Baselight not reading metadata correctly

[Eugene Lehnert]

Hello-

If there are any Baselight users out there. What is your experience with Baselight reading Epic camera metadata? The Baselight is reading the camera metadata but it is not interpreting it correctly. It is getting close but not %100 there. I notice that the client has curve grading in RedCine-X. I don't think this is being carried across.

Does anyone have any experience with this?

Thanks,

Eugene

 

[Nick Shaw]

Just checked on my Baselight, and I am seeing a perfect match between REDCINE-X and Baselight. If you load an RMD file in Baselight it will take all the parameters, including the curves. However the curves will not be visible in Baselight, only the result of them. Indeed there is nothing in Baselight to indicate that curves are being applied in the RED SDK decode, which is a bit dangerous IMHO.

Are they definitely saving RMDs from REDCINE-X? And are you definitely choosing "Reset All To Metadata (Read RMD)" in Baselight?

 

[M Most]

Just checked on my Baselight, and I am seeing a perfect match between REDCINE-X and Baselight. If you load an RMD file in Baselight it will take all the parameters, including the curves. However the curves will not be visible in Baselight, only the result of them. Indeed there is nothing in Baselight to indicate that curves are being applied in the RED SDK decode, which is a bit dangerous IMHO.

Are they definitely saving RMDs from REDCINE-X? And are you definitely choosing "Reset All To Metadata (Read RMD)" in Baselight?

It was more than a bit dangerous for me on a job about 2 years ago in which we were asked to deliver DPX's with "just" the metadata settings. What they really wanted was everything EXCEPT the curve correction, and with RedlogFilm gamma curve applied. Some other programs allow you to specifically ignore curve corrections, but in general, curve corrections should only be used when the result is intended to be a largely final grade. Another option in Baselight (as I recall, it's been a while) is to reset to camera settings. That will not get you the RMD values, but will likely get you what they were seeing on set when the material was shot. You can then change the color and gamma settings if you wish, but you won't have to deal with curve corrections that you can't undo.

 

[Nick Pasquariello]

It was more than a bit dangerous for me on a job about 2 years ago in which we were asked to deliver DPX's with "just" the metadata settings. What they really wanted was everything EXCEPT the curve correction, and with RedlogFilm gamma curve applied. Some other programs allow you to specifically ignore curve corrections, but in general, curve corrections should only be used when the result is intended to be a largely final grade. Another option in Baselight (as I recall, it's been a while) is to reset to camera settings. That will not get you the RMD values, but will likely get you what they were seeing on set when the material was shot. You can then change the color and gamma settings if you wish, but you won't have to deal with curve corrections that you can't undo.

Unless curves were set in-camera. Not super common, but possible now. I think it was added in the 4.0 update? Or maybe the late 3's. Either way, best bet is to take the Reds into RedCine-X, and create a preset that just affects curves, Colorspace, and Gammaspace, and use that to set those parameters to what you want, while ignoring every other parameter. (For anyone unfamiliar: in RedCine-X, create a Look Preset with the desired values for Color Space, Gamma Space, and Curves. Right click on the Look Preset you just made. Uncheck everything you wish to stay unaltered. In this case, you would only leave Color Space, Gamma Space, and Curves selected. Maybe the LGG and other gains as well, depending. Then apply this new preset to everything you want to prepare for grading. BAM. Kelvin, Tint, ISO, FLUT, etc. is untouched. Curves, and relevant spaces are set to the way you want them.)

 

[M Most]

Not necessarily the way you want them if you want to undo anything the curves are doing. Resolve has a check box to ignore curve corrections. That's a good option for all color correction programs to have.

 

[Nick Shaw]

Not necessarily the way you want them if you want to undo anything the curves are doing.

That's what Nick means, isn't it?

In the case you are talking about you make a preset with curves set to do nothing, which is what you want. I do exactly that myself. I have a preset which leaves ISO, FLUT, Kelvin and tint untouched, changes gamma to REDlogFilm, and everything else to defaults. I apply this to all the clips to "clean" them then make new RMDs.

If people use LGG in camera or RCX there might be an argument for keeping that, because it is preserved in Baselight but also exposed and changeable. Although I would suggest it is not a good idea, as it could still be missed, and you might end up with a damaging effect from pushing the image one way and then back again in the grade. I'm sure most colourists would look momentarily at the result as a reference and then set it to defaults.

 

[Nick Pasquariello]

Mike, I agree that all grading programs should have a checkbox to ignore curves. But for those that don't, what I laid out is a pretty quick side-step to alter the RMDs for a lot of footage at once. I only grade in Resolve currently, and I always take my footage into RedCine-X to apply that preset first, because I just find it much easier for adjusting the Red Metadata than the way things are set up in Resolve.

 

[Nick Shaw]

Mike, I agree that all grading programs should have a checkbox to ignore curves.

As do I. Although what might be even better (assuming it is possible, depending on SDK processing order) would be an option to translate curves and LGG from the RED decode into the grading system's native processing tree, so they could be kept or not as desired, or even used as a start point but tweaked by the colourist. Maybe events could be generated in the log along the lines of "10:00:20:00 RED LGG was translated to a Baselight video grade" and "10:00:23:00 RED Alchemy setting was ignored"

I think it might be possible, as it looks at if curves and LGG are the last processes applied (before the post effects). Processing order in RCX appears to go roughly top to bottom in terms of the parameters (not quite, as the gamma curve seems to be applied after the main set of parameters, but before curves and LGG).

 

[Frank Cueto]

I had a similar request. One interesting thing to try would be to bring the footage as raw as u want and trying the match grade s3d function against a version with the RMD data applied. I know it worked for someone once when the DP wanted to start off his RMD data but that was creating a very crunchy "base memory" image for lack of a better term

 

[Nick Shaw]

I had a similar request. One interesting thing to try would be to bring the footage as raw as u want and trying the match grade s3d function against a version with the RMD data applied. I know it worked for someone once when the DP wanted to start off his RMD data but that was creating a very crunchy "base memory" image for lack of a better term

I don't know how necessary that would be. I obviously don't know exactly what Graeme's processing maths is, but I reckon I could have a fair crack at translating an RMD look numerically into a Baselight stack. Certainly for the basic parameters. Not necessarily for the more esoteric ones like alchemy etc.

 

[Frank Cueto]

Well. Unfortunately when u bring in r3ds and apply the metadata you will loose some dr, specially in the case of aggressive grades. The idea of using matchgrade is to be able to bring in the raw footage into baselights float space, therefore avoiding clipping, applyingatch grade which would in theory give you the look but not clip any values in case they want to do some changes.

Of course, this is all subjective to match grade working...

 

[Nick Shaw]

Well. Unfortunately when u bring in r3ds and apply the metadata you will loose some dr, specially in the case of aggressive grades.

Indeed. That's why I'm not talking about doing that. I'm talking about bringing in the R3D as REDlogFilm with only ISO, FLUT, Kelvin and tint applied from the meta-data, and translating all the other RED parameters into Baselight operations. Many of the RED parameters have effects which could be replicated precisely with Baselight operations, and I believe the others could be approximated quite well. Of course I won't know if I'm right about this until I actually try it.

By doing it this way you would get a start point which was a close match to the meta-data look, but fully adjustable using Baselight controls. And because the colourist will most likely want to a achieve a better version of the look by a different method, they are free to turn off any or all of the operations.