Activity › Forums › DaVinci Resolve › LUT applied in camera by mistake – can it be fixed?
-
LUT applied in camera by mistake – can it be fixed?
Posted by Rhys Bart on July 15, 2015 at 2:28 pmHello everyone. I’ve been given some rushes shot on a Sony FS7 (mxf files) that have had a LUT applied to them in camera (as in burnt into the image, not just in the viewfinder). The director did not know about this and obviously he’s not happy. He wants me to use Resolve to remove the LUT from the images, but I have no idea if this is even possible, and Google is throwing up no help.
Thanks for any advice.
Rhys
Tero Ahlfors replied 7 years, 7 months ago 8 Members · 16 Replies -
16 Replies
-
Blaise Douros
July 15, 2015 at 5:11 pmYou might be able to use https://cameramanben.github.io/LUTCalc/ to generate a LUT that you can use to reverse the effects of the one applied in-camera. Normally you’d go sLog3 > Rec709, for example, but you might be able to generate a Rec709 > sLog3 LUT with this LUT Calculator.
What LUT was baked in? One of the presets, or a custom LUT?
-
Rhys Bart
July 15, 2015 at 5:24 pmThank you. The LUT was baked in. I am assuming that it was one of the presets as they didn’t seem to know they were shooting with it…
Since my last post I’ve gone into Resolve and removed the metadata from the clip and the image does not change, which makes me think that it’s part of the image and can’t be fixed.
It would be impossible to adjust the dynamic range after this has happened, yeah? I guess that’s the issue here. The director want’s the full range of sLog3, but I think all is lost…
-
Blaise Douros
July 15, 2015 at 6:23 pmWith the FS7, when you bake in the LUT, it’s not just something that happens in metadata–it changes the footage that gets recorded. So yes, you are stuck, for now, with the baked-in look.
However, re-read my above. What I’m suggesting is that you create, via LUT Calc, a new LUT that does the reverse of the one baked into the footage. In theory, since the footage is recorded at 10-bit 4:2:2, the information is pretty much still there, but the baked-in LUT just changes the distribution of the luminance data.
So, what you MUST find out is the exact LUT that was baked into the footage. With that information, go to LUT Calc and see if you can create a new LUT that goes from the baked-in look to sLog3.
It will likely not be as good as if they recorded the footage in sLog3, but you may be able to get it looking pretty flat again. However, it is likely that you will not extract the full dynamic range from the footage, and the director’s just going to have to accept that–that is, simply, not your fault, and you can’t be held responsible for trying to reverse an irreversible mistake made on-set.
The other thing you need to let the director know is that, if the LUT that was baked in was just a basic Rec709 look or similar, then he hasn’t lost much–at some point during the coloring process, through a series of levels and curves adjustments, or through the application of a LUT, the colorist would have to get the footage to fit within the Rec709 color space anyway, so that it looks correct on consumer monitors. So working from a baked-in look is NOT the end of the world.
-
Michael Gissing
July 15, 2015 at 10:11 pmI can’t see a reverse LUT getting the dynamic range back. If it is baked in the recorded file then you can’t undo it.
-
Blaise Douros
July 15, 2015 at 10:43 pmIt depends on the LUT used to bake into the footage. If the LUT isn’t actually throwing away data by dropping the highlights and black clipping levels, they might have a fighting chance. But since what a LUT does in the first place is re-interpret the video color and luma data to be spread across a different curve, it stands to reason that you should be able to apply an inverse curve to re-distribute the data back to its original level, so long as none has been lost in the process.
-
Bill Ravens
July 16, 2015 at 1:48 amIt is possible to reverse the color correction by applying an inverse LUT. However, as noted, the DR is irreversible. That cannot be undone. Subsequent color correction will not perform as efficiently as it would have with an original, unmodified file.
-
Marc Wielage
July 16, 2015 at 4:55 amIf I can make a suggestion: always, always, always do a workflow test prior to starting a project, and take the shot through color-correction and sound to make sure a) it looks good, b) the levels are right, c) all the metadata is there, and d) it stays in sync with sound.
Once the levels are clipped and crushed, they’re gone forever. You can make it better, but you can’t get back the full Log range no matter what you do when the look is baked in.
-
Rhys Bart
July 16, 2015 at 8:29 pmThanks for such a detailed reply! The LUT they used in camera was REC709. I tried a few LUTs using the generator you suggested (thanks again) but the director wasn’t happy. We are going to cut with the REC709 rushes as they are and then he wants to try and see how far he can save it in the grade.
I spoke to the AC and apparently they thought the LUT was just for the viewfinder only. I guess this is something the FS7 does differently?
-
Rhys Bart
July 16, 2015 at 8:32 pmHey Marc. We’d done a workflow test – this is why the director was so shocked when he saw the rushes. Apparently the AC thought the LUT was being applied to the viewfinder only and nobody thought to question it. An unfortunate mistake.
-
Blaise Douros
July 16, 2015 at 9:13 pmMy company owns the FS7, and I work with it several times a week, both shooting and editing. It can apply LUTs selectively to different monitor outputs; but if you apply the LUT to the SDI-1 output, it will also bake it to the footage. This is pretty clearly labelled in the menu, so it sounds like this AC was not really paying attention. Not that I’m blaming and shaming, since I have done the same thing. My guess is that they were trying to get the waveform to display in the monitor, which it won’t unless you apply the LUT to all outputs.
Regardless, I think you’ll find that the footage is still eminently gradeable. At Rec709, it is still retaining a lot of usable color data at 10-bit 4:2:2. Good luck!
Reply to this Discussion! Login or Sign Up