am under the impression is under development -
 
You know, I believe this is exactly how the SI camera works, the Cineform compression applies a linear log curve to the images which can then "taken out" in post if desired, but by that rationale, you could modify that curve to be whatever you want it to be, or better yet, apply endless LUTs to the image, as I think is doable with the SI via Iridas color management and Cineform algorithm. It's funny that everything RED is working towards achieving the SI has already accomplished very well, alas, with only half the resolution, s16 DOF and a higher price, but if there's a camera system out there that RED should look at in terms of being the closest example to its own model, it should not be the F-23, Viper, Genesis, D-20 but the SI instead.
 
Sorry Stu, just realised I suggested the exact same thing you did a few posts earlier - should read your posts better, my bad!
 
Rudi Herbert said:
You know, I believe this is exactly how the SI camera works, the Cineform compression applies a linear log curve to the images which can then "taken out" in post if desired,
Click to expand...

Assuming the LUT is integer-based, you would loose precision (detail) by doing linear->lut->linear.
 
ProLost said:
I have no idea, but it doesn't matter, since we exposed via waveform as much as meter. Paul's here so maybe he'll give some insight. I left everything at the default for the conversion in RED Alert, but I dropped EV down until I wasn't getting any more detail from the highlights.
Click to expand...

As I said before, we recommend not treating the 709 output like a video camera for raw workflows (and for filmout) and putting it on a scope to measure where you should place your exposure. By definition a rec709 signal will do alot of DR compression and will be misleading on a scope when you're tying that back to a raw exposure. And because it's a raw workflow, how you rate the camera is important relative to how you process/develop raw in post. This will improve in the upcoming builds as the monitoring options evolve.

That was done in After Effects.
Click to expand...
Mapping rec709 directly to cineon log and capping at 685 and then comparing it to a film scan with overbrights seems confusing. If filmout is the intention then processing via REDlog would be more ideal especially if you're looking at mapping overbrights in relation to your raw exposure.

You can say that, but the only exposure assist RED One has is a (very clever) false-color mode that encourages one to place 18% gray at 45 IRE and skin tones somewhere north of 70%. If you branch out from there you're on your own. But yes, we have discussed here before the importance of branching out.
Click to expand...

That's goes back to same issue about not using the meters/monitoring in rec709 for a raw and film based workflows and why we recommend against doing so unless you're doing a strict rec709 output.

More options for metering and monitoring will appear soon but all this stuff is a work in progress and starting with rec709 is the best first choice as it covers the broadest usage cases. As a developer yourself I'm sure know quite well that if you add all the features you want, you'll never get something out there and you'll miss out on the organic evolution of a product.

Ultimately, I think we're saying the same things but from different perspectives.
 
Deanan said:
As I said before, we recommend not treating the 709 output like a video camera for raw workflows (and for filmout) and putting it on a scope to measure where you should place your exposure. By definition a rec709 signal will do alot of DR compression and will be misleading on a scope when you're tying that back to a raw exposure.
Click to expand...

Arg! But what choice do we have? Rec709 output is all we have to shoot with!

Deanan said:
Mapping rec709 directly to cineon log and capping at 685 and then comparing it to a film scan with overbrights seems confusing. If filmout is the intention then processing via REDlog would be more ideal especially if you're looking at mapping overbrights in relation to your raw exposure.
Click to expand...

It's purely an exercise meant to show that exposing using the Rec709 rulebook is inappropriate for a project destined for film.

It doesn't really matter which log format I use or how I get there, because either way I've got no detail in those windows, whether I place them at 685 or 1023. Nor is there much difference between processing to Rec709 and post-converting to Cineon log and using PD Log 685 in RED Alert.

RED Log sounds great, but it will make my shots look way overexposed. So I'll just darken them in the DI and reveal my flat windows again, so I'm in the same boat as if I'd used PD Log 685.

But if I had been monitoring in RED Log then I would have been encouraged to expose the scene appropriately for log processing.

RED Log monitoring will require its own assumptions about exposure best practices, and hopefully its own false-color mode.

Deanan said:
That's goes back to same issue about not using the meters/monitoring in rec709 for a raw and film based workflows and why we recommend against doing so unless you're doing a strict rec709 output.
Click to expand...

I don't think it's fair for you to say that you recommend against using the only monitoring option you currently offer. :)

But I understand what you're saying and I agree of course—my exploration is all about determining my own best practices with the current system, and giving you guys some windows into what I'll be wanting from future previewing optionsand how they will dramatically improve the camera's performance for film.

IF it can handle being underexposed! I must say, Paul pretty much had the fear of god put in him from the camera's owner about underexposing. There's a great fear of noise and compression artifacts showing up, and that's what Paul's next round of tests will be all about.

-Stu
 
ProLost said:
IF it can handle being underexposed! I must say, Paul pretty much had the fear of god put in him from the camera's owner about underexposing. There's a great fear of noise and compression artifacts showing up, and that's what Paul's next round of tests will be all about.

-Stu
Click to expand...

How would you know if you're underexposing if you don't know what iso you were exposing for?
 
Sorry if i get in the middle, but I think there are 2 prespectives in to discusion, Stu say´s from what i receive in the post this is how things look. There for Iso and what kind of light you are using is irrelevant, in the oposite deanan is going throw the Dp prescpective to try to have as everything as posible in the correct exposing area.
And both are saying pretty similar 709 is not trustworthy for going in to a posproduccion film out.
As how will you find out how to expose well, photometer or Spot in camera and doing some preproduction test to find out what goes well with the tests going all the way to positive if posible. pretty much what you will do to know how to light and mesure with negative.
If I got you guys wrong please say so. i am just trying to help
Jon Corcuera
 
Stu, I think you are taking the limitation of a typical digital camera that it may not be able to expose for over 600% bright object vs. film range which is just under 1250% bright (in the 10-bit DPX scan values) vis-a-vis the Rec 709 space.

Rec 709 space was designed to be viewed under a specific illumination to which our eyes are adapted to. To me it appears it is not the limitation of the space, but the particular implementation of a video camera and the corresponding display with limited range.
 
You must log in or register to reply here.
Back
Top